{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 11269,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 14.31354719582105,
      "learning_rate": 1.4749262536873157e-08,
      "loss": 0.5728,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.965879471635606,
      "learning_rate": 2.9498525073746314e-08,
      "loss": 0.6309,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.782665216858494,
      "learning_rate": 4.424778761061947e-08,
      "loss": 0.4926,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.139492372932114,
      "learning_rate": 5.899705014749263e-08,
      "loss": 0.7496,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.65911153128057,
      "learning_rate": 7.374631268436579e-08,
      "loss": 0.2972,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.64481074955557,
      "learning_rate": 8.849557522123894e-08,
      "loss": 0.7227,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.979666187367114,
      "learning_rate": 1.032448377581121e-07,
      "loss": 0.6882,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.824232853792573,
      "learning_rate": 1.1799410029498526e-07,
      "loss": 0.6311,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.781154479827578,
      "learning_rate": 1.327433628318584e-07,
      "loss": 0.4722,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.829163593848683,
      "learning_rate": 1.4749262536873157e-07,
      "loss": 0.7626,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.88456801792148,
      "learning_rate": 1.6224188790560474e-07,
      "loss": 0.4066,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.181456914113918,
      "learning_rate": 1.7699115044247788e-07,
      "loss": 0.7341,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.028646144455355,
      "learning_rate": 1.9174041297935104e-07,
      "loss": 0.7487,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.468818644075188,
      "learning_rate": 2.064896755162242e-07,
      "loss": 0.9078,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.948718984086417,
      "learning_rate": 2.2123893805309737e-07,
      "loss": 0.6477,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.965820225141114,
      "learning_rate": 2.359882005899705e-07,
      "loss": 0.7072,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.539199993492483,
      "learning_rate": 2.5073746312684365e-07,
      "loss": 0.7104,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.176127591392232,
      "learning_rate": 2.654867256637168e-07,
      "loss": 0.565,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.685470233448626,
      "learning_rate": 2.8023598820059e-07,
      "loss": 0.498,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.796564593276207,
      "learning_rate": 2.9498525073746315e-07,
      "loss": 0.8586,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.964882371548494,
      "learning_rate": 3.097345132743363e-07,
      "loss": 0.3302,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.007234026181623,
      "learning_rate": 3.244837758112095e-07,
      "loss": 0.5801,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.20011090134271,
      "learning_rate": 3.3923303834808264e-07,
      "loss": 0.848,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.80117695444336,
      "learning_rate": 3.5398230088495575e-07,
      "loss": 0.5546,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.377592982412377,
      "learning_rate": 3.687315634218289e-07,
      "loss": 0.7284,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.143104112129798,
      "learning_rate": 3.834808259587021e-07,
      "loss": 0.5385,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.372988817573354,
      "learning_rate": 3.9823008849557525e-07,
      "loss": 0.7779,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.493419785109115,
      "learning_rate": 4.129793510324484e-07,
      "loss": 0.6622,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.18233965554827,
      "learning_rate": 4.277286135693216e-07,
      "loss": 0.5622,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.268053392332689,
      "learning_rate": 4.4247787610619474e-07,
      "loss": 0.3426,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.696644712505401,
      "learning_rate": 4.5722713864306786e-07,
      "loss": 0.6257,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.437817366090727,
      "learning_rate": 4.71976401179941e-07,
      "loss": 0.6547,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.628440559263883,
      "learning_rate": 4.867256637168142e-07,
      "loss": 0.7116,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.931963045692111,
      "learning_rate": 5.014749262536873e-07,
      "loss": 0.6473,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.680631195265866,
      "learning_rate": 5.162241887905605e-07,
      "loss": 0.4345,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.684003266688723,
      "learning_rate": 5.309734513274336e-07,
      "loss": 0.7404,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.840260244759314,
      "learning_rate": 5.457227138643068e-07,
      "loss": 0.7142,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.756804548001467,
      "learning_rate": 5.6047197640118e-07,
      "loss": 0.3809,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.778979822133621,
      "learning_rate": 5.752212389380532e-07,
      "loss": 0.7537,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.800839918699953,
      "learning_rate": 5.899705014749263e-07,
      "loss": 0.3958,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.45664088403137,
      "learning_rate": 6.047197640117994e-07,
      "loss": 0.303,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.212679675599569,
      "learning_rate": 6.194690265486726e-07,
      "loss": 0.2025,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.439795832116872,
      "learning_rate": 6.342182890855458e-07,
      "loss": 0.6896,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.8661444303281,
      "learning_rate": 6.48967551622419e-07,
      "loss": 0.3688,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.744884155902624,
      "learning_rate": 6.637168141592922e-07,
      "loss": 0.3656,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.252511089488117,
      "learning_rate": 6.784660766961653e-07,
      "loss": 0.5614,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.276787295612664,
      "learning_rate": 6.932153392330385e-07,
      "loss": 0.8617,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.783025676984305,
      "learning_rate": 7.079646017699115e-07,
      "loss": 0.3914,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.153345360992281,
      "learning_rate": 7.227138643067848e-07,
      "loss": 0.5325,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.578266917992172,
      "learning_rate": 7.374631268436578e-07,
      "loss": 0.3553,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.516186323047946,
      "learning_rate": 7.522123893805311e-07,
      "loss": 0.2713,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.618919793461277,
      "learning_rate": 7.669616519174042e-07,
      "loss": 0.5757,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.011451536119191,
      "learning_rate": 7.817109144542774e-07,
      "loss": 0.6222,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.758464496000556,
      "learning_rate": 7.964601769911505e-07,
      "loss": 0.3804,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.329668832054149,
      "learning_rate": 8.112094395280237e-07,
      "loss": 0.3032,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.808134863332105,
      "learning_rate": 8.259587020648968e-07,
      "loss": 0.4371,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.485017443156223,
      "learning_rate": 8.4070796460177e-07,
      "loss": 0.2365,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.752360772410532,
      "learning_rate": 8.554572271386432e-07,
      "loss": 0.575,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.78323986651021,
      "learning_rate": 8.702064896755164e-07,
      "loss": 0.5786,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.266930965114382,
      "learning_rate": 8.849557522123895e-07,
      "loss": 0.8025,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.9785629262780455,
      "learning_rate": 8.997050147492627e-07,
      "loss": 0.4518,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.686699302644955,
      "learning_rate": 9.144542772861357e-07,
      "loss": 0.5705,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.981537611531389,
      "learning_rate": 9.292035398230089e-07,
      "loss": 0.2916,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.996499258838841,
      "learning_rate": 9.43952802359882e-07,
      "loss": 0.6264,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.372068731026832,
      "learning_rate": 9.587020648967554e-07,
      "loss": 0.2717,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.102770702345687,
      "learning_rate": 9.734513274336284e-07,
      "loss": 0.2966,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.011906138088363,
      "learning_rate": 9.882005899705016e-07,
      "loss": 0.4484,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.480450203555781,
      "learning_rate": 1.0029498525073746e-06,
      "loss": 0.5962,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.800454936743838,
      "learning_rate": 1.017699115044248e-06,
      "loss": 0.423,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.915215971347147,
      "learning_rate": 1.032448377581121e-06,
      "loss": 0.5069,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.495461616766264,
      "learning_rate": 1.0471976401179943e-06,
      "loss": 0.729,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.198124423804067,
      "learning_rate": 1.0619469026548673e-06,
      "loss": 0.4913,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.042077268708202,
      "learning_rate": 1.0766961651917405e-06,
      "loss": 0.5391,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.694935612522794,
      "learning_rate": 1.0914454277286137e-06,
      "loss": 0.6291,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.732611257597125,
      "learning_rate": 1.106194690265487e-06,
      "loss": 0.2033,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.4317508003774,
      "learning_rate": 1.12094395280236e-06,
      "loss": 0.4896,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.236894729426883,
      "learning_rate": 1.1356932153392331e-06,
      "loss": 0.5992,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.20702413849297,
      "learning_rate": 1.1504424778761064e-06,
      "loss": 0.4544,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.299589975189516,
      "learning_rate": 1.1651917404129796e-06,
      "loss": 0.2958,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.365715905215291,
      "learning_rate": 1.1799410029498526e-06,
      "loss": 0.3195,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.383576986822591,
      "learning_rate": 1.1946902654867258e-06,
      "loss": 0.2582,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.71070932863161,
      "learning_rate": 1.2094395280235988e-06,
      "loss": 0.3596,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.404423337125484,
      "learning_rate": 1.224188790560472e-06,
      "loss": 0.4669,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.894327103082441,
      "learning_rate": 1.2389380530973452e-06,
      "loss": 0.6169,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.676250128834968,
      "learning_rate": 1.2536873156342182e-06,
      "loss": 0.3099,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.244859893686236,
      "learning_rate": 1.2684365781710917e-06,
      "loss": 0.2854,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.311952125201904,
      "learning_rate": 1.2831858407079647e-06,
      "loss": 0.6007,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.7123857077694735,
      "learning_rate": 1.297935103244838e-06,
      "loss": 0.4484,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.434011779514033,
      "learning_rate": 1.312684365781711e-06,
      "loss": 0.4215,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.563904618248838,
      "learning_rate": 1.3274336283185843e-06,
      "loss": 0.2662,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.741379762801725,
      "learning_rate": 1.3421828908554573e-06,
      "loss": 0.5069,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.983252897020109,
      "learning_rate": 1.3569321533923306e-06,
      "loss": 0.4235,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.010389634611068,
      "learning_rate": 1.3716814159292036e-06,
      "loss": 0.4051,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.26068649148877,
      "learning_rate": 1.386430678466077e-06,
      "loss": 0.2749,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.30752549257276,
      "learning_rate": 1.40117994100295e-06,
      "loss": 0.3395,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.191635925940988,
      "learning_rate": 1.415929203539823e-06,
      "loss": 0.4564,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.8236210831642135,
      "learning_rate": 1.4306784660766962e-06,
      "loss": 0.5986,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.727952980967604,
      "learning_rate": 1.4454277286135697e-06,
      "loss": 0.4862,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.486446976251312,
      "learning_rate": 1.4601769911504427e-06,
      "loss": 0.1928,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.835728047679945,
      "learning_rate": 1.4749262536873157e-06,
      "loss": 0.509,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.8335504527409485,
      "learning_rate": 1.4896755162241889e-06,
      "loss": 0.3942,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.507719470893161,
      "learning_rate": 1.5044247787610621e-06,
      "loss": 0.3929,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.653618084644878,
      "learning_rate": 1.5191740412979353e-06,
      "loss": 0.6181,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.431825314452215,
      "learning_rate": 1.5339233038348083e-06,
      "loss": 0.3421,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.383276040634021,
      "learning_rate": 1.5486725663716816e-06,
      "loss": 0.6208,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.921985896966524,
      "learning_rate": 1.5634218289085548e-06,
      "loss": 0.4633,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.364667802374548,
      "learning_rate": 1.578171091445428e-06,
      "loss": 0.4333,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.20401347492339,
      "learning_rate": 1.592920353982301e-06,
      "loss": 0.5723,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.179755078220362,
      "learning_rate": 1.607669616519174e-06,
      "loss": 0.4635,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.231459544361524,
      "learning_rate": 1.6224188790560474e-06,
      "loss": 0.4887,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.462186100652572,
      "learning_rate": 1.6371681415929204e-06,
      "loss": 0.3714,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.363918420878127,
      "learning_rate": 1.6519174041297937e-06,
      "loss": 0.5687,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.208745891434966,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.3426,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.67064837534409,
      "learning_rate": 1.68141592920354e-06,
      "loss": 0.4113,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.164063177925497,
      "learning_rate": 1.696165191740413e-06,
      "loss": 0.5111,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.733360548855852,
      "learning_rate": 1.7109144542772863e-06,
      "loss": 0.3943,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.448915180530342,
      "learning_rate": 1.7256637168141593e-06,
      "loss": 0.2609,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.14701053302316,
      "learning_rate": 1.7404129793510328e-06,
      "loss": 0.3251,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.728507643734531,
      "learning_rate": 1.7551622418879058e-06,
      "loss": 0.6416,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.225241649991705,
      "learning_rate": 1.769911504424779e-06,
      "loss": 0.2969,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.5518417820779975,
      "learning_rate": 1.784660766961652e-06,
      "loss": 0.4277,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.797642123300228,
      "learning_rate": 1.7994100294985254e-06,
      "loss": 0.3754,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.000080510260502,
      "learning_rate": 1.8141592920353984e-06,
      "loss": 0.2144,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.789384747944488,
      "learning_rate": 1.8289085545722714e-06,
      "loss": 0.2494,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.80511834379678,
      "learning_rate": 1.8436578171091446e-06,
      "loss": 0.2874,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.475557132529528,
      "learning_rate": 1.8584070796460179e-06,
      "loss": 0.4063,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.875612001509973,
      "learning_rate": 1.873156342182891e-06,
      "loss": 0.4163,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.977418409494242,
      "learning_rate": 1.887905604719764e-06,
      "loss": 0.165,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.974121203163095,
      "learning_rate": 1.9026548672566373e-06,
      "loss": 0.5518,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.230795375802467,
      "learning_rate": 1.9174041297935107e-06,
      "loss": 0.3804,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.38469317652657,
      "learning_rate": 1.9321533923303837e-06,
      "loss": 0.4359,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.457247013005507,
      "learning_rate": 1.9469026548672567e-06,
      "loss": 0.1508,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.65543152844698,
      "learning_rate": 1.9616519174041298e-06,
      "loss": 0.4243,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.831761510691713,
      "learning_rate": 1.976401179941003e-06,
      "loss": 0.3118,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.498976173567721,
      "learning_rate": 1.991150442477876e-06,
      "loss": 0.3451,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.673411757580203,
      "learning_rate": 2.005899705014749e-06,
      "loss": 0.4181,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.068380121754199,
      "learning_rate": 2.0206489675516226e-06,
      "loss": 0.4074,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.252522103150605,
      "learning_rate": 2.035398230088496e-06,
      "loss": 0.4317,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.063460283062609,
      "learning_rate": 2.050147492625369e-06,
      "loss": 0.2996,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.395484090989445,
      "learning_rate": 2.064896755162242e-06,
      "loss": 0.3034,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.735572089278023,
      "learning_rate": 2.079646017699115e-06,
      "loss": 0.2888,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.045637379846663,
      "learning_rate": 2.0943952802359885e-06,
      "loss": 0.2423,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.198001474076925,
      "learning_rate": 2.1091445427728615e-06,
      "loss": 0.438,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.825438147422515,
      "learning_rate": 2.1238938053097345e-06,
      "loss": 0.4479,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.361902332474443,
      "learning_rate": 2.138643067846608e-06,
      "loss": 0.4161,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.8772908368937635,
      "learning_rate": 2.153392330383481e-06,
      "loss": 0.4433,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.852355869847926,
      "learning_rate": 2.1681415929203544e-06,
      "loss": 0.5018,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.630137808952474,
      "learning_rate": 2.1828908554572274e-06,
      "loss": 0.2009,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.907087057236669,
      "learning_rate": 2.1976401179941004e-06,
      "loss": 0.3305,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.988630072413649,
      "learning_rate": 2.212389380530974e-06,
      "loss": 0.2493,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.893387830712065,
      "learning_rate": 2.227138643067847e-06,
      "loss": 0.1838,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0622942531976545,
      "learning_rate": 2.24188790560472e-06,
      "loss": 0.2255,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.385407975474306,
      "learning_rate": 2.256637168141593e-06,
      "loss": 0.2971,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.04912557856553,
      "learning_rate": 2.2713864306784663e-06,
      "loss": 0.7353,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.264246594215466,
      "learning_rate": 2.2861356932153393e-06,
      "loss": 0.5947,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.145701997884185,
      "learning_rate": 2.3008849557522127e-06,
      "loss": 0.3852,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.914848029064771,
      "learning_rate": 2.3156342182890857e-06,
      "loss": 0.4172,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.503427918374818,
      "learning_rate": 2.330383480825959e-06,
      "loss": 0.2187,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.067783608585061,
      "learning_rate": 2.345132743362832e-06,
      "loss": 0.3633,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.639585729088271,
      "learning_rate": 2.359882005899705e-06,
      "loss": 0.3603,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.764487182516095,
      "learning_rate": 2.374631268436578e-06,
      "loss": 0.2625,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.354041894244881,
      "learning_rate": 2.3893805309734516e-06,
      "loss": 0.1635,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.854422557619191,
      "learning_rate": 2.4041297935103246e-06,
      "loss": 0.3714,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.853184217613694,
      "learning_rate": 2.4188790560471976e-06,
      "loss": 0.4865,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.4558487358189405,
      "learning_rate": 2.433628318584071e-06,
      "loss": 0.2746,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.5319461054001415,
      "learning_rate": 2.448377581120944e-06,
      "loss": 0.2268,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.014378589235187,
      "learning_rate": 2.4631268436578175e-06,
      "loss": 0.2429,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.100245366989213,
      "learning_rate": 2.4778761061946905e-06,
      "loss": 0.2648,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.9454562889554925,
      "learning_rate": 2.4926253687315635e-06,
      "loss": 0.5273,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.117676330585173,
      "learning_rate": 2.5073746312684365e-06,
      "loss": 0.414,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.359780458775346,
      "learning_rate": 2.52212389380531e-06,
      "loss": 0.1094,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.202224893086955,
      "learning_rate": 2.5368731563421834e-06,
      "loss": 0.181,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3465172838957917,
      "learning_rate": 2.5516224188790564e-06,
      "loss": 0.1066,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.780838868874581,
      "learning_rate": 2.5663716814159294e-06,
      "loss": 0.3776,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.144309176995577,
      "learning_rate": 2.5811209439528024e-06,
      "loss": 0.1941,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.621063735666506,
      "learning_rate": 2.595870206489676e-06,
      "loss": 0.6111,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.388918366225008,
      "learning_rate": 2.6106194690265492e-06,
      "loss": 0.462,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.7885416125833,
      "learning_rate": 2.625368731563422e-06,
      "loss": 0.3032,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.563413588016434,
      "learning_rate": 2.6401179941002952e-06,
      "loss": 0.3081,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.502515219699573,
      "learning_rate": 2.6548672566371687e-06,
      "loss": 0.3187,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.608810960955236,
      "learning_rate": 2.6696165191740413e-06,
      "loss": 0.137,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.477916910440584,
      "learning_rate": 2.6843657817109147e-06,
      "loss": 0.4292,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.01082908923766,
      "learning_rate": 2.6991150442477877e-06,
      "loss": 0.515,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.228159978977396,
      "learning_rate": 2.713864306784661e-06,
      "loss": 0.3523,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.213061485887999,
      "learning_rate": 2.728613569321534e-06,
      "loss": 0.38,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.677692134023223,
      "learning_rate": 2.743362831858407e-06,
      "loss": 0.3521,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.71872715522411,
      "learning_rate": 2.7581120943952806e-06,
      "loss": 0.6235,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.383474436231925,
      "learning_rate": 2.772861356932154e-06,
      "loss": 0.4489,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.9645736672554515,
      "learning_rate": 2.7876106194690266e-06,
      "loss": 0.4725,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.751693134536726,
      "learning_rate": 2.8023598820059e-06,
      "loss": 0.5848,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.784220853862607,
      "learning_rate": 2.817109144542773e-06,
      "loss": 0.2875,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.303962456997373,
      "learning_rate": 2.831858407079646e-06,
      "loss": 0.2553,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.810649322094012,
      "learning_rate": 2.8466076696165195e-06,
      "loss": 0.5371,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.082320731751332,
      "learning_rate": 2.8613569321533925e-06,
      "loss": 0.2914,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.994004886527222,
      "learning_rate": 2.876106194690266e-06,
      "loss": 0.3582,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.740424146845882,
      "learning_rate": 2.8908554572271393e-06,
      "loss": 0.5428,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.013950578650281,
      "learning_rate": 2.905604719764012e-06,
      "loss": 0.1979,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8942481565570484,
      "learning_rate": 2.9203539823008853e-06,
      "loss": 0.0316,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.8284828708958365,
      "learning_rate": 2.935103244837758e-06,
      "loss": 0.4329,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.086155417484154,
      "learning_rate": 2.9498525073746313e-06,
      "loss": 0.2731,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.8971086313834995,
      "learning_rate": 2.9646017699115048e-06,
      "loss": 0.4928,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.670321294826406,
      "learning_rate": 2.9793510324483778e-06,
      "loss": 0.7287,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.087694594764867,
      "learning_rate": 2.9941002949852512e-06,
      "loss": 0.0668,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.350814049637939,
      "learning_rate": 3.0088495575221242e-06,
      "loss": 0.2687,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.369420984834557,
      "learning_rate": 3.0235988200589972e-06,
      "loss": 0.4179,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.4914833516132555,
      "learning_rate": 3.0383480825958707e-06,
      "loss": 0.4497,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.953007966990798,
      "learning_rate": 3.0530973451327432e-06,
      "loss": 0.431,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.631909814993161,
      "learning_rate": 3.0678466076696167e-06,
      "loss": 0.3161,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.739675235469429,
      "learning_rate": 3.08259587020649e-06,
      "loss": 0.6028,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.791140060045366,
      "learning_rate": 3.097345132743363e-06,
      "loss": 0.4302,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.731479520224763,
      "learning_rate": 3.112094395280236e-06,
      "loss": 0.4764,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.568748295253279,
      "learning_rate": 3.1268436578171095e-06,
      "loss": 0.3726,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.620897850825795,
      "learning_rate": 3.1415929203539825e-06,
      "loss": 0.4044,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.392871342946025,
      "learning_rate": 3.156342182890856e-06,
      "loss": 0.4695,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.201327935184395,
      "learning_rate": 3.1710914454277286e-06,
      "loss": 0.3807,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.774971591423651,
      "learning_rate": 3.185840707964602e-06,
      "loss": 0.1672,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.558579208175288,
      "learning_rate": 3.2005899705014754e-06,
      "loss": 0.363,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.066426337977964,
      "learning_rate": 3.215339233038348e-06,
      "loss": 0.4059,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.191417116239285,
      "learning_rate": 3.2300884955752214e-06,
      "loss": 0.3803,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.242979275272407,
      "learning_rate": 3.244837758112095e-06,
      "loss": 0.3205,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.981517230913536,
      "learning_rate": 3.259587020648968e-06,
      "loss": 0.4517,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.723145471549654,
      "learning_rate": 3.274336283185841e-06,
      "loss": 0.3711,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.905826900556789,
      "learning_rate": 3.289085545722714e-06,
      "loss": 0.5338,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.2718742020556535,
      "learning_rate": 3.3038348082595873e-06,
      "loss": 0.3146,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.748275975341933,
      "learning_rate": 3.3185840707964607e-06,
      "loss": 0.5372,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.187418753157377,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.06,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.918682312974358,
      "learning_rate": 3.3480825958702068e-06,
      "loss": 0.2176,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.1998313613828575,
      "learning_rate": 3.36283185840708e-06,
      "loss": 0.2904,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.349046366602937,
      "learning_rate": 3.3775811209439528e-06,
      "loss": 0.1731,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.888886136722448,
      "learning_rate": 3.392330383480826e-06,
      "loss": 0.3958,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.528079446543147,
      "learning_rate": 3.407079646017699e-06,
      "loss": 0.1565,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.239170740833793,
      "learning_rate": 3.4218289085545726e-06,
      "loss": 0.1926,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.179889154507302,
      "learning_rate": 3.436578171091446e-06,
      "loss": 0.4578,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.90704903496395,
      "learning_rate": 3.4513274336283186e-06,
      "loss": 0.5481,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.326637381500092,
      "learning_rate": 3.466076696165192e-06,
      "loss": 0.4055,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.020144947315325,
      "learning_rate": 3.4808259587020655e-06,
      "loss": 0.2591,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.68640912539248,
      "learning_rate": 3.495575221238938e-06,
      "loss": 0.4017,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.75193269278809,
      "learning_rate": 3.5103244837758115e-06,
      "loss": 0.1174,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.89556185482438,
      "learning_rate": 3.5250737463126845e-06,
      "loss": 0.5534,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.2877745392887245,
      "learning_rate": 3.539823008849558e-06,
      "loss": 0.3808,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.606106223763138,
      "learning_rate": 3.554572271386431e-06,
      "loss": 0.3396,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.723795292257503,
      "learning_rate": 3.569321533923304e-06,
      "loss": 0.5487,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.808618229759299,
      "learning_rate": 3.5840707964601774e-06,
      "loss": 0.2026,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.841286530900835,
      "learning_rate": 3.598820058997051e-06,
      "loss": 0.402,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.918746996181583,
      "learning_rate": 3.6135693215339234e-06,
      "loss": 0.3487,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.392721901424792,
      "learning_rate": 3.628318584070797e-06,
      "loss": 0.6382,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.49861063596375,
      "learning_rate": 3.64306784660767e-06,
      "loss": 0.3402,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.886283342972637,
      "learning_rate": 3.657817109144543e-06,
      "loss": 0.7241,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.779257054111315,
      "learning_rate": 3.6725663716814163e-06,
      "loss": 0.4436,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.643324244155194,
      "learning_rate": 3.6873156342182893e-06,
      "loss": 0.462,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.717183270376607,
      "learning_rate": 3.7020648967551627e-06,
      "loss": 0.4532,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.00720216943617,
      "learning_rate": 3.7168141592920357e-06,
      "loss": 0.447,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.688072863438959,
      "learning_rate": 3.7315634218289087e-06,
      "loss": 0.2961,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.94381139104522,
      "learning_rate": 3.746312684365782e-06,
      "loss": 0.2002,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.925977193006524,
      "learning_rate": 3.7610619469026547e-06,
      "loss": 0.2585,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.4979292414158385,
      "learning_rate": 3.775811209439528e-06,
      "loss": 0.2778,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.334279214715469,
      "learning_rate": 3.7905604719764016e-06,
      "loss": 0.3079,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.577097403339071,
      "learning_rate": 3.8053097345132746e-06,
      "loss": 0.2878,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.8945927098436455,
      "learning_rate": 3.820058997050148e-06,
      "loss": 0.4268,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.68951743251673,
      "learning_rate": 3.8348082595870215e-06,
      "loss": 0.4526,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.187229759040664,
      "learning_rate": 3.849557522123894e-06,
      "loss": 0.4656,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.047647302297673,
      "learning_rate": 3.8643067846607675e-06,
      "loss": 0.4938,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.093936573855302,
      "learning_rate": 3.87905604719764e-06,
      "loss": 0.6037,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.41000672887728,
      "learning_rate": 3.8938053097345135e-06,
      "loss": 0.5481,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.72638376614309,
      "learning_rate": 3.908554572271387e-06,
      "loss": 0.5492,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.31091540664453,
      "learning_rate": 3.9233038348082595e-06,
      "loss": 0.561,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.366640133447396,
      "learning_rate": 3.938053097345133e-06,
      "loss": 0.2681,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.216520018096588,
      "learning_rate": 3.952802359882006e-06,
      "loss": 0.2798,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.857500664457439,
      "learning_rate": 3.967551622418879e-06,
      "loss": 0.46,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.861214564141401,
      "learning_rate": 3.982300884955752e-06,
      "loss": 0.406,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.992795695778138,
      "learning_rate": 3.997050147492626e-06,
      "loss": 0.704,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.328492890991043,
      "learning_rate": 4.011799410029498e-06,
      "loss": 0.5645,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.449083065902051,
      "learning_rate": 4.026548672566372e-06,
      "loss": 0.1717,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.668191199082496,
      "learning_rate": 4.041297935103245e-06,
      "loss": 0.5491,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.178745972168267,
      "learning_rate": 4.056047197640119e-06,
      "loss": 0.5802,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.892039183707798,
      "learning_rate": 4.070796460176992e-06,
      "loss": 0.4535,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.717905181566388,
      "learning_rate": 4.085545722713865e-06,
      "loss": 0.16,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.281718282158777,
      "learning_rate": 4.100294985250738e-06,
      "loss": 0.1636,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.574032246024608,
      "learning_rate": 4.115044247787611e-06,
      "loss": 0.446,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.910936601969521,
      "learning_rate": 4.129793510324484e-06,
      "loss": 0.4104,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.987258496890241,
      "learning_rate": 4.1445427728613576e-06,
      "loss": 0.3229,
      "step": 281
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.950139209899311,
      "learning_rate": 4.15929203539823e-06,
      "loss": 0.5216,
      "step": 282
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.5698981946290695,
      "learning_rate": 4.174041297935104e-06,
      "loss": 0.1977,
      "step": 283
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.413790313405681,
      "learning_rate": 4.188790560471977e-06,
      "loss": 0.3373,
      "step": 284
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.009097463266512,
      "learning_rate": 4.20353982300885e-06,
      "loss": 0.5371,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.371027190777878,
      "learning_rate": 4.218289085545723e-06,
      "loss": 0.2487,
      "step": 286
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.59363332913583,
      "learning_rate": 4.233038348082596e-06,
      "loss": 0.4501,
      "step": 287
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.621222472983674,
      "learning_rate": 4.247787610619469e-06,
      "loss": 0.179,
      "step": 288
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.242085263728761,
      "learning_rate": 4.2625368731563425e-06,
      "loss": 0.4551,
      "step": 289
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.3606824479963056,
      "learning_rate": 4.277286135693216e-06,
      "loss": 0.5426,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.017046865118748,
      "learning_rate": 4.2920353982300885e-06,
      "loss": 0.1184,
      "step": 291
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.810053730073689,
      "learning_rate": 4.306784660766962e-06,
      "loss": 0.144,
      "step": 292
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.626147705055332,
      "learning_rate": 4.321533923303835e-06,
      "loss": 0.3339,
      "step": 293
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.564793106048656,
      "learning_rate": 4.336283185840709e-06,
      "loss": 0.4327,
      "step": 294
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.705700296124938,
      "learning_rate": 4.351032448377581e-06,
      "loss": 0.3177,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.581964362912687,
      "learning_rate": 4.365781710914455e-06,
      "loss": 0.404,
      "step": 296
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.667526848596593,
      "learning_rate": 4.380530973451328e-06,
      "loss": 0.2752,
      "step": 297
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.558624411629653,
      "learning_rate": 4.395280235988201e-06,
      "loss": 0.5233,
      "step": 298
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.579854546347721,
      "learning_rate": 4.410029498525074e-06,
      "loss": 0.3606,
      "step": 299
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.768201071474902,
      "learning_rate": 4.424778761061948e-06,
      "loss": 0.1326,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.855152185596826,
      "learning_rate": 4.43952802359882e-06,
      "loss": 0.3423,
      "step": 301
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.730574981817511,
      "learning_rate": 4.454277286135694e-06,
      "loss": 0.597,
      "step": 302
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.636568294484113,
      "learning_rate": 4.469026548672566e-06,
      "loss": 0.3707,
      "step": 303
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.9093687766268275,
      "learning_rate": 4.48377581120944e-06,
      "loss": 0.3866,
      "step": 304
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9109499064325206,
      "learning_rate": 4.498525073746313e-06,
      "loss": 0.0546,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.109653570877494,
      "learning_rate": 4.513274336283186e-06,
      "loss": 0.5052,
      "step": 306
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.7634185179433,
      "learning_rate": 4.528023598820059e-06,
      "loss": 0.3395,
      "step": 307
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.25421169023348,
      "learning_rate": 4.5427728613569326e-06,
      "loss": 0.5127,
      "step": 308
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.250510366871729,
      "learning_rate": 4.557522123893805e-06,
      "loss": 0.305,
      "step": 309
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.259603731274524,
      "learning_rate": 4.5722713864306786e-06,
      "loss": 0.4984,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.475491030900222,
      "learning_rate": 4.587020648967552e-06,
      "loss": 0.1803,
      "step": 311
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.947981272019348,
      "learning_rate": 4.6017699115044254e-06,
      "loss": 0.3208,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.075437147061774,
      "learning_rate": 4.616519174041299e-06,
      "loss": 0.4247,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.872258261292472,
      "learning_rate": 4.6312684365781714e-06,
      "loss": 0.2662,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.518280661146363,
      "learning_rate": 4.646017699115045e-06,
      "loss": 0.6966,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.981045099684458,
      "learning_rate": 4.660766961651918e-06,
      "loss": 0.3499,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.225965212526031,
      "learning_rate": 4.675516224188791e-06,
      "loss": 0.5036,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.88998923511416,
      "learning_rate": 4.690265486725664e-06,
      "loss": 0.4257,
      "step": 318
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.216256011318604,
      "learning_rate": 4.705014749262537e-06,
      "loss": 0.1714,
      "step": 319
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.715056954314634,
      "learning_rate": 4.71976401179941e-06,
      "loss": 0.4411,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.307603150674662,
      "learning_rate": 4.734513274336284e-06,
      "loss": 0.5203,
      "step": 321
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.724395538784852,
      "learning_rate": 4.749262536873156e-06,
      "loss": 0.2012,
      "step": 322
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.921535196572119,
      "learning_rate": 4.76401179941003e-06,
      "loss": 0.2886,
      "step": 323
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.8706958759393855,
      "learning_rate": 4.778761061946903e-06,
      "loss": 0.314,
      "step": 324
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.585720673771485,
      "learning_rate": 4.793510324483776e-06,
      "loss": 0.3061,
      "step": 325
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.4984534896829125,
      "learning_rate": 4.808259587020649e-06,
      "loss": 0.3473,
      "step": 326
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.8126887954738855,
      "learning_rate": 4.823008849557523e-06,
      "loss": 0.3092,
      "step": 327
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.361463612766768,
      "learning_rate": 4.837758112094395e-06,
      "loss": 0.2766,
      "step": 328
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.6061318894056,
      "learning_rate": 4.852507374631269e-06,
      "loss": 0.4218,
      "step": 329
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.231052692063322,
      "learning_rate": 4.867256637168142e-06,
      "loss": 0.7387,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.405974857867386,
      "learning_rate": 4.8820058997050155e-06,
      "loss": 0.3578,
      "step": 331
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.338545212586185,
      "learning_rate": 4.896755162241888e-06,
      "loss": 0.4064,
      "step": 332
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.601884035537584,
      "learning_rate": 4.9115044247787615e-06,
      "loss": 0.2393,
      "step": 333
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.80460410535148,
      "learning_rate": 4.926253687315635e-06,
      "loss": 0.3622,
      "step": 334
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.957379270699075,
      "learning_rate": 4.9410029498525075e-06,
      "loss": 0.5209,
      "step": 335
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.327956679369002,
      "learning_rate": 4.955752212389381e-06,
      "loss": 0.4521,
      "step": 336
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.490399845887929,
      "learning_rate": 4.970501474926254e-06,
      "loss": 0.3606,
      "step": 337
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.840445454541573,
      "learning_rate": 4.985250737463127e-06,
      "loss": 0.2728,
      "step": 338
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.348208451359714,
      "learning_rate": 5e-06,
      "loss": 0.2632,
      "step": 339
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.5455660852282613,
      "learning_rate": 4.999999896731129e-06,
      "loss": 0.0982,
      "step": 340
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3449930241711834,
      "learning_rate": 4.9999995869245225e-06,
      "loss": 0.153,
      "step": 341
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.504899212351084,
      "learning_rate": 4.999999070580207e-06,
      "loss": 0.3713,
      "step": 342
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.729351447832517,
      "learning_rate": 4.999998347698226e-06,
      "loss": 0.2369,
      "step": 343
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.553631413605947,
      "learning_rate": 4.9999974182786376e-06,
      "loss": 0.3509,
      "step": 344
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.012712318786689,
      "learning_rate": 4.9999962823215196e-06,
      "loss": 0.4717,
      "step": 345
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.712820388725436,
      "learning_rate": 4.999994939826965e-06,
      "loss": 0.2017,
      "step": 346
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.80345526153488,
      "learning_rate": 4.999993390795086e-06,
      "loss": 0.0234,
      "step": 347
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.7345499945017755,
      "learning_rate": 4.999991635226009e-06,
      "loss": 0.3634,
      "step": 348
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.766786214534938,
      "learning_rate": 4.999989673119881e-06,
      "loss": 0.2313,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.591563526787642,
      "learning_rate": 4.999987504476863e-06,
      "loss": 0.3712,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.728351799635282,
      "learning_rate": 4.999985129297133e-06,
      "loss": 0.416,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.13519912238862,
      "learning_rate": 4.999982547580889e-06,
      "loss": 0.422,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.015445522568022,
      "learning_rate": 4.999979759328344e-06,
      "loss": 0.4856,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.7586774397054095,
      "learning_rate": 4.999976764539727e-06,
      "loss": 0.2681,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.416254371601125,
      "learning_rate": 4.999973563215288e-06,
      "loss": 0.0203,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.43758318890487,
      "learning_rate": 4.999970155355289e-06,
      "loss": 0.3603,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.493827333865859,
      "learning_rate": 4.9999665409600125e-06,
      "loss": 0.2016,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.897395435989948,
      "learning_rate": 4.999962720029757e-06,
      "loss": 0.5675,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.077126283337919,
      "learning_rate": 4.999958692564838e-06,
      "loss": 0.3556,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.08883235282669,
      "learning_rate": 4.999954458565589e-06,
      "loss": 0.4734,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.778685576992799,
      "learning_rate": 4.999950018032359e-06,
      "loss": 0.2997,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.613829259527907,
      "learning_rate": 4.999945370965515e-06,
      "loss": 0.3882,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.5574761231647924,
      "learning_rate": 4.999940517365441e-06,
      "loss": 0.3642,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.626201184081092,
      "learning_rate": 4.999935457232539e-06,
      "loss": 0.4193,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.746623037753062,
      "learning_rate": 4.999930190567225e-06,
      "loss": 0.42,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.774183150545668,
      "learning_rate": 4.999924717369935e-06,
      "loss": 0.313,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.462932639712641,
      "learning_rate": 4.9999190376411215e-06,
      "loss": 0.2312,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.1944946709137385,
      "learning_rate": 4.999913151381254e-06,
      "loss": 0.2843,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.482978872667,
      "learning_rate": 4.999907058590818e-06,
      "loss": 0.6613,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.360757085885858,
      "learning_rate": 4.999900759270317e-06,
      "loss": 0.2794,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.821665318831147,
      "learning_rate": 4.999894253420272e-06,
      "loss": 0.8567,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.599061976442785,
      "learning_rate": 4.99988754104122e-06,
      "loss": 0.3162,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.810756426712397,
      "learning_rate": 4.9998806221337156e-06,
      "loss": 0.2801,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.452978242724487,
      "learning_rate": 4.999873496698331e-06,
      "loss": 0.6087,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.404511772769279,
      "learning_rate": 4.999866164735653e-06,
      "loss": 0.5423,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.086426048747888,
      "learning_rate": 4.99985862624629e-06,
      "loss": 0.2391,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.516185182129659,
      "learning_rate": 4.999850881230863e-06,
      "loss": 0.3237,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.952073725767943,
      "learning_rate": 4.999842929690012e-06,
      "loss": 0.3687,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.021838882253039,
      "learning_rate": 4.9998347716243946e-06,
      "loss": 0.4153,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.093213585308056,
      "learning_rate": 4.999826407034684e-06,
      "loss": 0.4912,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.098328153938281,
      "learning_rate": 4.999817835921572e-06,
      "loss": 0.513,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.226864605474398,
      "learning_rate": 4.9998090582857665e-06,
      "loss": 0.315,
      "step": 382
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.828873171665906,
      "learning_rate": 4.999800074127992e-06,
      "loss": 0.4847,
      "step": 383
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.818004874816558,
      "learning_rate": 4.999790883448992e-06,
      "loss": 0.3482,
      "step": 384
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.3540760352184815,
      "learning_rate": 4.999781486249524e-06,
      "loss": 0.4352,
      "step": 385
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.324884082877508,
      "learning_rate": 4.999771882530366e-06,
      "loss": 0.4271,
      "step": 386
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.784666868793869,
      "learning_rate": 4.999762072292311e-06,
      "loss": 0.3695,
      "step": 387
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.995923366175191,
      "learning_rate": 4.999752055536169e-06,
      "loss": 0.4253,
      "step": 388
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.418428538715507,
      "learning_rate": 4.999741832262767e-06,
      "loss": 0.3492,
      "step": 389
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.013789908749384,
      "learning_rate": 4.999731402472952e-06,
      "loss": 0.3782,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.362827070124288,
      "learning_rate": 4.999720766167583e-06,
      "loss": 0.4244,
      "step": 391
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.853997728026461,
      "learning_rate": 4.99970992334754e-06,
      "loss": 0.458,
      "step": 392
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.797534288431873,
      "learning_rate": 4.999698874013717e-06,
      "loss": 0.4463,
      "step": 393
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.2928786647661,
      "learning_rate": 4.999687618167031e-06,
      "loss": 0.7604,
      "step": 394
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.2107437835426405,
      "learning_rate": 4.999676155808408e-06,
      "loss": 0.1546,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.037612924102788,
      "learning_rate": 4.999664486938797e-06,
      "loss": 0.4366,
      "step": 396
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.538052650728863,
      "learning_rate": 4.99965261155916e-06,
      "loss": 0.2126,
      "step": 397
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.69921445408075,
      "learning_rate": 4.9996405296704794e-06,
      "loss": 0.3665,
      "step": 398
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5157976556503443,
      "learning_rate": 4.999628241273755e-06,
      "loss": 0.0252,
      "step": 399
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.6703907432376575,
      "learning_rate": 4.999615746369999e-06,
      "loss": 0.2096,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.714537807006354,
      "learning_rate": 4.999603044960245e-06,
      "loss": 0.3812,
      "step": 401
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.388907643688065,
      "learning_rate": 4.999590137045543e-06,
      "loss": 0.643,
      "step": 402
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.215717834414896,
      "learning_rate": 4.999577022626958e-06,
      "loss": 0.3939,
      "step": 403
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.4138423602192844,
      "learning_rate": 4.999563701705575e-06,
      "loss": 0.1799,
      "step": 404
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.8522785285167025,
      "learning_rate": 4.999550174282493e-06,
      "loss": 0.2763,
      "step": 405
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.524097374931975,
      "learning_rate": 4.9995364403588305e-06,
      "loss": 0.3527,
      "step": 406
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.730805814333051,
      "learning_rate": 4.999522499935722e-06,
      "loss": 0.4836,
      "step": 407
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.646963305872098,
      "learning_rate": 4.9995083530143185e-06,
      "loss": 0.4607,
      "step": 408
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.232912393464439,
      "learning_rate": 4.9994939995957895e-06,
      "loss": 0.2725,
      "step": 409
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.8190605075138695,
      "learning_rate": 4.999479439681321e-06,
      "loss": 0.373,
      "step": 410
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.697357663084856,
      "learning_rate": 4.9994646732721155e-06,
      "loss": 0.4972,
      "step": 411
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.736902225585279,
      "learning_rate": 4.999449700369392e-06,
      "loss": 0.3535,
      "step": 412
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.087348426815843,
      "learning_rate": 4.999434520974389e-06,
      "loss": 0.4532,
      "step": 413
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.285761156276342,
      "learning_rate": 4.99941913508836e-06,
      "loss": 0.5102,
      "step": 414
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.66498806842476,
      "learning_rate": 4.999403542712576e-06,
      "loss": 0.2254,
      "step": 415
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.666397741026113,
      "learning_rate": 4.999387743848325e-06,
      "loss": 0.37,
      "step": 416
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.586290561179673,
      "learning_rate": 4.999371738496913e-06,
      "loss": 0.1377,
      "step": 417
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.091767258310642,
      "learning_rate": 4.999355526659661e-06,
      "loss": 0.3701,
      "step": 418
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.613128507720976,
      "learning_rate": 4.99933910833791e-06,
      "loss": 0.3653,
      "step": 419
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.941159551869395,
      "learning_rate": 4.999322483533014e-06,
      "loss": 0.296,
      "step": 420
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.870781115787741,
      "learning_rate": 4.9993056522463505e-06,
      "loss": 0.5632,
      "step": 421
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.946129478987623,
      "learning_rate": 4.999288614479305e-06,
      "loss": 0.2431,
      "step": 422
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.594891464103957,
      "learning_rate": 4.999271370233289e-06,
      "loss": 0.2255,
      "step": 423
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.571777522449304,
      "learning_rate": 4.999253919509725e-06,
      "loss": 0.2741,
      "step": 424
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.100052939794932,
      "learning_rate": 4.999236262310055e-06,
      "loss": 0.3151,
      "step": 425
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.588855216319105,
      "learning_rate": 4.9992183986357376e-06,
      "loss": 0.5241,
      "step": 426
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.654175424877273,
      "learning_rate": 4.999200328488251e-06,
      "loss": 0.2149,
      "step": 427
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.137899541921036,
      "learning_rate": 4.999182051869085e-06,
      "loss": 0.3191,
      "step": 428
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.744723224929011,
      "learning_rate": 4.99916356877975e-06,
      "loss": 0.6184,
      "step": 429
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.3127171286978845,
      "learning_rate": 4.999144879221775e-06,
      "loss": 0.5707,
      "step": 430
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.2062846438557395,
      "learning_rate": 4.999125983196702e-06,
      "loss": 0.2214,
      "step": 431
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.8954721848640315,
      "learning_rate": 4.999106880706093e-06,
      "loss": 0.4524,
      "step": 432
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.0064866118773566,
      "learning_rate": 4.999087571751527e-06,
      "loss": 0.3023,
      "step": 433
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.34568862555559,
      "learning_rate": 4.999068056334597e-06,
      "loss": 0.6093,
      "step": 434
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.81193071941604,
      "learning_rate": 4.999048334456916e-06,
      "loss": 0.1789,
      "step": 435
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.006148873779516,
      "learning_rate": 4.999028406120115e-06,
      "loss": 0.3462,
      "step": 436
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.353439558619988,
      "learning_rate": 4.999008271325838e-06,
      "loss": 0.166,
      "step": 437
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.870660874883308,
      "learning_rate": 4.99898793007575e-06,
      "loss": 0.4825,
      "step": 438
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.951756093121901,
      "learning_rate": 4.998967382371533e-06,
      "loss": 0.3298,
      "step": 439
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.751623456331759,
      "learning_rate": 4.99894662821488e-06,
      "loss": 0.4092,
      "step": 440
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.572361350772929,
      "learning_rate": 4.9989256676075095e-06,
      "loss": 0.2775,
      "step": 441
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.625165553217044,
      "learning_rate": 4.998904500551152e-06,
      "loss": 0.3624,
      "step": 442
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.440787759722936,
      "learning_rate": 4.998883127047555e-06,
      "loss": 0.4108,
      "step": 443
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.366183723878273,
      "learning_rate": 4.998861547098486e-06,
      "loss": 0.3797,
      "step": 444
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0904145139357246,
      "learning_rate": 4.998839760705728e-06,
      "loss": 0.1236,
      "step": 445
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.4108059877729975,
      "learning_rate": 4.9988177678710794e-06,
      "loss": 0.2939,
      "step": 446
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.089185076028177,
      "learning_rate": 4.998795568596358e-06,
      "loss": 0.5047,
      "step": 447
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.869462783837914,
      "learning_rate": 4.998773162883398e-06,
      "loss": 0.2911,
      "step": 448
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.032162727470419,
      "learning_rate": 4.99875055073405e-06,
      "loss": 0.4058,
      "step": 449
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.728621751473672,
      "learning_rate": 4.998727732150182e-06,
      "loss": 0.4458,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.200648704145209,
      "learning_rate": 4.998704707133679e-06,
      "loss": 0.358,
      "step": 451
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.747261031047393,
      "learning_rate": 4.998681475686444e-06,
      "loss": 0.5386,
      "step": 452
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.757616777153422,
      "learning_rate": 4.998658037810396e-06,
      "loss": 0.5708,
      "step": 453
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.929190086252327,
      "learning_rate": 4.998634393507471e-06,
      "loss": 0.5977,
      "step": 454
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.809328348014042,
      "learning_rate": 4.998610542779622e-06,
      "loss": 0.0838,
      "step": 455
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.968716875656413,
      "learning_rate": 4.998586485628821e-06,
      "loss": 0.2533,
      "step": 456
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.808727138253011,
      "learning_rate": 4.998562222057054e-06,
      "loss": 0.208,
      "step": 457
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.584596992978992,
      "learning_rate": 4.9985377520663255e-06,
      "loss": 0.5649,
      "step": 458
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.057455605389969,
      "learning_rate": 4.9985130756586585e-06,
      "loss": 0.3754,
      "step": 459
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.340297062454588,
      "learning_rate": 4.99848819283609e-06,
      "loss": 0.3853,
      "step": 460
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.202871378187258,
      "learning_rate": 4.998463103600677e-06,
      "loss": 0.3294,
      "step": 461
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.853487425076116,
      "learning_rate": 4.998437807954492e-06,
      "loss": 0.4974,
      "step": 462
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.499412211735712,
      "learning_rate": 4.998412305899623e-06,
      "loss": 0.4282,
      "step": 463
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.068671835202401,
      "learning_rate": 4.998386597438179e-06,
      "loss": 0.3569,
      "step": 464
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.975173230042677,
      "learning_rate": 4.998360682572284e-06,
      "loss": 0.6401,
      "step": 465
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.0536843094592845,
      "learning_rate": 4.998334561304078e-06,
      "loss": 0.3467,
      "step": 466
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.0258890055263,
      "learning_rate": 4.998308233635718e-06,
      "loss": 0.2831,
      "step": 467
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.174738424499288,
      "learning_rate": 4.998281699569381e-06,
      "loss": 0.253,
      "step": 468
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.03118164700265,
      "learning_rate": 4.99825495910726e-06,
      "loss": 0.7141,
      "step": 469
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.3470876771530955,
      "learning_rate": 4.9982280122515606e-06,
      "loss": 0.4784,
      "step": 470
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.480444876210304,
      "learning_rate": 4.998200859004512e-06,
      "loss": 0.3516,
      "step": 471
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.8182630127140555,
      "learning_rate": 4.9981734993683555e-06,
      "loss": 0.6094,
      "step": 472
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.636069370084782,
      "learning_rate": 4.998145933345353e-06,
      "loss": 0.3465,
      "step": 473
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.919606151985379,
      "learning_rate": 4.998118160937782e-06,
      "loss": 0.608,
      "step": 474
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.064852932113094,
      "learning_rate": 4.998090182147934e-06,
      "loss": 0.2433,
      "step": 475
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.330960096622529,
      "learning_rate": 4.998061996978124e-06,
      "loss": 0.4947,
      "step": 476
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.086526518045042,
      "learning_rate": 4.998033605430679e-06,
      "loss": 0.4722,
      "step": 477
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.628044118778697,
      "learning_rate": 4.998005007507946e-06,
      "loss": 0.2368,
      "step": 478
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.5819387483432745,
      "learning_rate": 4.997976203212285e-06,
      "loss": 0.1958,
      "step": 479
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.716577384865029,
      "learning_rate": 4.997947192546076e-06,
      "loss": 0.5442,
      "step": 480
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.286338300714259,
      "learning_rate": 4.997917975511718e-06,
      "loss": 0.3031,
      "step": 481
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.2500598017500675,
      "learning_rate": 4.997888552111624e-06,
      "loss": 0.3968,
      "step": 482
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.647986747363238,
      "learning_rate": 4.997858922348224e-06,
      "loss": 0.2777,
      "step": 483
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.279203499094539,
      "learning_rate": 4.997829086223965e-06,
      "loss": 0.3065,
      "step": 484
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.393534187958327,
      "learning_rate": 4.997799043741314e-06,
      "loss": 0.376,
      "step": 485
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.831502743664822,
      "learning_rate": 4.997768794902752e-06,
      "loss": 0.4806,
      "step": 486
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.403217571543578,
      "learning_rate": 4.997738339710778e-06,
      "loss": 0.3434,
      "step": 487
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.559685267869535,
      "learning_rate": 4.997707678167909e-06,
      "loss": 0.0823,
      "step": 488
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.031986294123999,
      "learning_rate": 4.997676810276676e-06,
      "loss": 0.5132,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.222283142494859,
      "learning_rate": 4.99764573603963e-06,
      "loss": 0.2091,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.4019748105418275,
      "learning_rate": 4.99761445545934e-06,
      "loss": 0.3043,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.135655730707593,
      "learning_rate": 4.997582968538388e-06,
      "loss": 0.4779,
      "step": 492
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.185012738094043,
      "learning_rate": 4.997551275279377e-06,
      "loss": 0.6587,
      "step": 493
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.932208824791579,
      "learning_rate": 4.997519375684923e-06,
      "loss": 0.316,
      "step": 494
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.758750078716067,
      "learning_rate": 4.997487269757664e-06,
      "loss": 0.3403,
      "step": 495
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.479662576397544,
      "learning_rate": 4.997454957500251e-06,
      "loss": 0.328,
      "step": 496
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.174702776632983,
      "learning_rate": 4.997422438915354e-06,
      "loss": 0.2877,
      "step": 497
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.3185639540135785,
      "learning_rate": 4.9973897140056586e-06,
      "loss": 0.1819,
      "step": 498
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.783104462834797,
      "learning_rate": 4.99735678277387e-06,
      "loss": 0.2266,
      "step": 499
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.6070088538914975,
      "learning_rate": 4.997323645222707e-06,
      "loss": 0.2389,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.58400048487137,
      "learning_rate": 4.997290301354909e-06,
      "loss": 0.2965,
      "step": 501
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.297200928062214,
      "learning_rate": 4.997256751173229e-06,
      "loss": 0.4587,
      "step": 502
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.183862262246363,
      "learning_rate": 4.99722299468044e-06,
      "loss": 0.4111,
      "step": 503
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.354343025873604,
      "learning_rate": 4.99718903187933e-06,
      "loss": 0.1884,
      "step": 504
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.130663001474344,
      "learning_rate": 4.997154862772706e-06,
      "loss": 0.3716,
      "step": 505
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.429185737788216,
      "learning_rate": 4.99712048736339e-06,
      "loss": 0.3301,
      "step": 506
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.250435781001297,
      "learning_rate": 4.997085905654222e-06,
      "loss": 0.2608,
      "step": 507
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9873515801350266,
      "learning_rate": 4.9970511176480586e-06,
      "loss": 0.1257,
      "step": 508
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.093229426717078,
      "learning_rate": 4.997016123347775e-06,
      "loss": 0.5469,
      "step": 509
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.022248052061403,
      "learning_rate": 4.9969809227562605e-06,
      "loss": 0.133,
      "step": 510
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.793147164285434,
      "learning_rate": 4.9969455158764246e-06,
      "loss": 0.4875,
      "step": 511
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.471424851909347,
      "learning_rate": 4.996909902711192e-06,
      "loss": 0.481,
      "step": 512
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.521375925122893,
      "learning_rate": 4.996874083263506e-06,
      "loss": 0.2172,
      "step": 513
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.599350649957783,
      "learning_rate": 4.996838057536323e-06,
      "loss": 0.4658,
      "step": 514
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1994129482377005,
      "learning_rate": 4.996801825532622e-06,
      "loss": 0.1089,
      "step": 515
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.456021463579534,
      "learning_rate": 4.996765387255396e-06,
      "loss": 0.4776,
      "step": 516
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.33538813136177,
      "learning_rate": 4.996728742707653e-06,
      "loss": 0.4329,
      "step": 517
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2625370863678382,
      "learning_rate": 4.996691891892423e-06,
      "loss": 0.1674,
      "step": 518
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.386881005457414,
      "learning_rate": 4.9966548348127495e-06,
      "loss": 0.4872,
      "step": 519
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.817495800486912,
      "learning_rate": 4.996617571471694e-06,
      "loss": 0.3827,
      "step": 520
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.820956408645621,
      "learning_rate": 4.996580101872335e-06,
      "loss": 0.3844,
      "step": 521
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.701726876012715,
      "learning_rate": 4.996542426017768e-06,
      "loss": 0.2115,
      "step": 522
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.1915726870077785,
      "learning_rate": 4.996504543911106e-06,
      "loss": 0.3317,
      "step": 523
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.328253516513202,
      "learning_rate": 4.996466455555478e-06,
      "loss": 0.3442,
      "step": 524
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.154319898618326,
      "learning_rate": 4.996428160954031e-06,
      "loss": 0.4013,
      "step": 525
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.89358747462921,
      "learning_rate": 4.996389660109928e-06,
      "loss": 0.3423,
      "step": 526
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.097273507889422,
      "learning_rate": 4.996350953026351e-06,
      "loss": 0.2066,
      "step": 527
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.636715614202374,
      "learning_rate": 4.9963120397064976e-06,
      "loss": 0.1388,
      "step": 528
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.494505811369369,
      "learning_rate": 4.996272920153582e-06,
      "loss": 0.1755,
      "step": 529
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.5290964420759,
      "learning_rate": 4.996233594370836e-06,
      "loss": 0.3609,
      "step": 530
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.535642387540622,
      "learning_rate": 4.996194062361509e-06,
      "loss": 0.5188,
      "step": 531
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.62895298787732,
      "learning_rate": 4.996154324128867e-06,
      "loss": 0.2673,
      "step": 532
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.137277500064123,
      "learning_rate": 4.996114379676192e-06,
      "loss": 0.328,
      "step": 533
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.281419952635379,
      "learning_rate": 4.9960742290067855e-06,
      "loss": 0.3499,
      "step": 534
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.802480217583315,
      "learning_rate": 4.9960338721239634e-06,
      "loss": 0.3878,
      "step": 535
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.25416605839524,
      "learning_rate": 4.99599330903106e-06,
      "loss": 0.4943,
      "step": 536
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.271389675608033,
      "learning_rate": 4.9959525397314265e-06,
      "loss": 0.418,
      "step": 537
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.668737918777067,
      "learning_rate": 4.995911564228432e-06,
      "loss": 0.4464,
      "step": 538
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.3486232261952456,
      "learning_rate": 4.9958703825254605e-06,
      "loss": 0.4056,
      "step": 539
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.122348062935166,
      "learning_rate": 4.995828994625915e-06,
      "loss": 0.3725,
      "step": 540
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.664446581594326,
      "learning_rate": 4.995787400533213e-06,
      "loss": 0.4651,
      "step": 541
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.85925431532091,
      "learning_rate": 4.9957456002507934e-06,
      "loss": 0.2227,
      "step": 542
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.055537880197035,
      "learning_rate": 4.995703593782107e-06,
      "loss": 0.1902,
      "step": 543
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.61270252770125,
      "learning_rate": 4.995661381130627e-06,
      "loss": 0.4701,
      "step": 544
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.875721066891864,
      "learning_rate": 4.995618962299839e-06,
      "loss": 0.2565,
      "step": 545
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.376816815601363,
      "learning_rate": 4.995576337293246e-06,
      "loss": 0.3331,
      "step": 546
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.223589434075748,
      "learning_rate": 4.995533506114373e-06,
      "loss": 0.5675,
      "step": 547
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.729106270970082,
      "learning_rate": 4.995490468766756e-06,
      "loss": 0.6038,
      "step": 548
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.714677738758223,
      "learning_rate": 4.995447225253952e-06,
      "loss": 0.3444,
      "step": 549
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.188068526353272,
      "learning_rate": 4.995403775579532e-06,
      "loss": 0.355,
      "step": 550
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.5697179227273015,
      "learning_rate": 4.995360119747086e-06,
      "loss": 0.5554,
      "step": 551
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.783105106107803,
      "learning_rate": 4.995316257760222e-06,
      "loss": 0.4041,
      "step": 552
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.194088039023633,
      "learning_rate": 4.995272189622562e-06,
      "loss": 0.6415,
      "step": 553
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.488853095427772,
      "learning_rate": 4.995227915337747e-06,
      "loss": 0.4571,
      "step": 554
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.305781007469251,
      "learning_rate": 4.9951834349094354e-06,
      "loss": 0.3381,
      "step": 555
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.773456997597374,
      "learning_rate": 4.995138748341302e-06,
      "loss": 0.3131,
      "step": 556
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.872594554879847,
      "learning_rate": 4.995093855637038e-06,
      "loss": 0.309,
      "step": 557
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.5793027515626,
      "learning_rate": 4.995048756800353e-06,
      "loss": 0.3433,
      "step": 558
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.1898040559664755,
      "learning_rate": 4.995003451834971e-06,
      "loss": 0.2916,
      "step": 559
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.264734901615739,
      "learning_rate": 4.994957940744637e-06,
      "loss": 0.3839,
      "step": 560
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.122085322627052,
      "learning_rate": 4.99491222353311e-06,
      "loss": 0.4991,
      "step": 561
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.614144035477857,
      "learning_rate": 4.994866300204167e-06,
      "loss": 0.0659,
      "step": 562
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.860757372809934,
      "learning_rate": 4.994820170761601e-06,
      "loss": 0.4912,
      "step": 563
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.275498836330784,
      "learning_rate": 4.994773835209226e-06,
      "loss": 0.3844,
      "step": 564
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.503609528577412,
      "learning_rate": 4.994727293550866e-06,
      "loss": 0.4654,
      "step": 565
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.614619720390818,
      "learning_rate": 4.994680545790369e-06,
      "loss": 0.3102,
      "step": 566
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.143998728295609,
      "learning_rate": 4.994633591931596e-06,
      "loss": 0.6021,
      "step": 567
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.216106690988294,
      "learning_rate": 4.9945864319784256e-06,
      "loss": 0.3658,
      "step": 568
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.760897212743643,
      "learning_rate": 4.994539065934755e-06,
      "loss": 0.4654,
      "step": 569
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.143945526141823,
      "learning_rate": 4.994491493804496e-06,
      "loss": 0.421,
      "step": 570
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.1160183488007025,
      "learning_rate": 4.994443715591582e-06,
      "loss": 0.504,
      "step": 571
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.903903860934759,
      "learning_rate": 4.994395731299955e-06,
      "loss": 0.4,
      "step": 572
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.887313473246122,
      "learning_rate": 4.9943475409335835e-06,
      "loss": 0.4324,
      "step": 573
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.076902754749434,
      "learning_rate": 4.994299144496447e-06,
      "loss": 0.5291,
      "step": 574
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.127988007547308,
      "learning_rate": 4.994250541992544e-06,
      "loss": 0.3187,
      "step": 575
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.638406466442166,
      "learning_rate": 4.99420173342589e-06,
      "loss": 0.1721,
      "step": 576
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.502261290155829,
      "learning_rate": 4.994152718800517e-06,
      "loss": 0.4619,
      "step": 577
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.7039369116971335,
      "learning_rate": 4.994103498120474e-06,
      "loss": 0.3539,
      "step": 578
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.415081194962708,
      "learning_rate": 4.994054071389829e-06,
      "loss": 0.3021,
      "step": 579
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.8462809962193605,
      "learning_rate": 4.994004438612664e-06,
      "loss": 0.2258,
      "step": 580
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.884821146430121,
      "learning_rate": 4.99395459979308e-06,
      "loss": 0.5129,
      "step": 581
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.946416168999551,
      "learning_rate": 4.993904554935193e-06,
      "loss": 0.2357,
      "step": 582
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.892622265116859,
      "learning_rate": 4.993854304043139e-06,
      "loss": 0.4759,
      "step": 583
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.027673095752436,
      "learning_rate": 4.99380384712107e-06,
      "loss": 0.2678,
      "step": 584
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.427746768630235,
      "learning_rate": 4.993753184173154e-06,
      "loss": 0.3378,
      "step": 585
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.00297747785127,
      "learning_rate": 4.993702315203574e-06,
      "loss": 0.3498,
      "step": 586
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.486060252846052,
      "learning_rate": 4.993651240216537e-06,
      "loss": 0.737,
      "step": 587
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.13987676353716,
      "learning_rate": 4.993599959216259e-06,
      "loss": 0.4793,
      "step": 588
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.022848713010218,
      "learning_rate": 4.993548472206978e-06,
      "loss": 0.1173,
      "step": 589
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.084228190918327,
      "learning_rate": 4.993496779192947e-06,
      "loss": 0.4247,
      "step": 590
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.9004574428060215,
      "learning_rate": 4.993444880178437e-06,
      "loss": 0.3315,
      "step": 591
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.070321357553123,
      "learning_rate": 4.993392775167737e-06,
      "loss": 0.3473,
      "step": 592
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.826481878726085,
      "learning_rate": 4.99334046416515e-06,
      "loss": 0.4956,
      "step": 593
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.987398471768236,
      "learning_rate": 4.9932879471749976e-06,
      "loss": 0.395,
      "step": 594
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.8352103287921615,
      "learning_rate": 4.993235224201618e-06,
      "loss": 0.3998,
      "step": 595
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.759066038204785,
      "learning_rate": 4.993182295249369e-06,
      "loss": 0.4124,
      "step": 596
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.6034699321494,
      "learning_rate": 4.9931291603226226e-06,
      "loss": 0.4124,
      "step": 597
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.594767026825527,
      "learning_rate": 4.993075819425767e-06,
      "loss": 0.235,
      "step": 598
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.28251724710278,
      "learning_rate": 4.993022272563211e-06,
      "loss": 0.3274,
      "step": 599
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.296829467253172,
      "learning_rate": 4.992968519739377e-06,
      "loss": 0.2856,
      "step": 600
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.804652825974466,
      "learning_rate": 4.992914560958706e-06,
      "loss": 0.3632,
      "step": 601
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.915688426574023,
      "learning_rate": 4.992860396225656e-06,
      "loss": 0.7008,
      "step": 602
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.276533089082179,
      "learning_rate": 4.992806025544703e-06,
      "loss": 0.3753,
      "step": 603
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.635130252780867,
      "learning_rate": 4.992751448920337e-06,
      "loss": 0.5633,
      "step": 604
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.2770165153237,
      "learning_rate": 4.992696666357068e-06,
      "loss": 0.4398,
      "step": 605
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.636574564915326,
      "learning_rate": 4.9926416778594205e-06,
      "loss": 0.5059,
      "step": 606
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.891378624775306,
      "learning_rate": 4.992586483431939e-06,
      "loss": 0.364,
      "step": 607
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.381649905561892,
      "learning_rate": 4.992531083079183e-06,
      "loss": 0.4278,
      "step": 608
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.618380781344579,
      "learning_rate": 4.992475476805729e-06,
      "loss": 0.3308,
      "step": 609
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.666218554926143,
      "learning_rate": 4.992419664616171e-06,
      "loss": 0.4987,
      "step": 610
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.783664832358752,
      "learning_rate": 4.992363646515119e-06,
      "loss": 0.2319,
      "step": 611
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.30279978839127,
      "learning_rate": 4.992307422507203e-06,
      "loss": 0.423,
      "step": 612
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.496145071686213,
      "learning_rate": 4.9922509925970665e-06,
      "loss": 0.3947,
      "step": 613
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.903778086925277,
      "learning_rate": 4.9921943567893726e-06,
      "loss": 0.545,
      "step": 614
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.281595514578019,
      "learning_rate": 4.9921375150887985e-06,
      "loss": 0.3078,
      "step": 615
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.65111545171332,
      "learning_rate": 4.992080467500042e-06,
      "loss": 0.3259,
      "step": 616
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.455835521929184,
      "learning_rate": 4.992023214027814e-06,
      "loss": 0.3462,
      "step": 617
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.4523240779264714,
      "learning_rate": 4.991965754676848e-06,
      "loss": 0.5593,
      "step": 618
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.7422965206087415,
      "learning_rate": 4.991908089451887e-06,
      "loss": 0.4354,
      "step": 619
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.857977668905956,
      "learning_rate": 4.991850218357697e-06,
      "loss": 0.4384,
      "step": 620
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.307164057323436,
      "learning_rate": 4.991792141399059e-06,
      "loss": 0.3087,
      "step": 621
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.827933003200096,
      "learning_rate": 4.991733858580771e-06,
      "loss": 0.6488,
      "step": 622
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.732084768236774,
      "learning_rate": 4.991675369907648e-06,
      "loss": 0.4742,
      "step": 623
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.77654347387765,
      "learning_rate": 4.991616675384522e-06,
      "loss": 0.4998,
      "step": 624
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.6942989824257,
      "learning_rate": 4.991557775016242e-06,
      "loss": 0.3271,
      "step": 625
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.2935516106392,
      "learning_rate": 4.991498668807674e-06,
      "loss": 0.1901,
      "step": 626
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.952205490251467,
      "learning_rate": 4.991439356763701e-06,
      "loss": 0.1978,
      "step": 627
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.544031253407698,
      "learning_rate": 4.991379838889223e-06,
      "loss": 0.4571,
      "step": 628
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.4596822046705045,
      "learning_rate": 4.991320115189158e-06,
      "loss": 0.2577,
      "step": 629
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.443907887998791,
      "learning_rate": 4.991260185668439e-06,
      "loss": 0.1723,
      "step": 630
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.391487056757394,
      "learning_rate": 4.991200050332018e-06,
      "loss": 0.5163,
      "step": 631
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.448811821699414,
      "learning_rate": 4.991139709184861e-06,
      "loss": 0.2362,
      "step": 632
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.25925792405568,
      "learning_rate": 4.9910791622319565e-06,
      "loss": 0.1983,
      "step": 633
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.176875322608217,
      "learning_rate": 4.991018409478303e-06,
      "loss": 0.3115,
      "step": 634
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.019875812247772,
      "learning_rate": 4.990957450928923e-06,
      "loss": 0.2008,
      "step": 635
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.679982485563525,
      "learning_rate": 4.99089628658885e-06,
      "loss": 0.4661,
      "step": 636
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.1140943933565,
      "learning_rate": 4.990834916463138e-06,
      "loss": 0.383,
      "step": 637
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.540321575697741,
      "learning_rate": 4.990773340556857e-06,
      "loss": 0.5539,
      "step": 638
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.599042742616538,
      "learning_rate": 4.9907115588750944e-06,
      "loss": 0.3602,
      "step": 639
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.256809915000877,
      "learning_rate": 4.9906495714229545e-06,
      "loss": 0.2736,
      "step": 640
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.630360879352501,
      "learning_rate": 4.990587378205557e-06,
      "loss": 0.4159,
      "step": 641
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.173467979533397,
      "learning_rate": 4.990524979228042e-06,
      "loss": 0.3235,
      "step": 642
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.915381510178831,
      "learning_rate": 4.9904623744955626e-06,
      "loss": 0.3339,
      "step": 643
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.37803997131279,
      "learning_rate": 4.990399564013293e-06,
      "loss": 0.3142,
      "step": 644
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.50483874593242,
      "learning_rate": 4.990336547786421e-06,
      "loss": 0.3062,
      "step": 645
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.605929737038529,
      "learning_rate": 4.9902733258201515e-06,
      "loss": 0.2266,
      "step": 646
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.602181988442946,
      "learning_rate": 4.990209898119711e-06,
      "loss": 0.5254,
      "step": 647
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.901337174007983,
      "learning_rate": 4.990146264690336e-06,
      "loss": 0.2397,
      "step": 648
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.987923668032799,
      "learning_rate": 4.990082425537286e-06,
      "loss": 0.4499,
      "step": 649
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.580158134505028,
      "learning_rate": 4.990018380665834e-06,
      "loss": 0.4471,
      "step": 650
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.728436673864426,
      "learning_rate": 4.989954130081272e-06,
      "loss": 0.2698,
      "step": 651
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.806231030587149,
      "learning_rate": 4.989889673788906e-06,
      "loss": 0.3218,
      "step": 652
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.346541605876197,
      "learning_rate": 4.989825011794063e-06,
      "loss": 0.4772,
      "step": 653
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.330464867082852,
      "learning_rate": 4.989760144102085e-06,
      "loss": 0.4158,
      "step": 654
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.356880503356142,
      "learning_rate": 4.98969507071833e-06,
      "loss": 0.2917,
      "step": 655
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.866520189539124,
      "learning_rate": 4.989629791648174e-06,
      "loss": 0.306,
      "step": 656
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.326162392032411,
      "learning_rate": 4.989564306897011e-06,
      "loss": 0.4267,
      "step": 657
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.326751498487382,
      "learning_rate": 4.989498616470251e-06,
      "loss": 0.4912,
      "step": 658
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.0944823747941035,
      "learning_rate": 4.9894327203733214e-06,
      "loss": 0.2103,
      "step": 659
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.0392597533959105,
      "learning_rate": 4.989366618611664e-06,
      "loss": 0.3616,
      "step": 660
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.654308200099705,
      "learning_rate": 4.989300311190742e-06,
      "loss": 0.4667,
      "step": 661
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.947488527204091,
      "learning_rate": 4.989233798116033e-06,
      "loss": 0.3381,
      "step": 662
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.187634579605834,
      "learning_rate": 4.989167079393031e-06,
      "loss": 0.3559,
      "step": 663
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.331868540088735,
      "learning_rate": 4.989100155027249e-06,
      "loss": 0.2539,
      "step": 664
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.936775536143662,
      "learning_rate": 4.989033025024216e-06,
      "loss": 0.36,
      "step": 665
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.949183640709516,
      "learning_rate": 4.988965689389476e-06,
      "loss": 0.2779,
      "step": 666
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.341954019344076,
      "learning_rate": 4.988898148128595e-06,
      "loss": 0.2737,
      "step": 667
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.708976837982273,
      "learning_rate": 4.988830401247151e-06,
      "loss": 0.5187,
      "step": 668
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.4197195626156445,
      "learning_rate": 4.988762448750741e-06,
      "loss": 0.4904,
      "step": 669
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.118775513022373,
      "learning_rate": 4.98869429064498e-06,
      "loss": 0.2659,
      "step": 670
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.995381560707275,
      "learning_rate": 4.988625926935497e-06,
      "loss": 0.3283,
      "step": 671
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.465565934911391,
      "learning_rate": 4.988557357627941e-06,
      "loss": 0.5889,
      "step": 672
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.841273049896508,
      "learning_rate": 4.9884885827279774e-06,
      "loss": 0.2221,
      "step": 673
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.927536416639761,
      "learning_rate": 4.988419602241288e-06,
      "loss": 0.397,
      "step": 674
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.347327998409497,
      "learning_rate": 4.988350416173571e-06,
      "loss": 0.2516,
      "step": 675
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.546908770777795,
      "learning_rate": 4.9882810245305416e-06,
      "loss": 0.41,
      "step": 676
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.689448669496554,
      "learning_rate": 4.988211427317934e-06,
      "loss": 0.3612,
      "step": 677
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.495724702534899,
      "learning_rate": 4.988141624541497e-06,
      "loss": 0.4856,
      "step": 678
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.732839432070167,
      "learning_rate": 4.988071616206997e-06,
      "loss": 0.4734,
      "step": 679
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.086272752902296,
      "learning_rate": 4.988001402320219e-06,
      "loss": 0.2694,
      "step": 680
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.989547061447947,
      "learning_rate": 4.987930982886964e-06,
      "loss": 0.179,
      "step": 681
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.032265594436711,
      "learning_rate": 4.987860357913048e-06,
      "loss": 0.247,
      "step": 682
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.533646247471228,
      "learning_rate": 4.987789527404306e-06,
      "loss": 0.2942,
      "step": 683
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.392314174319122,
      "learning_rate": 4.9877184913665924e-06,
      "loss": 0.3691,
      "step": 684
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.9258961964832695,
      "learning_rate": 4.9876472498057715e-06,
      "loss": 0.3643,
      "step": 685
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.294996430434361,
      "learning_rate": 4.987575802727732e-06,
      "loss": 0.3258,
      "step": 686
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.901380409789979,
      "learning_rate": 4.987504150138376e-06,
      "loss": 0.3048,
      "step": 687
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.18961558677958,
      "learning_rate": 4.987432292043621e-06,
      "loss": 0.3206,
      "step": 688
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.224698097582956,
      "learning_rate": 4.987360228449407e-06,
      "loss": 0.3658,
      "step": 689
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.73846658475888,
      "learning_rate": 4.987287959361685e-06,
      "loss": 0.3837,
      "step": 690
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.227323272789492,
      "learning_rate": 4.987215484786426e-06,
      "loss": 0.2676,
      "step": 691
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.390024137441155,
      "learning_rate": 4.987142804729619e-06,
      "loss": 0.3399,
      "step": 692
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.4916302724485755,
      "learning_rate": 4.987069919197266e-06,
      "loss": 0.3178,
      "step": 693
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.97751498810212,
      "learning_rate": 4.986996828195391e-06,
      "loss": 0.1565,
      "step": 694
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.408555043450299,
      "learning_rate": 4.98692353173003e-06,
      "loss": 0.2804,
      "step": 695
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.339998519504957,
      "learning_rate": 4.98685002980724e-06,
      "loss": 0.508,
      "step": 696
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.291852796693744,
      "learning_rate": 4.986776322433092e-06,
      "loss": 0.373,
      "step": 697
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.290256736119538,
      "learning_rate": 4.986702409613678e-06,
      "loss": 0.4572,
      "step": 698
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.344491259394742,
      "learning_rate": 4.986628291355101e-06,
      "loss": 0.3114,
      "step": 699
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.888593726347257,
      "learning_rate": 4.986553967663487e-06,
      "loss": 0.4531,
      "step": 700
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4315481797216507,
      "learning_rate": 4.986479438544974e-06,
      "loss": 0.1714,
      "step": 701
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.1420898929393255,
      "learning_rate": 4.986404704005722e-06,
      "loss": 0.1984,
      "step": 702
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.06568319200799,
      "learning_rate": 4.986329764051902e-06,
      "loss": 0.3389,
      "step": 703
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.626576983110791,
      "learning_rate": 4.986254618689708e-06,
      "loss": 0.3575,
      "step": 704
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.8464086951896945,
      "learning_rate": 4.9861792679253455e-06,
      "loss": 0.3274,
      "step": 705
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.552546766294312,
      "learning_rate": 4.986103711765041e-06,
      "loss": 0.382,
      "step": 706
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.457491834707048,
      "learning_rate": 4.986027950215038e-06,
      "loss": 0.4674,
      "step": 707
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.347001512129924,
      "learning_rate": 4.985951983281593e-06,
      "loss": 0.1538,
      "step": 708
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.346472551186413,
      "learning_rate": 4.985875810970983e-06,
      "loss": 0.37,
      "step": 709
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.583157257428118,
      "learning_rate": 4.985799433289501e-06,
      "loss": 0.3099,
      "step": 710
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.892831934358203,
      "learning_rate": 4.985722850243458e-06,
      "loss": 0.2338,
      "step": 711
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.349630849261461,
      "learning_rate": 4.985646061839179e-06,
      "loss": 0.2276,
      "step": 712
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.83796324809018,
      "learning_rate": 4.9855690680830094e-06,
      "loss": 0.5167,
      "step": 713
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6991903751033586,
      "learning_rate": 4.985491868981309e-06,
      "loss": 0.0362,
      "step": 714
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.73781544304384,
      "learning_rate": 4.985414464540457e-06,
      "loss": 0.3693,
      "step": 715
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.219379980218504,
      "learning_rate": 4.9853368547668465e-06,
      "loss": 0.4984,
      "step": 716
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.542610905132617,
      "learning_rate": 4.98525903966689e-06,
      "loss": 0.4005,
      "step": 717
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.046680434173873,
      "learning_rate": 4.985181019247016e-06,
      "loss": 0.5163,
      "step": 718
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.421433987580532,
      "learning_rate": 4.985102793513671e-06,
      "loss": 0.6261,
      "step": 719
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.365490350504943,
      "learning_rate": 4.985024362473317e-06,
      "loss": 0.2882,
      "step": 720
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.879641680960923,
      "learning_rate": 4.984945726132433e-06,
      "loss": 0.2166,
      "step": 721
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.902357186416891,
      "learning_rate": 4.984866884497517e-06,
      "loss": 0.5572,
      "step": 722
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.543845026805887,
      "learning_rate": 4.984787837575081e-06,
      "loss": 0.6275,
      "step": 723
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.666822125398447,
      "learning_rate": 4.984708585371656e-06,
      "loss": 0.419,
      "step": 724
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.440138386005689,
      "learning_rate": 4.98462912789379e-06,
      "loss": 0.2077,
      "step": 725
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.994818314334752,
      "learning_rate": 4.984549465148047e-06,
      "loss": 0.4837,
      "step": 726
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.06818620070806,
      "learning_rate": 4.984469597141007e-06,
      "loss": 0.6486,
      "step": 727
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.946670425697041,
      "learning_rate": 4.984389523879271e-06,
      "loss": 0.0625,
      "step": 728
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.835520462170364,
      "learning_rate": 4.984309245369453e-06,
      "loss": 0.6111,
      "step": 729
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.894527222278872,
      "learning_rate": 4.984228761618184e-06,
      "loss": 0.1043,
      "step": 730
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.734978607420163,
      "learning_rate": 4.984148072632115e-06,
      "loss": 0.5727,
      "step": 731
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.154803616414234,
      "learning_rate": 4.98406717841791e-06,
      "loss": 0.3514,
      "step": 732
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.971324720368878,
      "learning_rate": 4.9839860789822556e-06,
      "loss": 0.3579,
      "step": 733
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.637605849864982,
      "learning_rate": 4.9839047743318485e-06,
      "loss": 0.1719,
      "step": 734
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.422137884906374,
      "learning_rate": 4.983823264473408e-06,
      "loss": 0.4142,
      "step": 735
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.185185349107533,
      "learning_rate": 4.9837415494136656e-06,
      "loss": 0.4418,
      "step": 736
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.535545009795047,
      "learning_rate": 4.983659629159373e-06,
      "loss": 0.3548,
      "step": 737
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.501332040417036,
      "learning_rate": 4.9835775037173e-06,
      "loss": 0.4607,
      "step": 738
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.084187498422502,
      "learning_rate": 4.983495173094229e-06,
      "loss": 0.5265,
      "step": 739
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.964834749389983,
      "learning_rate": 4.9834126372969626e-06,
      "loss": 0.4012,
      "step": 740
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.08098155869979,
      "learning_rate": 4.983329896332321e-06,
      "loss": 0.1554,
      "step": 741
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.8026265944058366,
      "learning_rate": 4.983246950207137e-06,
      "loss": 0.0535,
      "step": 742
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3776162346955245,
      "learning_rate": 4.983163798928265e-06,
      "loss": 0.112,
      "step": 743
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.35591111130559,
      "learning_rate": 4.983080442502575e-06,
      "loss": 0.3405,
      "step": 744
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.462627120567767,
      "learning_rate": 4.982996880936952e-06,
      "loss": 0.3,
      "step": 745
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.968565171054131,
      "learning_rate": 4.9829131142383e-06,
      "loss": 0.3834,
      "step": 746
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.538084714736161,
      "learning_rate": 4.98282914241354e-06,
      "loss": 0.3726,
      "step": 747
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.5278386996355975,
      "learning_rate": 4.98274496546961e-06,
      "loss": 0.4013,
      "step": 748
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.866840215876987,
      "learning_rate": 4.982660583413461e-06,
      "loss": 0.3248,
      "step": 749
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.181523079359963,
      "learning_rate": 4.982575996252068e-06,
      "loss": 0.1846,
      "step": 750
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.441389934637184,
      "learning_rate": 4.9824912039924165e-06,
      "loss": 0.57,
      "step": 751
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.638051756113612,
      "learning_rate": 4.982406206641514e-06,
      "loss": 0.2446,
      "step": 752
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.192351681091069,
      "learning_rate": 4.9823210042063805e-06,
      "loss": 0.5429,
      "step": 753
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.941197539279536,
      "learning_rate": 4.982235596694056e-06,
      "loss": 0.4503,
      "step": 754
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.037835917929743,
      "learning_rate": 4.982149984111596e-06,
      "loss": 0.4337,
      "step": 755
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.739492797793943,
      "learning_rate": 4.982064166466074e-06,
      "loss": 0.3614,
      "step": 756
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.669003241384868,
      "learning_rate": 4.9819781437645784e-06,
      "loss": 0.4315,
      "step": 757
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.4128433710422215,
      "learning_rate": 4.981891916014219e-06,
      "loss": 0.4124,
      "step": 758
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.826525288491135,
      "learning_rate": 4.981805483222117e-06,
      "loss": 0.3476,
      "step": 759
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.456446866086912,
      "learning_rate": 4.9817188453954125e-06,
      "loss": 0.2988,
      "step": 760
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.373136892655504,
      "learning_rate": 4.981632002541265e-06,
      "loss": 0.2884,
      "step": 761
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.736140075914428,
      "learning_rate": 4.981544954666847e-06,
      "loss": 0.4673,
      "step": 762
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.853623517985942,
      "learning_rate": 4.981457701779353e-06,
      "loss": 0.3146,
      "step": 763
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.742512362292617,
      "learning_rate": 4.981370243885989e-06,
      "loss": 0.4357,
      "step": 764
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7107066837792355,
      "learning_rate": 4.9812825809939804e-06,
      "loss": 0.2134,
      "step": 765
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4525840887969466,
      "learning_rate": 4.9811947131105705e-06,
      "loss": 0.097,
      "step": 766
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.926454101761235,
      "learning_rate": 4.981106640243017e-06,
      "loss": 0.3173,
      "step": 767
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.24509233688707,
      "learning_rate": 4.981018362398598e-06,
      "loss": 0.5048,
      "step": 768
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.3443462030347275,
      "learning_rate": 4.9809298795846066e-06,
      "loss": 0.4787,
      "step": 769
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.725819777323659,
      "learning_rate": 4.980841191808351e-06,
      "loss": 0.4133,
      "step": 770
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.674768993649168,
      "learning_rate": 4.980752299077159e-06,
      "loss": 0.3861,
      "step": 771
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.056565374690054,
      "learning_rate": 4.980663201398374e-06,
      "loss": 0.4744,
      "step": 772
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.634628042809716,
      "learning_rate": 4.980573898779359e-06,
      "loss": 0.4387,
      "step": 773
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.8697327858519985,
      "learning_rate": 4.980484391227488e-06,
      "loss": 0.3573,
      "step": 774
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.552667965863953,
      "learning_rate": 4.98039467875016e-06,
      "loss": 0.1473,
      "step": 775
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.759590754243705,
      "learning_rate": 4.9803047613547825e-06,
      "loss": 0.6198,
      "step": 776
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.905042009796894,
      "learning_rate": 4.980214639048787e-06,
      "loss": 0.4183,
      "step": 777
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.856836765197414,
      "learning_rate": 4.980124311839617e-06,
      "loss": 0.6381,
      "step": 778
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.583235132807263,
      "learning_rate": 4.980033779734736e-06,
      "loss": 0.2976,
      "step": 779
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.640131465211307,
      "learning_rate": 4.9799430427416235e-06,
      "loss": 0.3522,
      "step": 780
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.732787966827545,
      "learning_rate": 4.979852100867775e-06,
      "loss": 0.302,
      "step": 781
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.3015677489123805,
      "learning_rate": 4.979760954120704e-06,
      "loss": 0.4088,
      "step": 782
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.282317255695599,
      "learning_rate": 4.979669602507941e-06,
      "loss": 0.4759,
      "step": 783
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.443664262359987,
      "learning_rate": 4.979578046037032e-06,
      "loss": 0.3383,
      "step": 784
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.163545541893899,
      "learning_rate": 4.979486284715541e-06,
      "loss": 0.5248,
      "step": 785
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.33297976041109,
      "learning_rate": 4.979394318551051e-06,
      "loss": 0.3068,
      "step": 786
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.4199971435085015,
      "learning_rate": 4.979302147551157e-06,
      "loss": 0.4425,
      "step": 787
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.753760539930669,
      "learning_rate": 4.979209771723476e-06,
      "loss": 0.1573,
      "step": 788
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.8897109229936495,
      "learning_rate": 4.979117191075637e-06,
      "loss": 0.3735,
      "step": 789
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.9909457166078415,
      "learning_rate": 4.979024405615291e-06,
      "loss": 0.5218,
      "step": 790
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.613013865582627,
      "learning_rate": 4.978931415350103e-06,
      "loss": 0.2332,
      "step": 791
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.940643944662163,
      "learning_rate": 4.978838220287755e-06,
      "loss": 0.4721,
      "step": 792
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.281740897932103,
      "learning_rate": 4.978744820435946e-06,
      "loss": 0.6087,
      "step": 793
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.5793970726755115,
      "learning_rate": 4.978651215802393e-06,
      "loss": 0.0221,
      "step": 794
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.451511708992536,
      "learning_rate": 4.978557406394827e-06,
      "loss": 0.2204,
      "step": 795
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.631481714390161,
      "learning_rate": 4.978463392221001e-06,
      "loss": 0.093,
      "step": 796
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.002151219884524,
      "learning_rate": 4.978369173288682e-06,
      "loss": 0.194,
      "step": 797
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.38401075704614,
      "learning_rate": 4.9782747496056515e-06,
      "loss": 0.3348,
      "step": 798
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.650198112046276,
      "learning_rate": 4.978180121179711e-06,
      "loss": 0.2941,
      "step": 799
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.985395681798151,
      "learning_rate": 4.9780852880186794e-06,
      "loss": 0.0806,
      "step": 800
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.510362581473092,
      "learning_rate": 4.977990250130391e-06,
      "loss": 0.1283,
      "step": 801
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.245310164863651,
      "learning_rate": 4.977895007522696e-06,
      "loss": 0.3203,
      "step": 802
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.46200001568424,
      "learning_rate": 4.977799560203464e-06,
      "loss": 0.4211,
      "step": 803
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4901968443929423,
      "learning_rate": 4.977703908180582e-06,
      "loss": 0.034,
      "step": 804
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.038066269486616,
      "learning_rate": 4.97760805146195e-06,
      "loss": 0.6845,
      "step": 805
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.163478703746016,
      "learning_rate": 4.977511990055487e-06,
      "loss": 0.3198,
      "step": 806
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.973380512056832,
      "learning_rate": 4.97741572396913e-06,
      "loss": 0.4471,
      "step": 807
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.982164455804094,
      "learning_rate": 4.9773192532108325e-06,
      "loss": 0.1884,
      "step": 808
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.1893648504636865,
      "learning_rate": 4.977222577788564e-06,
      "loss": 0.367,
      "step": 809
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.434452215612489,
      "learning_rate": 4.977125697710311e-06,
      "loss": 0.4632,
      "step": 810
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.231300392879415,
      "learning_rate": 4.977028612984078e-06,
      "loss": 0.2675,
      "step": 811
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.314127689320212,
      "learning_rate": 4.976931323617885e-06,
      "loss": 0.4741,
      "step": 812
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.31417890092407,
      "learning_rate": 4.976833829619769e-06,
      "loss": 0.1054,
      "step": 813
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.188161177852326,
      "learning_rate": 4.976736130997786e-06,
      "loss": 0.304,
      "step": 814
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.432729648580248,
      "learning_rate": 4.976638227760007e-06,
      "loss": 0.3716,
      "step": 815
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.60133642772163,
      "learning_rate": 4.9765401199145195e-06,
      "loss": 0.2771,
      "step": 816
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.197629557553334,
      "learning_rate": 4.976441807469429e-06,
      "loss": 0.2752,
      "step": 817
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3622734976488013,
      "learning_rate": 4.976343290432858e-06,
      "loss": 0.1849,
      "step": 818
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.42588472075848,
      "learning_rate": 4.976244568812945e-06,
      "loss": 0.5354,
      "step": 819
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8900759757067402,
      "learning_rate": 4.976145642617846e-06,
      "loss": 0.0514,
      "step": 820
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.542827383952137,
      "learning_rate": 4.976046511855735e-06,
      "loss": 0.4864,
      "step": 821
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.325805078767318,
      "learning_rate": 4.975947176534799e-06,
      "loss": 0.1962,
      "step": 822
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.148217366736404,
      "learning_rate": 4.9758476366632466e-06,
      "loss": 0.3497,
      "step": 823
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.147955589788913,
      "learning_rate": 4.975747892249301e-06,
      "loss": 0.2522,
      "step": 824
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.575255936676909,
      "learning_rate": 4.975647943301203e-06,
      "loss": 0.3748,
      "step": 825
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.8741768325755395,
      "learning_rate": 4.975547789827209e-06,
      "loss": 0.3431,
      "step": 826
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.474849985701583,
      "learning_rate": 4.9754474318355925e-06,
      "loss": 0.4052,
      "step": 827
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.702708659882531,
      "learning_rate": 4.975346869334648e-06,
      "loss": 0.3634,
      "step": 828
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.04910081309933,
      "learning_rate": 4.975246102332679e-06,
      "loss": 0.272,
      "step": 829
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.51308449578091,
      "learning_rate": 4.975145130838013e-06,
      "loss": 0.4707,
      "step": 830
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.706928605392194,
      "learning_rate": 4.975043954858992e-06,
      "loss": 0.3678,
      "step": 831
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.250516139204984,
      "learning_rate": 4.974942574403974e-06,
      "loss": 0.2317,
      "step": 832
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.649512219943744,
      "learning_rate": 4.974840989481333e-06,
      "loss": 0.5217,
      "step": 833
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.400145453874358,
      "learning_rate": 4.9747392000994644e-06,
      "loss": 0.3944,
      "step": 834
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.681850208678285,
      "learning_rate": 4.974637206266776e-06,
      "loss": 0.3822,
      "step": 835
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.862696079933064,
      "learning_rate": 4.9745350079916935e-06,
      "loss": 0.3535,
      "step": 836
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.2380755404088415,
      "learning_rate": 4.974432605282661e-06,
      "loss": 0.2738,
      "step": 837
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.247973879345016,
      "learning_rate": 4.974329998148137e-06,
      "loss": 0.2595,
      "step": 838
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.0417571099827905,
      "learning_rate": 4.974227186596601e-06,
      "loss": 0.6248,
      "step": 839
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.681730739698144,
      "learning_rate": 4.974124170636545e-06,
      "loss": 0.4868,
      "step": 840
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.893355620357823,
      "learning_rate": 4.974020950276479e-06,
      "loss": 0.1238,
      "step": 841
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.883546067380728,
      "learning_rate": 4.9739175255249324e-06,
      "loss": 0.3353,
      "step": 842
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.956663105307953,
      "learning_rate": 4.97381389639045e-06,
      "loss": 0.2832,
      "step": 843
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.233128658480574,
      "learning_rate": 4.973710062881591e-06,
      "loss": 0.2975,
      "step": 844
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.235012842768355,
      "learning_rate": 4.973606025006934e-06,
      "loss": 0.308,
      "step": 845
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.718938470412191,
      "learning_rate": 4.973501782775075e-06,
      "loss": 0.429,
      "step": 846
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.016717833966988,
      "learning_rate": 4.9733973361946255e-06,
      "loss": 0.2642,
      "step": 847
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.116402460100625,
      "learning_rate": 4.973292685274215e-06,
      "loss": 0.4338,
      "step": 848
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.063493374073193,
      "learning_rate": 4.973187830022488e-06,
      "loss": 0.6093,
      "step": 849
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.461209518270583,
      "learning_rate": 4.973082770448107e-06,
      "loss": 0.4538,
      "step": 850
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7210028961800243,
      "learning_rate": 4.972977506559754e-06,
      "loss": 0.2091,
      "step": 851
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.647200089954085,
      "learning_rate": 4.972872038366124e-06,
      "loss": 0.3351,
      "step": 852
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.035430342469435,
      "learning_rate": 4.972766365875929e-06,
      "loss": 0.1256,
      "step": 853
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.785481871816953,
      "learning_rate": 4.972660489097901e-06,
      "loss": 0.3348,
      "step": 854
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.766591833123959,
      "learning_rate": 4.972554408040786e-06,
      "loss": 0.5106,
      "step": 855
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.160752137986519,
      "learning_rate": 4.972448122713347e-06,
      "loss": 0.3126,
      "step": 856
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.973349356978723,
      "learning_rate": 4.972341633124367e-06,
      "loss": 0.4707,
      "step": 857
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.352781350692913,
      "learning_rate": 4.9722349392826416e-06,
      "loss": 0.3492,
      "step": 858
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.215644600318088,
      "learning_rate": 4.9721280411969874e-06,
      "loss": 0.2865,
      "step": 859
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.375807633187805,
      "learning_rate": 4.972020938876234e-06,
      "loss": 0.6115,
      "step": 860
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.90473748255133,
      "learning_rate": 4.971913632329231e-06,
      "loss": 0.2756,
      "step": 861
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.11941436476218,
      "learning_rate": 4.9718061215648414e-06,
      "loss": 0.3309,
      "step": 862
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.301821554596965,
      "learning_rate": 4.9716984065919495e-06,
      "loss": 0.1248,
      "step": 863
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.695644930040999,
      "learning_rate": 4.971590487419453e-06,
      "loss": 0.4266,
      "step": 864
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.310332785829851,
      "learning_rate": 4.971482364056269e-06,
      "loss": 0.4113,
      "step": 865
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1601300737432894,
      "learning_rate": 4.971374036511328e-06,
      "loss": 0.0545,
      "step": 866
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.65600319417142,
      "learning_rate": 4.971265504793581e-06,
      "loss": 0.4062,
      "step": 867
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.588311998458038,
      "learning_rate": 4.971156768911994e-06,
      "loss": 0.1692,
      "step": 868
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.4805176777581,
      "learning_rate": 4.971047828875551e-06,
      "loss": 0.1713,
      "step": 869
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.996265775039968,
      "learning_rate": 4.970938684693249e-06,
      "loss": 0.5993,
      "step": 870
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.167723843044812,
      "learning_rate": 4.970829336374109e-06,
      "loss": 0.4294,
      "step": 871
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.75429446503125,
      "learning_rate": 4.970719783927162e-06,
      "loss": 0.2262,
      "step": 872
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.666762093295619,
      "learning_rate": 4.970610027361461e-06,
      "loss": 0.2825,
      "step": 873
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.595532044468351,
      "learning_rate": 4.970500066686071e-06,
      "loss": 0.2937,
      "step": 874
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.376813116444104,
      "learning_rate": 4.97038990191008e-06,
      "loss": 0.4419,
      "step": 875
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.16980355084688,
      "learning_rate": 4.9702795330425844e-06,
      "loss": 0.2849,
      "step": 876
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.209578398036085,
      "learning_rate": 4.970168960092706e-06,
      "loss": 0.4748,
      "step": 877
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.867205300478844,
      "learning_rate": 4.970058183069578e-06,
      "loss": 0.4057,
      "step": 878
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.980691776882161,
      "learning_rate": 4.969947201982354e-06,
      "loss": 0.2722,
      "step": 879
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.355572838356053,
      "learning_rate": 4.969836016840201e-06,
      "loss": 0.5711,
      "step": 880
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.179615206520523,
      "learning_rate": 4.969724627652306e-06,
      "loss": 0.2978,
      "step": 881
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.939478555807708,
      "learning_rate": 4.96961303442787e-06,
      "loss": 0.5085,
      "step": 882
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.82062168413814,
      "learning_rate": 4.969501237176113e-06,
      "loss": 0.3784,
      "step": 883
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.650699257787719,
      "learning_rate": 4.969389235906271e-06,
      "loss": 0.3546,
      "step": 884
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.472176713378931,
      "learning_rate": 4.969277030627597e-06,
      "loss": 0.3468,
      "step": 885
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.8092979132364695,
      "learning_rate": 4.969164621349361e-06,
      "loss": 0.3635,
      "step": 886
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.6038316817599725,
      "learning_rate": 4.96905200808085e-06,
      "loss": 0.3144,
      "step": 887
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.226320899766654,
      "learning_rate": 4.968939190831368e-06,
      "loss": 0.1322,
      "step": 888
    },
    {
      "epoch": 0.08,
      "grad_norm": 8.942030679071168,
      "learning_rate": 4.968826169610233e-06,
      "loss": 0.6064,
      "step": 889
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.708524956979403,
      "learning_rate": 4.968712944426784e-06,
      "loss": 0.266,
      "step": 890
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.755026819689841,
      "learning_rate": 4.968599515290377e-06,
      "loss": 0.5113,
      "step": 891
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.977813861244442,
      "learning_rate": 4.968485882210379e-06,
      "loss": 0.6108,
      "step": 892
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.747506882506638,
      "learning_rate": 4.9683720451961795e-06,
      "loss": 0.3348,
      "step": 893
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0668454230315625,
      "learning_rate": 4.968258004257185e-06,
      "loss": 0.0669,
      "step": 894
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.905601267479151,
      "learning_rate": 4.968143759402814e-06,
      "loss": 0.3821,
      "step": 895
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.489635135965179,
      "learning_rate": 4.968029310642507e-06,
      "loss": 0.1933,
      "step": 896
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.0712463629606725,
      "learning_rate": 4.967914657985718e-06,
      "loss": 0.2388,
      "step": 897
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.4768743153143635,
      "learning_rate": 4.96779980144192e-06,
      "loss": 0.8064,
      "step": 898
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.718005739368291,
      "learning_rate": 4.967684741020601e-06,
      "loss": 0.2454,
      "step": 899
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.05263255160719,
      "learning_rate": 4.967569476731267e-06,
      "loss": 0.3194,
      "step": 900
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.375609227460908,
      "learning_rate": 4.967454008583441e-06,
      "loss": 0.5964,
      "step": 901
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.318106830417689,
      "learning_rate": 4.967338336586662e-06,
      "loss": 0.3864,
      "step": 902
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.808976063498439,
      "learning_rate": 4.967222460750486e-06,
      "loss": 0.1147,
      "step": 903
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.640244436126009,
      "learning_rate": 4.967106381084487e-06,
      "loss": 0.5413,
      "step": 904
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.378798658733776,
      "learning_rate": 4.9669900975982535e-06,
      "loss": 0.203,
      "step": 905
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.70874041605021,
      "learning_rate": 4.966873610301393e-06,
      "loss": 0.3211,
      "step": 906
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.646972122213877,
      "learning_rate": 4.966756919203531e-06,
      "loss": 0.5284,
      "step": 907
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.821962084598155,
      "learning_rate": 4.9666400243143055e-06,
      "loss": 0.4704,
      "step": 908
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.3566197645577525,
      "learning_rate": 4.966522925643373e-06,
      "loss": 0.3906,
      "step": 909
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.34533711277818,
      "learning_rate": 4.966405623200411e-06,
      "loss": 0.2278,
      "step": 910
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.608814806159728,
      "learning_rate": 4.966288116995108e-06,
      "loss": 0.5991,
      "step": 911
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.154168707362406,
      "learning_rate": 4.966170407037172e-06,
      "loss": 0.282,
      "step": 912
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.0134762804376996,
      "learning_rate": 4.966052493336327e-06,
      "loss": 0.3966,
      "step": 913
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.629510691173181,
      "learning_rate": 4.965934375902317e-06,
      "loss": 0.6752,
      "step": 914
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.968646728058386,
      "learning_rate": 4.9658160547448985e-06,
      "loss": 0.4413,
      "step": 915
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.288697006405744,
      "learning_rate": 4.965697529873846e-06,
      "loss": 0.3586,
      "step": 916
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.365938710436579,
      "learning_rate": 4.965578801298953e-06,
      "loss": 0.2501,
      "step": 917
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.2082736906534555,
      "learning_rate": 4.9654598690300275e-06,
      "loss": 0.1954,
      "step": 918
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.847538036865891,
      "learning_rate": 4.9653407330768945e-06,
      "loss": 0.4023,
      "step": 919
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.275548191035905,
      "learning_rate": 4.965221393449398e-06,
      "loss": 0.2376,
      "step": 920
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.369839191401426,
      "learning_rate": 4.9651018501573955e-06,
      "loss": 0.3017,
      "step": 921
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.7981850280695175,
      "learning_rate": 4.964982103210764e-06,
      "loss": 0.1915,
      "step": 922
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4630169166527685,
      "learning_rate": 4.964862152619397e-06,
      "loss": 0.0754,
      "step": 923
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.11027247503684,
      "learning_rate": 4.964741998393202e-06,
      "loss": 0.4044,
      "step": 924
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.429747734515996,
      "learning_rate": 4.9646216405421075e-06,
      "loss": 0.3448,
      "step": 925
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.083632186298042,
      "learning_rate": 4.9645010790760575e-06,
      "loss": 0.391,
      "step": 926
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.407218745728452,
      "learning_rate": 4.964380314005011e-06,
      "loss": 0.5015,
      "step": 927
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.04726931300016,
      "learning_rate": 4.964259345338944e-06,
      "loss": 0.339,
      "step": 928
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.0812727123702714,
      "learning_rate": 4.9641381730878525e-06,
      "loss": 0.2349,
      "step": 929
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.51792818705954,
      "learning_rate": 4.964016797261746e-06,
      "loss": 0.3823,
      "step": 930
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.244265651313615,
      "learning_rate": 4.9638952178706515e-06,
      "loss": 0.2322,
      "step": 931
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.008646542281431,
      "learning_rate": 4.963773434924615e-06,
      "loss": 0.2926,
      "step": 932
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.348703452871472,
      "learning_rate": 4.963651448433696e-06,
      "loss": 0.4899,
      "step": 933
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.932777058570783,
      "learning_rate": 4.963529258407973e-06,
      "loss": 0.316,
      "step": 934
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.250591518713116,
      "learning_rate": 4.9634068648575405e-06,
      "loss": 0.3293,
      "step": 935
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.845638400067904,
      "learning_rate": 4.96328426779251e-06,
      "loss": 0.5829,
      "step": 936
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.109416044971239,
      "learning_rate": 4.9631614672230105e-06,
      "loss": 0.4058,
      "step": 937
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.753815553845626,
      "learning_rate": 4.963038463159186e-06,
      "loss": 0.2617,
      "step": 938
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.717483778986392,
      "learning_rate": 4.962915255611201e-06,
      "loss": 0.597,
      "step": 939
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.276991973932953,
      "learning_rate": 4.962791844589232e-06,
      "loss": 0.1682,
      "step": 940
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.369243207532881,
      "learning_rate": 4.9626682301034745e-06,
      "loss": 0.4698,
      "step": 941
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.366613830823923,
      "learning_rate": 4.962544412164142e-06,
      "loss": 0.4747,
      "step": 942
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.305201906323133,
      "learning_rate": 4.962420390781464e-06,
      "loss": 0.4198,
      "step": 943
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.228164003520596,
      "learning_rate": 4.962296165965686e-06,
      "loss": 0.5069,
      "step": 944
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.78380716493057,
      "learning_rate": 4.962171737727069e-06,
      "loss": 0.2581,
      "step": 945
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.997442618223293,
      "learning_rate": 4.9620471060758955e-06,
      "loss": 0.4245,
      "step": 946
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.184906922744991,
      "learning_rate": 4.9619222710224615e-06,
      "loss": 0.3508,
      "step": 947
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.2543859209840775,
      "learning_rate": 4.961797232577078e-06,
      "loss": 0.2544,
      "step": 948
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.1014514259690795,
      "learning_rate": 4.961671990750078e-06,
      "loss": 0.1997,
      "step": 949
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.7987879332925676,
      "learning_rate": 4.961546545551807e-06,
      "loss": 0.1112,
      "step": 950
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.424728800006951,
      "learning_rate": 4.961420896992629e-06,
      "loss": 0.3149,
      "step": 951
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.126159210830009,
      "learning_rate": 4.961295045082924e-06,
      "loss": 0.3304,
      "step": 952
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.690231154778905,
      "learning_rate": 4.961168989833088e-06,
      "loss": 0.5697,
      "step": 953
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.593808594984215,
      "learning_rate": 4.961042731253538e-06,
      "loss": 0.3341,
      "step": 954
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.116237651931205,
      "learning_rate": 4.960916269354703e-06,
      "loss": 0.2567,
      "step": 955
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.36594205484873,
      "learning_rate": 4.960789604147031e-06,
      "loss": 0.4923,
      "step": 956
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.303168551970379,
      "learning_rate": 4.960662735640986e-06,
      "loss": 0.281,
      "step": 957
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.4633767954863135,
      "learning_rate": 4.960535663847051e-06,
      "loss": 0.317,
      "step": 958
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5228930446495377,
      "learning_rate": 4.960408388775722e-06,
      "loss": 0.1702,
      "step": 959
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.0969424433572925,
      "learning_rate": 4.960280910437515e-06,
      "loss": 0.4022,
      "step": 960
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.444089320370083,
      "learning_rate": 4.960153228842961e-06,
      "loss": 0.1604,
      "step": 961
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.011673970870251,
      "learning_rate": 4.960025344002609e-06,
      "loss": 0.3366,
      "step": 962
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.980797607859079,
      "learning_rate": 4.959897255927024e-06,
      "loss": 0.5307,
      "step": 963
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.687777955699838,
      "learning_rate": 4.9597689646267874e-06,
      "loss": 0.2456,
      "step": 964
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.355385574552104,
      "learning_rate": 4.959640470112499e-06,
      "loss": 0.358,
      "step": 965
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.49378166137148,
      "learning_rate": 4.959511772394774e-06,
      "loss": 0.3361,
      "step": 966
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.804908146496613,
      "learning_rate": 4.959382871484244e-06,
      "loss": 0.4329,
      "step": 967
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.513320143321792,
      "learning_rate": 4.959253767391559e-06,
      "loss": 0.3937,
      "step": 968
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.695276256007862,
      "learning_rate": 4.959124460127385e-06,
      "loss": 0.5979,
      "step": 969
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.393535477928443,
      "learning_rate": 4.958994949702405e-06,
      "loss": 0.3847,
      "step": 970
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.369594589073238,
      "learning_rate": 4.958865236127317e-06,
      "loss": 0.1949,
      "step": 971
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.062769094637351,
      "learning_rate": 4.95873531941284e-06,
      "loss": 0.6363,
      "step": 972
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.6334595097884,
      "learning_rate": 4.958605199569704e-06,
      "loss": 0.4866,
      "step": 973
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.647438489802034,
      "learning_rate": 4.958474876608661e-06,
      "loss": 0.3292,
      "step": 974
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.462982158066824,
      "learning_rate": 4.958344350540477e-06,
      "loss": 0.2518,
      "step": 975
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.411664303256436,
      "learning_rate": 4.958213621375935e-06,
      "loss": 0.5358,
      "step": 976
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.701656774734289,
      "learning_rate": 4.958082689125836e-06,
      "loss": 0.3754,
      "step": 977
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.866034039075973,
      "learning_rate": 4.9579515538009956e-06,
      "loss": 0.2418,
      "step": 978
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.854019305999059,
      "learning_rate": 4.95782021541225e-06,
      "loss": 0.4513,
      "step": 979
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.01208045128382,
      "learning_rate": 4.957688673970447e-06,
      "loss": 0.2996,
      "step": 980
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.42695611237201,
      "learning_rate": 4.957556929486456e-06,
      "loss": 0.2822,
      "step": 981
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.968237154355726,
      "learning_rate": 4.95742498197116e-06,
      "loss": 0.3654,
      "step": 982
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.825909415302943,
      "learning_rate": 4.9572928314354595e-06,
      "loss": 0.1193,
      "step": 983
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.2503053300037195,
      "learning_rate": 4.9571604778902725e-06,
      "loss": 0.3518,
      "step": 984
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.388580451933672,
      "learning_rate": 4.957027921346535e-06,
      "loss": 0.3194,
      "step": 985
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.473108477868533,
      "learning_rate": 4.956895161815196e-06,
      "loss": 0.4816,
      "step": 986
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.023992283888757,
      "learning_rate": 4.956762199307224e-06,
      "loss": 0.4997,
      "step": 987
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.786944776626021,
      "learning_rate": 4.956629033833604e-06,
      "loss": 0.4661,
      "step": 988
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.332088105176019,
      "learning_rate": 4.956495665405338e-06,
      "loss": 0.2403,
      "step": 989
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.404945970772277,
      "learning_rate": 4.956362094033443e-06,
      "loss": 0.4984,
      "step": 990
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.23315594545397,
      "learning_rate": 4.956228319728955e-06,
      "loss": 0.1382,
      "step": 991
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.812029364113967,
      "learning_rate": 4.9560943425029265e-06,
      "loss": 0.4135,
      "step": 992
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.542091588832867,
      "learning_rate": 4.955960162366424e-06,
      "loss": 0.453,
      "step": 993
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.088140072049043,
      "learning_rate": 4.955825779330533e-06,
      "loss": 0.4479,
      "step": 994
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.893349885279304,
      "learning_rate": 4.955691193406358e-06,
      "loss": 0.2506,
      "step": 995
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.425317912679648,
      "learning_rate": 4.955556404605016e-06,
      "loss": 0.3339,
      "step": 996
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.220390606293297,
      "learning_rate": 4.955421412937641e-06,
      "loss": 0.2075,
      "step": 997
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.260256566538155,
      "learning_rate": 4.955286218415388e-06,
      "loss": 0.1959,
      "step": 998
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.933468324077996,
      "learning_rate": 4.955150821049426e-06,
      "loss": 0.3843,
      "step": 999
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.774368130962512,
      "learning_rate": 4.955015220850939e-06,
      "loss": 0.2879,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.962477877460655,
      "learning_rate": 4.954879417831132e-06,
      "loss": 0.4908,
      "step": 1001
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.888378000030824,
      "learning_rate": 4.954743412001223e-06,
      "loss": 0.631,
      "step": 1002
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.776096897579773,
      "learning_rate": 4.954607203372448e-06,
      "loss": 0.3804,
      "step": 1003
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.824980935374355,
      "learning_rate": 4.95447079195606e-06,
      "loss": 0.3104,
      "step": 1004
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.317757711623021,
      "learning_rate": 4.954334177763328e-06,
      "loss": 0.5554,
      "step": 1005
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.239747535081235,
      "learning_rate": 4.954197360805541e-06,
      "loss": 0.1911,
      "step": 1006
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.557465334118495,
      "learning_rate": 4.954060341093999e-06,
      "loss": 0.3556,
      "step": 1007
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.852239816522973,
      "learning_rate": 4.9539231186400235e-06,
      "loss": 0.3267,
      "step": 1008
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.760226275216049,
      "learning_rate": 4.953785693454951e-06,
      "loss": 0.3937,
      "step": 1009
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.94451970887655,
      "learning_rate": 4.953648065550135e-06,
      "loss": 0.4282,
      "step": 1010
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.9497809880737287,
      "learning_rate": 4.953510234936944e-06,
      "loss": 0.2182,
      "step": 1011
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.05421017085235,
      "learning_rate": 4.953372201626768e-06,
      "loss": 0.2342,
      "step": 1012
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.378319606822431,
      "learning_rate": 4.953233965631008e-06,
      "loss": 0.2314,
      "step": 1013
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.730729319253587,
      "learning_rate": 4.953095526961086e-06,
      "loss": 0.5099,
      "step": 1014
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.193338964499809,
      "learning_rate": 4.952956885628438e-06,
      "loss": 0.553,
      "step": 1015
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.700828158591889,
      "learning_rate": 4.952818041644518e-06,
      "loss": 0.5517,
      "step": 1016
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.631891685675143,
      "learning_rate": 4.952678995020797e-06,
      "loss": 0.5076,
      "step": 1017
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.562414789357485,
      "learning_rate": 4.952539745768763e-06,
      "loss": 0.3903,
      "step": 1018
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.221414666914377,
      "learning_rate": 4.9524002938999195e-06,
      "loss": 0.4584,
      "step": 1019
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3538533133344726,
      "learning_rate": 4.9522606394257866e-06,
      "loss": 0.1619,
      "step": 1020
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.855541743273111,
      "learning_rate": 4.952120782357902e-06,
      "loss": 0.4085,
      "step": 1021
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.554662727366842,
      "learning_rate": 4.951980722707821e-06,
      "loss": 0.2978,
      "step": 1022
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3013628444491623,
      "learning_rate": 4.951840460487114e-06,
      "loss": 0.0699,
      "step": 1023
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.603937778223343,
      "learning_rate": 4.9516999957073695e-06,
      "loss": 0.4655,
      "step": 1024
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.764512823590926,
      "learning_rate": 4.951559328380191e-06,
      "loss": 0.362,
      "step": 1025
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.429375156192606,
      "learning_rate": 4.9514184585171996e-06,
      "loss": 0.082,
      "step": 1026
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.801873646597372,
      "learning_rate": 4.951277386130034e-06,
      "loss": 0.5925,
      "step": 1027
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.353317985085587,
      "learning_rate": 4.951136111230349e-06,
      "loss": 0.1598,
      "step": 1028
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.999444910814274,
      "learning_rate": 4.950994633829815e-06,
      "loss": 0.4994,
      "step": 1029
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.635562419211423,
      "learning_rate": 4.950852953940122e-06,
      "loss": 0.3858,
      "step": 1030
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.2085282550320775,
      "learning_rate": 4.950711071572974e-06,
      "loss": 0.2266,
      "step": 1031
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.408153345771418,
      "learning_rate": 4.950568986740092e-06,
      "loss": 0.23,
      "step": 1032
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.520124366089255,
      "learning_rate": 4.950426699453215e-06,
      "loss": 0.2619,
      "step": 1033
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.375085575497177,
      "learning_rate": 4.950284209724097e-06,
      "loss": 0.2405,
      "step": 1034
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2240043111326404,
      "learning_rate": 4.950141517564511e-06,
      "loss": 0.0507,
      "step": 1035
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.291393403646438,
      "learning_rate": 4.949998622986246e-06,
      "loss": 0.2948,
      "step": 1036
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.1048976003542474,
      "learning_rate": 4.949855526001106e-06,
      "loss": 0.4192,
      "step": 1037
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.541025770340522,
      "learning_rate": 4.949712226620913e-06,
      "loss": 0.4244,
      "step": 1038
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.403108458398658,
      "learning_rate": 4.949568724857506e-06,
      "loss": 0.2874,
      "step": 1039
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.313308066644358,
      "learning_rate": 4.949425020722742e-06,
      "loss": 0.2839,
      "step": 1040
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.34612508130463,
      "learning_rate": 4.949281114228491e-06,
      "loss": 0.4016,
      "step": 1041
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.249122179838972,
      "learning_rate": 4.949137005386642e-06,
      "loss": 0.3385,
      "step": 1042
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.903621754208549,
      "learning_rate": 4.948992694209101e-06,
      "loss": 0.4634,
      "step": 1043
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.923646691286421,
      "learning_rate": 4.948848180707791e-06,
      "loss": 0.3343,
      "step": 1044
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.472991829531103,
      "learning_rate": 4.948703464894649e-06,
      "loss": 0.3872,
      "step": 1045
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.646075996913297,
      "learning_rate": 4.948558546781634e-06,
      "loss": 0.4274,
      "step": 1046
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.931771807455757,
      "learning_rate": 4.948413426380716e-06,
      "loss": 0.3252,
      "step": 1047
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.166765424058561,
      "learning_rate": 4.948268103703884e-06,
      "loss": 0.1444,
      "step": 1048
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.572168674274038,
      "learning_rate": 4.948122578763145e-06,
      "loss": 0.1067,
      "step": 1049
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.075452933274743,
      "learning_rate": 4.947976851570521e-06,
      "loss": 0.2637,
      "step": 1050
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.702386018734275,
      "learning_rate": 4.947830922138051e-06,
      "loss": 0.4434,
      "step": 1051
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.684082326210763,
      "learning_rate": 4.947684790477791e-06,
      "loss": 0.3455,
      "step": 1052
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.863776862754598,
      "learning_rate": 4.947538456601815e-06,
      "loss": 0.3929,
      "step": 1053
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.339768997600262,
      "learning_rate": 4.94739192052221e-06,
      "loss": 0.3555,
      "step": 1054
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.642674471628039,
      "learning_rate": 4.947245182251084e-06,
      "loss": 0.4263,
      "step": 1055
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.119368579940979,
      "learning_rate": 4.94709824180056e-06,
      "loss": 0.2806,
      "step": 1056
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.797584865718189,
      "learning_rate": 4.946951099182775e-06,
      "loss": 0.4317,
      "step": 1057
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.490533599379392,
      "learning_rate": 4.946803754409889e-06,
      "loss": 0.2639,
      "step": 1058
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.554191971083522,
      "learning_rate": 4.9466562074940714e-06,
      "loss": 0.5638,
      "step": 1059
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.356852120781216,
      "learning_rate": 4.946508458447515e-06,
      "loss": 0.1215,
      "step": 1060
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.745248275379132,
      "learning_rate": 4.9463605072824226e-06,
      "loss": 0.3259,
      "step": 1061
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.996274403416067,
      "learning_rate": 4.946212354011019e-06,
      "loss": 0.6782,
      "step": 1062
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.1469524162962585,
      "learning_rate": 4.946063998645544e-06,
      "loss": 0.3128,
      "step": 1063
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.575773360891477,
      "learning_rate": 4.945915441198255e-06,
      "loss": 0.6513,
      "step": 1064
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.815346014589078,
      "learning_rate": 4.9457666816814235e-06,
      "loss": 0.5446,
      "step": 1065
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.09887323964124,
      "learning_rate": 4.945617720107338e-06,
      "loss": 0.3246,
      "step": 1066
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.0295481549712715,
      "learning_rate": 4.945468556488309e-06,
      "loss": 0.3304,
      "step": 1067
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.837326474065224,
      "learning_rate": 4.945319190836656e-06,
      "loss": 0.4717,
      "step": 1068
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.802292221558989,
      "learning_rate": 4.945169623164721e-06,
      "loss": 0.4318,
      "step": 1069
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.695058023275336,
      "learning_rate": 4.945019853484859e-06,
      "loss": 0.3588,
      "step": 1070
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.312370628893169,
      "learning_rate": 4.944869881809444e-06,
      "loss": 0.2903,
      "step": 1071
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.001582997601393,
      "learning_rate": 4.944719708150866e-06,
      "loss": 0.1863,
      "step": 1072
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.555585674008557,
      "learning_rate": 4.944569332521532e-06,
      "loss": 0.1646,
      "step": 1073
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.931788971623673,
      "learning_rate": 4.944418754933864e-06,
      "loss": 0.2255,
      "step": 1074
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.738295566286143,
      "learning_rate": 4.944267975400303e-06,
      "loss": 0.6073,
      "step": 1075
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.663479765555172,
      "learning_rate": 4.944116993933307e-06,
      "loss": 0.4782,
      "step": 1076
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.461164448968103,
      "learning_rate": 4.943965810545346e-06,
      "loss": 0.3891,
      "step": 1077
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.454412990231613,
      "learning_rate": 4.943814425248913e-06,
      "loss": 0.1351,
      "step": 1078
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.9213956452562,
      "learning_rate": 4.943662838056513e-06,
      "loss": 0.2709,
      "step": 1079
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.550563466692943,
      "learning_rate": 4.94351104898067e-06,
      "loss": 0.351,
      "step": 1080
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.885145934026491,
      "learning_rate": 4.943359058033924e-06,
      "loss": 0.4477,
      "step": 1081
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.569684810349401,
      "learning_rate": 4.943206865228831e-06,
      "loss": 0.2638,
      "step": 1082
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.600744422759881,
      "learning_rate": 4.9430544705779665e-06,
      "loss": 0.5306,
      "step": 1083
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.3877139667118565,
      "learning_rate": 4.9429018740939185e-06,
      "loss": 0.1506,
      "step": 1084
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.498117563103084,
      "learning_rate": 4.9427490757892945e-06,
      "loss": 0.3499,
      "step": 1085
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.8301473167111935,
      "learning_rate": 4.9425960756767185e-06,
      "loss": 0.4862,
      "step": 1086
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.079104616897807,
      "learning_rate": 4.94244287376883e-06,
      "loss": 0.3579,
      "step": 1087
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.063073670601033,
      "learning_rate": 4.942289470078286e-06,
      "loss": 0.2949,
      "step": 1088
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.7052889349766795,
      "learning_rate": 4.94213586461776e-06,
      "loss": 0.1991,
      "step": 1089
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.861894605197201,
      "learning_rate": 4.941982057399943e-06,
      "loss": 0.2048,
      "step": 1090
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.124574226874624,
      "learning_rate": 4.941828048437538e-06,
      "loss": 0.2211,
      "step": 1091
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.9308993034032134,
      "learning_rate": 4.9416738377432734e-06,
      "loss": 0.3395,
      "step": 1092
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.876490307839925,
      "learning_rate": 4.941519425329887e-06,
      "loss": 0.1567,
      "step": 1093
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.556950470600164,
      "learning_rate": 4.9413648112101355e-06,
      "loss": 0.3567,
      "step": 1094
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.251495180581067,
      "learning_rate": 4.941209995396793e-06,
      "loss": 0.4352,
      "step": 1095
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.998652832661758,
      "learning_rate": 4.941054977902649e-06,
      "loss": 0.3994,
      "step": 1096
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.151708238392076,
      "learning_rate": 4.94089975874051e-06,
      "loss": 0.4155,
      "step": 1097
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.294409818157684,
      "learning_rate": 4.940744337923201e-06,
      "loss": 0.3779,
      "step": 1098
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.343018532026159,
      "learning_rate": 4.94058871546356e-06,
      "loss": 0.5161,
      "step": 1099
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.918973721577168,
      "learning_rate": 4.940432891374446e-06,
      "loss": 0.1851,
      "step": 1100
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.593305405781647,
      "learning_rate": 4.94027686566873e-06,
      "loss": 0.4976,
      "step": 1101
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.44729623910843,
      "learning_rate": 4.940120638359305e-06,
      "loss": 0.5053,
      "step": 1102
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.9203112094419734,
      "learning_rate": 4.939964209459075e-06,
      "loss": 0.2131,
      "step": 1103
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0167896935903247,
      "learning_rate": 4.939807578980965e-06,
      "loss": 0.0073,
      "step": 1104
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.264188962600838,
      "learning_rate": 4.9396507469379155e-06,
      "loss": 0.5172,
      "step": 1105
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0322942503119044,
      "learning_rate": 4.939493713342881e-06,
      "loss": 0.1528,
      "step": 1106
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.954820163064762,
      "learning_rate": 4.939336478208836e-06,
      "loss": 0.3572,
      "step": 1107
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.070364294261725,
      "learning_rate": 4.93917904154877e-06,
      "loss": 0.5443,
      "step": 1108
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.545031478901175,
      "learning_rate": 4.9390214033756924e-06,
      "loss": 0.4306,
      "step": 1109
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.640944831019814,
      "learning_rate": 4.938863563702623e-06,
      "loss": 0.3482,
      "step": 1110
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.859100219587046,
      "learning_rate": 4.938705522542603e-06,
      "loss": 0.2738,
      "step": 1111
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.660895344175958,
      "learning_rate": 4.938547279908689e-06,
      "loss": 0.1989,
      "step": 1112
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.562445583652174,
      "learning_rate": 4.938388835813954e-06,
      "loss": 0.3621,
      "step": 1113
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.919726611413628,
      "learning_rate": 4.938230190271489e-06,
      "loss": 0.6655,
      "step": 1114
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.108661454740646,
      "learning_rate": 4.938071343294399e-06,
      "loss": 0.3115,
      "step": 1115
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.440497348717015,
      "learning_rate": 4.937912294895808e-06,
      "loss": 0.2502,
      "step": 1116
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.232525904572355,
      "learning_rate": 4.9377530450888565e-06,
      "loss": 0.4099,
      "step": 1117
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8537981128897916,
      "learning_rate": 4.937593593886699e-06,
      "loss": 0.1555,
      "step": 1118
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.265594746010287,
      "learning_rate": 4.937433941302511e-06,
      "loss": 0.4244,
      "step": 1119
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.1997464368330855,
      "learning_rate": 4.93727408734948e-06,
      "loss": 0.1865,
      "step": 1120
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.564923404372264,
      "learning_rate": 4.937114032040812e-06,
      "loss": 0.2865,
      "step": 1121
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.042032424579183,
      "learning_rate": 4.936953775389733e-06,
      "loss": 0.1584,
      "step": 1122
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.2521898886705465,
      "learning_rate": 4.9367933174094796e-06,
      "loss": 0.2518,
      "step": 1123
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.090163529287429,
      "learning_rate": 4.93663265811331e-06,
      "loss": 0.2451,
      "step": 1124
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.622045467833776,
      "learning_rate": 4.936471797514496e-06,
      "loss": 0.3813,
      "step": 1125
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.583836484701118,
      "learning_rate": 4.936310735626328e-06,
      "loss": 0.2714,
      "step": 1126
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.535319814805847,
      "learning_rate": 4.936149472462111e-06,
      "loss": 0.0991,
      "step": 1127
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.8905806457603354,
      "learning_rate": 4.935988008035168e-06,
      "loss": 0.1727,
      "step": 1128
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0198273408368315,
      "learning_rate": 4.935826342358841e-06,
      "loss": 0.1727,
      "step": 1129
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.517655708692484,
      "learning_rate": 4.935664475446481e-06,
      "loss": 0.1969,
      "step": 1130
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.561707147986619,
      "learning_rate": 4.9355024073114646e-06,
      "loss": 0.2846,
      "step": 1131
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.332684176252844,
      "learning_rate": 4.9353401379671806e-06,
      "loss": 0.3028,
      "step": 1132
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.107063246649002,
      "learning_rate": 4.9351776674270335e-06,
      "loss": 0.2957,
      "step": 1133
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.376916009104187,
      "learning_rate": 4.935014995704447e-06,
      "loss": 0.439,
      "step": 1134
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.7419451665062,
      "learning_rate": 4.934852122812859e-06,
      "loss": 0.2831,
      "step": 1135
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.148631719389048,
      "learning_rate": 4.934689048765726e-06,
      "loss": 0.294,
      "step": 1136
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.702536738173812,
      "learning_rate": 4.934525773576522e-06,
      "loss": 0.2018,
      "step": 1137
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.810323962627437,
      "learning_rate": 4.934362297258733e-06,
      "loss": 0.3373,
      "step": 1138
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.020644047394015,
      "learning_rate": 4.934198619825866e-06,
      "loss": 0.3594,
      "step": 1139
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.822320938741841,
      "learning_rate": 4.934034741291444e-06,
      "loss": 0.2876,
      "step": 1140
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.476743705884913,
      "learning_rate": 4.933870661669005e-06,
      "loss": 0.6271,
      "step": 1141
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.020238535247738,
      "learning_rate": 4.933706380972104e-06,
      "loss": 0.1922,
      "step": 1142
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.7986245744995895,
      "learning_rate": 4.933541899214314e-06,
      "loss": 0.4288,
      "step": 1143
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.724144036188058,
      "learning_rate": 4.933377216409223e-06,
      "loss": 0.1523,
      "step": 1144
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.402566668391758,
      "learning_rate": 4.933212332570438e-06,
      "loss": 0.3009,
      "step": 1145
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.012801164951059,
      "learning_rate": 4.933047247711578e-06,
      "loss": 0.3177,
      "step": 1146
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.155915667006767,
      "learning_rate": 4.932881961846283e-06,
      "loss": 0.6275,
      "step": 1147
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.297346334803083,
      "learning_rate": 4.932716474988208e-06,
      "loss": 0.5171,
      "step": 1148
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0282529174942456,
      "learning_rate": 4.9325507871510254e-06,
      "loss": 0.1551,
      "step": 1149
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.358126245227289,
      "learning_rate": 4.932384898348422e-06,
      "loss": 0.4248,
      "step": 1150
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.191454790058702,
      "learning_rate": 4.932218808594105e-06,
      "loss": 0.3734,
      "step": 1151
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.951597465535238,
      "learning_rate": 4.932052517901794e-06,
      "loss": 0.2402,
      "step": 1152
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.687616795948781,
      "learning_rate": 4.931886026285227e-06,
      "loss": 0.2149,
      "step": 1153
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.163475267816355,
      "learning_rate": 4.93171933375816e-06,
      "loss": 0.2484,
      "step": 1154
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.923458011052064,
      "learning_rate": 4.931552440334364e-06,
      "loss": 0.4,
      "step": 1155
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.860584178516449,
      "learning_rate": 4.931385346027627e-06,
      "loss": 0.3421,
      "step": 1156
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.09508256888116,
      "learning_rate": 4.931218050851752e-06,
      "loss": 0.4786,
      "step": 1157
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.397161523025571,
      "learning_rate": 4.931050554820562e-06,
      "loss": 0.4601,
      "step": 1158
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.015455562547345,
      "learning_rate": 4.930882857947894e-06,
      "loss": 0.6601,
      "step": 1159
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.3966467701186422,
      "learning_rate": 4.930714960247601e-06,
      "loss": 0.1609,
      "step": 1160
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.153367235342547,
      "learning_rate": 4.930546861733557e-06,
      "loss": 0.3468,
      "step": 1161
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.4312505229856,
      "learning_rate": 4.930378562419647e-06,
      "loss": 0.303,
      "step": 1162
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.900234261935695,
      "learning_rate": 4.9302100623197745e-06,
      "loss": 0.5012,
      "step": 1163
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8157358559765466,
      "learning_rate": 4.9300413614478626e-06,
      "loss": 0.1531,
      "step": 1164
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.382457765506917,
      "learning_rate": 4.929872459817847e-06,
      "loss": 0.4889,
      "step": 1165
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.6967900389733455,
      "learning_rate": 4.9297033574436816e-06,
      "loss": 0.4786,
      "step": 1166
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.4472702094986967,
      "learning_rate": 4.9295340543393376e-06,
      "loss": 0.2342,
      "step": 1167
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.597962311618219,
      "learning_rate": 4.9293645505188e-06,
      "loss": 0.3729,
      "step": 1168
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.862389630827546,
      "learning_rate": 4.929194845996075e-06,
      "loss": 0.2097,
      "step": 1169
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.266286473587009,
      "learning_rate": 4.929024940785182e-06,
      "loss": 0.3223,
      "step": 1170
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.285539097311399,
      "learning_rate": 4.928854834900156e-06,
      "loss": 0.3656,
      "step": 1171
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.224561786067671,
      "learning_rate": 4.928684528355052e-06,
      "loss": 0.1801,
      "step": 1172
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.206778291220932,
      "learning_rate": 4.928514021163939e-06,
      "loss": 0.3422,
      "step": 1173
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.677095931863252,
      "learning_rate": 4.928343313340905e-06,
      "loss": 0.4717,
      "step": 1174
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.589586972885153,
      "learning_rate": 4.928172404900051e-06,
      "loss": 0.3662,
      "step": 1175
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.614421230130441,
      "learning_rate": 4.9280012958554976e-06,
      "loss": 0.4387,
      "step": 1176
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.014875754070007,
      "learning_rate": 4.927829986221382e-06,
      "loss": 0.2817,
      "step": 1177
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.775616947028615,
      "learning_rate": 4.927658476011855e-06,
      "loss": 0.2956,
      "step": 1178
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.074337922985563,
      "learning_rate": 4.927486765241088e-06,
      "loss": 0.2298,
      "step": 1179
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.388373546356918,
      "learning_rate": 4.927314853923265e-06,
      "loss": 0.6809,
      "step": 1180
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2937144563642025,
      "learning_rate": 4.927142742072589e-06,
      "loss": 0.0124,
      "step": 1181
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.834545870636212,
      "learning_rate": 4.926970429703279e-06,
      "loss": 0.538,
      "step": 1182
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.201683553042719,
      "learning_rate": 4.926797916829572e-06,
      "loss": 0.5182,
      "step": 1183
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6820633525471673,
      "learning_rate": 4.926625203465718e-06,
      "loss": 0.1076,
      "step": 1184
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.363904010432387,
      "learning_rate": 4.926452289625987e-06,
      "loss": 0.6304,
      "step": 1185
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.57901582818345,
      "learning_rate": 4.926279175324665e-06,
      "loss": 0.2621,
      "step": 1186
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.013231564363306,
      "learning_rate": 4.9261058605760525e-06,
      "loss": 0.1256,
      "step": 1187
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.550186429404481,
      "learning_rate": 4.925932345394468e-06,
      "loss": 0.3503,
      "step": 1188
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.715950903079211,
      "learning_rate": 4.925758629794248e-06,
      "loss": 0.479,
      "step": 1189
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.113168071009456,
      "learning_rate": 4.92558471378974e-06,
      "loss": 0.4986,
      "step": 1190
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.6402751666420805,
      "learning_rate": 4.925410597395318e-06,
      "loss": 0.315,
      "step": 1191
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.097378454518381,
      "learning_rate": 4.925236280625362e-06,
      "loss": 0.303,
      "step": 1192
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9938086870848233,
      "learning_rate": 4.925061763494275e-06,
      "loss": 0.2497,
      "step": 1193
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.420657224084773,
      "learning_rate": 4.924887046016475e-06,
      "loss": 0.2246,
      "step": 1194
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.687018346285836,
      "learning_rate": 4.924712128206396e-06,
      "loss": 0.3021,
      "step": 1195
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.342161827550114,
      "learning_rate": 4.924537010078488e-06,
      "loss": 0.2943,
      "step": 1196
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.520694500377948,
      "learning_rate": 4.9243616916472194e-06,
      "loss": 0.1539,
      "step": 1197
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.870714072307103,
      "learning_rate": 4.924186172927074e-06,
      "loss": 0.2455,
      "step": 1198
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.457855836130011,
      "learning_rate": 4.924010453932551e-06,
      "loss": 0.1481,
      "step": 1199
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.504848360635611,
      "learning_rate": 4.9238345346781705e-06,
      "loss": 0.5816,
      "step": 1200
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.7356014424209203,
      "learning_rate": 4.923658415178463e-06,
      "loss": 0.1899,
      "step": 1201
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.8197459949594235,
      "learning_rate": 4.923482095447979e-06,
      "loss": 0.4784,
      "step": 1202
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.413480463546607,
      "learning_rate": 4.923305575501288e-06,
      "loss": 0.4124,
      "step": 1203
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.589878472281962,
      "learning_rate": 4.92312885535297e-06,
      "loss": 0.4862,
      "step": 1204
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.523289052191797,
      "learning_rate": 4.922951935017626e-06,
      "loss": 0.3186,
      "step": 1205
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.3929623684623005,
      "learning_rate": 4.922774814509872e-06,
      "loss": 0.2572,
      "step": 1206
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.373138672760857,
      "learning_rate": 4.922597493844341e-06,
      "loss": 0.3912,
      "step": 1207
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.5632237221889502,
      "learning_rate": 4.922419973035683e-06,
      "loss": 0.1992,
      "step": 1208
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.349322033068722,
      "learning_rate": 4.922242252098564e-06,
      "loss": 0.5838,
      "step": 1209
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.870024465297835,
      "learning_rate": 4.922064331047664e-06,
      "loss": 0.2907,
      "step": 1210
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.941652871621341,
      "learning_rate": 4.921886209897684e-06,
      "loss": 0.1291,
      "step": 1211
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9982252609295763,
      "learning_rate": 4.92170788866334e-06,
      "loss": 0.2519,
      "step": 1212
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.978998073459012,
      "learning_rate": 4.921529367359364e-06,
      "loss": 0.5585,
      "step": 1213
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.388018088585698,
      "learning_rate": 4.921350646000501e-06,
      "loss": 0.1706,
      "step": 1214
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.089563817222594,
      "learning_rate": 4.921171724601521e-06,
      "loss": 0.5498,
      "step": 1215
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.624666165931425,
      "learning_rate": 4.920992603177204e-06,
      "loss": 0.3629,
      "step": 1216
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.576398015626892,
      "learning_rate": 4.920813281742346e-06,
      "loss": 0.3123,
      "step": 1217
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.716716181954611,
      "learning_rate": 4.920633760311764e-06,
      "loss": 0.1641,
      "step": 1218
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.93184019315157,
      "learning_rate": 4.9204540389002885e-06,
      "loss": 0.5501,
      "step": 1219
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.573270326248066,
      "learning_rate": 4.920274117522768e-06,
      "loss": 0.4902,
      "step": 1220
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.774458691353789,
      "learning_rate": 4.920093996194064e-06,
      "loss": 0.1046,
      "step": 1221
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.441125116041495,
      "learning_rate": 4.91991367492906e-06,
      "loss": 0.311,
      "step": 1222
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.323559509172468,
      "learning_rate": 4.919733153742653e-06,
      "loss": 0.3986,
      "step": 1223
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.425550486424748,
      "learning_rate": 4.919552432649755e-06,
      "loss": 0.3412,
      "step": 1224
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.280286334978208,
      "learning_rate": 4.919371511665298e-06,
      "loss": 0.2507,
      "step": 1225
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.074366131144821,
      "learning_rate": 4.9191903908042285e-06,
      "loss": 0.2988,
      "step": 1226
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.6784207224787187,
      "learning_rate": 4.91900907008151e-06,
      "loss": 0.1811,
      "step": 1227
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.906980933514785,
      "learning_rate": 4.9188275495121205e-06,
      "loss": 0.4057,
      "step": 1228
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.5440687444735754,
      "learning_rate": 4.918645829111059e-06,
      "loss": 0.3596,
      "step": 1229
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2249940181880956,
      "learning_rate": 4.918463908893336e-06,
      "loss": 0.0444,
      "step": 1230
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.752910911520032,
      "learning_rate": 4.9182817888739826e-06,
      "loss": 0.277,
      "step": 1231
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.3509449779231755,
      "learning_rate": 4.918099469068044e-06,
      "loss": 0.3008,
      "step": 1232
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.748323449063028,
      "learning_rate": 4.917916949490582e-06,
      "loss": 0.3394,
      "step": 1233
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.650335169944918,
      "learning_rate": 4.917734230156677e-06,
      "loss": 0.2865,
      "step": 1234
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.399789949970793,
      "learning_rate": 4.917551311081423e-06,
      "loss": 0.3022,
      "step": 1235
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.0224695068874246,
      "learning_rate": 4.917368192279932e-06,
      "loss": 0.6919,
      "step": 1236
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.7086263085817555,
      "learning_rate": 4.917184873767333e-06,
      "loss": 0.2725,
      "step": 1237
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.3505517577694155,
      "learning_rate": 4.91700135555877e-06,
      "loss": 0.3571,
      "step": 1238
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.183389091052164,
      "learning_rate": 4.9168176376694055e-06,
      "loss": 0.3048,
      "step": 1239
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.929044019342033,
      "learning_rate": 4.916633720114416e-06,
      "loss": 0.4862,
      "step": 1240
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.898042011307218,
      "learning_rate": 4.916449602908998e-06,
      "loss": 0.1892,
      "step": 1241
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.021477439683825,
      "learning_rate": 4.91626528606836e-06,
      "loss": 0.5823,
      "step": 1242
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.5532309693835105,
      "learning_rate": 4.91608076960773e-06,
      "loss": 0.3682,
      "step": 1243
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.595963565579679,
      "learning_rate": 4.915896053542354e-06,
      "loss": 0.563,
      "step": 1244
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.665875975373176,
      "learning_rate": 4.915711137887489e-06,
      "loss": 0.4222,
      "step": 1245
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.0136458606074354,
      "learning_rate": 4.915526022658414e-06,
      "loss": 0.3022,
      "step": 1246
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.828488766094158,
      "learning_rate": 4.915340707870422e-06,
      "loss": 0.3286,
      "step": 1247
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.664302253587436,
      "learning_rate": 4.915155193538822e-06,
      "loss": 0.4579,
      "step": 1248
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.395481688117633,
      "learning_rate": 4.91496947967894e-06,
      "loss": 0.5459,
      "step": 1249
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.221204978461737,
      "learning_rate": 4.914783566306121e-06,
      "loss": 0.276,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.255291498232373,
      "learning_rate": 4.9145974534357225e-06,
      "loss": 0.5435,
      "step": 1251
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.9828366116584113,
      "learning_rate": 4.91441114108312e-06,
      "loss": 0.2375,
      "step": 1252
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.856918132473316,
      "learning_rate": 4.9142246292637065e-06,
      "loss": 0.6455,
      "step": 1253
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.770762859821151,
      "learning_rate": 4.9140379179928896e-06,
      "loss": 0.3148,
      "step": 1254
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.637203222991626,
      "learning_rate": 4.913851007286097e-06,
      "loss": 0.5709,
      "step": 1255
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.172390833390642,
      "learning_rate": 4.913663897158768e-06,
      "loss": 0.4654,
      "step": 1256
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.2107572046260495,
      "learning_rate": 4.91347658762636e-06,
      "loss": 0.4671,
      "step": 1257
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.254099038567297,
      "learning_rate": 4.913289078704351e-06,
      "loss": 0.1853,
      "step": 1258
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.139336261302221,
      "learning_rate": 4.9131013704082285e-06,
      "loss": 0.1266,
      "step": 1259
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.9132091511206655,
      "learning_rate": 4.912913462753503e-06,
      "loss": 0.3786,
      "step": 1260
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.0060214470747555,
      "learning_rate": 4.912725355755695e-06,
      "loss": 0.2123,
      "step": 1261
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.3993473899746665,
      "learning_rate": 4.9125370494303495e-06,
      "loss": 0.4025,
      "step": 1262
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.695895092954166,
      "learning_rate": 4.912348543793019e-06,
      "loss": 0.2399,
      "step": 1263
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9535662890166288,
      "learning_rate": 4.91215983885928e-06,
      "loss": 0.1058,
      "step": 1264
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.327409401141178,
      "learning_rate": 4.911970934644721e-06,
      "loss": 0.352,
      "step": 1265
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.824105760953562,
      "learning_rate": 4.911781831164949e-06,
      "loss": 0.4498,
      "step": 1266
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.5775063275505365,
      "learning_rate": 4.9115925284355855e-06,
      "loss": 0.443,
      "step": 1267
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.307331911396607,
      "learning_rate": 4.9114030264722715e-06,
      "loss": 0.2847,
      "step": 1268
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.0554463634713,
      "learning_rate": 4.911213325290662e-06,
      "loss": 0.4665,
      "step": 1269
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.225393070723593,
      "learning_rate": 4.911023424906429e-06,
      "loss": 0.1331,
      "step": 1270
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.925992615804011,
      "learning_rate": 4.910833325335262e-06,
      "loss": 0.2891,
      "step": 1271
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.187692282436508,
      "learning_rate": 4.910643026592864e-06,
      "loss": 0.4987,
      "step": 1272
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.4261052884834555,
      "learning_rate": 4.910452528694958e-06,
      "loss": 0.685,
      "step": 1273
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.801095068847434,
      "learning_rate": 4.910261831657282e-06,
      "loss": 0.3858,
      "step": 1274
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.354569510440954,
      "learning_rate": 4.9100709354955904e-06,
      "loss": 0.5317,
      "step": 1275
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.645667041246483,
      "learning_rate": 4.909879840225655e-06,
      "loss": 0.3637,
      "step": 1276
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.538662633029015,
      "learning_rate": 4.909688545863261e-06,
      "loss": 0.2994,
      "step": 1277
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.090896424303886,
      "learning_rate": 4.909497052424214e-06,
      "loss": 0.3632,
      "step": 1278
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.755482282144173,
      "learning_rate": 4.909305359924333e-06,
      "loss": 0.4792,
      "step": 1279
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.404149044018982,
      "learning_rate": 4.909113468379456e-06,
      "loss": 0.3117,
      "step": 1280
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.213923154465818,
      "learning_rate": 4.908921377805436e-06,
      "loss": 0.3858,
      "step": 1281
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.431344936230701,
      "learning_rate": 4.908729088218141e-06,
      "loss": 0.367,
      "step": 1282
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.012960616876207,
      "learning_rate": 4.908536599633459e-06,
      "loss": 0.3006,
      "step": 1283
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.407319253079615,
      "learning_rate": 4.90834391206729e-06,
      "loss": 0.344,
      "step": 1284
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.864821077403087,
      "learning_rate": 4.908151025535556e-06,
      "loss": 0.3135,
      "step": 1285
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.935704859259897,
      "learning_rate": 4.9079579400541895e-06,
      "loss": 0.4196,
      "step": 1286
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.729854980078456,
      "learning_rate": 4.907764655639146e-06,
      "loss": 0.3443,
      "step": 1287
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.785011594970722,
      "learning_rate": 4.907571172306389e-06,
      "loss": 0.3109,
      "step": 1288
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.81749983663714,
      "learning_rate": 4.907377490071906e-06,
      "loss": 0.437,
      "step": 1289
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8744838216072917,
      "learning_rate": 4.9071836089516975e-06,
      "loss": 0.2972,
      "step": 1290
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.410490007932799,
      "learning_rate": 4.906989528961781e-06,
      "loss": 0.4676,
      "step": 1291
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.491804591119035,
      "learning_rate": 4.906795250118191e-06,
      "loss": 0.2791,
      "step": 1292
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.074335219933526,
      "learning_rate": 4.906600772436977e-06,
      "loss": 0.3278,
      "step": 1293
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.457381616499063,
      "learning_rate": 4.906406095934206e-06,
      "loss": 0.2822,
      "step": 1294
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2876083745545683,
      "learning_rate": 4.906211220625961e-06,
      "loss": 0.0836,
      "step": 1295
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.552783758194143,
      "learning_rate": 4.906016146528343e-06,
      "loss": 0.4003,
      "step": 1296
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.633878281575212,
      "learning_rate": 4.905820873657466e-06,
      "loss": 0.3556,
      "step": 1297
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.7704687404324755,
      "learning_rate": 4.905625402029464e-06,
      "loss": 0.4685,
      "step": 1298
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.524320663463019,
      "learning_rate": 4.905429731660485e-06,
      "loss": 0.3657,
      "step": 1299
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.839656924484143,
      "learning_rate": 4.905233862566695e-06,
      "loss": 0.4151,
      "step": 1300
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5853674657169265,
      "learning_rate": 4.9050377947642754e-06,
      "loss": 0.1732,
      "step": 1301
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.070664811788367,
      "learning_rate": 4.904841528269425e-06,
      "loss": 0.3513,
      "step": 1302
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.638562830719029,
      "learning_rate": 4.904645063098358e-06,
      "loss": 0.3828,
      "step": 1303
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.630335646726862,
      "learning_rate": 4.904448399267304e-06,
      "loss": 0.3573,
      "step": 1304
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.907211861579737,
      "learning_rate": 4.904251536792512e-06,
      "loss": 0.557,
      "step": 1305
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.696095113469409,
      "learning_rate": 4.904054475690246e-06,
      "loss": 0.3551,
      "step": 1306
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.460091091559515,
      "learning_rate": 4.903857215976785e-06,
      "loss": 0.4241,
      "step": 1307
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.601824358058987,
      "learning_rate": 4.903659757668427e-06,
      "loss": 0.1377,
      "step": 1308
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.365990682208792,
      "learning_rate": 4.903462100781484e-06,
      "loss": 0.4529,
      "step": 1309
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.737333190778244,
      "learning_rate": 4.903264245332286e-06,
      "loss": 0.4234,
      "step": 1310
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.6067458111889583,
      "learning_rate": 4.903066191337178e-06,
      "loss": 0.1278,
      "step": 1311
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.806534834804124,
      "learning_rate": 4.902867938812524e-06,
      "loss": 0.4847,
      "step": 1312
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.313870084899383,
      "learning_rate": 4.9026694877747e-06,
      "loss": 0.3233,
      "step": 1313
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1265908717569917,
      "learning_rate": 4.902470838240104e-06,
      "loss": 0.1772,
      "step": 1314
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.4762728131413745,
      "learning_rate": 4.9022719902251455e-06,
      "loss": 0.2085,
      "step": 1315
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.401184635602929,
      "learning_rate": 4.902072943746252e-06,
      "loss": 0.328,
      "step": 1316
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.232866897092317,
      "learning_rate": 4.90187369881987e-06,
      "loss": 0.3181,
      "step": 1317
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.803586035293339,
      "learning_rate": 4.901674255462458e-06,
      "loss": 0.2699,
      "step": 1318
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.1929185569643455,
      "learning_rate": 4.901474613690494e-06,
      "loss": 0.5536,
      "step": 1319
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.166512836243729,
      "learning_rate": 4.901274773520471e-06,
      "loss": 0.3907,
      "step": 1320
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.172276106195313,
      "learning_rate": 4.901074734968899e-06,
      "loss": 0.6177,
      "step": 1321
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.409758000267503,
      "learning_rate": 4.900874498052305e-06,
      "loss": 0.4109,
      "step": 1322
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.46663073570602,
      "learning_rate": 4.90067406278723e-06,
      "loss": 0.3182,
      "step": 1323
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.78457949123076,
      "learning_rate": 4.900473429190234e-06,
      "loss": 0.4307,
      "step": 1324
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.755120534179274,
      "learning_rate": 4.900272597277893e-06,
      "loss": 0.3196,
      "step": 1325
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.998752972950404,
      "learning_rate": 4.900071567066797e-06,
      "loss": 0.4384,
      "step": 1326
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.04954108474418,
      "learning_rate": 4.899870338573555e-06,
      "loss": 0.4862,
      "step": 1327
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.07054854472423,
      "learning_rate": 4.899668911814792e-06,
      "loss": 0.4832,
      "step": 1328
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.211180364452602,
      "learning_rate": 4.899467286807148e-06,
      "loss": 0.3897,
      "step": 1329
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.007304529895474,
      "learning_rate": 4.899265463567281e-06,
      "loss": 0.2213,
      "step": 1330
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.9695579966045798,
      "learning_rate": 4.899063442111864e-06,
      "loss": 0.2168,
      "step": 1331
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.1046028967357975,
      "learning_rate": 4.8988612224575875e-06,
      "loss": 0.3141,
      "step": 1332
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.090204505036658,
      "learning_rate": 4.898658804621158e-06,
      "loss": 0.213,
      "step": 1333
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.251028629596293,
      "learning_rate": 4.8984561886192985e-06,
      "loss": 0.2544,
      "step": 1334
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.6995327812684478,
      "learning_rate": 4.898253374468747e-06,
      "loss": 0.2598,
      "step": 1335
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.4225400031697815,
      "learning_rate": 4.89805036218626e-06,
      "loss": 0.4384,
      "step": 1336
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.8265165660734124,
      "learning_rate": 4.897847151788608e-06,
      "loss": 0.3773,
      "step": 1337
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.6142224198235215,
      "learning_rate": 4.897643743292582e-06,
      "loss": 0.2317,
      "step": 1338
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.637931804197387,
      "learning_rate": 4.8974401367149845e-06,
      "loss": 0.5173,
      "step": 1339
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.35847839495454,
      "learning_rate": 4.897236332072636e-06,
      "loss": 0.5276,
      "step": 1340
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.0991895983596836,
      "learning_rate": 4.897032329382376e-06,
      "loss": 0.264,
      "step": 1341
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.149002984269706,
      "learning_rate": 4.8968281286610565e-06,
      "loss": 0.2583,
      "step": 1342
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.60028227593281,
      "learning_rate": 4.896623729925548e-06,
      "loss": 0.181,
      "step": 1343
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.068401600545275,
      "learning_rate": 4.8964191331927365e-06,
      "loss": 0.1806,
      "step": 1344
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.5988908531360515,
      "learning_rate": 4.896214338479526e-06,
      "loss": 0.4151,
      "step": 1345
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.36072324478676,
      "learning_rate": 4.896009345802834e-06,
      "loss": 0.6232,
      "step": 1346
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.246009607516789,
      "learning_rate": 4.895804155179598e-06,
      "loss": 0.3818,
      "step": 1347
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.70139252322416,
      "learning_rate": 4.895598766626767e-06,
      "loss": 0.3353,
      "step": 1348
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.9146226353778966,
      "learning_rate": 4.895393180161312e-06,
      "loss": 0.2692,
      "step": 1349
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.591515796095818,
      "learning_rate": 4.895187395800217e-06,
      "loss": 0.5247,
      "step": 1350
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.945848496621707,
      "learning_rate": 4.894981413560481e-06,
      "loss": 0.2541,
      "step": 1351
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.424962856667558,
      "learning_rate": 4.894775233459124e-06,
      "loss": 0.2977,
      "step": 1352
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.736204968477876,
      "learning_rate": 4.8945688555131765e-06,
      "loss": 0.4457,
      "step": 1353
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.107002524588832,
      "learning_rate": 4.894362279739692e-06,
      "loss": 0.1251,
      "step": 1354
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.357434857782764,
      "learning_rate": 4.894155506155734e-06,
      "loss": 0.3945,
      "step": 1355
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.992037277970906,
      "learning_rate": 4.893948534778385e-06,
      "loss": 0.4578,
      "step": 1356
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.4623778023890655,
      "learning_rate": 4.8937413656247455e-06,
      "loss": 0.6488,
      "step": 1357
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.096675275547006,
      "learning_rate": 4.893533998711931e-06,
      "loss": 0.3653,
      "step": 1358
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.824950218679396,
      "learning_rate": 4.8933264340570716e-06,
      "loss": 0.301,
      "step": 1359
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5539966611388243,
      "learning_rate": 4.8931186716773175e-06,
      "loss": 0.2556,
      "step": 1360
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.179224433325234,
      "learning_rate": 4.89291071158983e-06,
      "loss": 0.1659,
      "step": 1361
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.748834540701216,
      "learning_rate": 4.892702553811791e-06,
      "loss": 0.448,
      "step": 1362
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.576403705456475,
      "learning_rate": 4.8924941983603995e-06,
      "loss": 0.3588,
      "step": 1363
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.942109977584922,
      "learning_rate": 4.892285645252866e-06,
      "loss": 0.3873,
      "step": 1364
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.2098232945889174,
      "learning_rate": 4.892076894506421e-06,
      "loss": 0.0913,
      "step": 1365
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.820588687728016,
      "learning_rate": 4.891867946138311e-06,
      "loss": 0.4171,
      "step": 1366
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.3727799109873,
      "learning_rate": 4.891658800165798e-06,
      "loss": 0.5322,
      "step": 1367
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.7583958173884096,
      "learning_rate": 4.891449456606161e-06,
      "loss": 0.2324,
      "step": 1368
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.4496895557270655,
      "learning_rate": 4.891239915476694e-06,
      "loss": 0.2929,
      "step": 1369
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.6178322289131764,
      "learning_rate": 4.89103017679471e-06,
      "loss": 0.2762,
      "step": 1370
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.460640791276219,
      "learning_rate": 4.890820240577534e-06,
      "loss": 0.285,
      "step": 1371
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.255586284260099,
      "learning_rate": 4.890610106842512e-06,
      "loss": 0.2347,
      "step": 1372
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.377091207390276,
      "learning_rate": 4.890399775607003e-06,
      "loss": 0.2831,
      "step": 1373
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.750567050082062,
      "learning_rate": 4.890189246888385e-06,
      "loss": 0.51,
      "step": 1374
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.700836882904084,
      "learning_rate": 4.8899785207040485e-06,
      "loss": 0.7061,
      "step": 1375
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.371764085185224,
      "learning_rate": 4.889767597071405e-06,
      "loss": 0.4594,
      "step": 1376
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.8247853715670805,
      "learning_rate": 4.889556476007878e-06,
      "loss": 0.3329,
      "step": 1377
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.8693922358026294,
      "learning_rate": 4.889345157530911e-06,
      "loss": 0.2866,
      "step": 1378
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.008724888101017,
      "learning_rate": 4.889133641657962e-06,
      "loss": 0.1796,
      "step": 1379
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.390299582282176,
      "learning_rate": 4.8889219284065045e-06,
      "loss": 0.4529,
      "step": 1380
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.905423143454793,
      "learning_rate": 4.8887100177940296e-06,
      "loss": 0.2876,
      "step": 1381
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.635997461449608,
      "learning_rate": 4.888497909838044e-06,
      "loss": 0.3617,
      "step": 1382
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.352434945665413,
      "learning_rate": 4.888285604556071e-06,
      "loss": 0.4534,
      "step": 1383
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.410945326087265,
      "learning_rate": 4.88807310196565e-06,
      "loss": 0.3512,
      "step": 1384
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.617680435641105,
      "learning_rate": 4.887860402084339e-06,
      "loss": 0.2743,
      "step": 1385
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.9132293911530978,
      "learning_rate": 4.8876475049297076e-06,
      "loss": 0.2774,
      "step": 1386
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4696375207090098,
      "learning_rate": 4.887434410519345e-06,
      "loss": 0.2077,
      "step": 1387
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.977787463948352,
      "learning_rate": 4.887221118870858e-06,
      "loss": 0.3139,
      "step": 1388
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.205880341347019,
      "learning_rate": 4.887007630001864e-06,
      "loss": 0.2872,
      "step": 1389
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.031005686206253,
      "learning_rate": 4.886793943930004e-06,
      "loss": 0.1216,
      "step": 1390
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.7000777039028554,
      "learning_rate": 4.886580060672929e-06,
      "loss": 0.2253,
      "step": 1391
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.239683178968669,
      "learning_rate": 4.886365980248311e-06,
      "loss": 0.3218,
      "step": 1392
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.847745619109806,
      "learning_rate": 4.8861517026738344e-06,
      "loss": 0.6475,
      "step": 1393
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.112481639829281,
      "learning_rate": 4.885937227967205e-06,
      "loss": 0.4281,
      "step": 1394
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.274980429971524,
      "learning_rate": 4.8857225561461375e-06,
      "loss": 0.2853,
      "step": 1395
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.537991655119224,
      "learning_rate": 4.885507687228369e-06,
      "loss": 0.2435,
      "step": 1396
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.950605937638363,
      "learning_rate": 4.885292621231652e-06,
      "loss": 0.298,
      "step": 1397
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.677588451736301,
      "learning_rate": 4.885077358173752e-06,
      "loss": 0.4075,
      "step": 1398
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.440991457114333,
      "learning_rate": 4.884861898072455e-06,
      "loss": 0.3293,
      "step": 1399
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.437780153795077,
      "learning_rate": 4.88464624094556e-06,
      "loss": 0.5207,
      "step": 1400
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.847927560246554,
      "learning_rate": 4.884430386810883e-06,
      "loss": 0.2882,
      "step": 1401
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.3766731736411226,
      "learning_rate": 4.884214335686259e-06,
      "loss": 0.1673,
      "step": 1402
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.066614696174603,
      "learning_rate": 4.8839980875895355e-06,
      "loss": 0.2455,
      "step": 1403
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.697266431745929,
      "learning_rate": 4.883781642538577e-06,
      "loss": 0.1069,
      "step": 1404
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.19368058017248,
      "learning_rate": 4.883565000551267e-06,
      "loss": 0.4254,
      "step": 1405
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.3955223836112105,
      "learning_rate": 4.8833481616455035e-06,
      "loss": 0.196,
      "step": 1406
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.742999110243388,
      "learning_rate": 4.883131125839199e-06,
      "loss": 0.0136,
      "step": 1407
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.0370655622418425,
      "learning_rate": 4.882913893150285e-06,
      "loss": 0.4244,
      "step": 1408
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.451957402717778,
      "learning_rate": 4.882696463596707e-06,
      "loss": 0.4641,
      "step": 1409
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4863113534174377,
      "learning_rate": 4.882478837196429e-06,
      "loss": 0.1391,
      "step": 1410
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.515431145019314,
      "learning_rate": 4.882261013967432e-06,
      "loss": 0.3322,
      "step": 1411
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.475210995287374,
      "learning_rate": 4.882042993927708e-06,
      "loss": 0.3541,
      "step": 1412
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.345793658366856,
      "learning_rate": 4.881824777095271e-06,
      "loss": 0.4334,
      "step": 1413
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.875950385731004,
      "learning_rate": 4.881606363488148e-06,
      "loss": 0.453,
      "step": 1414
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.079086390690707,
      "learning_rate": 4.881387753124384e-06,
      "loss": 0.3425,
      "step": 1415
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.1464938850573665,
      "learning_rate": 4.881168946022038e-06,
      "loss": 0.2803,
      "step": 1416
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.013710985618428,
      "learning_rate": 4.880949942199189e-06,
      "loss": 0.4059,
      "step": 1417
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.336930817335986,
      "learning_rate": 4.880730741673929e-06,
      "loss": 0.5306,
      "step": 1418
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.200648933707356,
      "learning_rate": 4.880511344464367e-06,
      "loss": 0.4288,
      "step": 1419
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.038046676438841,
      "learning_rate": 4.880291750588628e-06,
      "loss": 0.2561,
      "step": 1420
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.167943880397058,
      "learning_rate": 4.8800719600648565e-06,
      "loss": 0.1923,
      "step": 1421
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.5992518839616725,
      "learning_rate": 4.879851972911207e-06,
      "loss": 0.2294,
      "step": 1422
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.548146366319858,
      "learning_rate": 4.8796317891458555e-06,
      "loss": 0.226,
      "step": 1423
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.584115111705035,
      "learning_rate": 4.879411408786992e-06,
      "loss": 0.2968,
      "step": 1424
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.769341421961014,
      "learning_rate": 4.879190831852823e-06,
      "loss": 0.5551,
      "step": 1425
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.786540290167373,
      "learning_rate": 4.878970058361574e-06,
      "loss": 0.4308,
      "step": 1426
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.7917776459400283,
      "learning_rate": 4.878749088331481e-06,
      "loss": 0.2535,
      "step": 1427
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.232122724475042,
      "learning_rate": 4.878527921780801e-06,
      "loss": 0.494,
      "step": 1428
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.0420329675808855,
      "learning_rate": 4.878306558727804e-06,
      "loss": 0.3095,
      "step": 1429
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.682593741044674,
      "learning_rate": 4.878084999190782e-06,
      "loss": 0.2426,
      "step": 1430
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.706040450065104,
      "learning_rate": 4.877863243188035e-06,
      "loss": 0.1568,
      "step": 1431
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.413344338988046,
      "learning_rate": 4.8776412907378845e-06,
      "loss": 0.2566,
      "step": 1432
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.947921358383704,
      "learning_rate": 4.877419141858668e-06,
      "loss": 0.308,
      "step": 1433
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.146720524640976,
      "learning_rate": 4.877196796568737e-06,
      "loss": 0.5782,
      "step": 1434
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.250173219837029,
      "learning_rate": 4.876974254886463e-06,
      "loss": 0.2831,
      "step": 1435
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.864656707082475,
      "learning_rate": 4.876751516830229e-06,
      "loss": 0.6558,
      "step": 1436
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.942940125417463,
      "learning_rate": 4.876528582418437e-06,
      "loss": 0.3791,
      "step": 1437
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.670305969643168,
      "learning_rate": 4.876305451669505e-06,
      "loss": 0.4867,
      "step": 1438
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.990619025177092,
      "learning_rate": 4.8760821246018675e-06,
      "loss": 0.388,
      "step": 1439
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.625451328574593,
      "learning_rate": 4.875858601233974e-06,
      "loss": 0.4064,
      "step": 1440
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.9151769261412825,
      "learning_rate": 4.875634881584291e-06,
      "loss": 0.2453,
      "step": 1441
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.006856757494431,
      "learning_rate": 4.875410965671301e-06,
      "loss": 0.1982,
      "step": 1442
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.9644062142034695,
      "learning_rate": 4.875186853513503e-06,
      "loss": 0.3774,
      "step": 1443
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.350590961410725,
      "learning_rate": 4.874962545129412e-06,
      "loss": 0.2729,
      "step": 1444
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.212129045545784,
      "learning_rate": 4.87473804053756e-06,
      "loss": 0.4604,
      "step": 1445
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.566519132793905,
      "learning_rate": 4.8745133397564935e-06,
      "loss": 0.3381,
      "step": 1446
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.108209795131333,
      "learning_rate": 4.8742884428047765e-06,
      "loss": 0.3031,
      "step": 1447
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.616867634899011,
      "learning_rate": 4.874063349700988e-06,
      "loss": 0.4533,
      "step": 1448
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.34720775273612,
      "learning_rate": 4.873838060463726e-06,
      "loss": 0.2995,
      "step": 1449
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.843685202030465,
      "learning_rate": 4.8736125751116015e-06,
      "loss": 0.2586,
      "step": 1450
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.107546241240651,
      "learning_rate": 4.873386893663243e-06,
      "loss": 0.4727,
      "step": 1451
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.068386228570674,
      "learning_rate": 4.873161016137296e-06,
      "loss": 0.2358,
      "step": 1452
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.209217893421984,
      "learning_rate": 4.87293494255242e-06,
      "loss": 0.3497,
      "step": 1453
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.4792428278189664,
      "learning_rate": 4.872708672927294e-06,
      "loss": 0.2123,
      "step": 1454
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.145692888982796,
      "learning_rate": 4.872482207280609e-06,
      "loss": 0.3588,
      "step": 1455
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.622237500746232,
      "learning_rate": 4.872255545631076e-06,
      "loss": 0.2794,
      "step": 1456
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.2703830349883685,
      "learning_rate": 4.872028687997421e-06,
      "loss": 0.3783,
      "step": 1457
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.538922657300181,
      "learning_rate": 4.8718016343983845e-06,
      "loss": 0.5522,
      "step": 1458
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.0119442216304915,
      "learning_rate": 4.871574384852726e-06,
      "loss": 0.3989,
      "step": 1459
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.5158843056885156,
      "learning_rate": 4.871346939379219e-06,
      "loss": 0.1474,
      "step": 1460
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.473791941552364,
      "learning_rate": 4.8711192979966535e-06,
      "loss": 0.4896,
      "step": 1461
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.494706305136444,
      "learning_rate": 4.870891460723836e-06,
      "loss": 0.2783,
      "step": 1462
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.592412955094521,
      "learning_rate": 4.870663427579591e-06,
      "loss": 0.297,
      "step": 1463
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.402444879730027,
      "learning_rate": 4.870435198582756e-06,
      "loss": 0.3197,
      "step": 1464
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.3688334395097685,
      "learning_rate": 4.870206773752187e-06,
      "loss": 0.305,
      "step": 1465
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.0163107484319935,
      "learning_rate": 4.869978153106754e-06,
      "loss": 0.244,
      "step": 1466
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.172847722621922,
      "learning_rate": 4.869749336665346e-06,
      "loss": 0.2862,
      "step": 1467
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.871934722228981,
      "learning_rate": 4.869520324446865e-06,
      "loss": 0.3911,
      "step": 1468
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.0867020925053,
      "learning_rate": 4.8692911164702325e-06,
      "loss": 0.2932,
      "step": 1469
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.25551667513736,
      "learning_rate": 4.869061712754384e-06,
      "loss": 0.4791,
      "step": 1470
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3665068860629197,
      "learning_rate": 4.8688321133182716e-06,
      "loss": 0.2815,
      "step": 1471
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.661207614140692,
      "learning_rate": 4.868602318180864e-06,
      "loss": 0.5742,
      "step": 1472
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.528355888288467,
      "learning_rate": 4.868372327361145e-06,
      "loss": 0.1941,
      "step": 1473
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.0518948739497,
      "learning_rate": 4.868142140878117e-06,
      "loss": 0.4963,
      "step": 1474
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.019013071323205,
      "learning_rate": 4.867911758750795e-06,
      "loss": 0.0637,
      "step": 1475
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.136138892279936,
      "learning_rate": 4.867681180998213e-06,
      "loss": 0.4711,
      "step": 1476
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.203654345968982,
      "learning_rate": 4.867450407639419e-06,
      "loss": 0.5809,
      "step": 1477
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.982355194456822,
      "learning_rate": 4.8672194386934804e-06,
      "loss": 0.2428,
      "step": 1478
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.1913449988637606,
      "learning_rate": 4.866988274179477e-06,
      "loss": 0.1698,
      "step": 1479
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.055550768662337,
      "learning_rate": 4.866756914116508e-06,
      "loss": 0.3959,
      "step": 1480
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.459592775383734,
      "learning_rate": 4.866525358523685e-06,
      "loss": 0.4001,
      "step": 1481
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.9271338782417646,
      "learning_rate": 4.866293607420139e-06,
      "loss": 0.1368,
      "step": 1482
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.972726051075272,
      "learning_rate": 4.866061660825018e-06,
      "loss": 0.3944,
      "step": 1483
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.928472845929809,
      "learning_rate": 4.865829518757482e-06,
      "loss": 0.3409,
      "step": 1484
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.630031694520759,
      "learning_rate": 4.8655971812367104e-06,
      "loss": 0.3759,
      "step": 1485
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.545494193804064,
      "learning_rate": 4.865364648281897e-06,
      "loss": 0.2604,
      "step": 1486
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.5890275098115527,
      "learning_rate": 4.865131919912254e-06,
      "loss": 0.1838,
      "step": 1487
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.829406525618208,
      "learning_rate": 4.864898996147008e-06,
      "loss": 0.1663,
      "step": 1488
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.121427710839816,
      "learning_rate": 4.8646658770054e-06,
      "loss": 0.3364,
      "step": 1489
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8315573117828965,
      "learning_rate": 4.864432562506691e-06,
      "loss": 0.0944,
      "step": 1490
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.9378223691290506,
      "learning_rate": 4.864199052670156e-06,
      "loss": 0.0739,
      "step": 1491
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.165110165681748,
      "learning_rate": 4.863965347515086e-06,
      "loss": 0.4277,
      "step": 1492
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.519634213505566,
      "learning_rate": 4.863731447060789e-06,
      "loss": 0.2551,
      "step": 1493
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.114885039372019,
      "learning_rate": 4.863497351326588e-06,
      "loss": 0.3966,
      "step": 1494
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.9177050662233324,
      "learning_rate": 4.863263060331824e-06,
      "loss": 0.2167,
      "step": 1495
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.073385104545578,
      "learning_rate": 4.863028574095853e-06,
      "loss": 0.1237,
      "step": 1496
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.080711219371721,
      "learning_rate": 4.862793892638047e-06,
      "loss": 0.309,
      "step": 1497
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6120354794196126,
      "learning_rate": 4.862559015977793e-06,
      "loss": 0.257,
      "step": 1498
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.714061242874837,
      "learning_rate": 4.862323944134495e-06,
      "loss": 0.4916,
      "step": 1499
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.1612425650579805,
      "learning_rate": 4.862088677127576e-06,
      "loss": 0.3329,
      "step": 1500
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.9650056133264275,
      "learning_rate": 4.8618532149764705e-06,
      "loss": 0.5083,
      "step": 1501
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.0688793258895535,
      "learning_rate": 4.861617557700632e-06,
      "loss": 0.115,
      "step": 1502
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.459713509245477,
      "learning_rate": 4.8613817053195285e-06,
      "loss": 0.4093,
      "step": 1503
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.501291859038678,
      "learning_rate": 4.861145657852647e-06,
      "loss": 0.1807,
      "step": 1504
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.581992247235775,
      "learning_rate": 4.860909415319487e-06,
      "loss": 0.2004,
      "step": 1505
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.070228910557373,
      "learning_rate": 4.860672977739566e-06,
      "loss": 0.3379,
      "step": 1506
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.140699446696341,
      "learning_rate": 4.860436345132418e-06,
      "loss": 0.3086,
      "step": 1507
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.022567065686537,
      "learning_rate": 4.86019951751759e-06,
      "loss": 0.5937,
      "step": 1508
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.026663896282319,
      "learning_rate": 4.8599624949146505e-06,
      "loss": 0.4436,
      "step": 1509
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.1690975327265365,
      "learning_rate": 4.85972527734318e-06,
      "loss": 0.2692,
      "step": 1510
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.330956560589934,
      "learning_rate": 4.859487864822776e-06,
      "loss": 0.2689,
      "step": 1511
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.726609926028263,
      "learning_rate": 4.8592502573730525e-06,
      "loss": 0.2704,
      "step": 1512
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.7145211120585575,
      "learning_rate": 4.859012455013639e-06,
      "loss": 0.359,
      "step": 1513
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.684090668683447,
      "learning_rate": 4.858774457764183e-06,
      "loss": 0.3499,
      "step": 1514
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.877891166252586,
      "learning_rate": 4.8585362656443455e-06,
      "loss": 0.3041,
      "step": 1515
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.493926608971406,
      "learning_rate": 4.858297878673805e-06,
      "loss": 0.3179,
      "step": 1516
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.649779679657409,
      "learning_rate": 4.858059296872257e-06,
      "loss": 0.2421,
      "step": 1517
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.165878302872649,
      "learning_rate": 4.857820520259408e-06,
      "loss": 0.2043,
      "step": 1518
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.196715998403068,
      "learning_rate": 4.857581548854989e-06,
      "loss": 0.318,
      "step": 1519
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.477163352297193,
      "learning_rate": 4.857342382678742e-06,
      "loss": 0.5083,
      "step": 1520
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.662863667451964,
      "learning_rate": 4.857103021750423e-06,
      "loss": 0.4563,
      "step": 1521
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.508272576604038,
      "learning_rate": 4.8568634660898095e-06,
      "loss": 0.5332,
      "step": 1522
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8009905791626903,
      "learning_rate": 4.8566237157166915e-06,
      "loss": 0.1336,
      "step": 1523
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.065927429535158,
      "learning_rate": 4.856383770650876e-06,
      "loss": 0.2156,
      "step": 1524
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.892264530066868,
      "learning_rate": 4.856143630912186e-06,
      "loss": 0.2395,
      "step": 1525
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.489846267422401,
      "learning_rate": 4.855903296520461e-06,
      "loss": 0.4102,
      "step": 1526
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.026185600967683,
      "learning_rate": 4.855662767495556e-06,
      "loss": 0.3771,
      "step": 1527
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.019761887614458,
      "learning_rate": 4.855422043857342e-06,
      "loss": 0.1773,
      "step": 1528
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.549274227429701,
      "learning_rate": 4.855181125625706e-06,
      "loss": 0.5073,
      "step": 1529
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.614864155444496,
      "learning_rate": 4.854940012820554e-06,
      "loss": 0.0986,
      "step": 1530
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.0605286194242485,
      "learning_rate": 4.8546987054618025e-06,
      "loss": 0.4601,
      "step": 1531
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.140307183178633,
      "learning_rate": 4.85445720356939e-06,
      "loss": 0.538,
      "step": 1532
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.544374308539927,
      "learning_rate": 4.854215507163264e-06,
      "loss": 0.4086,
      "step": 1533
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.018018308081106,
      "learning_rate": 4.853973616263396e-06,
      "loss": 0.2584,
      "step": 1534
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.858312675713498,
      "learning_rate": 4.85373153088977e-06,
      "loss": 0.4325,
      "step": 1535
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.450956803995387,
      "learning_rate": 4.853489251062383e-06,
      "loss": 0.5094,
      "step": 1536
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.098682095804232,
      "learning_rate": 4.853246776801254e-06,
      "loss": 0.4489,
      "step": 1537
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.799095906875707,
      "learning_rate": 4.853004108126413e-06,
      "loss": 0.3305,
      "step": 1538
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.263223462810356,
      "learning_rate": 4.8527612450579085e-06,
      "loss": 0.3911,
      "step": 1539
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.0869541104946325,
      "learning_rate": 4.852518187615805e-06,
      "loss": 0.4148,
      "step": 1540
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.885680871145673,
      "learning_rate": 4.852274935820182e-06,
      "loss": 0.4312,
      "step": 1541
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.011933946235737,
      "learning_rate": 4.852031489691138e-06,
      "loss": 0.4462,
      "step": 1542
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.504856383648752,
      "learning_rate": 4.851787849248783e-06,
      "loss": 0.5562,
      "step": 1543
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.5962684829781235,
      "learning_rate": 4.851544014513246e-06,
      "loss": 0.2905,
      "step": 1544
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.053887099549039,
      "learning_rate": 4.8512999855046714e-06,
      "loss": 0.1541,
      "step": 1545
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.883024808054571,
      "learning_rate": 4.85105576224322e-06,
      "loss": 0.2999,
      "step": 1546
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.988972991203576,
      "learning_rate": 4.8508113447490675e-06,
      "loss": 0.348,
      "step": 1547
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.350206837801943,
      "learning_rate": 4.850566733042408e-06,
      "loss": 0.5818,
      "step": 1548
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.572039170763437,
      "learning_rate": 4.850321927143449e-06,
      "loss": 0.4295,
      "step": 1549
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.829679182916999,
      "learning_rate": 4.850076927072415e-06,
      "loss": 0.3508,
      "step": 1550
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.592851020388913,
      "learning_rate": 4.849831732849547e-06,
      "loss": 0.3252,
      "step": 1551
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.898176763037151,
      "learning_rate": 4.849586344495102e-06,
      "loss": 0.3342,
      "step": 1552
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.554527287011423,
      "learning_rate": 4.849340762029353e-06,
      "loss": 0.2073,
      "step": 1553
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.520579627240912,
      "learning_rate": 4.849094985472588e-06,
      "loss": 0.6091,
      "step": 1554
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.59537695623124,
      "learning_rate": 4.848849014845113e-06,
      "loss": 0.3433,
      "step": 1555
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.363393994850539,
      "learning_rate": 4.848602850167247e-06,
      "loss": 0.5299,
      "step": 1556
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.56959735601239,
      "learning_rate": 4.848356491459329e-06,
      "loss": 0.3864,
      "step": 1557
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.409204475395978,
      "learning_rate": 4.848109938741711e-06,
      "loss": 0.3997,
      "step": 1558
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.8544299146310923,
      "learning_rate": 4.8478631920347615e-06,
      "loss": 0.2968,
      "step": 1559
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4689535340497186,
      "learning_rate": 4.847616251358866e-06,
      "loss": 0.1227,
      "step": 1560
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.521517034090262,
      "learning_rate": 4.8473691167344264e-06,
      "loss": 0.4356,
      "step": 1561
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.169294642990855,
      "learning_rate": 4.847121788181859e-06,
      "loss": 0.2964,
      "step": 1562
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.142053793272121,
      "learning_rate": 4.846874265721595e-06,
      "loss": 0.4968,
      "step": 1563
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.0177258350977825,
      "learning_rate": 4.846626549374087e-06,
      "loss": 0.3795,
      "step": 1564
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.862859975997768,
      "learning_rate": 4.846378639159798e-06,
      "loss": 0.4474,
      "step": 1565
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.389298205025913,
      "learning_rate": 4.846130535099209e-06,
      "loss": 0.5435,
      "step": 1566
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.478326789743864,
      "learning_rate": 4.845882237212818e-06,
      "loss": 0.5042,
      "step": 1567
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.644283017290516,
      "learning_rate": 4.845633745521138e-06,
      "loss": 0.1807,
      "step": 1568
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.741002219293276,
      "learning_rate": 4.845385060044698e-06,
      "loss": 0.2871,
      "step": 1569
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.359754426862071,
      "learning_rate": 4.8451361808040434e-06,
      "loss": 0.4501,
      "step": 1570
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.168692131131604,
      "learning_rate": 4.844887107819734e-06,
      "loss": 0.1335,
      "step": 1571
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.96988957061784,
      "learning_rate": 4.8446378411123484e-06,
      "loss": 0.2738,
      "step": 1572
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.575154529136756,
      "learning_rate": 4.84438838070248e-06,
      "loss": 0.4545,
      "step": 1573
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.629441996908707,
      "learning_rate": 4.844138726610738e-06,
      "loss": 0.4447,
      "step": 1574
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.045681951928446,
      "learning_rate": 4.843888878857746e-06,
      "loss": 0.1989,
      "step": 1575
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.388512808074125,
      "learning_rate": 4.843638837464147e-06,
      "loss": 0.4343,
      "step": 1576
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.00372588679398,
      "learning_rate": 4.843388602450597e-06,
      "loss": 0.4127,
      "step": 1577
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.502777011966711,
      "learning_rate": 4.84313817383777e-06,
      "loss": 0.4506,
      "step": 1578
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.935054001276233,
      "learning_rate": 4.842887551646355e-06,
      "loss": 0.4254,
      "step": 1579
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.294372805369906,
      "learning_rate": 4.842636735897057e-06,
      "loss": 0.4404,
      "step": 1580
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.427831854603716,
      "learning_rate": 4.842385726610597e-06,
      "loss": 0.5918,
      "step": 1581
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.418444658332386,
      "learning_rate": 4.842134523807713e-06,
      "loss": 0.1747,
      "step": 1582
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.425586890537586,
      "learning_rate": 4.841883127509156e-06,
      "loss": 0.4371,
      "step": 1583
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.42311771833548,
      "learning_rate": 4.841631537735698e-06,
      "loss": 0.3783,
      "step": 1584
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.896184593371718,
      "learning_rate": 4.841379754508122e-06,
      "loss": 0.517,
      "step": 1585
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.363639877693155,
      "learning_rate": 4.84112777784723e-06,
      "loss": 0.2653,
      "step": 1586
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.191570223951271,
      "learning_rate": 4.84087560777384e-06,
      "loss": 0.2948,
      "step": 1587
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.3535363993678615,
      "learning_rate": 4.8406232443087815e-06,
      "loss": 0.2944,
      "step": 1588
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.175224074485378,
      "learning_rate": 4.840370687472908e-06,
      "loss": 0.5244,
      "step": 1589
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.241077724407053,
      "learning_rate": 4.840117937287082e-06,
      "loss": 0.4824,
      "step": 1590
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.364537134237818,
      "learning_rate": 4.839864993772185e-06,
      "loss": 0.2226,
      "step": 1591
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.402787755799058,
      "learning_rate": 4.839611856949115e-06,
      "loss": 0.3414,
      "step": 1592
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.071162064911199,
      "learning_rate": 4.839358526838782e-06,
      "loss": 0.3321,
      "step": 1593
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.825135913418499,
      "learning_rate": 4.839105003462119e-06,
      "loss": 0.2065,
      "step": 1594
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.28640776824394,
      "learning_rate": 4.838851286840067e-06,
      "loss": 0.4908,
      "step": 1595
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.522571996770473,
      "learning_rate": 4.838597376993589e-06,
      "loss": 0.1027,
      "step": 1596
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.173217562958282,
      "learning_rate": 4.838343273943661e-06,
      "loss": 0.2895,
      "step": 1597
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.218377546775241,
      "learning_rate": 4.838088977711275e-06,
      "loss": 0.2115,
      "step": 1598
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.9239781514642535,
      "learning_rate": 4.837834488317442e-06,
      "loss": 0.4252,
      "step": 1599
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.272980558520837,
      "learning_rate": 4.837579805783185e-06,
      "loss": 0.4264,
      "step": 1600
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.774441700512969,
      "learning_rate": 4.837324930129546e-06,
      "loss": 0.3291,
      "step": 1601
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.063674163448097,
      "learning_rate": 4.837069861377579e-06,
      "loss": 0.2429,
      "step": 1602
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.92528245474382,
      "learning_rate": 4.836814599548358e-06,
      "loss": 0.1011,
      "step": 1603
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.6494353894400207,
      "learning_rate": 4.836559144662973e-06,
      "loss": 0.249,
      "step": 1604
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.525151136403611,
      "learning_rate": 4.836303496742526e-06,
      "loss": 0.4751,
      "step": 1605
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.109682956406212,
      "learning_rate": 4.836047655808139e-06,
      "loss": 0.2949,
      "step": 1606
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.166935848755377,
      "learning_rate": 4.835791621880948e-06,
      "loss": 0.2543,
      "step": 1607
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.99013295621999,
      "learning_rate": 4.835535394982104e-06,
      "loss": 0.4309,
      "step": 1608
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.9010749408867245,
      "learning_rate": 4.835278975132776e-06,
      "loss": 0.1845,
      "step": 1609
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.4708128187508067,
      "learning_rate": 4.8350223623541494e-06,
      "loss": 0.1779,
      "step": 1610
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.9728970533065153,
      "learning_rate": 4.834765556667422e-06,
      "loss": 0.2634,
      "step": 1611
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.73106548286095,
      "learning_rate": 4.834508558093811e-06,
      "loss": 0.0477,
      "step": 1612
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.544107619425711,
      "learning_rate": 4.83425136665455e-06,
      "loss": 0.043,
      "step": 1613
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.884101396597218,
      "learning_rate": 4.833993982370885e-06,
      "loss": 0.2099,
      "step": 1614
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.645200176622244,
      "learning_rate": 4.833736405264079e-06,
      "loss": 0.3628,
      "step": 1615
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.68750934500531,
      "learning_rate": 4.833478635355413e-06,
      "loss": 0.4249,
      "step": 1616
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.2716238179485035,
      "learning_rate": 4.833220672666183e-06,
      "loss": 0.477,
      "step": 1617
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.7715432711291688,
      "learning_rate": 4.832962517217701e-06,
      "loss": 0.2315,
      "step": 1618
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.530251878609868,
      "learning_rate": 4.8327041690312925e-06,
      "loss": 0.3609,
      "step": 1619
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1054567026886093,
      "learning_rate": 4.832445628128303e-06,
      "loss": 0.2564,
      "step": 1620
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.9763205133443265,
      "learning_rate": 4.83218689453009e-06,
      "loss": 0.4395,
      "step": 1621
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.845437161663246,
      "learning_rate": 4.831927968258031e-06,
      "loss": 0.3975,
      "step": 1622
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.264854301715616,
      "learning_rate": 4.831668849333516e-06,
      "loss": 0.253,
      "step": 1623
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7286059500583808,
      "learning_rate": 4.831409537777952e-06,
      "loss": 0.0126,
      "step": 1624
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.011431165158598,
      "learning_rate": 4.831150033612761e-06,
      "loss": 0.3493,
      "step": 1625
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.132207858006234,
      "learning_rate": 4.8308903368593855e-06,
      "loss": 0.1813,
      "step": 1626
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.301926772981657,
      "learning_rate": 4.830630447539276e-06,
      "loss": 0.3562,
      "step": 1627
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.365403607937093,
      "learning_rate": 4.8303703656739065e-06,
      "loss": 0.2685,
      "step": 1628
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.637391785970282,
      "learning_rate": 4.830110091284763e-06,
      "loss": 0.2587,
      "step": 1629
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.721593221290869,
      "learning_rate": 4.829849624393347e-06,
      "loss": 0.1837,
      "step": 1630
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.531688442616048,
      "learning_rate": 4.829588965021178e-06,
      "loss": 0.2561,
      "step": 1631
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.685504194763894,
      "learning_rate": 4.82932811318979e-06,
      "loss": 0.282,
      "step": 1632
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.360278035423687,
      "learning_rate": 4.829067068920733e-06,
      "loss": 0.4137,
      "step": 1633
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.673187260576411,
      "learning_rate": 4.828805832235574e-06,
      "loss": 0.2403,
      "step": 1634
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.821355734040323,
      "learning_rate": 4.828544403155894e-06,
      "loss": 0.538,
      "step": 1635
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.283783788945022,
      "learning_rate": 4.8282827817032916e-06,
      "loss": 0.4631,
      "step": 1636
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.981301362653537,
      "learning_rate": 4.828020967899382e-06,
      "loss": 0.3001,
      "step": 1637
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.3938613679066942,
      "learning_rate": 4.827758961765793e-06,
      "loss": 0.1825,
      "step": 1638
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.98154013435295,
      "learning_rate": 4.827496763324171e-06,
      "loss": 0.3041,
      "step": 1639
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.197463935981693,
      "learning_rate": 4.827234372596176e-06,
      "loss": 0.138,
      "step": 1640
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.070439685244517,
      "learning_rate": 4.8269717896034895e-06,
      "loss": 0.3493,
      "step": 1641
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.534037646206441,
      "learning_rate": 4.826709014367801e-06,
      "loss": 0.2521,
      "step": 1642
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.121461708877521,
      "learning_rate": 4.826446046910821e-06,
      "loss": 0.554,
      "step": 1643
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.175414892222637,
      "learning_rate": 4.826182887254275e-06,
      "loss": 0.0661,
      "step": 1644
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.860871954069593,
      "learning_rate": 4.825919535419903e-06,
      "loss": 0.5128,
      "step": 1645
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.233877693719967,
      "learning_rate": 4.825655991429463e-06,
      "loss": 0.493,
      "step": 1646
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.2613537393126055,
      "learning_rate": 4.825392255304726e-06,
      "loss": 0.3277,
      "step": 1647
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.590754150575229,
      "learning_rate": 4.825128327067483e-06,
      "loss": 0.4591,
      "step": 1648
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.306248184013304,
      "learning_rate": 4.824864206739535e-06,
      "loss": 0.0174,
      "step": 1649
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4692343571773923,
      "learning_rate": 4.824599894342707e-06,
      "loss": 0.1699,
      "step": 1650
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.074585127824686,
      "learning_rate": 4.824335389898831e-06,
      "loss": 0.2048,
      "step": 1651
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.6982846621829575,
      "learning_rate": 4.824070693429761e-06,
      "loss": 0.5898,
      "step": 1652
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.705486383378405,
      "learning_rate": 4.8238058049573655e-06,
      "loss": 0.3016,
      "step": 1653
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.192668888563532,
      "learning_rate": 4.823540724503526e-06,
      "loss": 0.3331,
      "step": 1654
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.971110238468136,
      "learning_rate": 4.823275452090145e-06,
      "loss": 0.0304,
      "step": 1655
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.511383644253186,
      "learning_rate": 4.8230099877391355e-06,
      "loss": 0.1538,
      "step": 1656
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.927532980816818,
      "learning_rate": 4.82274433147243e-06,
      "loss": 0.5448,
      "step": 1657
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.231552926409218,
      "learning_rate": 4.822478483311976e-06,
      "loss": 0.3883,
      "step": 1658
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.99986893057774,
      "learning_rate": 4.822212443279736e-06,
      "loss": 0.3082,
      "step": 1659
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7647449029938023,
      "learning_rate": 4.821946211397689e-06,
      "loss": 0.3114,
      "step": 1660
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.787502260948815,
      "learning_rate": 4.821679787687829e-06,
      "loss": 0.3389,
      "step": 1661
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.552495728887681,
      "learning_rate": 4.82141317217217e-06,
      "loss": 0.281,
      "step": 1662
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.289562182086702,
      "learning_rate": 4.821146364872734e-06,
      "loss": 0.2111,
      "step": 1663
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.203271630701889,
      "learning_rate": 4.820879365811566e-06,
      "loss": 0.1902,
      "step": 1664
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.5473709973444745,
      "learning_rate": 4.820612175010723e-06,
      "loss": 0.3911,
      "step": 1665
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.873044824858172,
      "learning_rate": 4.8203447924922796e-06,
      "loss": 0.4933,
      "step": 1666
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.077351022825108,
      "learning_rate": 4.8200772182783265e-06,
      "loss": 0.3468,
      "step": 1667
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.733558622489228,
      "learning_rate": 4.819809452390967e-06,
      "loss": 0.4419,
      "step": 1668
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.1360751153011956,
      "learning_rate": 4.8195414948523245e-06,
      "loss": 0.1475,
      "step": 1669
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7793248993515087,
      "learning_rate": 4.8192733456845355e-06,
      "loss": 0.3679,
      "step": 1670
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.661387471161434,
      "learning_rate": 4.819005004909754e-06,
      "loss": 0.3673,
      "step": 1671
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9152318448056795,
      "learning_rate": 4.818736472550148e-06,
      "loss": 0.3403,
      "step": 1672
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.822383259198448,
      "learning_rate": 4.818467748627903e-06,
      "loss": 0.1914,
      "step": 1673
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9424266204325686,
      "learning_rate": 4.81819883316522e-06,
      "loss": 0.3151,
      "step": 1674
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.358203934594817,
      "learning_rate": 4.817929726184314e-06,
      "loss": 0.0828,
      "step": 1675
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.855072515578874,
      "learning_rate": 4.817660427707418e-06,
      "loss": 0.4622,
      "step": 1676
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5305604165970257,
      "learning_rate": 4.817390937756781e-06,
      "loss": 0.2048,
      "step": 1677
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.909268856298574,
      "learning_rate": 4.817121256354666e-06,
      "loss": 0.3359,
      "step": 1678
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.240680253759838,
      "learning_rate": 4.816851383523354e-06,
      "loss": 0.363,
      "step": 1679
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.192183355456815,
      "learning_rate": 4.816581319285139e-06,
      "loss": 0.5185,
      "step": 1680
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.045403224944036,
      "learning_rate": 4.816311063662332e-06,
      "loss": 0.2555,
      "step": 1681
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.089249171113186,
      "learning_rate": 4.816040616677262e-06,
      "loss": 0.313,
      "step": 1682
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.230000308740259,
      "learning_rate": 4.8157699783522715e-06,
      "loss": 0.221,
      "step": 1683
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.93976533042171,
      "learning_rate": 4.8154991487097185e-06,
      "loss": 0.4412,
      "step": 1684
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.8055428007295213,
      "learning_rate": 4.815228127771978e-06,
      "loss": 0.1078,
      "step": 1685
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.027451040455886,
      "learning_rate": 4.8149569155614415e-06,
      "loss": 0.3853,
      "step": 1686
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.377590459904981,
      "learning_rate": 4.814685512100513e-06,
      "loss": 0.3435,
      "step": 1687
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.124391082058077,
      "learning_rate": 4.814413917411617e-06,
      "loss": 0.5076,
      "step": 1688
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.803967919321809,
      "learning_rate": 4.814142131517189e-06,
      "loss": 0.4672,
      "step": 1689
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.561654214080879,
      "learning_rate": 4.813870154439684e-06,
      "loss": 0.3045,
      "step": 1690
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.283794992387649,
      "learning_rate": 4.813597986201571e-06,
      "loss": 0.3191,
      "step": 1691
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.64100922879455,
      "learning_rate": 4.813325626825336e-06,
      "loss": 0.4725,
      "step": 1692
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.4875726379251955,
      "learning_rate": 4.81305307633348e-06,
      "loss": 0.3912,
      "step": 1693
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.171069672803588,
      "learning_rate": 4.812780334748517e-06,
      "loss": 0.3122,
      "step": 1694
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.3428517693207445,
      "learning_rate": 4.812507402092983e-06,
      "loss": 0.3068,
      "step": 1695
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.337746221026062,
      "learning_rate": 4.8122342783894246e-06,
      "loss": 0.1835,
      "step": 1696
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.601381043825855,
      "learning_rate": 4.811960963660407e-06,
      "loss": 0.178,
      "step": 1697
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.378259977732885,
      "learning_rate": 4.811687457928509e-06,
      "loss": 0.2135,
      "step": 1698
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.495562803522631,
      "learning_rate": 4.811413761216327e-06,
      "loss": 0.4402,
      "step": 1699
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9852829736888897,
      "learning_rate": 4.811139873546472e-06,
      "loss": 0.2095,
      "step": 1700
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.243118604451611,
      "learning_rate": 4.810865794941573e-06,
      "loss": 0.1885,
      "step": 1701
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.404055764345063,
      "learning_rate": 4.810591525424269e-06,
      "loss": 0.4824,
      "step": 1702
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7627202101107406,
      "learning_rate": 4.8103170650172235e-06,
      "loss": 0.2508,
      "step": 1703
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.310638893095209,
      "learning_rate": 4.810042413743108e-06,
      "loss": 0.3149,
      "step": 1704
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.641008960380941,
      "learning_rate": 4.809767571624614e-06,
      "loss": 0.1359,
      "step": 1705
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.849415562810203,
      "learning_rate": 4.8094925386844485e-06,
      "loss": 0.2701,
      "step": 1706
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.261767986766788,
      "learning_rate": 4.809217314945331e-06,
      "loss": 0.3154,
      "step": 1707
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.246028205113838,
      "learning_rate": 4.80894190043e-06,
      "loss": 0.3287,
      "step": 1708
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.058867378801684,
      "learning_rate": 4.80866629516121e-06,
      "loss": 0.4692,
      "step": 1709
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.437336558860687,
      "learning_rate": 4.80839049916173e-06,
      "loss": 0.3487,
      "step": 1710
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.202333624522988,
      "learning_rate": 4.808114512454344e-06,
      "loss": 0.3605,
      "step": 1711
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7202510096805628,
      "learning_rate": 4.807838335061853e-06,
      "loss": 0.2924,
      "step": 1712
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.667835123490464,
      "learning_rate": 4.807561967007073e-06,
      "loss": 0.2757,
      "step": 1713
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.706591409175506,
      "learning_rate": 4.807285408312837e-06,
      "loss": 0.4984,
      "step": 1714
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.315117566305738,
      "learning_rate": 4.807008659001992e-06,
      "loss": 0.392,
      "step": 1715
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.751805952868362,
      "learning_rate": 4.806731719097404e-06,
      "loss": 0.2911,
      "step": 1716
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.7894787745618688,
      "learning_rate": 4.80645458862195e-06,
      "loss": 0.267,
      "step": 1717
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.236740067608775,
      "learning_rate": 4.806177267598525e-06,
      "loss": 0.375,
      "step": 1718
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.477036337324568,
      "learning_rate": 4.8058997560500406e-06,
      "loss": 0.4107,
      "step": 1719
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.002539785210008,
      "learning_rate": 4.805622053999425e-06,
      "loss": 0.2195,
      "step": 1720
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.195385310504636,
      "learning_rate": 4.805344161469617e-06,
      "loss": 0.3648,
      "step": 1721
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.4067162141576235,
      "learning_rate": 4.805066078483579e-06,
      "loss": 0.3705,
      "step": 1722
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.449742311523641,
      "learning_rate": 4.804787805064281e-06,
      "loss": 0.3725,
      "step": 1723
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.627018038605281,
      "learning_rate": 4.804509341234716e-06,
      "loss": 0.5056,
      "step": 1724
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.4801925985776725,
      "learning_rate": 4.8042306870178865e-06,
      "loss": 0.4078,
      "step": 1725
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.0470853483216285,
      "learning_rate": 4.8039518424368146e-06,
      "loss": 0.2926,
      "step": 1726
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.498043389891289,
      "learning_rate": 4.803672807514538e-06,
      "loss": 0.2658,
      "step": 1727
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.582122144564207,
      "learning_rate": 4.803393582274108e-06,
      "loss": 0.2195,
      "step": 1728
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.47635744793843,
      "learning_rate": 4.803114166738593e-06,
      "loss": 0.4403,
      "step": 1729
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.204437861116536,
      "learning_rate": 4.802834560931076e-06,
      "loss": 0.4307,
      "step": 1730
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.67297058641945,
      "learning_rate": 4.802554764874659e-06,
      "loss": 0.4311,
      "step": 1731
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.327818072602281,
      "learning_rate": 4.802274778592457e-06,
      "loss": 0.3954,
      "step": 1732
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.4498320212543425,
      "learning_rate": 4.801994602107599e-06,
      "loss": 0.2904,
      "step": 1733
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.247951124785325,
      "learning_rate": 4.801714235443233e-06,
      "loss": 0.1854,
      "step": 1734
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.049795686187666,
      "learning_rate": 4.8014336786225225e-06,
      "loss": 0.4759,
      "step": 1735
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5029619535980516,
      "learning_rate": 4.801152931668644e-06,
      "loss": 0.1139,
      "step": 1736
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.590181369566527,
      "learning_rate": 4.8008719946047924e-06,
      "loss": 0.3542,
      "step": 1737
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.870685268838023,
      "learning_rate": 4.800590867454177e-06,
      "loss": 0.3326,
      "step": 1738
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.904896190163302,
      "learning_rate": 4.8003095502400234e-06,
      "loss": 0.3935,
      "step": 1739
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.590759081512935,
      "learning_rate": 4.8000280429855725e-06,
      "loss": 0.449,
      "step": 1740
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.850352050607795,
      "learning_rate": 4.799746345714082e-06,
      "loss": 0.2789,
      "step": 1741
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.436561446145114,
      "learning_rate": 4.799464458448822e-06,
      "loss": 0.2507,
      "step": 1742
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.574411777860661,
      "learning_rate": 4.799182381213083e-06,
      "loss": 0.3396,
      "step": 1743
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.432605925920885,
      "learning_rate": 4.798900114030168e-06,
      "loss": 0.4997,
      "step": 1744
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.65985277618257,
      "learning_rate": 4.798617656923397e-06,
      "loss": 0.2916,
      "step": 1745
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.082391128390032,
      "learning_rate": 4.798335009916103e-06,
      "loss": 0.3262,
      "step": 1746
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.686111087165405,
      "learning_rate": 4.79805217303164e-06,
      "loss": 0.2726,
      "step": 1747
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.4411632228885205,
      "learning_rate": 4.797769146293373e-06,
      "loss": 0.425,
      "step": 1748
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5767004245301464,
      "learning_rate": 4.797485929724684e-06,
      "loss": 0.252,
      "step": 1749
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.001873219579046,
      "learning_rate": 4.797202523348972e-06,
      "loss": 0.1338,
      "step": 1750
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.596381061409276,
      "learning_rate": 4.79691892718965e-06,
      "loss": 0.2661,
      "step": 1751
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.198125037290187,
      "learning_rate": 4.796635141270147e-06,
      "loss": 0.3092,
      "step": 1752
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.371896888396021,
      "learning_rate": 4.7963511656139084e-06,
      "loss": 0.2844,
      "step": 1753
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.283066792152147,
      "learning_rate": 4.7960670002443955e-06,
      "loss": 0.473,
      "step": 1754
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.639573919500299,
      "learning_rate": 4.795782645185084e-06,
      "loss": 0.2345,
      "step": 1755
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.6575066941876964,
      "learning_rate": 4.795498100459465e-06,
      "loss": 0.1847,
      "step": 1756
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.375321536246233,
      "learning_rate": 4.795213366091048e-06,
      "loss": 0.2866,
      "step": 1757
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.044909990277055,
      "learning_rate": 4.794928442103355e-06,
      "loss": 0.1967,
      "step": 1758
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.416107071632289,
      "learning_rate": 4.794643328519927e-06,
      "loss": 0.2096,
      "step": 1759
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.103200496306532,
      "learning_rate": 4.7943580253643155e-06,
      "loss": 0.2343,
      "step": 1760
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.878915537230614,
      "learning_rate": 4.794072532660093e-06,
      "loss": 0.4943,
      "step": 1761
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.601140529717997,
      "learning_rate": 4.793786850430846e-06,
      "loss": 0.3094,
      "step": 1762
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.83496571348455,
      "learning_rate": 4.793500978700174e-06,
      "loss": 0.1889,
      "step": 1763
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5508861287347155,
      "learning_rate": 4.793214917491697e-06,
      "loss": 0.1454,
      "step": 1764
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.283627827499599,
      "learning_rate": 4.792928666829045e-06,
      "loss": 0.2331,
      "step": 1765
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.9493211780419006,
      "learning_rate": 4.7926422267358685e-06,
      "loss": 0.0566,
      "step": 1766
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.700989153433952,
      "learning_rate": 4.7923555972358324e-06,
      "loss": 0.3965,
      "step": 1767
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.493155668141879,
      "learning_rate": 4.792068778352615e-06,
      "loss": 0.2494,
      "step": 1768
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.909851919712548,
      "learning_rate": 4.791781770109912e-06,
      "loss": 0.1884,
      "step": 1769
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.62390924133257,
      "learning_rate": 4.791494572531437e-06,
      "loss": 0.4525,
      "step": 1770
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.699820493737302,
      "learning_rate": 4.7912071856409135e-06,
      "loss": 0.3264,
      "step": 1771
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.123692118476406,
      "learning_rate": 4.790919609462085e-06,
      "loss": 0.4421,
      "step": 1772
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.475982117838994,
      "learning_rate": 4.790631844018711e-06,
      "loss": 0.3394,
      "step": 1773
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.212010249025954,
      "learning_rate": 4.7903438893345644e-06,
      "loss": 0.1937,
      "step": 1774
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.661536742046328,
      "learning_rate": 4.790055745433434e-06,
      "loss": 0.4487,
      "step": 1775
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.530730920679976,
      "learning_rate": 4.789767412339127e-06,
      "loss": 0.2862,
      "step": 1776
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.3246241564145205,
      "learning_rate": 4.789478890075461e-06,
      "loss": 0.3156,
      "step": 1777
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.140834133856293,
      "learning_rate": 4.789190178666274e-06,
      "loss": 0.4837,
      "step": 1778
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8402187555846004,
      "learning_rate": 4.788901278135418e-06,
      "loss": 0.2562,
      "step": 1779
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.149158162354407,
      "learning_rate": 4.788612188506761e-06,
      "loss": 0.4519,
      "step": 1780
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.432815771971586,
      "learning_rate": 4.788322909804185e-06,
      "loss": 0.3308,
      "step": 1781
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.871782400812437,
      "learning_rate": 4.788033442051588e-06,
      "loss": 0.2349,
      "step": 1782
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.457295230099407,
      "learning_rate": 4.787743785272888e-06,
      "loss": 0.243,
      "step": 1783
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.660022163158276,
      "learning_rate": 4.78745393949201e-06,
      "loss": 0.4336,
      "step": 1784
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.639118382217684,
      "learning_rate": 4.787163904732905e-06,
      "loss": 0.4416,
      "step": 1785
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.597360141925622,
      "learning_rate": 4.78687368101953e-06,
      "loss": 0.1686,
      "step": 1786
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.953048072698487,
      "learning_rate": 4.786583268375863e-06,
      "loss": 0.4407,
      "step": 1787
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.884791420299992,
      "learning_rate": 4.786292666825898e-06,
      "loss": 0.3983,
      "step": 1788
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.704519821627583,
      "learning_rate": 4.786001876393641e-06,
      "loss": 0.222,
      "step": 1789
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.397316658973507,
      "learning_rate": 4.785710897103118e-06,
      "loss": 0.3605,
      "step": 1790
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.928165632426735,
      "learning_rate": 4.785419728978366e-06,
      "loss": 0.6169,
      "step": 1791
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.186080721429695,
      "learning_rate": 4.78512837204344e-06,
      "loss": 0.3076,
      "step": 1792
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.488844932933939,
      "learning_rate": 4.784836826322412e-06,
      "loss": 0.0345,
      "step": 1793
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.371942617820949,
      "learning_rate": 4.784545091839368e-06,
      "loss": 0.2709,
      "step": 1794
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.129104664897584,
      "learning_rate": 4.784253168618408e-06,
      "loss": 0.385,
      "step": 1795
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9215436608813388,
      "learning_rate": 4.78396105668365e-06,
      "loss": 0.2209,
      "step": 1796
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0468947729749414,
      "learning_rate": 4.783668756059229e-06,
      "loss": 0.078,
      "step": 1797
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.062646369256378,
      "learning_rate": 4.7833762667692905e-06,
      "loss": 0.2461,
      "step": 1798
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.596200465445684,
      "learning_rate": 4.783083588837999e-06,
      "loss": 0.3668,
      "step": 1799
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4808108289388584,
      "learning_rate": 4.7827907222895355e-06,
      "loss": 0.174,
      "step": 1800
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.478115921308195,
      "learning_rate": 4.782497667148094e-06,
      "loss": 0.5333,
      "step": 1801
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.720503793450639,
      "learning_rate": 4.782204423437886e-06,
      "loss": 0.2168,
      "step": 1802
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.757003930862618,
      "learning_rate": 4.781910991183138e-06,
      "loss": 0.4235,
      "step": 1803
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.2082629331125325,
      "learning_rate": 4.781617370408091e-06,
      "loss": 0.2794,
      "step": 1804
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.9584496482493,
      "learning_rate": 4.781323561137004e-06,
      "loss": 0.4724,
      "step": 1805
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.149755735890815,
      "learning_rate": 4.7810295633941475e-06,
      "loss": 0.3357,
      "step": 1806
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.86466870001872,
      "learning_rate": 4.780735377203813e-06,
      "loss": 0.3831,
      "step": 1807
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.284920868986309,
      "learning_rate": 4.780441002590303e-06,
      "loss": 0.3081,
      "step": 1808
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.101137770914744,
      "learning_rate": 4.780146439577937e-06,
      "loss": 0.2324,
      "step": 1809
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.253812215665532,
      "learning_rate": 4.779851688191053e-06,
      "loss": 0.5583,
      "step": 1810
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.05171326391651,
      "learning_rate": 4.779556748453999e-06,
      "loss": 0.4095,
      "step": 1811
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.89669342686593,
      "learning_rate": 4.7792616203911414e-06,
      "loss": 0.3637,
      "step": 1812
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.781199391330098,
      "learning_rate": 4.7789663040268646e-06,
      "loss": 0.2126,
      "step": 1813
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.468054473805599,
      "learning_rate": 4.778670799385565e-06,
      "loss": 0.6174,
      "step": 1814
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.030522375592649,
      "learning_rate": 4.778375106491655e-06,
      "loss": 0.2349,
      "step": 1815
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.302252165149779,
      "learning_rate": 4.778079225369564e-06,
      "loss": 0.4057,
      "step": 1816
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5410863074477796,
      "learning_rate": 4.777783156043737e-06,
      "loss": 0.2001,
      "step": 1817
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.619291173699032,
      "learning_rate": 4.777486898538633e-06,
      "loss": 0.4977,
      "step": 1818
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.345809185982555,
      "learning_rate": 4.7771904528787275e-06,
      "loss": 0.5362,
      "step": 1819
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4313352950621323,
      "learning_rate": 4.77689381908851e-06,
      "loss": 0.2166,
      "step": 1820
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3291989387708063,
      "learning_rate": 4.776596997192488e-06,
      "loss": 0.1188,
      "step": 1821
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.54204554050337,
      "learning_rate": 4.776299987215186e-06,
      "loss": 0.412,
      "step": 1822
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.317907440357903,
      "learning_rate": 4.7760027891811375e-06,
      "loss": 0.6025,
      "step": 1823
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.102901015413308,
      "learning_rate": 4.775705403114897e-06,
      "loss": 0.4399,
      "step": 1824
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.024259099012777,
      "learning_rate": 4.775407829041033e-06,
      "loss": 0.3001,
      "step": 1825
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.791649828800089,
      "learning_rate": 4.775110066984131e-06,
      "loss": 0.4271,
      "step": 1826
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9732649416923556,
      "learning_rate": 4.77481211696879e-06,
      "loss": 0.33,
      "step": 1827
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.589478269765623,
      "learning_rate": 4.774513979019623e-06,
      "loss": 0.3371,
      "step": 1828
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.635648178745653,
      "learning_rate": 4.774215653161263e-06,
      "loss": 0.3132,
      "step": 1829
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1970769185488987,
      "learning_rate": 4.773917139418356e-06,
      "loss": 0.1655,
      "step": 1830
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.5137334447530697,
      "learning_rate": 4.773618437815563e-06,
      "loss": 0.2027,
      "step": 1831
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.094112227886149,
      "learning_rate": 4.773319548377562e-06,
      "loss": 0.3697,
      "step": 1832
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.168267530289933,
      "learning_rate": 4.773020471129045e-06,
      "loss": 0.2984,
      "step": 1833
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.2543208550848535,
      "learning_rate": 4.772721206094721e-06,
      "loss": 0.1725,
      "step": 1834
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.234593987302693,
      "learning_rate": 4.772421753299313e-06,
      "loss": 0.3016,
      "step": 1835
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.951463900026712,
      "learning_rate": 4.772122112767562e-06,
      "loss": 0.1368,
      "step": 1836
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.714739184255372,
      "learning_rate": 4.77182228452422e-06,
      "loss": 0.4153,
      "step": 1837
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.434271751709683,
      "learning_rate": 4.77152226859406e-06,
      "loss": 0.3809,
      "step": 1838
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.8758160939102138,
      "learning_rate": 4.771222065001867e-06,
      "loss": 0.0953,
      "step": 1839
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.761965018310133,
      "learning_rate": 4.770921673772442e-06,
      "loss": 0.1336,
      "step": 1840
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.298870682470986,
      "learning_rate": 4.770621094930602e-06,
      "loss": 0.3815,
      "step": 1841
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.064610508687042,
      "learning_rate": 4.770320328501179e-06,
      "loss": 0.384,
      "step": 1842
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.474473529278588,
      "learning_rate": 4.770019374509022e-06,
      "loss": 0.3073,
      "step": 1843
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.589332364643142,
      "learning_rate": 4.7697182329789925e-06,
      "loss": 0.5423,
      "step": 1844
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.042551073348723,
      "learning_rate": 4.769416903935972e-06,
      "loss": 0.1103,
      "step": 1845
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.482378385584036,
      "learning_rate": 4.769115387404851e-06,
      "loss": 0.3602,
      "step": 1846
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.353889231807109,
      "learning_rate": 4.768813683410544e-06,
      "loss": 0.5435,
      "step": 1847
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.727058714256918,
      "learning_rate": 4.768511791977972e-06,
      "loss": 0.4439,
      "step": 1848
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.377562412127414,
      "learning_rate": 4.768209713132079e-06,
      "loss": 0.2605,
      "step": 1849
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.9781017683126425,
      "learning_rate": 4.767907446897819e-06,
      "loss": 0.271,
      "step": 1850
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.9512377786076214,
      "learning_rate": 4.767604993300165e-06,
      "loss": 0.2492,
      "step": 1851
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.219765805968633,
      "learning_rate": 4.767302352364104e-06,
      "loss": 0.3973,
      "step": 1852
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.58346145911279,
      "learning_rate": 4.766999524114638e-06,
      "loss": 0.4558,
      "step": 1853
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.227407889429635,
      "learning_rate": 4.766696508576787e-06,
      "loss": 0.3572,
      "step": 1854
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.1020127945624845,
      "learning_rate": 4.766393305775583e-06,
      "loss": 0.3303,
      "step": 1855
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.20007118434225,
      "learning_rate": 4.766089915736075e-06,
      "loss": 0.1997,
      "step": 1856
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.798850429592407,
      "learning_rate": 4.765786338483329e-06,
      "loss": 0.4159,
      "step": 1857
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.4890039953573795,
      "learning_rate": 4.765482574042424e-06,
      "loss": 0.0715,
      "step": 1858
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.354966966128631,
      "learning_rate": 4.765178622438456e-06,
      "loss": 0.1686,
      "step": 1859
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.121066073865178,
      "learning_rate": 4.764874483696535e-06,
      "loss": 0.3741,
      "step": 1860
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.261503923521855,
      "learning_rate": 4.764570157841789e-06,
      "loss": 0.3897,
      "step": 1861
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.474631868900998,
      "learning_rate": 4.764265644899358e-06,
      "loss": 0.2173,
      "step": 1862
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.772334571512609,
      "learning_rate": 4.763960944894403e-06,
      "loss": 0.4574,
      "step": 1863
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8038417827515154,
      "learning_rate": 4.763656057852092e-06,
      "loss": 0.1008,
      "step": 1864
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.905783577236055,
      "learning_rate": 4.763350983797617e-06,
      "loss": 0.4239,
      "step": 1865
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.78171603240206,
      "learning_rate": 4.763045722756179e-06,
      "loss": 0.4312,
      "step": 1866
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.207291199212523,
      "learning_rate": 4.762740274752999e-06,
      "loss": 0.3164,
      "step": 1867
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.691025971681893,
      "learning_rate": 4.762434639813311e-06,
      "loss": 0.4884,
      "step": 1868
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.8974234743374225,
      "learning_rate": 4.762128817962366e-06,
      "loss": 0.4791,
      "step": 1869
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.022524447406445,
      "learning_rate": 4.7618228092254284e-06,
      "loss": 0.3057,
      "step": 1870
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.7188152120004885,
      "learning_rate": 4.761516613627779e-06,
      "loss": 0.5852,
      "step": 1871
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.239231428103924,
      "learning_rate": 4.761210231194715e-06,
      "loss": 0.2775,
      "step": 1872
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.074425339893904,
      "learning_rate": 4.7609036619515476e-06,
      "loss": 0.4238,
      "step": 1873
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.5983368701771195,
      "learning_rate": 4.760596905923605e-06,
      "loss": 0.4901,
      "step": 1874
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.519654342047812,
      "learning_rate": 4.7602899631362276e-06,
      "loss": 0.3984,
      "step": 1875
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.9044158173090984,
      "learning_rate": 4.759982833614776e-06,
      "loss": 0.2918,
      "step": 1876
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.698584611550869,
      "learning_rate": 4.759675517384623e-06,
      "loss": 0.4126,
      "step": 1877
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.499620605790925,
      "learning_rate": 4.759368014471157e-06,
      "loss": 0.3747,
      "step": 1878
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.7977517741400604,
      "learning_rate": 4.759060324899782e-06,
      "loss": 0.0448,
      "step": 1879
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.072002753787068,
      "learning_rate": 4.758752448695919e-06,
      "loss": 0.4329,
      "step": 1880
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.484512772907381,
      "learning_rate": 4.758444385885004e-06,
      "loss": 0.3667,
      "step": 1881
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.133905044706734,
      "learning_rate": 4.758136136492485e-06,
      "loss": 0.417,
      "step": 1882
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.870928426394677,
      "learning_rate": 4.75782770054383e-06,
      "loss": 0.407,
      "step": 1883
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.148958543673607,
      "learning_rate": 4.75751907806452e-06,
      "loss": 0.331,
      "step": 1884
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.096590630697008,
      "learning_rate": 4.757210269080051e-06,
      "loss": 0.3419,
      "step": 1885
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.123762004102234,
      "learning_rate": 4.756901273615937e-06,
      "loss": 0.3308,
      "step": 1886
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.1101297240350245,
      "learning_rate": 4.756592091697704e-06,
      "loss": 0.2267,
      "step": 1887
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.218045674139313,
      "learning_rate": 4.756282723350897e-06,
      "loss": 0.3222,
      "step": 1888
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.913158408424876,
      "learning_rate": 4.755973168601072e-06,
      "loss": 0.4275,
      "step": 1889
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.436020996123322,
      "learning_rate": 4.755663427473805e-06,
      "loss": 0.3046,
      "step": 1890
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.60789987792555,
      "learning_rate": 4.7553534999946846e-06,
      "loss": 0.509,
      "step": 1891
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.8126448747953114,
      "learning_rate": 4.755043386189315e-06,
      "loss": 0.2304,
      "step": 1892
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.506006810918753,
      "learning_rate": 4.754733086083317e-06,
      "loss": 0.1041,
      "step": 1893
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.127059451689439,
      "learning_rate": 4.754422599702326e-06,
      "loss": 0.3806,
      "step": 1894
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3100286580742786,
      "learning_rate": 4.754111927071992e-06,
      "loss": 0.2346,
      "step": 1895
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.181176569680955,
      "learning_rate": 4.753801068217982e-06,
      "loss": 0.4395,
      "step": 1896
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.484726952908934,
      "learning_rate": 4.753490023165979e-06,
      "loss": 0.2202,
      "step": 1897
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.29889451895835,
      "learning_rate": 4.753178791941677e-06,
      "loss": 0.3891,
      "step": 1898
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.715188028763644,
      "learning_rate": 4.75286737457079e-06,
      "loss": 0.4498,
      "step": 1899
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.350475017837337,
      "learning_rate": 4.752555771079047e-06,
      "loss": 0.1076,
      "step": 1900
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.6076765389282635,
      "learning_rate": 4.752243981492188e-06,
      "loss": 0.4985,
      "step": 1901
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.993238292437422,
      "learning_rate": 4.751932005835975e-06,
      "loss": 0.3493,
      "step": 1902
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.72838732776006,
      "learning_rate": 4.751619844136179e-06,
      "loss": 0.2306,
      "step": 1903
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.009932301597608,
      "learning_rate": 4.751307496418592e-06,
      "loss": 0.342,
      "step": 1904
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.813068107547624,
      "learning_rate": 4.750994962709017e-06,
      "loss": 0.1737,
      "step": 1905
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3274454728237965,
      "learning_rate": 4.7506822430332725e-06,
      "loss": 0.2083,
      "step": 1906
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.544520582674794,
      "learning_rate": 4.750369337417197e-06,
      "loss": 0.4974,
      "step": 1907
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.896214749447545,
      "learning_rate": 4.750056245886639e-06,
      "loss": 0.3175,
      "step": 1908
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.01606294955305,
      "learning_rate": 4.749742968467467e-06,
      "loss": 0.3031,
      "step": 1909
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.797510470059728,
      "learning_rate": 4.749429505185559e-06,
      "loss": 0.4287,
      "step": 1910
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4812995536718,
      "learning_rate": 4.749115856066815e-06,
      "loss": 0.1266,
      "step": 1911
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.779100622480183,
      "learning_rate": 4.7488020211371446e-06,
      "loss": 0.2097,
      "step": 1912
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.554126060145534,
      "learning_rate": 4.748488000422477e-06,
      "loss": 0.3144,
      "step": 1913
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.169675065590191,
      "learning_rate": 4.748173793948755e-06,
      "loss": 0.3664,
      "step": 1914
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.018386973744518,
      "learning_rate": 4.747859401741937e-06,
      "loss": 0.1225,
      "step": 1915
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.314617424283985,
      "learning_rate": 4.7475448238279945e-06,
      "loss": 0.3922,
      "step": 1916
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.893933478065889,
      "learning_rate": 4.747230060232918e-06,
      "loss": 0.285,
      "step": 1917
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.503438612978081,
      "learning_rate": 4.746915110982712e-06,
      "loss": 0.2274,
      "step": 1918
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.4881092669036144,
      "learning_rate": 4.746599976103395e-06,
      "loss": 0.1564,
      "step": 1919
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.459150595638113,
      "learning_rate": 4.746284655621003e-06,
      "loss": 0.4057,
      "step": 1920
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0461308670228595,
      "learning_rate": 4.745969149561585e-06,
      "loss": 0.155,
      "step": 1921
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.34496511083270825,
      "learning_rate": 4.745653457951207e-06,
      "loss": 0.0037,
      "step": 1922
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0328509903927907,
      "learning_rate": 4.7453375808159516e-06,
      "loss": 0.2115,
      "step": 1923
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.399415504819255,
      "learning_rate": 4.745021518181913e-06,
      "loss": 0.402,
      "step": 1924
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.3500061393078853,
      "learning_rate": 4.7447052700752025e-06,
      "loss": 0.0433,
      "step": 1925
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.427339007985305,
      "learning_rate": 4.744388836521949e-06,
      "loss": 0.4503,
      "step": 1926
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.876375985930535,
      "learning_rate": 4.744072217548292e-06,
      "loss": 0.2269,
      "step": 1927
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.496836953067217,
      "learning_rate": 4.743755413180392e-06,
      "loss": 0.2746,
      "step": 1928
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.3559176072255505,
      "learning_rate": 4.74343842344442e-06,
      "loss": 0.2784,
      "step": 1929
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.72864308532654,
      "learning_rate": 4.743121248366563e-06,
      "loss": 0.221,
      "step": 1930
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.2199396770706405,
      "learning_rate": 4.742803887973027e-06,
      "loss": 0.232,
      "step": 1931
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.2022184924824835,
      "learning_rate": 4.7424863422900306e-06,
      "loss": 0.3677,
      "step": 1932
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.7364102277479474,
      "learning_rate": 4.7421686113438054e-06,
      "loss": 0.1479,
      "step": 1933
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.944904725613318,
      "learning_rate": 4.741850695160602e-06,
      "loss": 0.2729,
      "step": 1934
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.067022951896485,
      "learning_rate": 4.741532593766687e-06,
      "loss": 0.255,
      "step": 1935
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.679668235576928,
      "learning_rate": 4.741214307188338e-06,
      "loss": 0.5111,
      "step": 1936
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.241462961074427,
      "learning_rate": 4.740895835451851e-06,
      "loss": 0.4206,
      "step": 1937
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.895458954066052,
      "learning_rate": 4.740577178583536e-06,
      "loss": 0.348,
      "step": 1938
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.767380879961586,
      "learning_rate": 4.74025833660972e-06,
      "loss": 0.2985,
      "step": 1939
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.689313497293588,
      "learning_rate": 4.739939309556744e-06,
      "loss": 0.3649,
      "step": 1940
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.283923443517927,
      "learning_rate": 4.7396200974509644e-06,
      "loss": 0.3779,
      "step": 1941
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.853011527724012,
      "learning_rate": 4.739300700318751e-06,
      "loss": 0.2569,
      "step": 1942
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.368932212238413,
      "learning_rate": 4.738981118186494e-06,
      "loss": 0.4784,
      "step": 1943
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.681440424177366,
      "learning_rate": 4.738661351080594e-06,
      "loss": 0.3952,
      "step": 1944
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.644586711483248,
      "learning_rate": 4.7383413990274675e-06,
      "loss": 0.4633,
      "step": 1945
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.9524904856933047,
      "learning_rate": 4.738021262053549e-06,
      "loss": 0.2057,
      "step": 1946
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.798762755299082,
      "learning_rate": 4.737700940185287e-06,
      "loss": 0.5706,
      "step": 1947
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.142354081083817,
      "learning_rate": 4.737380433449144e-06,
      "loss": 0.3478,
      "step": 1948
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.369446892590385,
      "learning_rate": 4.737059741871599e-06,
      "loss": 0.7859,
      "step": 1949
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.0141185548349,
      "learning_rate": 4.736738865479145e-06,
      "loss": 0.4586,
      "step": 1950
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.938649950804296,
      "learning_rate": 4.736417804298293e-06,
      "loss": 0.4431,
      "step": 1951
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.12572304877085,
      "learning_rate": 4.7360965583555665e-06,
      "loss": 0.6355,
      "step": 1952
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.272220460079509,
      "learning_rate": 4.735775127677504e-06,
      "loss": 0.3174,
      "step": 1953
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.4751460395101668,
      "learning_rate": 4.735453512290663e-06,
      "loss": 0.2791,
      "step": 1954
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.290551311697918,
      "learning_rate": 4.735131712221613e-06,
      "loss": 0.4301,
      "step": 1955
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.663190839098286,
      "learning_rate": 4.734809727496938e-06,
      "loss": 0.4804,
      "step": 1956
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.485922879904412,
      "learning_rate": 4.734487558143242e-06,
      "loss": 0.2643,
      "step": 1957
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.354694259985322,
      "learning_rate": 4.734165204187137e-06,
      "loss": 0.3561,
      "step": 1958
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.47579390560048,
      "learning_rate": 4.7338426656552575e-06,
      "loss": 0.4666,
      "step": 1959
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.235972494345486,
      "learning_rate": 4.733519942574249e-06,
      "loss": 0.486,
      "step": 1960
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.118626315449896,
      "learning_rate": 4.733197034970772e-06,
      "loss": 0.2793,
      "step": 1961
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.958963156188324,
      "learning_rate": 4.732873942871505e-06,
      "loss": 0.2469,
      "step": 1962
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.701922201620421,
      "learning_rate": 4.732550666303141e-06,
      "loss": 0.3027,
      "step": 1963
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.953918009643117,
      "learning_rate": 4.732227205292385e-06,
      "loss": 0.208,
      "step": 1964
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.466359427447186,
      "learning_rate": 4.7319035598659636e-06,
      "loss": 0.3528,
      "step": 1965
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.0743154253774465,
      "learning_rate": 4.731579730050611e-06,
      "loss": 0.5053,
      "step": 1966
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.222850084551865,
      "learning_rate": 4.731255715873082e-06,
      "loss": 0.4198,
      "step": 1967
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.176608459688928,
      "learning_rate": 4.7309315173601455e-06,
      "loss": 0.3848,
      "step": 1968
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.1169973671469573,
      "learning_rate": 4.730607134538585e-06,
      "loss": 0.1305,
      "step": 1969
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.175036624753402,
      "learning_rate": 4.730282567435198e-06,
      "loss": 0.3914,
      "step": 1970
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.648931554797861,
      "learning_rate": 4.7299578160768e-06,
      "loss": 0.2995,
      "step": 1971
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.3043700314355835,
      "learning_rate": 4.729632880490221e-06,
      "loss": 0.5553,
      "step": 1972
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.987939773524044,
      "learning_rate": 4.729307760702304e-06,
      "loss": 0.3464,
      "step": 1973
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.973499468684706,
      "learning_rate": 4.72898245673991e-06,
      "loss": 0.4572,
      "step": 1974
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.0503871264747335,
      "learning_rate": 4.728656968629913e-06,
      "loss": 0.2704,
      "step": 1975
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.347251173315286,
      "learning_rate": 4.728331296399204e-06,
      "loss": 0.4138,
      "step": 1976
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.1729089134999593,
      "learning_rate": 4.7280054400746885e-06,
      "loss": 0.1298,
      "step": 1977
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.567118028222025,
      "learning_rate": 4.727679399683286e-06,
      "loss": 0.571,
      "step": 1978
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.763473800786063,
      "learning_rate": 4.727353175251934e-06,
      "loss": 0.344,
      "step": 1979
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.399397218715444,
      "learning_rate": 4.727026766807583e-06,
      "loss": 0.2676,
      "step": 1980
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.186235678244633,
      "learning_rate": 4.726700174377198e-06,
      "loss": 0.3522,
      "step": 1981
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.415581025514767,
      "learning_rate": 4.7263733979877625e-06,
      "loss": 0.4,
      "step": 1982
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.7212549688248484,
      "learning_rate": 4.726046437666272e-06,
      "loss": 0.4045,
      "step": 1983
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.865867127495284,
      "learning_rate": 4.725719293439738e-06,
      "loss": 0.3107,
      "step": 1984
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.392545314703321,
      "learning_rate": 4.725391965335189e-06,
      "loss": 0.2832,
      "step": 1985
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.245988079686076,
      "learning_rate": 4.725064453379666e-06,
      "loss": 0.4247,
      "step": 1986
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.234595477876955,
      "learning_rate": 4.724736757600226e-06,
      "loss": 0.4411,
      "step": 1987
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.666058399255909,
      "learning_rate": 4.724408878023943e-06,
      "loss": 0.5962,
      "step": 1988
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.470595680717126,
      "learning_rate": 4.7240808146779045e-06,
      "loss": 0.203,
      "step": 1989
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.02606035846316,
      "learning_rate": 4.723752567589213e-06,
      "loss": 0.3469,
      "step": 1990
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.7023096987519475,
      "learning_rate": 4.723424136784987e-06,
      "loss": 0.4592,
      "step": 1991
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.3211802739141145,
      "learning_rate": 4.723095522292359e-06,
      "loss": 0.4054,
      "step": 1992
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.683690971983575,
      "learning_rate": 4.722766724138479e-06,
      "loss": 0.4346,
      "step": 1993
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0198887958978067,
      "learning_rate": 4.722437742350509e-06,
      "loss": 0.1741,
      "step": 1994
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.61050970895164,
      "learning_rate": 4.722108576955628e-06,
      "loss": 0.2317,
      "step": 1995
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.704736668340132,
      "learning_rate": 4.721779227981032e-06,
      "loss": 0.3578,
      "step": 1996
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.814540540477167,
      "learning_rate": 4.72144969545393e-06,
      "loss": 0.298,
      "step": 1997
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.843870224555012,
      "learning_rate": 4.721119979401544e-06,
      "loss": 0.2275,
      "step": 1998
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.373181558128303,
      "learning_rate": 4.7207900798511155e-06,
      "loss": 0.2735,
      "step": 1999
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.212353156326022,
      "learning_rate": 4.720459996829898e-06,
      "loss": 0.5067,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.400847043885809,
      "learning_rate": 4.720129730365162e-06,
      "loss": 0.2377,
      "step": 2001
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.732764502868214,
      "learning_rate": 4.719799280484192e-06,
      "loss": 0.3025,
      "step": 2002
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.323357176931031,
      "learning_rate": 4.719468647214289e-06,
      "loss": 0.2629,
      "step": 2003
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.605501914533404,
      "learning_rate": 4.7191378305827685e-06,
      "loss": 0.8634,
      "step": 2004
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8454036841089876,
      "learning_rate": 4.71880683061696e-06,
      "loss": 0.0241,
      "step": 2005
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.712607311960154,
      "learning_rate": 4.718475647344208e-06,
      "loss": 0.2596,
      "step": 2006
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.46305130359245,
      "learning_rate": 4.718144280791876e-06,
      "loss": 0.7606,
      "step": 2007
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.172731594645225,
      "learning_rate": 4.717812730987338e-06,
      "loss": 0.5002,
      "step": 2008
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.191551078360433,
      "learning_rate": 4.717480997957986e-06,
      "loss": 0.3952,
      "step": 2009
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.013938426439505,
      "learning_rate": 4.717149081731225e-06,
      "loss": 0.4924,
      "step": 2010
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.632007619940815,
      "learning_rate": 4.7168169823344775e-06,
      "loss": 0.2483,
      "step": 2011
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.246097420723695,
      "learning_rate": 4.7164846997951794e-06,
      "loss": 0.4802,
      "step": 2012
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.186180942591034,
      "learning_rate": 4.716152234140782e-06,
      "loss": 0.3292,
      "step": 2013
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.746666041651345,
      "learning_rate": 4.715819585398752e-06,
      "loss": 0.4982,
      "step": 2014
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.4801632683340324,
      "learning_rate": 4.715486753596572e-06,
      "loss": 0.381,
      "step": 2015
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.408357912632044,
      "learning_rate": 4.7151537387617375e-06,
      "loss": 0.3515,
      "step": 2016
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.7256053278260444,
      "learning_rate": 4.714820540921762e-06,
      "loss": 0.2591,
      "step": 2017
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.6041946110728595,
      "learning_rate": 4.714487160104172e-06,
      "loss": 0.4883,
      "step": 2018
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.310424084895188,
      "learning_rate": 4.7141535963365104e-06,
      "loss": 0.1477,
      "step": 2019
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.113358200500003,
      "learning_rate": 4.7138198496463336e-06,
      "loss": 0.3294,
      "step": 2020
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.506902104251425,
      "learning_rate": 4.713485920061215e-06,
      "loss": 0.8111,
      "step": 2021
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.403087289705214,
      "learning_rate": 4.713151807608741e-06,
      "loss": 0.7862,
      "step": 2022
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.323015791251663,
      "learning_rate": 4.712817512316515e-06,
      "loss": 0.4439,
      "step": 2023
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.008810964654374,
      "learning_rate": 4.712483034212156e-06,
      "loss": 0.3222,
      "step": 2024
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.5394141653331466,
      "learning_rate": 4.712148373323296e-06,
      "loss": 0.1346,
      "step": 2025
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.833458621782274,
      "learning_rate": 4.711813529677582e-06,
      "loss": 0.3088,
      "step": 2026
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.975163921514783,
      "learning_rate": 4.711478503302678e-06,
      "loss": 0.39,
      "step": 2027
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.722077336368479,
      "learning_rate": 4.711143294226264e-06,
      "loss": 0.311,
      "step": 2028
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.250451898637062,
      "learning_rate": 4.71080790247603e-06,
      "loss": 0.2652,
      "step": 2029
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.7163487051433455,
      "learning_rate": 4.710472328079687e-06,
      "loss": 0.5634,
      "step": 2030
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.0220842425157053,
      "learning_rate": 4.710136571064957e-06,
      "loss": 0.1977,
      "step": 2031
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.585935070532135,
      "learning_rate": 4.709800631459579e-06,
      "loss": 0.3674,
      "step": 2032
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.88087777927586,
      "learning_rate": 4.7094645092913085e-06,
      "loss": 0.1844,
      "step": 2033
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.073366637223225,
      "learning_rate": 4.709128204587912e-06,
      "loss": 0.3703,
      "step": 2034
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.0966519492459454,
      "learning_rate": 4.708791717377173e-06,
      "loss": 0.3746,
      "step": 2035
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.260371610041602,
      "learning_rate": 4.708455047686892e-06,
      "loss": 0.0838,
      "step": 2036
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.8977223849813725,
      "learning_rate": 4.708118195544883e-06,
      "loss": 0.2313,
      "step": 2037
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.592698099132235,
      "learning_rate": 4.707781160978974e-06,
      "loss": 0.3355,
      "step": 2038
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.843478257839081,
      "learning_rate": 4.707443944017009e-06,
      "loss": 0.4418,
      "step": 2039
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.109734884447254,
      "learning_rate": 4.707106544686848e-06,
      "loss": 0.5207,
      "step": 2040
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.890596821475505,
      "learning_rate": 4.706768963016366e-06,
      "loss": 0.5337,
      "step": 2041
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.4059663200723005,
      "learning_rate": 4.706431199033451e-06,
      "loss": 0.424,
      "step": 2042
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.413203325196151,
      "learning_rate": 4.7060932527660085e-06,
      "loss": 0.2599,
      "step": 2043
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.200632819812429,
      "learning_rate": 4.705755124241956e-06,
      "loss": 0.3443,
      "step": 2044
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.558285308988678,
      "learning_rate": 4.705416813489231e-06,
      "loss": 0.16,
      "step": 2045
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.867434842111195,
      "learning_rate": 4.705078320535781e-06,
      "loss": 0.4265,
      "step": 2046
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.4689709626470595,
      "learning_rate": 4.704739645409571e-06,
      "loss": 0.1873,
      "step": 2047
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.233170608783925,
      "learning_rate": 4.70440078813858e-06,
      "loss": 0.2743,
      "step": 2048
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.8934099371595643,
      "learning_rate": 4.704061748750803e-06,
      "loss": 0.3167,
      "step": 2049
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.339176256698062,
      "learning_rate": 4.703722527274252e-06,
      "loss": 0.4591,
      "step": 2050
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.955831413625042,
      "learning_rate": 4.70338312373695e-06,
      "loss": 0.374,
      "step": 2051
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.463975114307998,
      "learning_rate": 4.703043538166936e-06,
      "loss": 0.3207,
      "step": 2052
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.8170998971753862,
      "learning_rate": 4.7027037705922656e-06,
      "loss": 0.1214,
      "step": 2053
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.506807166770844,
      "learning_rate": 4.702363821041009e-06,
      "loss": 0.2451,
      "step": 2054
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.953120740326698,
      "learning_rate": 4.702023689541252e-06,
      "loss": 0.4983,
      "step": 2055
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.193465982650527,
      "learning_rate": 4.701683376121094e-06,
      "loss": 0.3298,
      "step": 2056
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.4420528082340454,
      "learning_rate": 4.701342880808648e-06,
      "loss": 0.3249,
      "step": 2057
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.392203127586814,
      "learning_rate": 4.701002203632047e-06,
      "loss": 0.4847,
      "step": 2058
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.115129593078125,
      "learning_rate": 4.7006613446194345e-06,
      "loss": 0.3544,
      "step": 2059
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.597946946968969,
      "learning_rate": 4.700320303798971e-06,
      "loss": 0.4292,
      "step": 2060
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.073680988067178,
      "learning_rate": 4.699979081198831e-06,
      "loss": 0.6904,
      "step": 2061
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.649562005018555,
      "learning_rate": 4.699637676847206e-06,
      "loss": 0.4883,
      "step": 2062
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.65769510606456,
      "learning_rate": 4.699296090772301e-06,
      "loss": 0.3826,
      "step": 2063
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.4617202647027865,
      "learning_rate": 4.698954323002335e-06,
      "loss": 0.2888,
      "step": 2064
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.182559146618751,
      "learning_rate": 4.698612373565543e-06,
      "loss": 0.4185,
      "step": 2065
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.054571720437092,
      "learning_rate": 4.698270242490177e-06,
      "loss": 0.4054,
      "step": 2066
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.5869669703751,
      "learning_rate": 4.6979279298044995e-06,
      "loss": 0.2516,
      "step": 2067
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.404971893616793,
      "learning_rate": 4.697585435536794e-06,
      "loss": 0.2046,
      "step": 2068
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.660732814519622,
      "learning_rate": 4.6972427597153535e-06,
      "loss": 0.1524,
      "step": 2069
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.629567650828624,
      "learning_rate": 4.696899902368488e-06,
      "loss": 0.386,
      "step": 2070
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.014527031004351,
      "learning_rate": 4.696556863524524e-06,
      "loss": 0.4028,
      "step": 2071
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.7367775699312755,
      "learning_rate": 4.696213643211801e-06,
      "loss": 0.5333,
      "step": 2072
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.211766648590792,
      "learning_rate": 4.6958702414586744e-06,
      "loss": 0.5181,
      "step": 2073
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.10436360757955,
      "learning_rate": 4.695526658293514e-06,
      "loss": 0.3234,
      "step": 2074
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.001223079239237,
      "learning_rate": 4.6951828937447055e-06,
      "loss": 0.1563,
      "step": 2075
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.125458690278839,
      "learning_rate": 4.694838947840648e-06,
      "loss": 0.2092,
      "step": 2076
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.219621294126167,
      "learning_rate": 4.694494820609759e-06,
      "loss": 0.4011,
      "step": 2077
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.793998728142006,
      "learning_rate": 4.6941505120804655e-06,
      "loss": 0.6054,
      "step": 2078
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.653193104399593,
      "learning_rate": 4.693806022281214e-06,
      "loss": 0.4711,
      "step": 2079
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.146878704543705,
      "learning_rate": 4.693461351240465e-06,
      "loss": 0.2385,
      "step": 2080
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6275513398416597,
      "learning_rate": 4.693116498986693e-06,
      "loss": 0.3089,
      "step": 2081
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.144468662723922,
      "learning_rate": 4.692771465548388e-06,
      "loss": 0.2893,
      "step": 2082
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.987023447442332,
      "learning_rate": 4.692426250954055e-06,
      "loss": 0.1702,
      "step": 2083
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.6753536665556945,
      "learning_rate": 4.692080855232215e-06,
      "loss": 0.5177,
      "step": 2084
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.540081751908243,
      "learning_rate": 4.691735278411401e-06,
      "loss": 0.2832,
      "step": 2085
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.88118046293137,
      "learning_rate": 4.691389520520164e-06,
      "loss": 0.4898,
      "step": 2086
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.875518860174292,
      "learning_rate": 4.6910435815870684e-06,
      "loss": 0.3185,
      "step": 2087
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.322326209028281,
      "learning_rate": 4.690697461640695e-06,
      "loss": 0.4651,
      "step": 2088
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.425559628820769,
      "learning_rate": 4.6903511607096365e-06,
      "loss": 0.4529,
      "step": 2089
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3478269351053216,
      "learning_rate": 4.690004678822505e-06,
      "loss": 0.2072,
      "step": 2090
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.177860039949339,
      "learning_rate": 4.689658016007923e-06,
      "loss": 0.326,
      "step": 2091
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.746727151478234,
      "learning_rate": 4.689311172294532e-06,
      "loss": 0.2945,
      "step": 2092
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.415432702266085,
      "learning_rate": 4.688964147710985e-06,
      "loss": 0.4802,
      "step": 2093
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.75431230158945,
      "learning_rate": 4.688616942285952e-06,
      "loss": 0.5523,
      "step": 2094
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.162910853186948,
      "learning_rate": 4.688269556048117e-06,
      "loss": 0.2931,
      "step": 2095
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.6065493436195,
      "learning_rate": 4.687921989026181e-06,
      "loss": 0.2307,
      "step": 2096
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.067592705604944,
      "learning_rate": 4.687574241248856e-06,
      "loss": 0.3001,
      "step": 2097
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.448773113325218,
      "learning_rate": 4.687226312744873e-06,
      "loss": 0.5082,
      "step": 2098
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.7907356996154045,
      "learning_rate": 4.686878203542974e-06,
      "loss": 0.34,
      "step": 2099
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.92605002579798,
      "learning_rate": 4.686529913671921e-06,
      "loss": 0.5079,
      "step": 2100
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.925153824486819,
      "learning_rate": 4.686181443160486e-06,
      "loss": 0.4185,
      "step": 2101
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4289029364817756,
      "learning_rate": 4.685832792037459e-06,
      "loss": 0.1313,
      "step": 2102
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.695239552274567,
      "learning_rate": 4.685483960331643e-06,
      "loss": 0.6417,
      "step": 2103
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8673465714892306,
      "learning_rate": 4.685134948071857e-06,
      "loss": 0.2453,
      "step": 2104
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8002442630179742,
      "learning_rate": 4.684785755286934e-06,
      "loss": 0.2271,
      "step": 2105
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.269785316051476,
      "learning_rate": 4.684436382005724e-06,
      "loss": 0.2492,
      "step": 2106
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.817925190870965,
      "learning_rate": 4.68408682825709e-06,
      "loss": 0.3973,
      "step": 2107
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.723700478809504,
      "learning_rate": 4.68373709406991e-06,
      "loss": 0.3309,
      "step": 2108
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.957812995910252,
      "learning_rate": 4.683387179473077e-06,
      "loss": 0.2957,
      "step": 2109
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.6749302856449786,
      "learning_rate": 4.683037084495501e-06,
      "loss": 0.1515,
      "step": 2110
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4706916866888795,
      "learning_rate": 4.682686809166102e-06,
      "loss": 0.1017,
      "step": 2111
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.979294082633625,
      "learning_rate": 4.682336353513822e-06,
      "loss": 0.2667,
      "step": 2112
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8175592116810044,
      "learning_rate": 4.681985717567611e-06,
      "loss": 0.271,
      "step": 2113
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.7427989516365985,
      "learning_rate": 4.681634901356438e-06,
      "loss": 0.3975,
      "step": 2114
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.380516023476108,
      "learning_rate": 4.681283904909285e-06,
      "loss": 0.4416,
      "step": 2115
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.001960958416022,
      "learning_rate": 4.6809327282551505e-06,
      "loss": 0.1317,
      "step": 2116
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.042573657014198,
      "learning_rate": 4.680581371423047e-06,
      "loss": 0.3587,
      "step": 2117
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.586170737073957,
      "learning_rate": 4.6802298344420005e-06,
      "loss": 0.2415,
      "step": 2118
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.5627074030243993,
      "learning_rate": 4.6798781173410544e-06,
      "loss": 0.2984,
      "step": 2119
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.897376781962465,
      "learning_rate": 4.6795262201492655e-06,
      "loss": 0.4923,
      "step": 2120
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.958468166816521,
      "learning_rate": 4.6791741428957065e-06,
      "loss": 0.3465,
      "step": 2121
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.177103258312365,
      "learning_rate": 4.678821885609464e-06,
      "loss": 0.4144,
      "step": 2122
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.982118800644589,
      "learning_rate": 4.678469448319638e-06,
      "loss": 0.1833,
      "step": 2123
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.259887292920907,
      "learning_rate": 4.678116831055347e-06,
      "loss": 0.3144,
      "step": 2124
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.6878192161819072,
      "learning_rate": 4.677764033845723e-06,
      "loss": 0.2515,
      "step": 2125
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.309478289980715,
      "learning_rate": 4.677411056719912e-06,
      "loss": 0.4164,
      "step": 2126
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.510020180925843,
      "learning_rate": 4.677057899707073e-06,
      "loss": 0.3709,
      "step": 2127
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.87130329240313,
      "learning_rate": 4.676704562836385e-06,
      "loss": 0.4471,
      "step": 2128
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5723950728993366,
      "learning_rate": 4.6763510461370375e-06,
      "loss": 0.0997,
      "step": 2129
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8726734798845244,
      "learning_rate": 4.675997349638237e-06,
      "loss": 0.2779,
      "step": 2130
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.380948931515588,
      "learning_rate": 4.675643473369203e-06,
      "loss": 0.4084,
      "step": 2131
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.23799298110824,
      "learning_rate": 4.6752894173591725e-06,
      "loss": 0.3396,
      "step": 2132
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.510926523244565,
      "learning_rate": 4.674935181637395e-06,
      "loss": 0.4815,
      "step": 2133
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.833244103787142,
      "learning_rate": 4.6745807662331365e-06,
      "loss": 0.3366,
      "step": 2134
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.410514766118684,
      "learning_rate": 4.674226171175676e-06,
      "loss": 0.383,
      "step": 2135
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.563641042268081,
      "learning_rate": 4.673871396494308e-06,
      "loss": 0.5794,
      "step": 2136
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.997678888604391,
      "learning_rate": 4.673516442218344e-06,
      "loss": 0.4271,
      "step": 2137
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.54395918386902,
      "learning_rate": 4.673161308377109e-06,
      "loss": 0.3807,
      "step": 2138
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3926723321129746,
      "learning_rate": 4.672805994999939e-06,
      "loss": 0.1582,
      "step": 2139
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.880081096372845,
      "learning_rate": 4.672450502116192e-06,
      "loss": 0.5908,
      "step": 2140
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.262082361174646,
      "learning_rate": 4.672094829755234e-06,
      "loss": 0.3084,
      "step": 2141
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.0589189769197676,
      "learning_rate": 4.671738977946452e-06,
      "loss": 0.5823,
      "step": 2142
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.270728453235989,
      "learning_rate": 4.6713829467192416e-06,
      "loss": 0.261,
      "step": 2143
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.249958068001126,
      "learning_rate": 4.671026736103019e-06,
      "loss": 0.4695,
      "step": 2144
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.059304653017382,
      "learning_rate": 4.670670346127211e-06,
      "loss": 0.4078,
      "step": 2145
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.642386251059743,
      "learning_rate": 4.67031377682126e-06,
      "loss": 0.4276,
      "step": 2146
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.6294052617229475,
      "learning_rate": 4.669957028214627e-06,
      "loss": 0.2623,
      "step": 2147
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.971354945110564,
      "learning_rate": 4.669600100336782e-06,
      "loss": 0.3144,
      "step": 2148
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.27996824843089,
      "learning_rate": 4.6692429932172145e-06,
      "loss": 0.2871,
      "step": 2149
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.0203303395813785,
      "learning_rate": 4.668885706885426e-06,
      "loss": 0.5146,
      "step": 2150
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.5617052874256334,
      "learning_rate": 4.668528241370933e-06,
      "loss": 0.2619,
      "step": 2151
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.506197382818534,
      "learning_rate": 4.6681705967032695e-06,
      "loss": 0.4253,
      "step": 2152
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.836567411794888,
      "learning_rate": 4.667812772911981e-06,
      "loss": 0.6323,
      "step": 2153
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.54675844096817,
      "learning_rate": 4.66745477002663e-06,
      "loss": 0.3899,
      "step": 2154
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.729198348761671,
      "learning_rate": 4.667096588076791e-06,
      "loss": 0.343,
      "step": 2155
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.920978444437005,
      "learning_rate": 4.666738227092057e-06,
      "loss": 0.1474,
      "step": 2156
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.9835786993740108,
      "learning_rate": 4.666379687102034e-06,
      "loss": 0.2841,
      "step": 2157
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.496347984313075,
      "learning_rate": 4.666020968136342e-06,
      "loss": 0.4369,
      "step": 2158
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.740968642122763,
      "learning_rate": 4.665662070224617e-06,
      "loss": 0.412,
      "step": 2159
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.705770753934638,
      "learning_rate": 4.66530299339651e-06,
      "loss": 0.3785,
      "step": 2160
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.871913525480436,
      "learning_rate": 4.664943737681685e-06,
      "loss": 0.5574,
      "step": 2161
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.210619590738779,
      "learning_rate": 4.6645843031098234e-06,
      "loss": 0.2368,
      "step": 2162
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.391335987446885,
      "learning_rate": 4.664224689710618e-06,
      "loss": 0.4257,
      "step": 2163
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9624465724350535,
      "learning_rate": 4.66386489751378e-06,
      "loss": 0.2035,
      "step": 2164
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.497856709257478,
      "learning_rate": 4.663504926549033e-06,
      "loss": 0.2642,
      "step": 2165
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.778491613493397,
      "learning_rate": 4.663144776846116e-06,
      "loss": 0.3854,
      "step": 2166
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.113725868023575,
      "learning_rate": 4.662784448434783e-06,
      "loss": 0.4498,
      "step": 2167
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.731035999220729,
      "learning_rate": 4.662423941344802e-06,
      "loss": 0.3172,
      "step": 2168
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.788059137696408,
      "learning_rate": 4.662063255605957e-06,
      "loss": 0.202,
      "step": 2169
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.462575338869445,
      "learning_rate": 4.6617023912480465e-06,
      "loss": 0.2291,
      "step": 2170
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.300626461636346,
      "learning_rate": 4.6613413483008816e-06,
      "loss": 0.3332,
      "step": 2171
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.020523185886519,
      "learning_rate": 4.660980126794291e-06,
      "loss": 0.4265,
      "step": 2172
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.447087082241101,
      "learning_rate": 4.660618726758118e-06,
      "loss": 0.4496,
      "step": 2173
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.7482751559652767,
      "learning_rate": 4.660257148222218e-06,
      "loss": 0.1145,
      "step": 2174
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.910853405096754,
      "learning_rate": 4.6598953912164635e-06,
      "loss": 0.2498,
      "step": 2175
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.592493183452775,
      "learning_rate": 4.6595334557707415e-06,
      "loss": 0.6829,
      "step": 2176
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.392561241959458,
      "learning_rate": 4.6591713419149535e-06,
      "loss": 0.347,
      "step": 2177
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.53802592100612,
      "learning_rate": 4.658809049679014e-06,
      "loss": 0.3336,
      "step": 2178
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.55364483520814,
      "learning_rate": 4.658446579092856e-06,
      "loss": 0.2885,
      "step": 2179
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.8097557980598493,
      "learning_rate": 4.658083930186424e-06,
      "loss": 0.3516,
      "step": 2180
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.762033943909624,
      "learning_rate": 4.657721102989677e-06,
      "loss": 0.3223,
      "step": 2181
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.3485475204774025,
      "learning_rate": 4.657358097532592e-06,
      "loss": 0.308,
      "step": 2182
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.469800433604087,
      "learning_rate": 4.656994913845158e-06,
      "loss": 0.2649,
      "step": 2183
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.221577086457754,
      "learning_rate": 4.656631551957379e-06,
      "loss": 0.2238,
      "step": 2184
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.798237886408928,
      "learning_rate": 4.6562680118992744e-06,
      "loss": 0.2727,
      "step": 2185
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.889663569683427,
      "learning_rate": 4.655904293700878e-06,
      "loss": 0.4047,
      "step": 2186
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.047534755491498,
      "learning_rate": 4.655540397392239e-06,
      "loss": 0.3117,
      "step": 2187
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.916491893363081,
      "learning_rate": 4.655176323003421e-06,
      "loss": 0.531,
      "step": 2188
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.704651520749267,
      "learning_rate": 4.654812070564501e-06,
      "loss": 0.2764,
      "step": 2189
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.710047779206783,
      "learning_rate": 4.654447640105572e-06,
      "loss": 0.5383,
      "step": 2190
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.285493589516521,
      "learning_rate": 4.654083031656743e-06,
      "loss": 0.3533,
      "step": 2191
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.869562297448018,
      "learning_rate": 4.653718245248133e-06,
      "loss": 0.4083,
      "step": 2192
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.2898991560635302,
      "learning_rate": 4.653353280909882e-06,
      "loss": 0.1988,
      "step": 2193
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.904626202827074,
      "learning_rate": 4.652988138672139e-06,
      "loss": 0.399,
      "step": 2194
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.936521076592403,
      "learning_rate": 4.652622818565073e-06,
      "loss": 0.2484,
      "step": 2195
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.5173802324061407,
      "learning_rate": 4.652257320618862e-06,
      "loss": 0.1461,
      "step": 2196
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.512914763504493,
      "learning_rate": 4.651891644863704e-06,
      "loss": 0.635,
      "step": 2197
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.26241376397761,
      "learning_rate": 4.651525791329809e-06,
      "loss": 0.4127,
      "step": 2198
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.352716149070749,
      "learning_rate": 4.651159760047401e-06,
      "loss": 0.3945,
      "step": 2199
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.128954813795077,
      "learning_rate": 4.650793551046719e-06,
      "loss": 0.3862,
      "step": 2200
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.31241249338665,
      "learning_rate": 4.65042716435802e-06,
      "loss": 0.4903,
      "step": 2201
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9549304768162123,
      "learning_rate": 4.650060600011571e-06,
      "loss": 0.1455,
      "step": 2202
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.6798052670260555,
      "learning_rate": 4.6496938580376575e-06,
      "loss": 0.408,
      "step": 2203
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.958115942084431,
      "learning_rate": 4.649326938466575e-06,
      "loss": 0.4761,
      "step": 2204
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.26161611396669,
      "learning_rate": 4.64895984132864e-06,
      "loss": 0.2152,
      "step": 2205
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.737467106840222,
      "learning_rate": 4.648592566654177e-06,
      "loss": 0.5003,
      "step": 2206
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.484224199630592,
      "learning_rate": 4.648225114473532e-06,
      "loss": 0.4305,
      "step": 2207
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.1476141006529823,
      "learning_rate": 4.64785748481706e-06,
      "loss": 0.2422,
      "step": 2208
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.358374898806271,
      "learning_rate": 4.647489677715133e-06,
      "loss": 0.3453,
      "step": 2209
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.837714775329737,
      "learning_rate": 4.647121693198135e-06,
      "loss": 0.5076,
      "step": 2210
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.966743193209042,
      "learning_rate": 4.646753531296472e-06,
      "loss": 0.4084,
      "step": 2211
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.590819627624936,
      "learning_rate": 4.6463851920405565e-06,
      "loss": 0.6544,
      "step": 2212
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.126228753606711,
      "learning_rate": 4.6460166754608195e-06,
      "loss": 0.4215,
      "step": 2213
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.7641840225018273,
      "learning_rate": 4.6456479815877065e-06,
      "loss": 0.2493,
      "step": 2214
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.332405871378078,
      "learning_rate": 4.645279110451676e-06,
      "loss": 0.4772,
      "step": 2215
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.562666386292056,
      "learning_rate": 4.6449100620832045e-06,
      "loss": 0.1897,
      "step": 2216
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.245241182285973,
      "learning_rate": 4.644540836512779e-06,
      "loss": 0.2959,
      "step": 2217
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.3770574595876734,
      "learning_rate": 4.644171433770903e-06,
      "loss": 0.1494,
      "step": 2218
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.1903711099715695,
      "learning_rate": 4.643801853888096e-06,
      "loss": 0.1026,
      "step": 2219
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.819647710790244,
      "learning_rate": 4.64343209689489e-06,
      "loss": 0.325,
      "step": 2220
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.656735447380207,
      "learning_rate": 4.643062162821833e-06,
      "loss": 0.3501,
      "step": 2221
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.127453308387496,
      "learning_rate": 4.642692051699487e-06,
      "loss": 0.2365,
      "step": 2222
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.903778245348387,
      "learning_rate": 4.642321763558429e-06,
      "loss": 0.2829,
      "step": 2223
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.403643794572614,
      "learning_rate": 4.64195129842925e-06,
      "loss": 0.3078,
      "step": 2224
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.374740564195117,
      "learning_rate": 4.641580656342556e-06,
      "loss": 0.2366,
      "step": 2225
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.037658359558236,
      "learning_rate": 4.641209837328968e-06,
      "loss": 0.2922,
      "step": 2226
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.384248943864956,
      "learning_rate": 4.640838841419121e-06,
      "loss": 0.204,
      "step": 2227
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.746160822669017,
      "learning_rate": 4.6404676686436645e-06,
      "loss": 0.3069,
      "step": 2228
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.617321356201202,
      "learning_rate": 4.6400963190332636e-06,
      "loss": 0.5138,
      "step": 2229
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.773335450336333,
      "learning_rate": 4.6397247926185974e-06,
      "loss": 0.3283,
      "step": 2230
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.434880550897557,
      "learning_rate": 4.639353089430359e-06,
      "loss": 0.4176,
      "step": 2231
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.597963340692231,
      "learning_rate": 4.638981209499257e-06,
      "loss": 0.2372,
      "step": 2232
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4806913148116876,
      "learning_rate": 4.638609152856014e-06,
      "loss": 0.2058,
      "step": 2233
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.44416998026606,
      "learning_rate": 4.638236919531369e-06,
      "loss": 0.2028,
      "step": 2234
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.957293735630923,
      "learning_rate": 4.637864509556073e-06,
      "loss": 0.3362,
      "step": 2235
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.462294166475693,
      "learning_rate": 4.637491922960891e-06,
      "loss": 0.4784,
      "step": 2236
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.822709756486112,
      "learning_rate": 4.637119159776606e-06,
      "loss": 0.4129,
      "step": 2237
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.211823901643818,
      "learning_rate": 4.6367462200340155e-06,
      "loss": 0.1645,
      "step": 2238
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.520749595826988,
      "learning_rate": 4.636373103763927e-06,
      "loss": 0.4393,
      "step": 2239
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.500427903335377,
      "learning_rate": 4.635999810997167e-06,
      "loss": 0.2611,
      "step": 2240
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.096723711542244,
      "learning_rate": 4.635626341764574e-06,
      "loss": 0.5671,
      "step": 2241
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.146323772129426,
      "learning_rate": 4.635252696097005e-06,
      "loss": 0.3848,
      "step": 2242
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.983417029651616,
      "learning_rate": 4.634878874025325e-06,
      "loss": 0.3151,
      "step": 2243
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.914129523526848,
      "learning_rate": 4.63450487558042e-06,
      "loss": 0.4365,
      "step": 2244
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.083764210025057,
      "learning_rate": 4.6341307007931864e-06,
      "loss": 0.4632,
      "step": 2245
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.492319184391597,
      "learning_rate": 4.633756349694538e-06,
      "loss": 0.3128,
      "step": 2246
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.705461020086718,
      "learning_rate": 4.633381822315401e-06,
      "loss": 0.3297,
      "step": 2247
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.339710834943245,
      "learning_rate": 4.633007118686718e-06,
      "loss": 0.3902,
      "step": 2248
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4995211974977316,
      "learning_rate": 4.632632238839443e-06,
      "loss": 0.3262,
      "step": 2249
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.701331163725999,
      "learning_rate": 4.63225718280455e-06,
      "loss": 0.3451,
      "step": 2250
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.650219444089011,
      "learning_rate": 4.631881950613021e-06,
      "loss": 0.3648,
      "step": 2251
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.7643249227599105,
      "learning_rate": 4.6315065422958574e-06,
      "loss": 0.4462,
      "step": 2252
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.598212509532075,
      "learning_rate": 4.631130957884074e-06,
      "loss": 0.6078,
      "step": 2253
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.789584450672207,
      "learning_rate": 4.630755197408699e-06,
      "loss": 0.4237,
      "step": 2254
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.219229198731275,
      "learning_rate": 4.630379260900777e-06,
      "loss": 0.0922,
      "step": 2255
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.946874526342851,
      "learning_rate": 4.630003148391363e-06,
      "loss": 0.2746,
      "step": 2256
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.039551625463302,
      "learning_rate": 4.6296268599115335e-06,
      "loss": 0.2212,
      "step": 2257
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.136337832464934,
      "learning_rate": 4.629250395492373e-06,
      "loss": 0.3103,
      "step": 2258
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.168448856904572,
      "learning_rate": 4.628873755164985e-06,
      "loss": 0.2475,
      "step": 2259
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8127711759668017,
      "learning_rate": 4.628496938960484e-06,
      "loss": 0.3098,
      "step": 2260
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.112040829123306,
      "learning_rate": 4.628119946910001e-06,
      "loss": 0.487,
      "step": 2261
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.672075642903573,
      "learning_rate": 4.627742779044682e-06,
      "loss": 0.2746,
      "step": 2262
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8746520720739404,
      "learning_rate": 4.627365435395686e-06,
      "loss": 0.2753,
      "step": 2263
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.763200723381225,
      "learning_rate": 4.626987915994189e-06,
      "loss": 0.2323,
      "step": 2264
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.454355319580722,
      "learning_rate": 4.626610220871377e-06,
      "loss": 0.3953,
      "step": 2265
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.284209515567419,
      "learning_rate": 4.626232350058455e-06,
      "loss": 0.2619,
      "step": 2266
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.028813186221561,
      "learning_rate": 4.625854303586641e-06,
      "loss": 0.3588,
      "step": 2267
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9510689966617996,
      "learning_rate": 4.625476081487167e-06,
      "loss": 0.201,
      "step": 2268
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.520247310983519,
      "learning_rate": 4.625097683791279e-06,
      "loss": 0.1818,
      "step": 2269
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.873275349320121,
      "learning_rate": 4.624719110530239e-06,
      "loss": 0.2276,
      "step": 2270
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.80568417013838,
      "learning_rate": 4.624340361735325e-06,
      "loss": 0.3426,
      "step": 2271
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.231707320570546,
      "learning_rate": 4.623961437437824e-06,
      "loss": 0.3282,
      "step": 2272
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.2832825035624893,
      "learning_rate": 4.6235823376690415e-06,
      "loss": 0.1702,
      "step": 2273
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.7750064020742995,
      "learning_rate": 4.623203062460298e-06,
      "loss": 0.3617,
      "step": 2274
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.232530354665683,
      "learning_rate": 4.6228236118429285e-06,
      "loss": 0.5601,
      "step": 2275
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.2363087527159795,
      "learning_rate": 4.622443985848279e-06,
      "loss": 0.3879,
      "step": 2276
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.70145785777701,
      "learning_rate": 4.622064184507712e-06,
      "loss": 0.3099,
      "step": 2277
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.989225475355167,
      "learning_rate": 4.6216842078526066e-06,
      "loss": 0.5032,
      "step": 2278
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.48092454105925,
      "learning_rate": 4.621304055914355e-06,
      "loss": 0.1786,
      "step": 2279
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.860581553184095,
      "learning_rate": 4.620923728724361e-06,
      "loss": 0.536,
      "step": 2280
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.802401592490863,
      "learning_rate": 4.6205432263140484e-06,
      "loss": 0.5183,
      "step": 2281
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6144706499096304,
      "learning_rate": 4.6201625487148496e-06,
      "loss": 0.1323,
      "step": 2282
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.529997328056945,
      "learning_rate": 4.619781695958217e-06,
      "loss": 0.4764,
      "step": 2283
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.540820880063197,
      "learning_rate": 4.619400668075612e-06,
      "loss": 0.2969,
      "step": 2284
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.244096778755608,
      "learning_rate": 4.619019465098516e-06,
      "loss": 0.424,
      "step": 2285
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8260321155119894,
      "learning_rate": 4.61863808705842e-06,
      "loss": 0.1766,
      "step": 2286
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.290302481111041,
      "learning_rate": 4.6182565339868324e-06,
      "loss": 0.2261,
      "step": 2287
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8090254823585665,
      "learning_rate": 4.617874805915275e-06,
      "loss": 0.2842,
      "step": 2288
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.550801650337606,
      "learning_rate": 4.617492902875285e-06,
      "loss": 0.4223,
      "step": 2289
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.825545594668776,
      "learning_rate": 4.617110824898413e-06,
      "loss": 0.2103,
      "step": 2290
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.742342298988285,
      "learning_rate": 4.616728572016223e-06,
      "loss": 0.2531,
      "step": 2291
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.695598242504971,
      "learning_rate": 4.616346144260297e-06,
      "loss": 0.3051,
      "step": 2292
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.814592786507794,
      "learning_rate": 4.615963541662229e-06,
      "loss": 0.2939,
      "step": 2293
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.374129699586291,
      "learning_rate": 4.615580764253627e-06,
      "loss": 0.1709,
      "step": 2294
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.450319677610274,
      "learning_rate": 4.615197812066114e-06,
      "loss": 0.4891,
      "step": 2295
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.671828964175316,
      "learning_rate": 4.614814685131328e-06,
      "loss": 0.3263,
      "step": 2296
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.8585934852270904,
      "learning_rate": 4.614431383480921e-06,
      "loss": 0.325,
      "step": 2297
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.568134624927768,
      "learning_rate": 4.61404790714656e-06,
      "loss": 0.4821,
      "step": 2298
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.530972653551891,
      "learning_rate": 4.613664256159925e-06,
      "loss": 0.3299,
      "step": 2299
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.9097898840566168,
      "learning_rate": 4.613280430552713e-06,
      "loss": 0.304,
      "step": 2300
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5288478430978127,
      "learning_rate": 4.612896430356632e-06,
      "loss": 0.1022,
      "step": 2301
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.035355430124421,
      "learning_rate": 4.612512255603406e-06,
      "loss": 0.248,
      "step": 2302
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.613775466433762,
      "learning_rate": 4.612127906324776e-06,
      "loss": 0.3363,
      "step": 2303
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4746007245928285,
      "learning_rate": 4.611743382552493e-06,
      "loss": 0.3003,
      "step": 2304
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.650705395063684,
      "learning_rate": 4.611358684318324e-06,
      "loss": 0.2783,
      "step": 2305
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.132898582035551,
      "learning_rate": 4.610973811654054e-06,
      "loss": 0.3106,
      "step": 2306
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.1789269521587755,
      "learning_rate": 4.610588764591476e-06,
      "loss": 0.2793,
      "step": 2307
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.930730283465864,
      "learning_rate": 4.610203543162402e-06,
      "loss": 0.3109,
      "step": 2308
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.9545979179780657,
      "learning_rate": 4.609818147398657e-06,
      "loss": 0.3299,
      "step": 2309
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.0627192222549153,
      "learning_rate": 4.60943257733208e-06,
      "loss": 0.106,
      "step": 2310
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.517099387735254,
      "learning_rate": 4.609046832994527e-06,
      "loss": 0.757,
      "step": 2311
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.291992283154247,
      "learning_rate": 4.608660914417864e-06,
      "loss": 0.2748,
      "step": 2312
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.910997062843323,
      "learning_rate": 4.608274821633973e-06,
      "loss": 0.4933,
      "step": 2313
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.060376471546727,
      "learning_rate": 4.607888554674755e-06,
      "loss": 0.3756,
      "step": 2314
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7509106276057627,
      "learning_rate": 4.607502113572117e-06,
      "loss": 0.1148,
      "step": 2315
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.650947930290655,
      "learning_rate": 4.607115498357989e-06,
      "loss": 0.4795,
      "step": 2316
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0973861978207027,
      "learning_rate": 4.606728709064308e-06,
      "loss": 0.0313,
      "step": 2317
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.350813624124402,
      "learning_rate": 4.606341745723029e-06,
      "loss": 0.1989,
      "step": 2318
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.307257461755724,
      "learning_rate": 4.6059546083661225e-06,
      "loss": 0.4934,
      "step": 2319
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7446227503123395,
      "learning_rate": 4.605567297025572e-06,
      "loss": 0.163,
      "step": 2320
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.428918230580227,
      "learning_rate": 4.605179811733373e-06,
      "loss": 0.364,
      "step": 2321
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.766484357736966,
      "learning_rate": 4.60479215252154e-06,
      "loss": 0.4674,
      "step": 2322
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.195418742882745,
      "learning_rate": 4.604404319422099e-06,
      "loss": 0.6585,
      "step": 2323
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.884335608468501,
      "learning_rate": 4.60401631246709e-06,
      "loss": 0.2633,
      "step": 2324
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.915985557691775,
      "learning_rate": 4.603628131688567e-06,
      "loss": 0.4591,
      "step": 2325
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.573059137408107,
      "learning_rate": 4.603239777118604e-06,
      "loss": 0.2969,
      "step": 2326
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.851053505506103,
      "learning_rate": 4.602851248789281e-06,
      "loss": 0.4571,
      "step": 2327
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.404013230062476,
      "learning_rate": 4.602462546732698e-06,
      "loss": 0.533,
      "step": 2328
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.819758349526005,
      "learning_rate": 4.602073670980967e-06,
      "loss": 0.2767,
      "step": 2329
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.634572838881517,
      "learning_rate": 4.601684621566215e-06,
      "loss": 0.3115,
      "step": 2330
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.759605381739065,
      "learning_rate": 4.601295398520583e-06,
      "loss": 0.3633,
      "step": 2331
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.398254062982764,
      "learning_rate": 4.600906001876229e-06,
      "loss": 0.3695,
      "step": 2332
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.523017118989881,
      "learning_rate": 4.600516431665319e-06,
      "loss": 0.4549,
      "step": 2333
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.9612100550398,
      "learning_rate": 4.600126687920041e-06,
      "loss": 0.4674,
      "step": 2334
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.085009906875327,
      "learning_rate": 4.599736770672593e-06,
      "loss": 0.3167,
      "step": 2335
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.40114295466725,
      "learning_rate": 4.599346679955187e-06,
      "loss": 0.263,
      "step": 2336
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.342179242984702,
      "learning_rate": 4.598956415800051e-06,
      "loss": 0.3529,
      "step": 2337
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.817512111507233,
      "learning_rate": 4.598565978239425e-06,
      "loss": 0.2742,
      "step": 2338
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.406452151092829,
      "learning_rate": 4.598175367305568e-06,
      "loss": 0.3762,
      "step": 2339
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.118874850178238,
      "learning_rate": 4.59778458303075e-06,
      "loss": 0.2113,
      "step": 2340
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.977244509639108,
      "learning_rate": 4.597393625447253e-06,
      "loss": 0.3402,
      "step": 2341
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.940638912634737,
      "learning_rate": 4.5970024945873785e-06,
      "loss": 0.5004,
      "step": 2342
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.240054297388842,
      "learning_rate": 4.596611190483438e-06,
      "loss": 0.2121,
      "step": 2343
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.9948980105558265,
      "learning_rate": 4.59621971316776e-06,
      "loss": 0.2607,
      "step": 2344
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.603272118781721,
      "learning_rate": 4.595828062672687e-06,
      "loss": 0.0875,
      "step": 2345
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.159854627245724,
      "learning_rate": 4.595436239030575e-06,
      "loss": 0.4843,
      "step": 2346
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.874036206149626,
      "learning_rate": 4.595044242273794e-06,
      "loss": 0.2138,
      "step": 2347
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.688061701339252,
      "learning_rate": 4.5946520724347295e-06,
      "loss": 0.5043,
      "step": 2348
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.26608825712956,
      "learning_rate": 4.59425972954578e-06,
      "loss": 0.3465,
      "step": 2349
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3763248755561674,
      "learning_rate": 4.593867213639359e-06,
      "loss": 0.2229,
      "step": 2350
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.304640331227816,
      "learning_rate": 4.593474524747895e-06,
      "loss": 0.2491,
      "step": 2351
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.586867239049424,
      "learning_rate": 4.59308166290383e-06,
      "loss": 0.4344,
      "step": 2352
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.417942238114141,
      "learning_rate": 4.592688628139619e-06,
      "loss": 0.4794,
      "step": 2353
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.255721087310218,
      "learning_rate": 4.592295420487733e-06,
      "loss": 0.2217,
      "step": 2354
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.713159028810783,
      "learning_rate": 4.591902039980659e-06,
      "loss": 0.435,
      "step": 2355
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.227193529455461,
      "learning_rate": 4.591508486650893e-06,
      "loss": 0.2965,
      "step": 2356
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.724881522268267,
      "learning_rate": 4.59111476053095e-06,
      "loss": 0.254,
      "step": 2357
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.5076845745138785,
      "learning_rate": 4.590720861653358e-06,
      "loss": 0.4844,
      "step": 2358
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.131278713776767,
      "learning_rate": 4.590326790050658e-06,
      "loss": 0.1793,
      "step": 2359
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.4970127182466,
      "learning_rate": 4.589932545755409e-06,
      "loss": 0.5287,
      "step": 2360
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.335127164365867,
      "learning_rate": 4.589538128800177e-06,
      "loss": 0.3264,
      "step": 2361
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.700023710457166,
      "learning_rate": 4.5891435392175506e-06,
      "loss": 0.1434,
      "step": 2362
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.633793600616888,
      "learning_rate": 4.588748777040127e-06,
      "loss": 0.5244,
      "step": 2363
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.168915583584949,
      "learning_rate": 4.5883538423005206e-06,
      "loss": 0.4286,
      "step": 2364
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.9435348914216797,
      "learning_rate": 4.587958735031359e-06,
      "loss": 0.317,
      "step": 2365
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.430105251878706,
      "learning_rate": 4.587563455265282e-06,
      "loss": 0.3025,
      "step": 2366
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.321755633139072,
      "learning_rate": 4.587168003034948e-06,
      "loss": 0.332,
      "step": 2367
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.747603499422815,
      "learning_rate": 4.586772378373026e-06,
      "loss": 0.5772,
      "step": 2368
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.7717384346947718,
      "learning_rate": 4.5863765813122006e-06,
      "loss": 0.2216,
      "step": 2369
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.270388639144521,
      "learning_rate": 4.585980611885171e-06,
      "loss": 0.3483,
      "step": 2370
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.745820065907126,
      "learning_rate": 4.585584470124651e-06,
      "loss": 0.2941,
      "step": 2371
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.187565840198555,
      "learning_rate": 4.585188156063367e-06,
      "loss": 0.5381,
      "step": 2372
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.08486144555888,
      "learning_rate": 4.58479166973406e-06,
      "loss": 0.277,
      "step": 2373
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.7121466122984113,
      "learning_rate": 4.584395011169487e-06,
      "loss": 0.3171,
      "step": 2374
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.892859709981345,
      "learning_rate": 4.583998180402418e-06,
      "loss": 0.5337,
      "step": 2375
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.374534521120348,
      "learning_rate": 4.583601177465636e-06,
      "loss": 0.2278,
      "step": 2376
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.2405612620027044,
      "learning_rate": 4.58320400239194e-06,
      "loss": 0.5374,
      "step": 2377
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.3264090116420375,
      "learning_rate": 4.582806655214142e-06,
      "loss": 0.3527,
      "step": 2378
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.42996360546406,
      "learning_rate": 4.582409135965071e-06,
      "loss": 0.4978,
      "step": 2379
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.729998469341578,
      "learning_rate": 4.582011444677566e-06,
      "loss": 0.3233,
      "step": 2380
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.644476250317528,
      "learning_rate": 4.581613581384483e-06,
      "loss": 0.3817,
      "step": 2381
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.834199564616082,
      "learning_rate": 4.581215546118691e-06,
      "loss": 0.3621,
      "step": 2382
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.268500278995007,
      "learning_rate": 4.580817338913075e-06,
      "loss": 0.2589,
      "step": 2383
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.940812266504626,
      "learning_rate": 4.580418959800533e-06,
      "loss": 0.5076,
      "step": 2384
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.9223175098156857,
      "learning_rate": 4.580020408813975e-06,
      "loss": 0.1962,
      "step": 2385
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.311488791994285,
      "learning_rate": 4.579621685986328e-06,
      "loss": 0.4649,
      "step": 2386
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.262599717430774,
      "learning_rate": 4.579222791350535e-06,
      "loss": 0.6206,
      "step": 2387
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3120512979153918,
      "learning_rate": 4.578823724939547e-06,
      "loss": 0.2154,
      "step": 2388
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.7982039344921055,
      "learning_rate": 4.578424486786336e-06,
      "loss": 0.2801,
      "step": 2389
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.245709643516535,
      "learning_rate": 4.578025076923883e-06,
      "loss": 0.3131,
      "step": 2390
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.666052928981795,
      "learning_rate": 4.577625495385186e-06,
      "loss": 0.3745,
      "step": 2391
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.195359868122345,
      "learning_rate": 4.577225742203259e-06,
      "loss": 0.2564,
      "step": 2392
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.773943989359534,
      "learning_rate": 4.576825817411123e-06,
      "loss": 0.3032,
      "step": 2393
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.781305415456491,
      "learning_rate": 4.57642572104182e-06,
      "loss": 0.3299,
      "step": 2394
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.274550351133367,
      "learning_rate": 4.576025453128405e-06,
      "loss": 0.1294,
      "step": 2395
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.977933658846086,
      "learning_rate": 4.575625013703945e-06,
      "loss": 0.3839,
      "step": 2396
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5913162156619842,
      "learning_rate": 4.575224402801523e-06,
      "loss": 0.3134,
      "step": 2397
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1488681252684985,
      "learning_rate": 4.574823620454234e-06,
      "loss": 0.054,
      "step": 2398
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.289069541840168,
      "learning_rate": 4.574422666695191e-06,
      "loss": 0.3307,
      "step": 2399
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.008072839818109,
      "learning_rate": 4.574021541557517e-06,
      "loss": 0.3915,
      "step": 2400
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.0398426580651146,
      "learning_rate": 4.573620245074352e-06,
      "loss": 0.5294,
      "step": 2401
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.806734880686079,
      "learning_rate": 4.573218777278849e-06,
      "loss": 0.5202,
      "step": 2402
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.429737609595951,
      "learning_rate": 4.572817138204174e-06,
      "loss": 0.4265,
      "step": 2403
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.287607622963159,
      "learning_rate": 4.572415327883509e-06,
      "loss": 0.1858,
      "step": 2404
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.644149941184763,
      "learning_rate": 4.572013346350052e-06,
      "loss": 0.3641,
      "step": 2405
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.8988177760620335,
      "learning_rate": 4.57161119363701e-06,
      "loss": 0.3649,
      "step": 2406
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.860369313273391,
      "learning_rate": 4.571208869777608e-06,
      "loss": 0.4931,
      "step": 2407
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.09726037813417,
      "learning_rate": 4.570806374805084e-06,
      "loss": 0.054,
      "step": 2408
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.437640947949627,
      "learning_rate": 4.5704037087526885e-06,
      "loss": 0.7189,
      "step": 2409
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.574857516167142,
      "learning_rate": 4.570000871653691e-06,
      "loss": 0.259,
      "step": 2410
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.045191313290229,
      "learning_rate": 4.569597863541368e-06,
      "loss": 0.5365,
      "step": 2411
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.779349666897344,
      "learning_rate": 4.569194684449019e-06,
      "loss": 0.2926,
      "step": 2412
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.445964603473931,
      "learning_rate": 4.568791334409947e-06,
      "loss": 0.3394,
      "step": 2413
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.976585999769182,
      "learning_rate": 4.56838781345748e-06,
      "loss": 0.3471,
      "step": 2414
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.504402893521912,
      "learning_rate": 4.567984121624952e-06,
      "loss": 0.3001,
      "step": 2415
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.840216974035632,
      "learning_rate": 4.567580258945714e-06,
      "loss": 0.2859,
      "step": 2416
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.968548677900109,
      "learning_rate": 4.567176225453133e-06,
      "loss": 0.4253,
      "step": 2417
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.3797173370448705,
      "learning_rate": 4.566772021180586e-06,
      "loss": 0.0967,
      "step": 2418
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.887850560469798,
      "learning_rate": 4.566367646161468e-06,
      "loss": 0.3442,
      "step": 2419
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5060513551596606,
      "learning_rate": 4.565963100429185e-06,
      "loss": 0.1734,
      "step": 2420
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.295393337509885,
      "learning_rate": 4.56555838401716e-06,
      "loss": 0.6304,
      "step": 2421
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.33439632006394,
      "learning_rate": 4.565153496958828e-06,
      "loss": 0.4707,
      "step": 2422
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.037450939477209,
      "learning_rate": 4.564748439287639e-06,
      "loss": 0.2514,
      "step": 2423
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.695297508236177,
      "learning_rate": 4.564343211037058e-06,
      "loss": 0.5419,
      "step": 2424
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.935559127672845,
      "learning_rate": 4.563937812240562e-06,
      "loss": 0.0822,
      "step": 2425
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.093214076438625,
      "learning_rate": 4.563532242931641e-06,
      "loss": 0.6639,
      "step": 2426
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.4934602893029485,
      "learning_rate": 4.563126503143805e-06,
      "loss": 0.2472,
      "step": 2427
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.498472214152876,
      "learning_rate": 4.56272059291057e-06,
      "loss": 0.3528,
      "step": 2428
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.468551795247169,
      "learning_rate": 4.562314512265474e-06,
      "loss": 0.3112,
      "step": 2429
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7517513460993297,
      "learning_rate": 4.561908261242064e-06,
      "loss": 0.112,
      "step": 2430
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.9364331870105715,
      "learning_rate": 4.561501839873902e-06,
      "loss": 0.4943,
      "step": 2431
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.508073633581208,
      "learning_rate": 4.561095248194566e-06,
      "loss": 0.42,
      "step": 2432
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.518280649554243,
      "learning_rate": 4.560688486237645e-06,
      "loss": 0.371,
      "step": 2433
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.500701521779485,
      "learning_rate": 4.560281554036744e-06,
      "loss": 0.1102,
      "step": 2434
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.0635350504469168,
      "learning_rate": 4.559874451625482e-06,
      "loss": 0.1433,
      "step": 2435
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.012998171689021,
      "learning_rate": 4.559467179037492e-06,
      "loss": 0.4351,
      "step": 2436
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.765802979525937,
      "learning_rate": 4.559059736306421e-06,
      "loss": 0.3009,
      "step": 2437
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.617131051852058,
      "learning_rate": 4.558652123465928e-06,
      "loss": 0.4112,
      "step": 2438
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.505531751355133,
      "learning_rate": 4.5582443405496915e-06,
      "loss": 0.4152,
      "step": 2439
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.822397887391677,
      "learning_rate": 4.557836387591399e-06,
      "loss": 0.3129,
      "step": 2440
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.750573402439328,
      "learning_rate": 4.557428264624752e-06,
      "loss": 0.463,
      "step": 2441
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.383734557714368,
      "learning_rate": 4.557019971683469e-06,
      "loss": 0.496,
      "step": 2442
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1106118721439664,
      "learning_rate": 4.556611508801281e-06,
      "loss": 0.0535,
      "step": 2443
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.752859799139099,
      "learning_rate": 4.5562028760119325e-06,
      "loss": 0.4692,
      "step": 2444
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.097708595987667,
      "learning_rate": 4.555794073349185e-06,
      "loss": 0.1662,
      "step": 2445
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.0199573950826184,
      "learning_rate": 4.555385100846809e-06,
      "loss": 0.1439,
      "step": 2446
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.077814009470854,
      "learning_rate": 4.5549759585385925e-06,
      "loss": 0.3293,
      "step": 2447
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.654151992114951,
      "learning_rate": 4.554566646458338e-06,
      "loss": 0.2644,
      "step": 2448
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.461063780825909,
      "learning_rate": 4.55415716463986e-06,
      "loss": 0.2627,
      "step": 2449
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.5926627897312424,
      "learning_rate": 4.553747513116988e-06,
      "loss": 0.2842,
      "step": 2450
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.448335042855297,
      "learning_rate": 4.553337691923566e-06,
      "loss": 0.3338,
      "step": 2451
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3688983217041217,
      "learning_rate": 4.552927701093451e-06,
      "loss": 0.1834,
      "step": 2452
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.9066301396453245,
      "learning_rate": 4.5525175406605135e-06,
      "loss": 0.214,
      "step": 2453
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.874287842939503,
      "learning_rate": 4.55210721065864e-06,
      "loss": 0.2876,
      "step": 2454
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.520396892670774,
      "learning_rate": 4.55169671112173e-06,
      "loss": 0.4314,
      "step": 2455
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.8859209888593584,
      "learning_rate": 4.551286042083696e-06,
      "loss": 0.3275,
      "step": 2456
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.645247322707181,
      "learning_rate": 4.550875203578467e-06,
      "loss": 0.2469,
      "step": 2457
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.090788685764076,
      "learning_rate": 4.550464195639983e-06,
      "loss": 0.4607,
      "step": 2458
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.8097794897513375,
      "learning_rate": 4.5500530183022005e-06,
      "loss": 0.2368,
      "step": 2459
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.917639987706663,
      "learning_rate": 4.5496416715990874e-06,
      "loss": 0.4486,
      "step": 2460
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.739045505897998,
      "learning_rate": 4.5492301555646286e-06,
      "loss": 0.6301,
      "step": 2461
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.466211819458655,
      "learning_rate": 4.548818470232823e-06,
      "loss": 0.1976,
      "step": 2462
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.584829582831233,
      "learning_rate": 4.548406615637678e-06,
      "loss": 0.5217,
      "step": 2463
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.268580999228525,
      "learning_rate": 4.547994591813222e-06,
      "loss": 0.3468,
      "step": 2464
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.02997212299298,
      "learning_rate": 4.547582398793494e-06,
      "loss": 0.5063,
      "step": 2465
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.694704093255404,
      "learning_rate": 4.547170036612546e-06,
      "loss": 0.2648,
      "step": 2466
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.39325480111951,
      "learning_rate": 4.546757505304448e-06,
      "loss": 0.3612,
      "step": 2467
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.319846366158071,
      "learning_rate": 4.546344804903278e-06,
      "loss": 0.3889,
      "step": 2468
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.490113785332644,
      "learning_rate": 4.545931935443133e-06,
      "loss": 0.2037,
      "step": 2469
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.275871797957014,
      "learning_rate": 4.545518896958123e-06,
      "loss": 0.5786,
      "step": 2470
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.880086878511817,
      "learning_rate": 4.54510568948237e-06,
      "loss": 0.1974,
      "step": 2471
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3135354735466587,
      "learning_rate": 4.544692313050011e-06,
      "loss": 0.1443,
      "step": 2472
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.519704501614332,
      "learning_rate": 4.5442787676951985e-06,
      "loss": 0.6047,
      "step": 2473
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.5989772568693175,
      "learning_rate": 4.543865053452096e-06,
      "loss": 0.4239,
      "step": 2474
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.6935011989976303,
      "learning_rate": 4.543451170354884e-06,
      "loss": 0.1443,
      "step": 2475
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7015080010158985,
      "learning_rate": 4.543037118437754e-06,
      "loss": 0.1436,
      "step": 2476
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.341103555909388,
      "learning_rate": 4.542622897734915e-06,
      "loss": 0.3413,
      "step": 2477
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.984682619155498,
      "learning_rate": 4.542208508280586e-06,
      "loss": 0.312,
      "step": 2478
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.186048302550299,
      "learning_rate": 4.541793950109003e-06,
      "loss": 0.297,
      "step": 2479
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.7492588415780226,
      "learning_rate": 4.541379223254413e-06,
      "loss": 0.2464,
      "step": 2480
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.719800169538461,
      "learning_rate": 4.54096432775108e-06,
      "loss": 0.5107,
      "step": 2481
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.07806357115908,
      "learning_rate": 4.540549263633281e-06,
      "loss": 0.2362,
      "step": 2482
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.159841099274561,
      "learning_rate": 4.5401340309353064e-06,
      "loss": 0.2297,
      "step": 2483
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.9386018441116906,
      "learning_rate": 4.5397186296914606e-06,
      "loss": 0.2465,
      "step": 2484
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.087405725494863,
      "learning_rate": 4.539303059936061e-06,
      "loss": 0.0959,
      "step": 2485
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.153964021759104,
      "learning_rate": 4.538887321703441e-06,
      "loss": 0.3455,
      "step": 2486
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.096767318102341,
      "learning_rate": 4.538471415027947e-06,
      "loss": 0.3299,
      "step": 2487
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2511431543160065,
      "learning_rate": 4.538055339943938e-06,
      "loss": 0.1379,
      "step": 2488
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3449034994489595,
      "learning_rate": 4.537639096485789e-06,
      "loss": 0.1025,
      "step": 2489
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.7423964748665774,
      "learning_rate": 4.53722268468789e-06,
      "loss": 0.3456,
      "step": 2490
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.6958556741975235,
      "learning_rate": 4.536806104584639e-06,
      "loss": 0.3748,
      "step": 2491
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.746057270337468,
      "learning_rate": 4.536389356210453e-06,
      "loss": 0.248,
      "step": 2492
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.120546097654068,
      "learning_rate": 4.535972439599763e-06,
      "loss": 0.2554,
      "step": 2493
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.6002752793414725,
      "learning_rate": 4.535555354787011e-06,
      "loss": 0.4052,
      "step": 2494
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.267764983575551,
      "learning_rate": 4.535138101806657e-06,
      "loss": 0.4451,
      "step": 2495
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.448871508987012,
      "learning_rate": 4.53472068069317e-06,
      "loss": 0.4135,
      "step": 2496
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.397062858096529,
      "learning_rate": 4.534303091481035e-06,
      "loss": 0.235,
      "step": 2497
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.538636734161793,
      "learning_rate": 4.5338853342047536e-06,
      "loss": 0.367,
      "step": 2498
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.1473474319842,
      "learning_rate": 4.533467408898838e-06,
      "loss": 0.4207,
      "step": 2499
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.909406704357924,
      "learning_rate": 4.533049315597813e-06,
      "loss": 0.3972,
      "step": 2500
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.575784807880218,
      "learning_rate": 4.532631054336221e-06,
      "loss": 0.361,
      "step": 2501
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.4634947964286,
      "learning_rate": 4.532212625148618e-06,
      "loss": 0.4099,
      "step": 2502
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.8308391675349664,
      "learning_rate": 4.5317940280695705e-06,
      "loss": 0.2132,
      "step": 2503
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.010805856523935,
      "learning_rate": 4.531375263133662e-06,
      "loss": 0.277,
      "step": 2504
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.51060010181404,
      "learning_rate": 4.530956330375489e-06,
      "loss": 0.1377,
      "step": 2505
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7866117052690194,
      "learning_rate": 4.53053722982966e-06,
      "loss": 0.1943,
      "step": 2506
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.279993702722202,
      "learning_rate": 4.530117961530802e-06,
      "loss": 0.4861,
      "step": 2507
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.628692256249441,
      "learning_rate": 4.52969852551355e-06,
      "loss": 0.31,
      "step": 2508
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.205293160598827,
      "learning_rate": 4.529278921812557e-06,
      "loss": 0.3874,
      "step": 2509
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.797264169983151,
      "learning_rate": 4.528859150462489e-06,
      "loss": 0.4431,
      "step": 2510
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.096835735735446,
      "learning_rate": 4.528439211498025e-06,
      "loss": 0.2474,
      "step": 2511
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.655250437874755,
      "learning_rate": 4.528019104953857e-06,
      "loss": 0.5724,
      "step": 2512
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.822316415357764,
      "learning_rate": 4.527598830864694e-06,
      "loss": 0.4757,
      "step": 2513
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.031614391796774,
      "learning_rate": 4.527178389265257e-06,
      "loss": 0.1592,
      "step": 2514
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.0530779915772936,
      "learning_rate": 4.526757780190279e-06,
      "loss": 0.1064,
      "step": 2515
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.126746929396739,
      "learning_rate": 4.526337003674512e-06,
      "loss": 0.3088,
      "step": 2516
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7150110563664875,
      "learning_rate": 4.525916059752715e-06,
      "loss": 0.1645,
      "step": 2517
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.624718615691014,
      "learning_rate": 4.525494948459665e-06,
      "loss": 0.2295,
      "step": 2518
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.6205570412524377,
      "learning_rate": 4.5250736698301535e-06,
      "loss": 0.2935,
      "step": 2519
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.924923924862713,
      "learning_rate": 4.524652223898984e-06,
      "loss": 0.2177,
      "step": 2520
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.049458938799767,
      "learning_rate": 4.524230610700974e-06,
      "loss": 0.4327,
      "step": 2521
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.015631486695566,
      "learning_rate": 4.523808830270955e-06,
      "loss": 0.2793,
      "step": 2522
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.728552014520234,
      "learning_rate": 4.523386882643772e-06,
      "loss": 0.3034,
      "step": 2523
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6818368517412432,
      "learning_rate": 4.5229647678542864e-06,
      "loss": 0.0371,
      "step": 2524
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.510169171655774,
      "learning_rate": 4.522542485937369e-06,
      "loss": 0.1726,
      "step": 2525
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.649250239550835,
      "learning_rate": 4.522120036927907e-06,
      "loss": 0.3692,
      "step": 2526
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.6223308107965293,
      "learning_rate": 4.521697420860802e-06,
      "loss": 0.1335,
      "step": 2527
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.505779976943792,
      "learning_rate": 4.521274637770968e-06,
      "loss": 0.4247,
      "step": 2528
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.814103234785583,
      "learning_rate": 4.520851687693333e-06,
      "loss": 0.5863,
      "step": 2529
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.373827540226567,
      "learning_rate": 4.520428570662839e-06,
      "loss": 0.6331,
      "step": 2530
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.7507206719627035,
      "learning_rate": 4.5200052867144425e-06,
      "loss": 0.2936,
      "step": 2531
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.746936370587256,
      "learning_rate": 4.5195818358831125e-06,
      "loss": 0.3216,
      "step": 2532
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.912204888295757,
      "learning_rate": 4.519158218203833e-06,
      "loss": 0.4273,
      "step": 2533
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.38287865541206,
      "learning_rate": 4.518734433711601e-06,
      "loss": 0.2369,
      "step": 2534
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.412146124529907,
      "learning_rate": 4.518310482441427e-06,
      "loss": 0.358,
      "step": 2535
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.603724093844889,
      "learning_rate": 4.517886364428337e-06,
      "loss": 0.5348,
      "step": 2536
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.3627346981556965,
      "learning_rate": 4.517462079707368e-06,
      "loss": 0.3321,
      "step": 2537
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.7835162573515353,
      "learning_rate": 4.517037628313572e-06,
      "loss": 0.0979,
      "step": 2538
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.8604345026399605,
      "learning_rate": 4.516613010282018e-06,
      "loss": 0.429,
      "step": 2539
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.976899885309566,
      "learning_rate": 4.516188225647784e-06,
      "loss": 0.3257,
      "step": 2540
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.237906993726197,
      "learning_rate": 4.5157632744459624e-06,
      "loss": 0.5386,
      "step": 2541
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.1076086742558795,
      "learning_rate": 4.515338156711662e-06,
      "loss": 0.4785,
      "step": 2542
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.266831132969493,
      "learning_rate": 4.514912872480004e-06,
      "loss": 0.4869,
      "step": 2543
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.195177602287892,
      "learning_rate": 4.514487421786123e-06,
      "loss": 0.3341,
      "step": 2544
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.031735852452952,
      "learning_rate": 4.514061804665169e-06,
      "loss": 0.3789,
      "step": 2545
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.501603108403889,
      "learning_rate": 4.513636021152301e-06,
      "loss": 0.4518,
      "step": 2546
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.092457031678811,
      "learning_rate": 4.513210071282699e-06,
      "loss": 0.1922,
      "step": 2547
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.172445810946436,
      "learning_rate": 4.51278395509155e-06,
      "loss": 0.1436,
      "step": 2548
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.446500325022914,
      "learning_rate": 4.512357672614059e-06,
      "loss": 0.3721,
      "step": 2549
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.710816537194437,
      "learning_rate": 4.5119312238854444e-06,
      "loss": 0.3682,
      "step": 2550
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.26513240046364,
      "learning_rate": 4.511504608940934e-06,
      "loss": 0.3466,
      "step": 2551
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.184553405856765,
      "learning_rate": 4.511077827815777e-06,
      "loss": 0.3154,
      "step": 2552
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.945422430585286,
      "learning_rate": 4.510650880545228e-06,
      "loss": 0.0947,
      "step": 2553
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.4958988759682095,
      "learning_rate": 4.510223767164561e-06,
      "loss": 0.2567,
      "step": 2554
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.100000352132465,
      "learning_rate": 4.5097964877090615e-06,
      "loss": 0.6382,
      "step": 2555
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.857275665537264,
      "learning_rate": 4.5093690422140305e-06,
      "loss": 0.3945,
      "step": 2556
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.212806197964467,
      "learning_rate": 4.5089414307147805e-06,
      "loss": 0.2302,
      "step": 2557
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.730294106297558,
      "learning_rate": 4.508513653246639e-06,
      "loss": 0.1633,
      "step": 2558
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.410943248345171,
      "learning_rate": 4.5080857098449465e-06,
      "loss": 0.4343,
      "step": 2559
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1262101661612087,
      "learning_rate": 4.507657600545058e-06,
      "loss": 0.0411,
      "step": 2560
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.554956670853777,
      "learning_rate": 4.5072293253823416e-06,
      "loss": 0.3277,
      "step": 2561
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.977319099930788,
      "learning_rate": 4.506800884392179e-06,
      "loss": 0.209,
      "step": 2562
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.4773213307574555,
      "learning_rate": 4.506372277609965e-06,
      "loss": 0.2533,
      "step": 2563
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.06149041966314,
      "learning_rate": 4.505943505071112e-06,
      "loss": 0.3031,
      "step": 2564
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.350458967655684,
      "learning_rate": 4.505514566811041e-06,
      "loss": 0.3266,
      "step": 2565
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8827592169070484,
      "learning_rate": 4.505085462865188e-06,
      "loss": 0.3155,
      "step": 2566
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.010542658597646,
      "learning_rate": 4.504656193269005e-06,
      "loss": 0.1217,
      "step": 2567
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.648825716139349,
      "learning_rate": 4.5042267580579556e-06,
      "loss": 0.2716,
      "step": 2568
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.606174003117234,
      "learning_rate": 4.503797157267517e-06,
      "loss": 0.4654,
      "step": 2569
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.5610353639497045,
      "learning_rate": 4.503367390933183e-06,
      "loss": 0.3814,
      "step": 2570
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8912785296201937,
      "learning_rate": 4.502937459090456e-06,
      "loss": 0.3737,
      "step": 2571
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.216816656872547,
      "learning_rate": 4.502507361774856e-06,
      "loss": 0.233,
      "step": 2572
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4893242259068713,
      "learning_rate": 4.502077099021915e-06,
      "loss": 0.1847,
      "step": 2573
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.310148849445667,
      "learning_rate": 4.501646670867182e-06,
      "loss": 0.2664,
      "step": 2574
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.251872605913346,
      "learning_rate": 4.501216077346212e-06,
      "loss": 0.2917,
      "step": 2575
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.266586690874091,
      "learning_rate": 4.500785318494582e-06,
      "loss": 0.3484,
      "step": 2576
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.30666580797211,
      "learning_rate": 4.500354394347879e-06,
      "loss": 0.1721,
      "step": 2577
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.570428722385268,
      "learning_rate": 4.499923304941702e-06,
      "loss": 0.4946,
      "step": 2578
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.5173132246759584,
      "learning_rate": 4.499492050311667e-06,
      "loss": 0.1498,
      "step": 2579
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.7459544682271195,
      "learning_rate": 4.499060630493403e-06,
      "loss": 0.3441,
      "step": 2580
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.393734634183107,
      "learning_rate": 4.498629045522549e-06,
      "loss": 0.2862,
      "step": 2581
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.797670051710912,
      "learning_rate": 4.498197295434762e-06,
      "loss": 0.2708,
      "step": 2582
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.437973312861167,
      "learning_rate": 4.497765380265711e-06,
      "loss": 0.5065,
      "step": 2583
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.040015276973492,
      "learning_rate": 4.497333300051081e-06,
      "loss": 0.2735,
      "step": 2584
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8806854315831787,
      "learning_rate": 4.496901054826564e-06,
      "loss": 0.2721,
      "step": 2585
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4294698044596226,
      "learning_rate": 4.496468644627873e-06,
      "loss": 0.1428,
      "step": 2586
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.224993163202815,
      "learning_rate": 4.49603606949073e-06,
      "loss": 0.2093,
      "step": 2587
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.7625778612328125,
      "learning_rate": 4.495603329450874e-06,
      "loss": 0.4614,
      "step": 2588
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.569338871042302,
      "learning_rate": 4.495170424544054e-06,
      "loss": 0.2512,
      "step": 2589
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.44824245306576,
      "learning_rate": 4.494737354806036e-06,
      "loss": 0.4241,
      "step": 2590
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.0755506568648805,
      "learning_rate": 4.494304120272596e-06,
      "loss": 0.3206,
      "step": 2591
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.108141678194941,
      "learning_rate": 4.493870720979529e-06,
      "loss": 0.5026,
      "step": 2592
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.6948900764809856,
      "learning_rate": 4.493437156962637e-06,
      "loss": 0.3334,
      "step": 2593
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.519371629446174,
      "learning_rate": 4.493003428257742e-06,
      "loss": 0.3704,
      "step": 2594
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.190709819818836,
      "learning_rate": 4.492569534900674e-06,
      "loss": 0.1287,
      "step": 2595
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8884598038892175,
      "learning_rate": 4.492135476927279e-06,
      "loss": 0.2747,
      "step": 2596
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.53835615651905,
      "learning_rate": 4.49170125437342e-06,
      "loss": 0.2193,
      "step": 2597
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.831222181036991,
      "learning_rate": 4.491266867274966e-06,
      "loss": 0.5643,
      "step": 2598
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.36524652145152,
      "learning_rate": 4.490832315667807e-06,
      "loss": 0.3087,
      "step": 2599
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.307909772731976,
      "learning_rate": 4.490397599587843e-06,
      "loss": 0.4343,
      "step": 2600
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.52552379476464,
      "learning_rate": 4.4899627190709875e-06,
      "loss": 0.5383,
      "step": 2601
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.36606360057587,
      "learning_rate": 4.489527674153168e-06,
      "loss": 0.3808,
      "step": 2602
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.6365503113546316,
      "learning_rate": 4.489092464870327e-06,
      "loss": 0.3126,
      "step": 2603
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.294811297854454,
      "learning_rate": 4.488657091258417e-06,
      "loss": 0.3061,
      "step": 2604
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.987781288056277,
      "learning_rate": 4.48822155335341e-06,
      "loss": 0.2188,
      "step": 2605
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.931641413222636,
      "learning_rate": 4.487785851191285e-06,
      "loss": 0.404,
      "step": 2606
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.379908923974259,
      "learning_rate": 4.487349984808038e-06,
      "loss": 0.5288,
      "step": 2607
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.817499576317827,
      "learning_rate": 4.48691395423968e-06,
      "loss": 0.3563,
      "step": 2608
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.246624688062002,
      "learning_rate": 4.486477759522231e-06,
      "loss": 0.4249,
      "step": 2609
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9320532592363366,
      "learning_rate": 4.48604140069173e-06,
      "loss": 0.1637,
      "step": 2610
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.602544038309347,
      "learning_rate": 4.485604877784224e-06,
      "loss": 0.4206,
      "step": 2611
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.4235267950468633,
      "learning_rate": 4.485168190835779e-06,
      "loss": 0.2511,
      "step": 2612
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.457038259579362,
      "learning_rate": 4.484731339882471e-06,
      "loss": 0.6449,
      "step": 2613
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.950250123996869,
      "learning_rate": 4.48429432496039e-06,
      "loss": 0.3367,
      "step": 2614
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8783829046791403,
      "learning_rate": 4.48385714610564e-06,
      "loss": 0.3351,
      "step": 2615
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.379481358340445,
      "learning_rate": 4.483419803354339e-06,
      "loss": 0.4498,
      "step": 2616
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.084268088226198,
      "learning_rate": 4.482982296742619e-06,
      "loss": 0.2195,
      "step": 2617
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.51076039396741,
      "learning_rate": 4.482544626306622e-06,
      "loss": 0.4928,
      "step": 2618
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.6658708568103355,
      "learning_rate": 4.482106792082509e-06,
      "loss": 0.4801,
      "step": 2619
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.242478594660299,
      "learning_rate": 4.48166879410645e-06,
      "loss": 0.4601,
      "step": 2620
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.292694326424632,
      "learning_rate": 4.481230632414632e-06,
      "loss": 0.5133,
      "step": 2621
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.17352291967602,
      "learning_rate": 4.480792307043252e-06,
      "loss": 0.5321,
      "step": 2622
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.346187467925938,
      "learning_rate": 4.480353818028522e-06,
      "loss": 0.2955,
      "step": 2623
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.064232287691346,
      "learning_rate": 4.479915165406669e-06,
      "loss": 0.3673,
      "step": 2624
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.027042270463525,
      "learning_rate": 4.479476349213932e-06,
      "loss": 0.3739,
      "step": 2625
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.074303252349789,
      "learning_rate": 4.479037369486565e-06,
      "loss": 0.4296,
      "step": 2626
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.9187914975832188,
      "learning_rate": 4.478598226260833e-06,
      "loss": 0.3258,
      "step": 2627
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.364508524905326,
      "learning_rate": 4.478158919573015e-06,
      "loss": 0.2948,
      "step": 2628
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.035650545083041,
      "learning_rate": 4.477719449459405e-06,
      "loss": 0.6076,
      "step": 2629
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8952208879880126,
      "learning_rate": 4.4772798159563125e-06,
      "loss": 0.2106,
      "step": 2630
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.6979443096830193,
      "learning_rate": 4.4768400191000546e-06,
      "loss": 0.3359,
      "step": 2631
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.021776337044665,
      "learning_rate": 4.476400058926967e-06,
      "loss": 0.3069,
      "step": 2632
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.50841714498847,
      "learning_rate": 4.475959935473394e-06,
      "loss": 0.3185,
      "step": 2633
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.731708955451524,
      "learning_rate": 4.475519648775701e-06,
      "loss": 0.7162,
      "step": 2634
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.3601336684095005,
      "learning_rate": 4.475079198870259e-06,
      "loss": 0.2997,
      "step": 2635
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.968500353436965,
      "learning_rate": 4.474638585793457e-06,
      "loss": 0.6258,
      "step": 2636
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.005316032085361,
      "learning_rate": 4.474197809581696e-06,
      "loss": 0.6456,
      "step": 2637
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.090807773752883,
      "learning_rate": 4.473756870271392e-06,
      "loss": 0.4771,
      "step": 2638
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.282926028951543,
      "learning_rate": 4.4733157678989715e-06,
      "loss": 0.2711,
      "step": 2639
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.241966258789398,
      "learning_rate": 4.472874502500877e-06,
      "loss": 0.3849,
      "step": 2640
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.632628286389102,
      "learning_rate": 4.4724330741135635e-06,
      "loss": 0.1487,
      "step": 2641
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.3089409908974545,
      "learning_rate": 4.471991482773499e-06,
      "loss": 0.4758,
      "step": 2642
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.38829449464978,
      "learning_rate": 4.471549728517167e-06,
      "loss": 0.5024,
      "step": 2643
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.6474438176221367,
      "learning_rate": 4.471107811381063e-06,
      "loss": 0.2232,
      "step": 2644
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.826028656621426,
      "learning_rate": 4.470665731401695e-06,
      "loss": 0.5012,
      "step": 2645
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.371395685057136,
      "learning_rate": 4.470223488615586e-06,
      "loss": 0.4418,
      "step": 2646
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.95489738836244,
      "learning_rate": 4.469781083059271e-06,
      "loss": 0.2985,
      "step": 2647
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.8731648621107957,
      "learning_rate": 4.469338514769301e-06,
      "loss": 0.3337,
      "step": 2648
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.540491723889279,
      "learning_rate": 4.4688957837822385e-06,
      "loss": 0.243,
      "step": 2649
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.8269401493535495,
      "learning_rate": 4.468452890134659e-06,
      "loss": 0.4223,
      "step": 2650
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.970994693641387,
      "learning_rate": 4.468009833863153e-06,
      "loss": 0.2213,
      "step": 2651
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.863517170536006,
      "learning_rate": 4.4675666150043215e-06,
      "loss": 0.0701,
      "step": 2652
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.430362962417776,
      "learning_rate": 4.467123233594783e-06,
      "loss": 0.2968,
      "step": 2653
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.139990632111642,
      "learning_rate": 4.466679689671168e-06,
      "loss": 0.2454,
      "step": 2654
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.229691178519221,
      "learning_rate": 4.46623598327012e-06,
      "loss": 0.1877,
      "step": 2655
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.750930628059434,
      "learning_rate": 4.465792114428293e-06,
      "loss": 0.4056,
      "step": 2656
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.356506916181838,
      "learning_rate": 4.465348083182361e-06,
      "loss": 0.405,
      "step": 2657
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.735352959390431,
      "learning_rate": 4.464903889569006e-06,
      "loss": 0.3145,
      "step": 2658
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.642619325662249,
      "learning_rate": 4.464459533624924e-06,
      "loss": 0.3597,
      "step": 2659
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.293522558754025,
      "learning_rate": 4.464015015386826e-06,
      "loss": 0.2366,
      "step": 2660
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.64970909568204,
      "learning_rate": 4.463570334891438e-06,
      "loss": 0.495,
      "step": 2661
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.729352443426049,
      "learning_rate": 4.463125492175495e-06,
      "loss": 0.3086,
      "step": 2662
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3778684935833043,
      "learning_rate": 4.462680487275749e-06,
      "loss": 0.0588,
      "step": 2663
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.760965861048904,
      "learning_rate": 4.462235320228964e-06,
      "loss": 0.1551,
      "step": 2664
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.839970359515806,
      "learning_rate": 4.4617899910719165e-06,
      "loss": 0.5367,
      "step": 2665
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.926103086350319,
      "learning_rate": 4.461344499841398e-06,
      "loss": 0.6641,
      "step": 2666
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.438661565900348,
      "learning_rate": 4.460898846574213e-06,
      "loss": 0.1763,
      "step": 2667
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.0633766467158705,
      "learning_rate": 4.4604530313071785e-06,
      "loss": 0.2168,
      "step": 2668
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.568218407292066,
      "learning_rate": 4.460007054077127e-06,
      "loss": 0.1936,
      "step": 2669
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.3815176821531,
      "learning_rate": 4.459560914920902e-06,
      "loss": 0.2579,
      "step": 2670
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.662645657604775,
      "learning_rate": 4.459114613875361e-06,
      "loss": 0.3074,
      "step": 2671
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.731396127125328,
      "learning_rate": 4.458668150977375e-06,
      "loss": 0.432,
      "step": 2672
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.250866486697598,
      "learning_rate": 4.458221526263831e-06,
      "loss": 0.2043,
      "step": 2673
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.113049737023145,
      "learning_rate": 4.457774739771624e-06,
      "loss": 0.2428,
      "step": 2674
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8937308941247566,
      "learning_rate": 4.457327791537667e-06,
      "loss": 0.1089,
      "step": 2675
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.222451442565336,
      "learning_rate": 4.456880681598883e-06,
      "loss": 0.6351,
      "step": 2676
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.2600284351664293,
      "learning_rate": 4.4564334099922125e-06,
      "loss": 0.2325,
      "step": 2677
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.636343275900572,
      "learning_rate": 4.455985976754604e-06,
      "loss": 0.3776,
      "step": 2678
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.163470707730324,
      "learning_rate": 4.4555383819230255e-06,
      "loss": 0.4362,
      "step": 2679
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.485732272591353,
      "learning_rate": 4.4550906255344525e-06,
      "loss": 0.1974,
      "step": 2680
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.069574232452924,
      "learning_rate": 4.4546427076258766e-06,
      "loss": 0.3643,
      "step": 2681
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.692179553973982,
      "learning_rate": 4.454194628234305e-06,
      "loss": 0.4916,
      "step": 2682
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.253179323252346,
      "learning_rate": 4.453746387396753e-06,
      "loss": 0.3534,
      "step": 2683
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8995252791567983,
      "learning_rate": 4.4532979851502535e-06,
      "loss": 0.1506,
      "step": 2684
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.207260358338526,
      "learning_rate": 4.452849421531851e-06,
      "loss": 0.2565,
      "step": 2685
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.1897778039372655,
      "learning_rate": 4.452400696578603e-06,
      "loss": 0.4069,
      "step": 2686
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.320319372558936,
      "learning_rate": 4.451951810327583e-06,
      "loss": 0.675,
      "step": 2687
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.517012317980802,
      "learning_rate": 4.4515027628158724e-06,
      "loss": 0.4481,
      "step": 2688
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.273173970284805,
      "learning_rate": 4.451053554080572e-06,
      "loss": 0.3957,
      "step": 2689
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.960515597187139,
      "learning_rate": 4.450604184158793e-06,
      "loss": 0.1814,
      "step": 2690
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.819624588745075,
      "learning_rate": 4.450154653087659e-06,
      "loss": 0.2719,
      "step": 2691
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.568321244469029,
      "learning_rate": 4.4497049609043095e-06,
      "loss": 0.3602,
      "step": 2692
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.323845223695206,
      "learning_rate": 4.449255107645894e-06,
      "loss": 0.2768,
      "step": 2693
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.10937306167472,
      "learning_rate": 4.448805093349578e-06,
      "loss": 0.414,
      "step": 2694
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.1270616867633745,
      "learning_rate": 4.448354918052541e-06,
      "loss": 0.1931,
      "step": 2695
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.717720593329794,
      "learning_rate": 4.447904581791972e-06,
      "loss": 0.2511,
      "step": 2696
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.4830926694743285,
      "learning_rate": 4.447454084605076e-06,
      "loss": 0.1404,
      "step": 2697
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.012465174476773,
      "learning_rate": 4.447003426529073e-06,
      "loss": 0.3406,
      "step": 2698
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.674083141439761,
      "learning_rate": 4.4465526076011914e-06,
      "loss": 0.3738,
      "step": 2699
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.095192834334127,
      "learning_rate": 4.446101627858678e-06,
      "loss": 0.5321,
      "step": 2700
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.301679018744749,
      "learning_rate": 4.445650487338787e-06,
      "loss": 0.3119,
      "step": 2701
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.266626334133536,
      "learning_rate": 4.445199186078793e-06,
      "loss": 0.2801,
      "step": 2702
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.5259342636600195,
      "learning_rate": 4.44474772411598e-06,
      "loss": 0.3911,
      "step": 2703
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.6178265970549113,
      "learning_rate": 4.444296101487644e-06,
      "loss": 0.124,
      "step": 2704
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.523787176530518,
      "learning_rate": 4.443844318231097e-06,
      "loss": 0.4471,
      "step": 2705
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.906661060279192,
      "learning_rate": 4.44339237438366e-06,
      "loss": 0.528,
      "step": 2706
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.720165073959827,
      "learning_rate": 4.442940269982676e-06,
      "loss": 0.4006,
      "step": 2707
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.539995378918902,
      "learning_rate": 4.4424880050654916e-06,
      "loss": 0.4809,
      "step": 2708
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.626536707536507,
      "learning_rate": 4.442035579669471e-06,
      "loss": 0.4143,
      "step": 2709
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.9458234338219587,
      "learning_rate": 4.441582993831993e-06,
      "loss": 0.2511,
      "step": 2710
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.651351815054388,
      "learning_rate": 4.441130247590446e-06,
      "loss": 0.1845,
      "step": 2711
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.962991060309609,
      "learning_rate": 4.4406773409822356e-06,
      "loss": 0.5375,
      "step": 2712
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.902819610433464,
      "learning_rate": 4.4402242740447784e-06,
      "loss": 0.2715,
      "step": 2713
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8646879353520878,
      "learning_rate": 4.4397710468155035e-06,
      "loss": 0.1283,
      "step": 2714
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.078363652689547,
      "learning_rate": 4.439317659331855e-06,
      "loss": 0.3085,
      "step": 2715
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.158215557719309,
      "learning_rate": 4.438864111631289e-06,
      "loss": 0.1622,
      "step": 2716
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.716028490742616,
      "learning_rate": 4.438410403751276e-06,
      "loss": 0.4421,
      "step": 2717
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.4217762830253085,
      "learning_rate": 4.4379565357293e-06,
      "loss": 0.6313,
      "step": 2718
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6723302592790965,
      "learning_rate": 4.4375025076028555e-06,
      "loss": 0.1397,
      "step": 2719
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.933052936298347,
      "learning_rate": 4.437048319409453e-06,
      "loss": 0.3504,
      "step": 2720
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.589052827067315,
      "learning_rate": 4.436593971186614e-06,
      "loss": 0.0134,
      "step": 2721
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.213713770278742,
      "learning_rate": 4.436139462971878e-06,
      "loss": 0.2925,
      "step": 2722
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.832894166070995,
      "learning_rate": 4.43568479480279e-06,
      "loss": 0.4347,
      "step": 2723
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9521123978292527,
      "learning_rate": 4.435229966716915e-06,
      "loss": 0.197,
      "step": 2724
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.570198769856312,
      "learning_rate": 4.434774978751828e-06,
      "loss": 0.2097,
      "step": 2725
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.360252059668541,
      "learning_rate": 4.434319830945119e-06,
      "loss": 0.4225,
      "step": 2726
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1607124937962743,
      "learning_rate": 4.433864523334388e-06,
      "loss": 0.1218,
      "step": 2727
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.616321233064526,
      "learning_rate": 4.4334090559572515e-06,
      "loss": 0.3114,
      "step": 2728
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.072216511968794,
      "learning_rate": 4.432953428851337e-06,
      "loss": 0.1892,
      "step": 2729
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.219675877344043,
      "learning_rate": 4.432497642054289e-06,
      "loss": 0.3006,
      "step": 2730
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.130184144953086,
      "learning_rate": 4.4320416956037586e-06,
      "loss": 0.4063,
      "step": 2731
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.103543708384295,
      "learning_rate": 4.431585589537416e-06,
      "loss": 0.5693,
      "step": 2732
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.497935922013931,
      "learning_rate": 4.431129323892942e-06,
      "loss": 0.3618,
      "step": 2733
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.3464347249606,
      "learning_rate": 4.430672898708032e-06,
      "loss": 0.1389,
      "step": 2734
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.899966069367769,
      "learning_rate": 4.4302163140203915e-06,
      "loss": 0.2405,
      "step": 2735
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.7418409344133816,
      "learning_rate": 4.429759569867743e-06,
      "loss": 0.2033,
      "step": 2736
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.917398693313885,
      "learning_rate": 4.4293026662878195e-06,
      "loss": 0.3229,
      "step": 2737
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.074042339457402,
      "learning_rate": 4.428845603318369e-06,
      "loss": 0.2269,
      "step": 2738
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.44219185791392,
      "learning_rate": 4.428388380997151e-06,
      "loss": 0.2833,
      "step": 2739
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3659165198515093,
      "learning_rate": 4.42793099936194e-06,
      "loss": 0.1433,
      "step": 2740
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.519687622187596,
      "learning_rate": 4.427473458450521e-06,
      "loss": 0.5453,
      "step": 2741
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.1568798989362126,
      "learning_rate": 4.427015758300696e-06,
      "loss": 0.2068,
      "step": 2742
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.458688570877728,
      "learning_rate": 4.426557898950276e-06,
      "loss": 0.1226,
      "step": 2743
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.40426535066736,
      "learning_rate": 4.426099880437088e-06,
      "loss": 0.3333,
      "step": 2744
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.104902599614021,
      "learning_rate": 4.425641702798971e-06,
      "loss": 0.4036,
      "step": 2745
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.9481272799001426,
      "learning_rate": 4.425183366073776e-06,
      "loss": 0.1722,
      "step": 2746
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9254557929970466,
      "learning_rate": 4.424724870299372e-06,
      "loss": 0.0997,
      "step": 2747
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.354208011795583,
      "learning_rate": 4.424266215513635e-06,
      "loss": 0.2523,
      "step": 2748
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.552091475829473,
      "learning_rate": 4.423807401754457e-06,
      "loss": 0.3315,
      "step": 2749
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.517787143360362,
      "learning_rate": 4.423348429059744e-06,
      "loss": 0.4006,
      "step": 2750
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.432676430559543,
      "learning_rate": 4.422889297467413e-06,
      "loss": 0.3052,
      "step": 2751
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.049151955121986,
      "learning_rate": 4.422430007015396e-06,
      "loss": 0.227,
      "step": 2752
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.8854347036445205,
      "learning_rate": 4.421970557741637e-06,
      "loss": 0.2308,
      "step": 2753
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.270713854813989,
      "learning_rate": 4.421510949684092e-06,
      "loss": 0.4454,
      "step": 2754
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.812528942752473,
      "learning_rate": 4.421051182880735e-06,
      "loss": 0.4046,
      "step": 2755
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.0043120760332815,
      "learning_rate": 4.420591257369547e-06,
      "loss": 0.6617,
      "step": 2756
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.477506928763934,
      "learning_rate": 4.4201311731885255e-06,
      "loss": 0.6125,
      "step": 2757
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.967753225259135,
      "learning_rate": 4.41967093037568e-06,
      "loss": 0.1441,
      "step": 2758
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.267528042084603,
      "learning_rate": 4.4192105289690345e-06,
      "loss": 0.3569,
      "step": 2759
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.021301311992869,
      "learning_rate": 4.418749969006624e-06,
      "loss": 0.237,
      "step": 2760
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.122875458631328,
      "learning_rate": 4.418289250526498e-06,
      "loss": 0.2196,
      "step": 2761
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.589116542206918,
      "learning_rate": 4.417828373566719e-06,
      "loss": 0.5338,
      "step": 2762
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.160369642677421,
      "learning_rate": 4.417367338165363e-06,
      "loss": 0.4124,
      "step": 2763
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.502901306362837,
      "learning_rate": 4.416906144360518e-06,
      "loss": 0.2299,
      "step": 2764
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.680616158736799,
      "learning_rate": 4.416444792190285e-06,
      "loss": 0.4926,
      "step": 2765
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.600690443702558,
      "learning_rate": 4.4159832816927785e-06,
      "loss": 0.3244,
      "step": 2766
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.33043757941139,
      "learning_rate": 4.4155216129061274e-06,
      "loss": 0.5248,
      "step": 2767
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.022769607696459,
      "learning_rate": 4.415059785868472e-06,
      "loss": 0.3591,
      "step": 2768
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.4967292622471,
      "learning_rate": 4.414597800617966e-06,
      "loss": 0.296,
      "step": 2769
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.150818711591611,
      "learning_rate": 4.414135657192778e-06,
      "loss": 0.3834,
      "step": 2770
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7487105899376134,
      "learning_rate": 4.413673355631084e-06,
      "loss": 0.182,
      "step": 2771
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.0824746321045,
      "learning_rate": 4.413210895971081e-06,
      "loss": 0.3451,
      "step": 2772
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3525486881651391,
      "learning_rate": 4.412748278250973e-06,
      "loss": 0.0048,
      "step": 2773
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.939868914991588,
      "learning_rate": 4.412285502508981e-06,
      "loss": 0.5725,
      "step": 2774
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.054896126508766,
      "learning_rate": 4.411822568783335e-06,
      "loss": 0.1748,
      "step": 2775
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.219699952353087,
      "learning_rate": 4.411359477112281e-06,
      "loss": 0.4318,
      "step": 2776
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.034328107482604,
      "learning_rate": 4.41089622753408e-06,
      "loss": 0.3486,
      "step": 2777
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.029370813087489,
      "learning_rate": 4.410432820087e-06,
      "loss": 0.3647,
      "step": 2778
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.792293975694076,
      "learning_rate": 4.409969254809327e-06,
      "loss": 0.3532,
      "step": 2779
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.297316402669383,
      "learning_rate": 4.409505531739359e-06,
      "loss": 0.2413,
      "step": 2780
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.269819736120861,
      "learning_rate": 4.409041650915404e-06,
      "loss": 0.4307,
      "step": 2781
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.319604765070743,
      "learning_rate": 4.408577612375788e-06,
      "loss": 0.3062,
      "step": 2782
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.221063187215686,
      "learning_rate": 4.408113416158848e-06,
      "loss": 0.5546,
      "step": 2783
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.474223167035721,
      "learning_rate": 4.407649062302932e-06,
      "loss": 0.3678,
      "step": 2784
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.383967018949842,
      "learning_rate": 4.407184550846403e-06,
      "loss": 0.3485,
      "step": 2785
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.464170281059849,
      "learning_rate": 4.406719881827637e-06,
      "loss": 0.3392,
      "step": 2786
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.7188465631862675,
      "learning_rate": 4.406255055285022e-06,
      "loss": 0.4728,
      "step": 2787
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.315803762819754,
      "learning_rate": 4.4057900712569605e-06,
      "loss": 0.3817,
      "step": 2788
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.884867149411901,
      "learning_rate": 4.4053249297818666e-06,
      "loss": 0.2739,
      "step": 2789
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.6235019965063096,
      "learning_rate": 4.404859630898169e-06,
      "loss": 0.3838,
      "step": 2790
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.389731032860871,
      "learning_rate": 4.4043941746443074e-06,
      "loss": 0.3811,
      "step": 2791
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.39164959614411,
      "learning_rate": 4.403928561058736e-06,
      "loss": 0.3688,
      "step": 2792
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.17550921027046,
      "learning_rate": 4.403462790179921e-06,
      "loss": 0.2563,
      "step": 2793
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.061205014506344,
      "learning_rate": 4.402996862046342e-06,
      "loss": 0.3005,
      "step": 2794
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.441881688252894,
      "learning_rate": 4.402530776696493e-06,
      "loss": 0.2541,
      "step": 2795
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.012587270093556,
      "learning_rate": 4.402064534168878e-06,
      "loss": 0.4173,
      "step": 2796
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.302205170554078,
      "learning_rate": 4.4015981345020175e-06,
      "loss": 0.1967,
      "step": 2797
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.778415948807388,
      "learning_rate": 4.401131577734442e-06,
      "loss": 0.4285,
      "step": 2798
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.4253165122399176,
      "learning_rate": 4.400664863904696e-06,
      "loss": 0.3033,
      "step": 2799
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.07170568998192,
      "learning_rate": 4.4001979930513374e-06,
      "loss": 0.4665,
      "step": 2800
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.7455495038383955,
      "learning_rate": 4.3997309652129364e-06,
      "loss": 0.5223,
      "step": 2801
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.675209475508069,
      "learning_rate": 4.399263780428078e-06,
      "loss": 0.4635,
      "step": 2802
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.391113454833994,
      "learning_rate": 4.398796438735357e-06,
      "loss": 0.1834,
      "step": 2803
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.419867501586172,
      "learning_rate": 4.398328940173383e-06,
      "loss": 0.3971,
      "step": 2804
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.716243260709992,
      "learning_rate": 4.397861284780779e-06,
      "loss": 0.2715,
      "step": 2805
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.783249590434248,
      "learning_rate": 4.397393472596181e-06,
      "loss": 0.3286,
      "step": 2806
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.468714147120069,
      "learning_rate": 4.396925503658237e-06,
      "loss": 0.254,
      "step": 2807
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.943144817250231,
      "learning_rate": 4.396457378005608e-06,
      "loss": 0.396,
      "step": 2808
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.987693845253818,
      "learning_rate": 4.395989095676967e-06,
      "loss": 0.1718,
      "step": 2809
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.5440054115113915,
      "learning_rate": 4.395520656711004e-06,
      "loss": 0.3306,
      "step": 2810
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.156451895762421,
      "learning_rate": 4.3950520611464166e-06,
      "loss": 0.2126,
      "step": 2811
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.5968665932175465,
      "learning_rate": 4.394583309021919e-06,
      "loss": 0.3129,
      "step": 2812
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.3524608490397885,
      "learning_rate": 4.394114400376237e-06,
      "loss": 0.2243,
      "step": 2813
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.4950458493248577,
      "learning_rate": 4.39364533524811e-06,
      "loss": 0.2181,
      "step": 2814
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.509269541633104,
      "learning_rate": 4.393176113676289e-06,
      "loss": 0.3279,
      "step": 2815
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.954231581946559,
      "learning_rate": 4.392706735699539e-06,
      "loss": 0.2923,
      "step": 2816
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.387269219343373,
      "learning_rate": 4.392237201356639e-06,
      "loss": 0.2719,
      "step": 2817
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.299720370713011,
      "learning_rate": 4.391767510686378e-06,
      "loss": 0.1599,
      "step": 2818
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.0261990450146,
      "learning_rate": 4.391297663727559e-06,
      "loss": 0.2596,
      "step": 2819
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.582395382200883,
      "learning_rate": 4.390827660519001e-06,
      "loss": 0.1472,
      "step": 2820
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.094915134029512,
      "learning_rate": 4.390357501099532e-06,
      "loss": 0.2701,
      "step": 2821
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9867789928648274,
      "learning_rate": 4.389887185507993e-06,
      "loss": 0.1337,
      "step": 2822
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.876615713516525,
      "learning_rate": 4.389416713783241e-06,
      "loss": 0.549,
      "step": 2823
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.485152952310536,
      "learning_rate": 4.388946085964144e-06,
      "loss": 0.4339,
      "step": 2824
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.89751955380998,
      "learning_rate": 4.388475302089581e-06,
      "loss": 0.3684,
      "step": 2825
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.8462946622720455,
      "learning_rate": 4.3880043621984485e-06,
      "loss": 0.3903,
      "step": 2826
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.860761360908152,
      "learning_rate": 4.387533266329652e-06,
      "loss": 0.3211,
      "step": 2827
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.349249096616833,
      "learning_rate": 4.38706201452211e-06,
      "loss": 0.4093,
      "step": 2828
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.691756951190696,
      "learning_rate": 4.386590606814757e-06,
      "loss": 0.4607,
      "step": 2829
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.539728859892109,
      "learning_rate": 4.386119043246537e-06,
      "loss": 0.301,
      "step": 2830
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.991137050320718,
      "learning_rate": 4.385647323856409e-06,
      "loss": 0.1809,
      "step": 2831
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.161955223033229,
      "learning_rate": 4.3851754486833445e-06,
      "loss": 0.5156,
      "step": 2832
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.9499650755185165,
      "learning_rate": 4.3847034177663265e-06,
      "loss": 0.2841,
      "step": 2833
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.363052966363307,
      "learning_rate": 4.384231231144353e-06,
      "loss": 0.1862,
      "step": 2834
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.109852179285309,
      "learning_rate": 4.383758888856432e-06,
      "loss": 0.3634,
      "step": 2835
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.996789454691579,
      "learning_rate": 4.383286390941587e-06,
      "loss": 0.3481,
      "step": 2836
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.597595800826433,
      "learning_rate": 4.3828137374388545e-06,
      "loss": 0.4174,
      "step": 2837
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.503525800511419,
      "learning_rate": 4.382340928387281e-06,
      "loss": 0.5112,
      "step": 2838
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.832621423798821,
      "learning_rate": 4.381867963825929e-06,
      "loss": 0.4193,
      "step": 2839
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.510671210070728,
      "learning_rate": 4.381394843793873e-06,
      "loss": 0.2568,
      "step": 2840
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.621170744869953,
      "learning_rate": 4.380921568330199e-06,
      "loss": 0.3787,
      "step": 2841
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.477713380594631,
      "learning_rate": 4.380448137474006e-06,
      "loss": 0.4848,
      "step": 2842
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.038857867018477,
      "learning_rate": 4.3799745512644065e-06,
      "loss": 0.3986,
      "step": 2843
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.2107990415750294,
      "learning_rate": 4.379500809740528e-06,
      "loss": 0.2014,
      "step": 2844
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.010169863190751,
      "learning_rate": 4.379026912941506e-06,
      "loss": 0.2396,
      "step": 2845
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.94099453953288,
      "learning_rate": 4.378552860906494e-06,
      "loss": 0.5286,
      "step": 2846
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.192890040720879,
      "learning_rate": 4.378078653674654e-06,
      "loss": 0.2547,
      "step": 2847
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.5913895762208536,
      "learning_rate": 4.377604291285164e-06,
      "loss": 0.2979,
      "step": 2848
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.700508457391877,
      "learning_rate": 4.377129773777212e-06,
      "loss": 0.3203,
      "step": 2849
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.1572236379452825,
      "learning_rate": 4.376655101190002e-06,
      "loss": 0.5595,
      "step": 2850
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.203091203015347,
      "learning_rate": 4.376180273562748e-06,
      "loss": 0.4215,
      "step": 2851
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.01347560941274,
      "learning_rate": 4.375705290934678e-06,
      "loss": 0.5443,
      "step": 2852
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.9229513328973225,
      "learning_rate": 4.375230153345033e-06,
      "loss": 0.3881,
      "step": 2853
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.687386588467924,
      "learning_rate": 4.3747548608330656e-06,
      "loss": 0.1641,
      "step": 2854
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.89791902516481,
      "learning_rate": 4.374279413438044e-06,
      "loss": 0.3047,
      "step": 2855
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.5436645065773003,
      "learning_rate": 4.373803811199247e-06,
      "loss": 0.2822,
      "step": 2856
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.01685773620382,
      "learning_rate": 4.373328054155965e-06,
      "loss": 0.646,
      "step": 2857
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.393716498568243,
      "learning_rate": 4.372852142347503e-06,
      "loss": 0.3743,
      "step": 2858
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.471372858522852,
      "learning_rate": 4.372376075813181e-06,
      "loss": 0.0866,
      "step": 2859
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.336787305109385,
      "learning_rate": 4.371899854592326e-06,
      "loss": 0.368,
      "step": 2860
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.075790646677383,
      "learning_rate": 4.371423478724282e-06,
      "loss": 0.1054,
      "step": 2861
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.7018439702701778,
      "learning_rate": 4.370946948248406e-06,
      "loss": 0.2159,
      "step": 2862
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.12475280975219,
      "learning_rate": 4.3704702632040665e-06,
      "loss": 0.4205,
      "step": 2863
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.8876698678096333,
      "learning_rate": 4.369993423630643e-06,
      "loss": 0.1557,
      "step": 2864
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.253119204716683,
      "learning_rate": 4.369516429567532e-06,
      "loss": 0.1829,
      "step": 2865
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.188223307726895,
      "learning_rate": 4.369039281054139e-06,
      "loss": 0.2684,
      "step": 2866
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.057635457382599,
      "learning_rate": 4.368561978129884e-06,
      "loss": 0.4115,
      "step": 2867
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.793385998748557,
      "learning_rate": 4.368084520834199e-06,
      "loss": 0.1324,
      "step": 2868
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.090414694292646,
      "learning_rate": 4.36760690920653e-06,
      "loss": 0.0992,
      "step": 2869
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.432207727987844,
      "learning_rate": 4.367129143286334e-06,
      "loss": 0.349,
      "step": 2870
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1010253155125667,
      "learning_rate": 4.366651223113083e-06,
      "loss": 0.1075,
      "step": 2871
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.036278329665872,
      "learning_rate": 4.366173148726259e-06,
      "loss": 0.3976,
      "step": 2872
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.891558067652309,
      "learning_rate": 4.365694920165359e-06,
      "loss": 0.3089,
      "step": 2873
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8813935440989855,
      "learning_rate": 4.365216537469893e-06,
      "loss": 0.1261,
      "step": 2874
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.189632170313995,
      "learning_rate": 4.364738000679379e-06,
      "loss": 0.3681,
      "step": 2875
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.179874592766125,
      "learning_rate": 4.3642593098333554e-06,
      "loss": 0.054,
      "step": 2876
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.403177668326677,
      "learning_rate": 4.363780464971367e-06,
      "loss": 0.3134,
      "step": 2877
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.286019224632928,
      "learning_rate": 4.363301466132974e-06,
      "loss": 0.3968,
      "step": 2878
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.084984876152737,
      "learning_rate": 4.362822313357749e-06,
      "loss": 0.2625,
      "step": 2879
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.737219558762453,
      "learning_rate": 4.362343006685279e-06,
      "loss": 0.5279,
      "step": 2880
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.070267351829396,
      "learning_rate": 4.361863546155159e-06,
      "loss": 0.4767,
      "step": 2881
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.000648578821882,
      "learning_rate": 4.361383931807002e-06,
      "loss": 0.3896,
      "step": 2882
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.8393938536128265,
      "learning_rate": 4.360904163680429e-06,
      "loss": 0.3319,
      "step": 2883
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8519314548897883,
      "learning_rate": 4.360424241815078e-06,
      "loss": 0.0663,
      "step": 2884
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.6345241680968075,
      "learning_rate": 4.359944166250598e-06,
      "loss": 0.359,
      "step": 2885
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.343595188892642,
      "learning_rate": 4.359463937026649e-06,
      "loss": 0.15,
      "step": 2886
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.619994899147726,
      "learning_rate": 4.358983554182906e-06,
      "loss": 0.2595,
      "step": 2887
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6302592195978114,
      "learning_rate": 4.358503017759056e-06,
      "loss": 0.0229,
      "step": 2888
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.423663752752018,
      "learning_rate": 4.358022327794799e-06,
      "loss": 0.3988,
      "step": 2889
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.370567749870277,
      "learning_rate": 4.357541484329846e-06,
      "loss": 0.4708,
      "step": 2890
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.322576336218652,
      "learning_rate": 4.357060487403923e-06,
      "loss": 0.3886,
      "step": 2891
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.068709500349168,
      "learning_rate": 4.356579337056766e-06,
      "loss": 0.2639,
      "step": 2892
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3433463569435378,
      "learning_rate": 4.356098033328129e-06,
      "loss": 0.0635,
      "step": 2893
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.519747204464452,
      "learning_rate": 4.355616576257771e-06,
      "loss": 0.48,
      "step": 2894
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5394129287100533,
      "learning_rate": 4.3551349658854685e-06,
      "loss": 0.1412,
      "step": 2895
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.524985968046199,
      "learning_rate": 4.354653202251011e-06,
      "loss": 0.4133,
      "step": 2896
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0892341775161776,
      "learning_rate": 4.354171285394199e-06,
      "loss": 0.2183,
      "step": 2897
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.17027210850744,
      "learning_rate": 4.353689215354845e-06,
      "loss": 0.4355,
      "step": 2898
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.789159787594821,
      "learning_rate": 4.353206992172777e-06,
      "loss": 0.1967,
      "step": 2899
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.692839819393644,
      "learning_rate": 4.3527246158878336e-06,
      "loss": 0.4303,
      "step": 2900
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.120284661794392,
      "learning_rate": 4.352242086539865e-06,
      "loss": 0.1833,
      "step": 2901
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.884486878500397,
      "learning_rate": 4.351759404168736e-06,
      "loss": 0.4493,
      "step": 2902
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.9502269212163212,
      "learning_rate": 4.351276568814325e-06,
      "loss": 0.3273,
      "step": 2903
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.542227431584874,
      "learning_rate": 4.350793580516518e-06,
      "loss": 0.3015,
      "step": 2904
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.9143590434555176,
      "learning_rate": 4.3503104393152216e-06,
      "loss": 0.3893,
      "step": 2905
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.8927652289578423,
      "learning_rate": 4.349827145250347e-06,
      "loss": 0.1558,
      "step": 2906
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.292034400844515,
      "learning_rate": 4.349343698361823e-06,
      "loss": 0.6322,
      "step": 2907
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.9125988044687245,
      "learning_rate": 4.34886009868959e-06,
      "loss": 0.4825,
      "step": 2908
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.975731569623501,
      "learning_rate": 4.348376346273599e-06,
      "loss": 0.4116,
      "step": 2909
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.852868577867937,
      "learning_rate": 4.347892441153818e-06,
      "loss": 0.408,
      "step": 2910
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.81207173249988,
      "learning_rate": 4.347408383370222e-06,
      "loss": 0.2942,
      "step": 2911
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.6566582135712045,
      "learning_rate": 4.346924172962803e-06,
      "loss": 0.7207,
      "step": 2912
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.881785765678679,
      "learning_rate": 4.346439809971564e-06,
      "loss": 0.2792,
      "step": 2913
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.394704708932192,
      "learning_rate": 4.34595529443652e-06,
      "loss": 0.4169,
      "step": 2914
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0157634352929774,
      "learning_rate": 4.345470626397701e-06,
      "loss": 0.2225,
      "step": 2915
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.8637665159141905,
      "learning_rate": 4.344985805895146e-06,
      "loss": 0.4145,
      "step": 2916
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.1676320556242,
      "learning_rate": 4.34450083296891e-06,
      "loss": 0.3511,
      "step": 2917
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.7301025519829105,
      "learning_rate": 4.344015707659057e-06,
      "loss": 0.5057,
      "step": 2918
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.8729174070376917,
      "learning_rate": 4.343530430005667e-06,
      "loss": 0.2131,
      "step": 2919
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.5969741446279535,
      "learning_rate": 4.343045000048832e-06,
      "loss": 0.5386,
      "step": 2920
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.516574144341772,
      "learning_rate": 4.342559417828654e-06,
      "loss": 0.5395,
      "step": 2921
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9418446640289238,
      "learning_rate": 4.342073683385252e-06,
      "loss": 0.0254,
      "step": 2922
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.47199858362478,
      "learning_rate": 4.341587796758752e-06,
      "loss": 0.2446,
      "step": 2923
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.9535888223804867,
      "learning_rate": 4.341101757989297e-06,
      "loss": 0.3036,
      "step": 2924
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.9959047352591757,
      "learning_rate": 4.340615567117043e-06,
      "loss": 0.2886,
      "step": 2925
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.438667535756947,
      "learning_rate": 4.340129224182154e-06,
      "loss": 0.2017,
      "step": 2926
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.718712888466576,
      "learning_rate": 4.33964272922481e-06,
      "loss": 0.4255,
      "step": 2927
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.610969812301758,
      "learning_rate": 4.339156082285202e-06,
      "loss": 0.6629,
      "step": 2928
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.7794668283981325,
      "learning_rate": 4.338669283403537e-06,
      "loss": 0.2327,
      "step": 2929
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.805000861560775,
      "learning_rate": 4.33818233262003e-06,
      "loss": 0.3811,
      "step": 2930
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.1429488275889215,
      "learning_rate": 4.33769522997491e-06,
      "loss": 0.3392,
      "step": 2931
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.461612022930191,
      "learning_rate": 4.337207975508421e-06,
      "loss": 0.2182,
      "step": 2932
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.19392701658002,
      "learning_rate": 4.336720569260815e-06,
      "loss": 0.3185,
      "step": 2933
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1502749498948255,
      "learning_rate": 4.336233011272361e-06,
      "loss": 0.083,
      "step": 2934
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.615737824942824,
      "learning_rate": 4.335745301583339e-06,
      "loss": 0.4537,
      "step": 2935
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.624850079015525,
      "learning_rate": 4.33525744023404e-06,
      "loss": 0.532,
      "step": 2936
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.499643127968376,
      "learning_rate": 4.334769427264768e-06,
      "loss": 0.2764,
      "step": 2937
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.3772085702178276,
      "learning_rate": 4.334281262715843e-06,
      "loss": 0.279,
      "step": 2938
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.191337612842258,
      "learning_rate": 4.333792946627592e-06,
      "loss": 0.2635,
      "step": 2939
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.669072807829677,
      "learning_rate": 4.333304479040358e-06,
      "loss": 0.4483,
      "step": 2940
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.8189550894627473,
      "learning_rate": 4.332815859994497e-06,
      "loss": 0.2542,
      "step": 2941
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.021568398414042,
      "learning_rate": 4.332327089530375e-06,
      "loss": 0.6094,
      "step": 2942
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.910294684169816,
      "learning_rate": 4.3318381676883725e-06,
      "loss": 0.564,
      "step": 2943
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.2636841312555536,
      "learning_rate": 4.331349094508881e-06,
      "loss": 0.2335,
      "step": 2944
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.086538044551177,
      "learning_rate": 4.3308598700323065e-06,
      "loss": 0.3476,
      "step": 2945
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.118066111420501,
      "learning_rate": 4.330370494299066e-06,
      "loss": 0.1777,
      "step": 2946
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.0596472254360725,
      "learning_rate": 4.329880967349589e-06,
      "loss": 0.5751,
      "step": 2947
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.891551881588166,
      "learning_rate": 4.329391289224318e-06,
      "loss": 0.4614,
      "step": 2948
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.534150200842685,
      "learning_rate": 4.328901459963707e-06,
      "loss": 0.3717,
      "step": 2949
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.649453432099121,
      "learning_rate": 4.3284114796082245e-06,
      "loss": 0.4566,
      "step": 2950
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.9566144114003867,
      "learning_rate": 4.327921348198349e-06,
      "loss": 0.1024,
      "step": 2951
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.431726075349464,
      "learning_rate": 4.327431065774575e-06,
      "loss": 0.2874,
      "step": 2952
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.716493127798827,
      "learning_rate": 4.326940632377405e-06,
      "loss": 0.2624,
      "step": 2953
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5181290801856588,
      "learning_rate": 4.326450048047357e-06,
      "loss": 0.0182,
      "step": 2954
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.878527723453733,
      "learning_rate": 4.3259593128249605e-06,
      "loss": 0.2029,
      "step": 2955
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.468854171122307,
      "learning_rate": 4.3254684267507574e-06,
      "loss": 0.3639,
      "step": 2956
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.577683611886482,
      "learning_rate": 4.324977389865304e-06,
      "loss": 0.2753,
      "step": 2957
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.910541812866893,
      "learning_rate": 4.324486202209165e-06,
      "loss": 0.2939,
      "step": 2958
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.077072074011594,
      "learning_rate": 4.323994863822921e-06,
      "loss": 0.5097,
      "step": 2959
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.608481010951339,
      "learning_rate": 4.3235033747471635e-06,
      "loss": 0.2815,
      "step": 2960
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.060224462667137,
      "learning_rate": 4.323011735022498e-06,
      "loss": 0.3197,
      "step": 2961
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.465533184888443,
      "learning_rate": 4.322519944689541e-06,
      "loss": 0.2549,
      "step": 2962
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.272176529185185,
      "learning_rate": 4.322028003788921e-06,
      "loss": 0.2015,
      "step": 2963
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.25197062707174,
      "learning_rate": 4.32153591236128e-06,
      "loss": 0.4883,
      "step": 2964
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.797337231855229,
      "learning_rate": 4.321043670447272e-06,
      "loss": 0.3052,
      "step": 2965
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.933139864361567,
      "learning_rate": 4.320551278087565e-06,
      "loss": 0.1633,
      "step": 2966
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.031507994544629,
      "learning_rate": 4.3200587353228376e-06,
      "loss": 0.3348,
      "step": 2967
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.9761080723748075,
      "learning_rate": 4.3195660421937795e-06,
      "loss": 0.3183,
      "step": 2968
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0828622079549883,
      "learning_rate": 4.3190731987410965e-06,
      "loss": 0.1077,
      "step": 2969
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.519490706749989,
      "learning_rate": 4.3185802050055046e-06,
      "loss": 0.4272,
      "step": 2970
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.181403581023159,
      "learning_rate": 4.318087061027732e-06,
      "loss": 0.3692,
      "step": 2971
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.0623466934404386,
      "learning_rate": 4.317593766848519e-06,
      "loss": 0.2519,
      "step": 2972
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.212450857727524,
      "learning_rate": 4.31710032250862e-06,
      "loss": 0.1656,
      "step": 2973
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.574817698112149,
      "learning_rate": 4.316606728048803e-06,
      "loss": 0.143,
      "step": 2974
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.4412649379108586,
      "learning_rate": 4.316112983509844e-06,
      "loss": 0.178,
      "step": 2975
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.887158923935688,
      "learning_rate": 4.315619088932533e-06,
      "loss": 0.5039,
      "step": 2976
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.0686367380441935,
      "learning_rate": 4.315125044357675e-06,
      "loss": 0.3996,
      "step": 2977
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.755517577707933,
      "learning_rate": 4.314630849826085e-06,
      "loss": 0.5052,
      "step": 2978
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.091566898639597,
      "learning_rate": 4.314136505378591e-06,
      "loss": 0.3502,
      "step": 2979
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.71067512202141,
      "learning_rate": 4.313642011056033e-06,
      "loss": 0.3655,
      "step": 2980
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.535457841354336,
      "learning_rate": 4.313147366899264e-06,
      "loss": 0.3784,
      "step": 2981
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.749889677589974,
      "learning_rate": 4.312652572949148e-06,
      "loss": 0.2012,
      "step": 2982
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.416999638312973,
      "learning_rate": 4.312157629246565e-06,
      "loss": 0.5042,
      "step": 2983
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.206382387127368,
      "learning_rate": 4.311662535832401e-06,
      "loss": 0.4051,
      "step": 2984
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.405389624418232,
      "learning_rate": 4.311167292747562e-06,
      "loss": 0.2523,
      "step": 2985
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.166122686501039,
      "learning_rate": 4.310671900032961e-06,
      "loss": 0.3032,
      "step": 2986
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.4789557584658524,
      "learning_rate": 4.310176357729525e-06,
      "loss": 0.2102,
      "step": 2987
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.528448425918501,
      "learning_rate": 4.309680665878193e-06,
      "loss": 0.5621,
      "step": 2988
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.183757749475222,
      "learning_rate": 4.309184824519917e-06,
      "loss": 0.4271,
      "step": 2989
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.4792429886402885,
      "learning_rate": 4.308688833695659e-06,
      "loss": 0.2535,
      "step": 2990
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.450059088774065,
      "learning_rate": 4.308192693446398e-06,
      "loss": 0.2761,
      "step": 2991
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.412083650421562,
      "learning_rate": 4.307696403813123e-06,
      "loss": 0.4037,
      "step": 2992
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7939338819114545,
      "learning_rate": 4.3071999648368325e-06,
      "loss": 0.3058,
      "step": 2993
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.026828729799026,
      "learning_rate": 4.306703376558541e-06,
      "loss": 0.4253,
      "step": 2994
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.703812654621273,
      "learning_rate": 4.3062066390192755e-06,
      "loss": 0.2984,
      "step": 2995
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.904013156246329,
      "learning_rate": 4.305709752260072e-06,
      "loss": 0.5356,
      "step": 2996
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.5628832382055164,
      "learning_rate": 4.305212716321982e-06,
      "loss": 0.1246,
      "step": 2997
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.364895035367767,
      "learning_rate": 4.3047155312460675e-06,
      "loss": 0.2286,
      "step": 2998
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.595019836948123,
      "learning_rate": 4.304218197073404e-06,
      "loss": 0.2918,
      "step": 2999
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.381354432306299,
      "learning_rate": 4.303720713845079e-06,
      "loss": 0.2792,
      "step": 3000
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.908832949498667,
      "learning_rate": 4.303223081602191e-06,
      "loss": 0.3122,
      "step": 3001
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7156917224249524,
      "learning_rate": 4.302725300385853e-06,
      "loss": 0.3286,
      "step": 3002
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.628443902019304,
      "learning_rate": 4.30222737023719e-06,
      "loss": 0.3799,
      "step": 3003
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.651319851690305,
      "learning_rate": 4.301729291197336e-06,
      "loss": 0.5781,
      "step": 3004
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.160330240876101,
      "learning_rate": 4.301231063307443e-06,
      "loss": 0.3479,
      "step": 3005
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.209417941439907,
      "learning_rate": 4.300732686608668e-06,
      "loss": 0.3641,
      "step": 3006
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.721723112601457,
      "learning_rate": 4.300234161142189e-06,
      "loss": 0.4584,
      "step": 3007
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.17057199186608,
      "learning_rate": 4.299735486949189e-06,
      "loss": 0.4127,
      "step": 3008
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.029569094720457,
      "learning_rate": 4.299236664070867e-06,
      "loss": 0.2338,
      "step": 3009
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.464016286613174,
      "learning_rate": 4.2987376925484335e-06,
      "loss": 0.3583,
      "step": 3010
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.341903796964639,
      "learning_rate": 4.298238572423109e-06,
      "loss": 0.473,
      "step": 3011
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.942427102606186,
      "learning_rate": 4.297739303736131e-06,
      "loss": 0.1847,
      "step": 3012
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.79268730344806,
      "learning_rate": 4.297239886528746e-06,
      "loss": 0.3646,
      "step": 3013
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.055826801994568,
      "learning_rate": 4.296740320842212e-06,
      "loss": 0.493,
      "step": 3014
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.934218963811229,
      "learning_rate": 4.2962406067178026e-06,
      "loss": 0.4893,
      "step": 3015
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.9015880342669136,
      "learning_rate": 4.2957407441968e-06,
      "loss": 0.3037,
      "step": 3016
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.457235584520425,
      "learning_rate": 4.295240733320501e-06,
      "loss": 0.3947,
      "step": 3017
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7786989120057353,
      "learning_rate": 4.294740574130215e-06,
      "loss": 0.1973,
      "step": 3018
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.981934786791738,
      "learning_rate": 4.294240266667261e-06,
      "loss": 0.4836,
      "step": 3019
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.830443406053916,
      "learning_rate": 4.2937398109729735e-06,
      "loss": 0.4026,
      "step": 3020
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.865744354714795,
      "learning_rate": 4.293239207088697e-06,
      "loss": 0.178,
      "step": 3021
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.503081726835048,
      "learning_rate": 4.29273845505579e-06,
      "loss": 0.289,
      "step": 3022
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.811228439903722,
      "learning_rate": 4.2922375549156196e-06,
      "loss": 0.3165,
      "step": 3023
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4892285424901153,
      "learning_rate": 4.29173650670957e-06,
      "loss": 0.1226,
      "step": 3024
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.322150577036969,
      "learning_rate": 4.291235310479034e-06,
      "loss": 0.325,
      "step": 3025
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.372799426176545,
      "learning_rate": 4.290733966265419e-06,
      "loss": 0.42,
      "step": 3026
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.8440544267199135,
      "learning_rate": 4.290232474110143e-06,
      "loss": 0.3388,
      "step": 3027
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.13644416863154,
      "learning_rate": 4.289730834054637e-06,
      "loss": 0.5306,
      "step": 3028
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.5232979679959215,
      "learning_rate": 4.289229046140344e-06,
      "loss": 0.1578,
      "step": 3029
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.500525887018999,
      "learning_rate": 4.28872711040872e-06,
      "loss": 0.2231,
      "step": 3030
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.913163112229652,
      "learning_rate": 4.288225026901231e-06,
      "loss": 0.5203,
      "step": 3031
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.6084722280555415,
      "learning_rate": 4.287722795659358e-06,
      "loss": 0.325,
      "step": 3032
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.7423066307063735,
      "learning_rate": 4.287220416724592e-06,
      "loss": 0.3767,
      "step": 3033
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.806165980268893,
      "learning_rate": 4.286717890138438e-06,
      "loss": 0.4173,
      "step": 3034
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.63350844348221,
      "learning_rate": 4.286215215942411e-06,
      "loss": 0.5365,
      "step": 3035
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.441216606600507,
      "learning_rate": 4.2857123941780406e-06,
      "loss": 0.2258,
      "step": 3036
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.870338344387919,
      "learning_rate": 4.285209424886867e-06,
      "loss": 0.365,
      "step": 3037
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.735054789925221,
      "learning_rate": 4.284706308110443e-06,
      "loss": 0.3731,
      "step": 3038
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.39246163565371,
      "learning_rate": 4.284203043890335e-06,
      "loss": 0.4171,
      "step": 3039
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.700633374784032,
      "learning_rate": 4.283699632268117e-06,
      "loss": 0.2673,
      "step": 3040
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8484157644154773,
      "learning_rate": 4.283196073285382e-06,
      "loss": 0.1447,
      "step": 3041
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.636846431241402,
      "learning_rate": 4.282692366983728e-06,
      "loss": 0.3401,
      "step": 3042
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.314244913623894,
      "learning_rate": 4.282188513404773e-06,
      "loss": 0.404,
      "step": 3043
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.786923808375183,
      "learning_rate": 4.2816845125901395e-06,
      "loss": 0.2358,
      "step": 3044
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.824366534739757,
      "learning_rate": 4.281180364581466e-06,
      "loss": 0.6061,
      "step": 3045
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.196786166255123,
      "learning_rate": 4.280676069420405e-06,
      "loss": 0.4663,
      "step": 3046
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.712128222060697,
      "learning_rate": 4.280171627148617e-06,
      "loss": 0.4412,
      "step": 3047
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.320298742438533,
      "learning_rate": 4.279667037807776e-06,
      "loss": 0.2774,
      "step": 3048
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.5640845418407805,
      "learning_rate": 4.27916230143957e-06,
      "loss": 0.2185,
      "step": 3049
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.744538867582682,
      "learning_rate": 4.278657418085697e-06,
      "loss": 0.3005,
      "step": 3050
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.004018187722163,
      "learning_rate": 4.2781523877878695e-06,
      "loss": 0.4075,
      "step": 3051
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.4421240044951436,
      "learning_rate": 4.277647210587809e-06,
      "loss": 0.3185,
      "step": 3052
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.4881305404197445,
      "learning_rate": 4.277141886527251e-06,
      "loss": 0.2555,
      "step": 3053
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.466427054874437,
      "learning_rate": 4.276636415647943e-06,
      "loss": 0.127,
      "step": 3054
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.428399874119739,
      "learning_rate": 4.2761307979916454e-06,
      "loss": 0.1743,
      "step": 3055
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.2587713935498748,
      "learning_rate": 4.275625033600129e-06,
      "loss": 0.245,
      "step": 3056
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.8363748885362194,
      "learning_rate": 4.275119122515177e-06,
      "loss": 0.3545,
      "step": 3057
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.840372587050301,
      "learning_rate": 4.274613064778586e-06,
      "loss": 0.3512,
      "step": 3058
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9638216564365738,
      "learning_rate": 4.274106860432166e-06,
      "loss": 0.0566,
      "step": 3059
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.498721346150111,
      "learning_rate": 4.273600509517734e-06,
      "loss": 0.2493,
      "step": 3060
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.198281895048378,
      "learning_rate": 4.273094012077123e-06,
      "loss": 0.4253,
      "step": 3061
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.2915316474864005,
      "learning_rate": 4.272587368152178e-06,
      "loss": 0.5147,
      "step": 3062
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.900462292575751,
      "learning_rate": 4.272080577784755e-06,
      "loss": 0.23,
      "step": 3063
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.3198764616260314,
      "learning_rate": 4.271573641016723e-06,
      "loss": 0.3065,
      "step": 3064
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.211609238458542,
      "learning_rate": 4.2710665578899625e-06,
      "loss": 0.2779,
      "step": 3065
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.428831467876001,
      "learning_rate": 4.270559328446365e-06,
      "loss": 0.2148,
      "step": 3066
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.331479542188101,
      "learning_rate": 4.270051952727837e-06,
      "loss": 0.2798,
      "step": 3067
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.567489801225529,
      "learning_rate": 4.2695444307762955e-06,
      "loss": 0.4701,
      "step": 3068
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.968119438038778,
      "learning_rate": 4.269036762633668e-06,
      "loss": 0.4289,
      "step": 3069
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.409368670812027,
      "learning_rate": 4.268528948341897e-06,
      "loss": 0.2486,
      "step": 3070
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.484039616965411,
      "learning_rate": 4.268020987942934e-06,
      "loss": 0.4205,
      "step": 3071
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.991658858559759,
      "learning_rate": 4.2675128814787445e-06,
      "loss": 0.438,
      "step": 3072
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.878283334712619,
      "learning_rate": 4.267004628991308e-06,
      "loss": 0.1587,
      "step": 3073
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6860745594655677,
      "learning_rate": 4.2664962305226115e-06,
      "loss": 0.1893,
      "step": 3074
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.28410870181224,
      "learning_rate": 4.265987686114657e-06,
      "loss": 0.358,
      "step": 3075
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.6185607587318147,
      "learning_rate": 4.265478995809458e-06,
      "loss": 0.2136,
      "step": 3076
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7756566337031794,
      "learning_rate": 4.26497015964904e-06,
      "loss": 0.2749,
      "step": 3077
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7339813877621377,
      "learning_rate": 4.264461177675441e-06,
      "loss": 0.0746,
      "step": 3078
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.495565141498009,
      "learning_rate": 4.26395204993071e-06,
      "loss": 0.3798,
      "step": 3079
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.358310766546136,
      "learning_rate": 4.263442776456909e-06,
      "loss": 0.4944,
      "step": 3080
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.239622870954128,
      "learning_rate": 4.262933357296112e-06,
      "loss": 0.2036,
      "step": 3081
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.9309188384190135,
      "learning_rate": 4.262423792490403e-06,
      "loss": 0.5083,
      "step": 3082
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.9896116514860847,
      "learning_rate": 4.261914082081882e-06,
      "loss": 0.3863,
      "step": 3083
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.407004541290456,
      "learning_rate": 4.2614042261126565e-06,
      "loss": 0.3364,
      "step": 3084
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.588950051442431,
      "learning_rate": 4.260894224624851e-06,
      "loss": 0.29,
      "step": 3085
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.828604086746591,
      "learning_rate": 4.260384077660596e-06,
      "loss": 0.4015,
      "step": 3086
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.542112057186561,
      "learning_rate": 4.259873785262041e-06,
      "loss": 0.4895,
      "step": 3087
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.803632086456449,
      "learning_rate": 4.259363347471341e-06,
      "loss": 0.3409,
      "step": 3088
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.756352577921671,
      "learning_rate": 4.2588527643306674e-06,
      "loss": 0.5268,
      "step": 3089
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.145800655180773,
      "learning_rate": 4.258342035882202e-06,
      "loss": 0.4239,
      "step": 3090
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.407502768300972,
      "learning_rate": 4.257831162168138e-06,
      "loss": 0.4379,
      "step": 3091
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.11100085282542,
      "learning_rate": 4.257320143230681e-06,
      "loss": 0.3918,
      "step": 3092
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.467306458943414,
      "learning_rate": 4.256808979112051e-06,
      "loss": 0.6204,
      "step": 3093
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.891782193518999,
      "learning_rate": 4.256297669854474e-06,
      "loss": 0.4183,
      "step": 3094
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.024660310093422,
      "learning_rate": 4.2557862155001955e-06,
      "loss": 0.3247,
      "step": 3095
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.504092775980563,
      "learning_rate": 4.255274616091468e-06,
      "loss": 0.3941,
      "step": 3096
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.453267316472454,
      "learning_rate": 4.254762871670557e-06,
      "loss": 0.5473,
      "step": 3097
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.7888587711538633,
      "learning_rate": 4.254250982279741e-06,
      "loss": 0.3373,
      "step": 3098
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3455854913855596,
      "learning_rate": 4.253738947961309e-06,
      "loss": 0.1564,
      "step": 3099
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.538153446062254,
      "learning_rate": 4.253226768757564e-06,
      "loss": 0.3157,
      "step": 3100
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.600979523169,
      "learning_rate": 4.2527144447108185e-06,
      "loss": 0.2647,
      "step": 3101
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.749959115663277,
      "learning_rate": 4.252201975863399e-06,
      "loss": 0.5034,
      "step": 3102
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.831610284601618,
      "learning_rate": 4.2516893622576425e-06,
      "loss": 0.4676,
      "step": 3103
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.689386356400991,
      "learning_rate": 4.251176603935899e-06,
      "loss": 0.4703,
      "step": 3104
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.205784825706818,
      "learning_rate": 4.250663700940531e-06,
      "loss": 0.3893,
      "step": 3105
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.753246956417301,
      "learning_rate": 4.25015065331391e-06,
      "loss": 0.4516,
      "step": 3106
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.8166589384634477,
      "learning_rate": 4.249637461098422e-06,
      "loss": 0.2118,
      "step": 3107
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.2137175913155245,
      "learning_rate": 4.249124124336466e-06,
      "loss": 0.3751,
      "step": 3108
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.59005967608162,
      "learning_rate": 4.24861064307045e-06,
      "loss": 0.3288,
      "step": 3109
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.906992602847943,
      "learning_rate": 4.248097017342796e-06,
      "loss": 0.2623,
      "step": 3110
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.436512104559211,
      "learning_rate": 4.247583247195936e-06,
      "loss": 0.3879,
      "step": 3111
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.242402300071471,
      "learning_rate": 4.247069332672317e-06,
      "loss": 0.2972,
      "step": 3112
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3756592751669077,
      "learning_rate": 4.246555273814394e-06,
      "loss": 0.0105,
      "step": 3113
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.321272616941412,
      "learning_rate": 4.246041070664639e-06,
      "loss": 0.3092,
      "step": 3114
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3350309853030464,
      "learning_rate": 4.2455267232655295e-06,
      "loss": 0.2809,
      "step": 3115
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.3164391989752247,
      "learning_rate": 4.245012231659561e-06,
      "loss": 0.2371,
      "step": 3116
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.847816842962599,
      "learning_rate": 4.244497595889237e-06,
      "loss": 0.4626,
      "step": 3117
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.116645671420943,
      "learning_rate": 4.243982815997073e-06,
      "loss": 0.3757,
      "step": 3118
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.983491687485204,
      "learning_rate": 4.243467892025601e-06,
      "loss": 0.3188,
      "step": 3119
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.158700791824309,
      "learning_rate": 4.242952824017359e-06,
      "loss": 0.3001,
      "step": 3120
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.945041899613963,
      "learning_rate": 4.242437612014899e-06,
      "loss": 0.2365,
      "step": 3121
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.340020787127969,
      "learning_rate": 4.241922256060788e-06,
      "loss": 0.3747,
      "step": 3122
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.538387815479213,
      "learning_rate": 4.241406756197599e-06,
      "loss": 0.2549,
      "step": 3123
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.521619209114294,
      "learning_rate": 4.240891112467922e-06,
      "loss": 0.3678,
      "step": 3124
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.399218317027395,
      "learning_rate": 4.240375324914355e-06,
      "loss": 0.3102,
      "step": 3125
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.508189457385382,
      "learning_rate": 4.239859393579513e-06,
      "loss": 0.2898,
      "step": 3126
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.10948580200985,
      "learning_rate": 4.2393433185060175e-06,
      "loss": 0.2434,
      "step": 3127
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.8534496815545145,
      "learning_rate": 4.238827099736505e-06,
      "loss": 0.382,
      "step": 3128
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.809471985895364,
      "learning_rate": 4.238310737313622e-06,
      "loss": 0.4174,
      "step": 3129
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.744834157768361,
      "learning_rate": 4.237794231280028e-06,
      "loss": 0.4895,
      "step": 3130
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.825817314061722,
      "learning_rate": 4.237277581678395e-06,
      "loss": 0.3477,
      "step": 3131
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.456622234958185,
      "learning_rate": 4.236760788551405e-06,
      "loss": 0.5914,
      "step": 3132
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.187200420607927,
      "learning_rate": 4.236243851941755e-06,
      "loss": 0.4275,
      "step": 3133
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.767549644370501,
      "learning_rate": 4.235726771892149e-06,
      "loss": 0.234,
      "step": 3134
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.091181369573874,
      "learning_rate": 4.235209548445307e-06,
      "loss": 0.1032,
      "step": 3135
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.135259590342545,
      "learning_rate": 4.23469218164396e-06,
      "loss": 0.3979,
      "step": 3136
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.934736398324178,
      "learning_rate": 4.234174671530849e-06,
      "loss": 0.4017,
      "step": 3137
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.826111701239201,
      "learning_rate": 4.23365701814873e-06,
      "loss": 0.3706,
      "step": 3138
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.59173162617746,
      "learning_rate": 4.233139221540366e-06,
      "loss": 0.4591,
      "step": 3139
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.604094773827162,
      "learning_rate": 4.232621281748538e-06,
      "loss": 0.2559,
      "step": 3140
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.975642196933942,
      "learning_rate": 4.2321031988160346e-06,
      "loss": 0.5065,
      "step": 3141
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.617697334902008,
      "learning_rate": 4.2315849727856565e-06,
      "loss": 0.343,
      "step": 3142
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.566195655521713,
      "learning_rate": 4.231066603700217e-06,
      "loss": 0.3764,
      "step": 3143
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.317656752023058,
      "learning_rate": 4.230548091602542e-06,
      "loss": 0.2912,
      "step": 3144
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.153631638253108,
      "learning_rate": 4.230029436535468e-06,
      "loss": 0.1872,
      "step": 3145
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.9134701960767604,
      "learning_rate": 4.229510638541844e-06,
      "loss": 0.3121,
      "step": 3146
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.630483112034675,
      "learning_rate": 4.22899169766453e-06,
      "loss": 0.3477,
      "step": 3147
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.3897288162348875,
      "learning_rate": 4.228472613946398e-06,
      "loss": 0.5787,
      "step": 3148
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.530851465167108,
      "learning_rate": 4.227953387430335e-06,
      "loss": 0.415,
      "step": 3149
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.721038273354481,
      "learning_rate": 4.227434018159234e-06,
      "loss": 0.1319,
      "step": 3150
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.735020903258649,
      "learning_rate": 4.2269145061760025e-06,
      "loss": 0.5047,
      "step": 3151
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.255229161432381,
      "learning_rate": 4.226394851523562e-06,
      "loss": 0.4718,
      "step": 3152
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.30144943650134,
      "learning_rate": 4.2258750542448425e-06,
      "loss": 0.4058,
      "step": 3153
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.675661656123729,
      "learning_rate": 4.2253551143827876e-06,
      "loss": 0.4943,
      "step": 3154
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.296089749772156,
      "learning_rate": 4.224835031980353e-06,
      "loss": 0.2759,
      "step": 3155
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.402798686341656,
      "learning_rate": 4.224314807080503e-06,
      "loss": 0.3578,
      "step": 3156
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.4398284226089215,
      "learning_rate": 4.223794439726219e-06,
      "loss": 0.469,
      "step": 3157
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.116693077361317,
      "learning_rate": 4.223273929960488e-06,
      "loss": 0.3086,
      "step": 3158
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.034076783769337,
      "learning_rate": 4.222753277826316e-06,
      "loss": 0.3938,
      "step": 3159
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.4834813427274978,
      "learning_rate": 4.222232483366713e-06,
      "loss": 0.2859,
      "step": 3160
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.668216569820696,
      "learning_rate": 4.221711546624705e-06,
      "loss": 0.3415,
      "step": 3161
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.5666132366768153,
      "learning_rate": 4.2211904676433314e-06,
      "loss": 0.2558,
      "step": 3162
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.606482964263229,
      "learning_rate": 4.22066924646564e-06,
      "loss": 0.2925,
      "step": 3163
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.5323608883530495,
      "learning_rate": 4.220147883134691e-06,
      "loss": 0.2799,
      "step": 3164
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.758040702853587,
      "learning_rate": 4.219626377693557e-06,
      "loss": 0.5606,
      "step": 3165
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.2813193413751085,
      "learning_rate": 4.219104730185322e-06,
      "loss": 0.36,
      "step": 3166
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.321393789851455,
      "learning_rate": 4.2185829406530845e-06,
      "loss": 0.3132,
      "step": 3167
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.960685003868751,
      "learning_rate": 4.218061009139949e-06,
      "loss": 0.2816,
      "step": 3168
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.870540340169016,
      "learning_rate": 4.217538935689036e-06,
      "loss": 0.0925,
      "step": 3169
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.032518720111812,
      "learning_rate": 4.217016720343477e-06,
      "loss": 0.1714,
      "step": 3170
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.346387550054793,
      "learning_rate": 4.216494363146414e-06,
      "loss": 0.4436,
      "step": 3171
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.821499834877528,
      "learning_rate": 4.215971864141003e-06,
      "loss": 0.5394,
      "step": 3172
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.878319872364689,
      "learning_rate": 4.215449223370409e-06,
      "loss": 0.2342,
      "step": 3173
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.988153821775201,
      "learning_rate": 4.214926440877811e-06,
      "loss": 0.3227,
      "step": 3174
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.37133085458173,
      "learning_rate": 4.214403516706399e-06,
      "loss": 0.3127,
      "step": 3175
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.569158534910589,
      "learning_rate": 4.213880450899372e-06,
      "loss": 0.3357,
      "step": 3176
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.853155029473078,
      "learning_rate": 4.213357243499946e-06,
      "loss": 0.3065,
      "step": 3177
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.761483227700859,
      "learning_rate": 4.212833894551345e-06,
      "loss": 0.2596,
      "step": 3178
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.033668326510002,
      "learning_rate": 4.212310404096804e-06,
      "loss": 0.3818,
      "step": 3179
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.671319189302898,
      "learning_rate": 4.2117867721795735e-06,
      "loss": 0.4237,
      "step": 3180
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.530178728717232,
      "learning_rate": 4.211262998842912e-06,
      "loss": 0.4181,
      "step": 3181
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.445844173715356,
      "learning_rate": 4.210739084130091e-06,
      "loss": 0.2895,
      "step": 3182
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.0151163409022415,
      "learning_rate": 4.210215028084395e-06,
      "loss": 0.6103,
      "step": 3183
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.70772310058771,
      "learning_rate": 4.209690830749118e-06,
      "loss": 0.2077,
      "step": 3184
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.97369409847565,
      "learning_rate": 4.2091664921675654e-06,
      "loss": 0.4067,
      "step": 3185
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.352453363927032,
      "learning_rate": 4.208642012383058e-06,
      "loss": 0.2059,
      "step": 3186
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6199273523211977,
      "learning_rate": 4.2081173914389245e-06,
      "loss": 0.089,
      "step": 3187
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.2038200567033805,
      "learning_rate": 4.207592629378506e-06,
      "loss": 0.214,
      "step": 3188
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.491416565385398,
      "learning_rate": 4.2070677262451575e-06,
      "loss": 0.3423,
      "step": 3189
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.608091096619177,
      "learning_rate": 4.206542682082242e-06,
      "loss": 0.3717,
      "step": 3190
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.264520173998606,
      "learning_rate": 4.206017496933136e-06,
      "loss": 0.3643,
      "step": 3191
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.858222403725331,
      "learning_rate": 4.20549217084123e-06,
      "loss": 0.3235,
      "step": 3192
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.795222399392967,
      "learning_rate": 4.204966703849922e-06,
      "loss": 0.1697,
      "step": 3193
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.67465321354133,
      "learning_rate": 4.204441096002623e-06,
      "loss": 0.1271,
      "step": 3194
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.113603141321843,
      "learning_rate": 4.203915347342758e-06,
      "loss": 0.401,
      "step": 3195
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.981023184633715,
      "learning_rate": 4.20338945791376e-06,
      "loss": 0.2183,
      "step": 3196
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.687003175057959,
      "learning_rate": 4.2028634277590775e-06,
      "loss": 0.5088,
      "step": 3197
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.212947923967325,
      "learning_rate": 4.2023372569221665e-06,
      "loss": 0.2784,
      "step": 3198
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.117714992945551,
      "learning_rate": 4.2018109454464985e-06,
      "loss": 0.4113,
      "step": 3199
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.580073981171509,
      "learning_rate": 4.201284493375553e-06,
      "loss": 0.3738,
      "step": 3200
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.924038761709532,
      "learning_rate": 4.200757900752822e-06,
      "loss": 0.1314,
      "step": 3201
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6161177400787317,
      "learning_rate": 4.200231167621814e-06,
      "loss": 0.0972,
      "step": 3202
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.544409817766956,
      "learning_rate": 4.1997042940260415e-06,
      "loss": 0.4516,
      "step": 3203
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.015183032642986,
      "learning_rate": 4.199177280009034e-06,
      "loss": 0.2636,
      "step": 3204
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.0473311087131085,
      "learning_rate": 4.19865012561433e-06,
      "loss": 0.4396,
      "step": 3205
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.869258464009267,
      "learning_rate": 4.19812283088548e-06,
      "loss": 0.124,
      "step": 3206
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.313740733038232,
      "learning_rate": 4.197595395866048e-06,
      "loss": 0.2674,
      "step": 3207
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.058835820592647,
      "learning_rate": 4.197067820599608e-06,
      "loss": 0.1726,
      "step": 3208
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.036264515310391,
      "learning_rate": 4.196540105129743e-06,
      "loss": 0.4089,
      "step": 3209
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.596841943001404,
      "learning_rate": 4.1960122495000535e-06,
      "loss": 0.5665,
      "step": 3210
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.1956306859238275,
      "learning_rate": 4.195484253754146e-06,
      "loss": 0.4026,
      "step": 3211
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.099232365838427,
      "learning_rate": 4.194956117935643e-06,
      "loss": 0.3802,
      "step": 3212
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.8878354083684883,
      "learning_rate": 4.194427842088175e-06,
      "loss": 0.2449,
      "step": 3213
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.194286532175739,
      "learning_rate": 4.193899426255386e-06,
      "loss": 0.333,
      "step": 3214
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2877531536747826,
      "learning_rate": 4.193370870480932e-06,
      "loss": 0.2194,
      "step": 3215
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.734613646153788,
      "learning_rate": 4.1928421748084775e-06,
      "loss": 0.416,
      "step": 3216
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.9660999379524022,
      "learning_rate": 4.192313339281703e-06,
      "loss": 0.3504,
      "step": 3217
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.98488490805747,
      "learning_rate": 4.1917843639442964e-06,
      "loss": 0.2908,
      "step": 3218
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0083145505025803,
      "learning_rate": 4.191255248839961e-06,
      "loss": 0.0772,
      "step": 3219
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.41181698002489,
      "learning_rate": 4.190725994012409e-06,
      "loss": 0.0782,
      "step": 3220
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.878432287930243,
      "learning_rate": 4.190196599505364e-06,
      "loss": 0.3658,
      "step": 3221
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.168670571952397,
      "learning_rate": 4.1896670653625624e-06,
      "loss": 0.2454,
      "step": 3222
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.1512775877891634,
      "learning_rate": 4.189137391627751e-06,
      "loss": 0.3111,
      "step": 3223
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.3034696331925257,
      "learning_rate": 4.188607578344691e-06,
      "loss": 0.1331,
      "step": 3224
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.416916790839534,
      "learning_rate": 4.188077625557151e-06,
      "loss": 0.2651,
      "step": 3225
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.811318427670221,
      "learning_rate": 4.187547533308914e-06,
      "loss": 0.2514,
      "step": 3226
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.61776362623313,
      "learning_rate": 4.1870173016437746e-06,
      "loss": 0.4896,
      "step": 3227
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.918579188839955,
      "learning_rate": 4.186486930605536e-06,
      "loss": 0.1924,
      "step": 3228
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.676382212946801,
      "learning_rate": 4.185956420238015e-06,
      "loss": 0.4859,
      "step": 3229
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.860739227140912,
      "learning_rate": 4.18542577058504e-06,
      "loss": 0.2815,
      "step": 3230
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.315368553147302,
      "learning_rate": 4.184894981690452e-06,
      "loss": 0.173,
      "step": 3231
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.704839188509637,
      "learning_rate": 4.1843640535981e-06,
      "loss": 0.2833,
      "step": 3232
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.581959860786494,
      "learning_rate": 4.183832986351849e-06,
      "loss": 0.4438,
      "step": 3233
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.243461529719013,
      "learning_rate": 4.183301779995572e-06,
      "loss": 0.3213,
      "step": 3234
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.856495862190246,
      "learning_rate": 4.182770434573154e-06,
      "loss": 0.3646,
      "step": 3235
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.458734642628158,
      "learning_rate": 4.182238950128494e-06,
      "loss": 0.5594,
      "step": 3236
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.263008272775373,
      "learning_rate": 4.181707326705499e-06,
      "loss": 0.1302,
      "step": 3237
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.446330991786299,
      "learning_rate": 4.181175564348089e-06,
      "loss": 0.2142,
      "step": 3238
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.8747161027939567,
      "learning_rate": 4.180643663100197e-06,
      "loss": 0.2256,
      "step": 3239
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.442700247971296,
      "learning_rate": 4.180111623005764e-06,
      "loss": 0.2493,
      "step": 3240
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.764621896511118,
      "learning_rate": 4.179579444108748e-06,
      "loss": 0.5055,
      "step": 3241
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4774799380455117,
      "learning_rate": 4.179047126453112e-06,
      "loss": 0.0828,
      "step": 3242
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.620599093942474,
      "learning_rate": 4.178514670082834e-06,
      "loss": 0.5231,
      "step": 3243
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.507378582304081,
      "learning_rate": 4.1779820750419035e-06,
      "loss": 0.3926,
      "step": 3244
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7837127160880963,
      "learning_rate": 4.17744934137432e-06,
      "loss": 0.3854,
      "step": 3245
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.7102720375939344,
      "learning_rate": 4.176916469124097e-06,
      "loss": 0.1315,
      "step": 3246
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.181863680689531,
      "learning_rate": 4.176383458335257e-06,
      "loss": 0.4021,
      "step": 3247
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.174747619181544,
      "learning_rate": 4.1758503090518335e-06,
      "loss": 0.3904,
      "step": 3248
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.4371986317558045,
      "learning_rate": 4.175317021317873e-06,
      "loss": 0.3186,
      "step": 3249
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.060400349638785,
      "learning_rate": 4.174783595177436e-06,
      "loss": 0.0885,
      "step": 3250
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.39109020657845,
      "learning_rate": 4.174250030674588e-06,
      "loss": 0.4257,
      "step": 3251
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.555675991101572,
      "learning_rate": 4.173716327853411e-06,
      "loss": 0.4064,
      "step": 3252
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.519314486869238,
      "learning_rate": 4.173182486757997e-06,
      "loss": 0.4298,
      "step": 3253
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.218133495831906,
      "learning_rate": 4.172648507432449e-06,
      "loss": 0.5902,
      "step": 3254
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.988357298301603,
      "learning_rate": 4.172114389920882e-06,
      "loss": 0.273,
      "step": 3255
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.678836668109109,
      "learning_rate": 4.171580134267422e-06,
      "loss": 0.1765,
      "step": 3256
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.652545648764838,
      "learning_rate": 4.171045740516207e-06,
      "loss": 0.3266,
      "step": 3257
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.176005985939848,
      "learning_rate": 4.170511208711385e-06,
      "loss": 0.2542,
      "step": 3258
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.8755683895095054,
      "learning_rate": 4.169976538897117e-06,
      "loss": 0.221,
      "step": 3259
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.5311628157865425,
      "learning_rate": 4.169441731117575e-06,
      "loss": 0.2189,
      "step": 3260
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.462433084793948,
      "learning_rate": 4.168906785416942e-06,
      "loss": 0.5607,
      "step": 3261
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.316314546602925,
      "learning_rate": 4.168371701839413e-06,
      "loss": 0.4285,
      "step": 3262
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.366912003443582,
      "learning_rate": 4.167836480429193e-06,
      "loss": 0.3416,
      "step": 3263
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.777820443932024,
      "learning_rate": 4.167301121230501e-06,
      "loss": 0.5295,
      "step": 3264
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.09795543259824,
      "learning_rate": 4.166765624287563e-06,
      "loss": 0.5399,
      "step": 3265
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.250533903061317,
      "learning_rate": 4.166229989644622e-06,
      "loss": 0.3847,
      "step": 3266
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.447801009207602,
      "learning_rate": 4.165694217345928e-06,
      "loss": 0.276,
      "step": 3267
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.520206617203525,
      "learning_rate": 4.165158307435744e-06,
      "loss": 0.4115,
      "step": 3268
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.581213855647317,
      "learning_rate": 4.164622259958345e-06,
      "loss": 0.2218,
      "step": 3269
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.042272065230615,
      "learning_rate": 4.164086074958016e-06,
      "loss": 0.1518,
      "step": 3270
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.004706665014975,
      "learning_rate": 4.163549752479054e-06,
      "loss": 0.2974,
      "step": 3271
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.691176464728607,
      "learning_rate": 4.163013292565767e-06,
      "loss": 0.2438,
      "step": 3272
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.670447348028557,
      "learning_rate": 4.162476695262475e-06,
      "loss": 0.5284,
      "step": 3273
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.502489718415662,
      "learning_rate": 4.16193996061351e-06,
      "loss": 0.4316,
      "step": 3274
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.103732871578812,
      "learning_rate": 4.161403088663213e-06,
      "loss": 0.2563,
      "step": 3275
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.022083914225554,
      "learning_rate": 4.160866079455939e-06,
      "loss": 0.6076,
      "step": 3276
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.362199290456773,
      "learning_rate": 4.160328933036051e-06,
      "loss": 0.3457,
      "step": 3277
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.1546699736792156,
      "learning_rate": 4.159791649447927e-06,
      "loss": 0.0937,
      "step": 3278
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.115573749947856,
      "learning_rate": 4.159254228735955e-06,
      "loss": 0.2841,
      "step": 3279
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.497477621036657,
      "learning_rate": 4.1587166709445334e-06,
      "loss": 0.3784,
      "step": 3280
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.370914510507119,
      "learning_rate": 4.158178976118072e-06,
      "loss": 0.2678,
      "step": 3281
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.7639300781791087,
      "learning_rate": 4.157641144300994e-06,
      "loss": 0.3323,
      "step": 3282
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.557365901777523,
      "learning_rate": 4.157103175537731e-06,
      "loss": 0.3246,
      "step": 3283
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.347318241086678,
      "learning_rate": 4.156565069872729e-06,
      "loss": 0.3224,
      "step": 3284
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.6440652411907815,
      "learning_rate": 4.156026827350442e-06,
      "loss": 0.4567,
      "step": 3285
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.3741159894857464,
      "learning_rate": 4.155488448015337e-06,
      "loss": 0.2029,
      "step": 3286
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1986710986806943,
      "learning_rate": 4.154949931911893e-06,
      "loss": 0.0452,
      "step": 3287
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.998974422673531,
      "learning_rate": 4.1544112790846e-06,
      "loss": 0.3485,
      "step": 3288
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.4212041349433004,
      "learning_rate": 4.153872489577958e-06,
      "loss": 0.3863,
      "step": 3289
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.414855327093977,
      "learning_rate": 4.15333356343648e-06,
      "loss": 0.2832,
      "step": 3290
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.990320064189403,
      "learning_rate": 4.152794500704688e-06,
      "loss": 0.2741,
      "step": 3291
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.852221665327771,
      "learning_rate": 4.1522553014271185e-06,
      "loss": 0.4725,
      "step": 3292
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.966634379325953,
      "learning_rate": 4.151715965648315e-06,
      "loss": 0.2645,
      "step": 3293
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.9438205760245197,
      "learning_rate": 4.151176493412838e-06,
      "loss": 0.1191,
      "step": 3294
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.009816754379008,
      "learning_rate": 4.150636884765254e-06,
      "loss": 0.5578,
      "step": 3295
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.888166938177667,
      "learning_rate": 4.150097139750142e-06,
      "loss": 0.4568,
      "step": 3296
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.3197648134001145,
      "learning_rate": 4.149557258412096e-06,
      "loss": 0.2587,
      "step": 3297
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.482297509263324,
      "learning_rate": 4.149017240795715e-06,
      "loss": 0.2802,
      "step": 3298
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.8274014700974544,
      "learning_rate": 4.148477086945616e-06,
      "loss": 0.2771,
      "step": 3299
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.061984933776137,
      "learning_rate": 4.147936796906421e-06,
      "loss": 0.3173,
      "step": 3300
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.14316968582094,
      "learning_rate": 4.147396370722767e-06,
      "loss": 0.4234,
      "step": 3301
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.204744995870366,
      "learning_rate": 4.146855808439303e-06,
      "loss": 0.203,
      "step": 3302
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.586430367147965,
      "learning_rate": 4.146315110100685e-06,
      "loss": 0.4183,
      "step": 3303
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.802902320922808,
      "learning_rate": 4.145774275751584e-06,
      "loss": 0.1602,
      "step": 3304
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.36583771855786,
      "learning_rate": 4.145233305436682e-06,
      "loss": 0.3769,
      "step": 3305
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.539173059881124,
      "learning_rate": 4.144692199200669e-06,
      "loss": 0.1298,
      "step": 3306
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.236494955046417,
      "learning_rate": 4.1441509570882514e-06,
      "loss": 0.3622,
      "step": 3307
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.743001251288188,
      "learning_rate": 4.143609579144142e-06,
      "loss": 0.2215,
      "step": 3308
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.408706177993014,
      "learning_rate": 4.143068065413068e-06,
      "loss": 0.3641,
      "step": 3309
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.6560697135012115,
      "learning_rate": 4.142526415939765e-06,
      "loss": 0.2607,
      "step": 3310
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.467300014739821,
      "learning_rate": 4.141984630768983e-06,
      "loss": 0.2668,
      "step": 3311
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.155943374924431,
      "learning_rate": 4.141442709945481e-06,
      "loss": 0.2407,
      "step": 3312
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.254735274172973,
      "learning_rate": 4.14090065351403e-06,
      "loss": 0.2705,
      "step": 3313
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.489104391501347,
      "learning_rate": 4.140358461519412e-06,
      "loss": 0.3377,
      "step": 3314
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.068438087909548,
      "learning_rate": 4.139816134006419e-06,
      "loss": 0.283,
      "step": 3315
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.232125591936127,
      "learning_rate": 4.139273671019856e-06,
      "loss": 0.4454,
      "step": 3316
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.865533523466102,
      "learning_rate": 4.138731072604542e-06,
      "loss": 0.1604,
      "step": 3317
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.240283277149715,
      "learning_rate": 4.138188338805299e-06,
      "loss": 0.4026,
      "step": 3318
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.3913182437776985,
      "learning_rate": 4.137645469666969e-06,
      "loss": 0.4124,
      "step": 3319
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.486604262233041,
      "learning_rate": 4.137102465234397e-06,
      "loss": 0.2569,
      "step": 3320
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.627284617325287,
      "learning_rate": 4.136559325552446e-06,
      "loss": 0.5038,
      "step": 3321
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.109238635746494,
      "learning_rate": 4.136016050665988e-06,
      "loss": 0.4643,
      "step": 3322
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.5032101328475127,
      "learning_rate": 4.135472640619905e-06,
      "loss": 0.1712,
      "step": 3323
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.791112967500542,
      "learning_rate": 4.13492909545909e-06,
      "loss": 0.2059,
      "step": 3324
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.690643898587492,
      "learning_rate": 4.134385415228449e-06,
      "loss": 0.4048,
      "step": 3325
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.705468600033234,
      "learning_rate": 4.133841599972898e-06,
      "loss": 0.3322,
      "step": 3326
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.08734820165134,
      "learning_rate": 4.133297649737364e-06,
      "loss": 0.4496,
      "step": 3327
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.503434406015387,
      "learning_rate": 4.132753564566786e-06,
      "loss": 0.2184,
      "step": 3328
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.707075441723288,
      "learning_rate": 4.132209344506113e-06,
      "loss": 0.2975,
      "step": 3329
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.144024185142441,
      "learning_rate": 4.131664989600307e-06,
      "loss": 0.2949,
      "step": 3330
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.422788710950601,
      "learning_rate": 4.131120499894339e-06,
      "loss": 0.4472,
      "step": 3331
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.501613577368323,
      "learning_rate": 4.130575875433193e-06,
      "loss": 0.2684,
      "step": 3332
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.62801051854706,
      "learning_rate": 4.130031116261861e-06,
      "loss": 0.5128,
      "step": 3333
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.264766016987173,
      "learning_rate": 4.1294862224253505e-06,
      "loss": 0.3275,
      "step": 3334
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.642374414525173,
      "learning_rate": 4.128941193968677e-06,
      "loss": 0.2241,
      "step": 3335
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.559212480988192,
      "learning_rate": 4.128396030936869e-06,
      "loss": 0.4543,
      "step": 3336
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.3045864750928753,
      "learning_rate": 4.127850733374964e-06,
      "loss": 0.2721,
      "step": 3337
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.156998194897921,
      "learning_rate": 4.127305301328012e-06,
      "loss": 0.0922,
      "step": 3338
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.5187947743701042,
      "learning_rate": 4.126759734841075e-06,
      "loss": 0.1999,
      "step": 3339
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.308754185986498,
      "learning_rate": 4.126214033959224e-06,
      "loss": 0.3825,
      "step": 3340
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.352191326718923,
      "learning_rate": 4.125668198727541e-06,
      "loss": 0.4295,
      "step": 3341
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.931533787452705,
      "learning_rate": 4.125122229191124e-06,
      "loss": 0.4117,
      "step": 3342
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.402026421620454,
      "learning_rate": 4.124576125395075e-06,
      "loss": 0.5774,
      "step": 3343
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.316699815242887,
      "learning_rate": 4.124029887384511e-06,
      "loss": 0.3151,
      "step": 3344
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.2648791595738205,
      "learning_rate": 4.12348351520456e-06,
      "loss": 0.3268,
      "step": 3345
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8254236464418776,
      "learning_rate": 4.122937008900361e-06,
      "loss": 0.2711,
      "step": 3346
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.16294376146496,
      "learning_rate": 4.122390368517063e-06,
      "loss": 0.3093,
      "step": 3347
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.299427102160675,
      "learning_rate": 4.121843594099827e-06,
      "loss": 0.5091,
      "step": 3348
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8047292133781117,
      "learning_rate": 4.121296685693824e-06,
      "loss": 0.2643,
      "step": 3349
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8782799007482445,
      "learning_rate": 4.1207496433442376e-06,
      "loss": 0.37,
      "step": 3350
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.741670060759917,
      "learning_rate": 4.1202024670962624e-06,
      "loss": 0.4058,
      "step": 3351
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.34518253892478,
      "learning_rate": 4.119655156995103e-06,
      "loss": 0.5425,
      "step": 3352
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.203046716419482,
      "learning_rate": 4.119107713085974e-06,
      "loss": 0.4761,
      "step": 3353
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.619864125951867,
      "learning_rate": 4.118560135414106e-06,
      "loss": 0.3286,
      "step": 3354
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.3806584394256185,
      "learning_rate": 4.118012424024733e-06,
      "loss": 0.25,
      "step": 3355
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.875103971723555,
      "learning_rate": 4.117464578963107e-06,
      "loss": 0.1872,
      "step": 3356
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.2379371626705753,
      "learning_rate": 4.116916600274487e-06,
      "loss": 0.1919,
      "step": 3357
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.090432072557759,
      "learning_rate": 4.1163684880041455e-06,
      "loss": 0.2202,
      "step": 3358
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.3466086451114596,
      "learning_rate": 4.115820242197364e-06,
      "loss": 0.1862,
      "step": 3359
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.090133446197999,
      "learning_rate": 4.115271862899436e-06,
      "loss": 0.1091,
      "step": 3360
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.820052360621459,
      "learning_rate": 4.114723350155666e-06,
      "loss": 0.3517,
      "step": 3361
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.2316077335394295,
      "learning_rate": 4.114174704011369e-06,
      "loss": 0.3961,
      "step": 3362
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.803515303020003,
      "learning_rate": 4.113625924511872e-06,
      "loss": 0.4647,
      "step": 3363
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.431498714465329,
      "learning_rate": 4.113077011702512e-06,
      "loss": 0.6164,
      "step": 3364
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.579422820661966,
      "learning_rate": 4.112527965628639e-06,
      "loss": 0.2182,
      "step": 3365
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.83567799535976,
      "learning_rate": 4.111978786335609e-06,
      "loss": 0.6127,
      "step": 3366
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4499257817314806,
      "learning_rate": 4.111429473868796e-06,
      "loss": 0.0148,
      "step": 3367
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.268848318291087,
      "learning_rate": 4.110880028273581e-06,
      "loss": 0.3341,
      "step": 3368
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.863924388797465,
      "learning_rate": 4.110330449595356e-06,
      "loss": 0.1332,
      "step": 3369
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.9629848275722637,
      "learning_rate": 4.1097807378795225e-06,
      "loss": 0.2534,
      "step": 3370
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.85134494986158,
      "learning_rate": 4.1092308931714985e-06,
      "loss": 0.2448,
      "step": 3371
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.926152362977245,
      "learning_rate": 4.108680915516707e-06,
      "loss": 0.4659,
      "step": 3372
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.101626248604234,
      "learning_rate": 4.108130804960585e-06,
      "loss": 0.2063,
      "step": 3373
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.723392454183246,
      "learning_rate": 4.1075805615485804e-06,
      "loss": 0.3571,
      "step": 3374
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.278193806058931,
      "learning_rate": 4.107030185326152e-06,
      "loss": 0.2733,
      "step": 3375
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.349189817144429,
      "learning_rate": 4.106479676338767e-06,
      "loss": 0.2304,
      "step": 3376
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.012663924975848,
      "learning_rate": 4.105929034631908e-06,
      "loss": 0.2786,
      "step": 3377
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.455159965800529,
      "learning_rate": 4.1053782602510665e-06,
      "loss": 0.3531,
      "step": 3378
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.341749665869069,
      "learning_rate": 4.104827353241743e-06,
      "loss": 0.3712,
      "step": 3379
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.636906811134324,
      "learning_rate": 4.104276313649451e-06,
      "loss": 0.6412,
      "step": 3380
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.1999482046529137,
      "learning_rate": 4.103725141519716e-06,
      "loss": 0.1951,
      "step": 3381
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.998852766425047,
      "learning_rate": 4.103173836898073e-06,
      "loss": 0.3421,
      "step": 3382
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.735453630348677,
      "learning_rate": 4.102622399830066e-06,
      "loss": 0.6189,
      "step": 3383
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.326048128964303,
      "learning_rate": 4.102070830361256e-06,
      "loss": 0.3254,
      "step": 3384
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.87249103740466,
      "learning_rate": 4.1015191285372065e-06,
      "loss": 0.1558,
      "step": 3385
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.346748604912007,
      "learning_rate": 4.100967294403498e-06,
      "loss": 0.409,
      "step": 3386
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.8192782120369957,
      "learning_rate": 4.100415328005722e-06,
      "loss": 0.1279,
      "step": 3387
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.887835876079366,
      "learning_rate": 4.099863229389478e-06,
      "loss": 0.3838,
      "step": 3388
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.329356846082889,
      "learning_rate": 4.099310998600378e-06,
      "loss": 0.2244,
      "step": 3389
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.931701061654482,
      "learning_rate": 4.098758635684044e-06,
      "loss": 0.219,
      "step": 3390
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.7311517177010574,
      "learning_rate": 4.098206140686109e-06,
      "loss": 0.3153,
      "step": 3391
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.785866139618991,
      "learning_rate": 4.097653513652219e-06,
      "loss": 0.2165,
      "step": 3392
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.0789621357351,
      "learning_rate": 4.0971007546280285e-06,
      "loss": 0.2137,
      "step": 3393
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.401390177543237,
      "learning_rate": 4.0965478636592036e-06,
      "loss": 0.3414,
      "step": 3394
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.740548888463525,
      "learning_rate": 4.095994840791422e-06,
      "loss": 0.4013,
      "step": 3395
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.136817806336687,
      "learning_rate": 4.095441686070371e-06,
      "loss": 0.2977,
      "step": 3396
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.700495873905837,
      "learning_rate": 4.094888399541751e-06,
      "loss": 0.4184,
      "step": 3397
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.21196485361704,
      "learning_rate": 4.094334981251269e-06,
      "loss": 0.2586,
      "step": 3398
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.170687513235782,
      "learning_rate": 4.093781431244649e-06,
      "loss": 0.304,
      "step": 3399
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6357762554933375,
      "learning_rate": 4.0932277495676194e-06,
      "loss": 0.2561,
      "step": 3400
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.123272847393897,
      "learning_rate": 4.092673936265925e-06,
      "loss": 0.4456,
      "step": 3401
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.59205916227652,
      "learning_rate": 4.092119991385319e-06,
      "loss": 0.2471,
      "step": 3402
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.778226271893621,
      "learning_rate": 4.091565914971565e-06,
      "loss": 0.5829,
      "step": 3403
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.9155898169373247,
      "learning_rate": 4.091011707070438e-06,
      "loss": 0.2932,
      "step": 3404
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.0357420919180305,
      "learning_rate": 4.090457367727723e-06,
      "loss": 0.335,
      "step": 3405
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.95478936533657,
      "learning_rate": 4.089902896989219e-06,
      "loss": 0.4115,
      "step": 3406
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.353385158663479,
      "learning_rate": 4.089348294900733e-06,
      "loss": 0.4939,
      "step": 3407
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.844532686965758,
      "learning_rate": 4.088793561508083e-06,
      "loss": 0.2703,
      "step": 3408
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6717405886167787,
      "learning_rate": 4.088238696857097e-06,
      "loss": 0.2319,
      "step": 3409
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.310120408410163,
      "learning_rate": 4.0876837009936175e-06,
      "loss": 0.2243,
      "step": 3410
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.525225898368687,
      "learning_rate": 4.0871285739634945e-06,
      "loss": 0.2715,
      "step": 3411
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.516909734687219,
      "learning_rate": 4.08657331581259e-06,
      "loss": 0.3347,
      "step": 3412
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.1677036953785525,
      "learning_rate": 4.0860179265867765e-06,
      "loss": 0.2317,
      "step": 3413
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.864978714612132,
      "learning_rate": 4.085462406331938e-06,
      "loss": 0.3552,
      "step": 3414
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.5022083294690445,
      "learning_rate": 4.084906755093969e-06,
      "loss": 0.1526,
      "step": 3415
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.5224826580080353,
      "learning_rate": 4.084350972918774e-06,
      "loss": 0.1406,
      "step": 3416
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.538982148227927,
      "learning_rate": 4.083795059852268e-06,
      "loss": 0.1987,
      "step": 3417
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.65252055105086,
      "learning_rate": 4.08323901594038e-06,
      "loss": 0.4937,
      "step": 3418
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.373452067676075,
      "learning_rate": 4.082682841229048e-06,
      "loss": 0.4918,
      "step": 3419
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.867630167351437,
      "learning_rate": 4.082126535764217e-06,
      "loss": 0.5664,
      "step": 3420
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.050562300497315,
      "learning_rate": 4.081570099591849e-06,
      "loss": 0.2799,
      "step": 3421
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.695098567368668,
      "learning_rate": 4.081013532757914e-06,
      "loss": 0.1365,
      "step": 3422
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.260630205107885,
      "learning_rate": 4.080456835308392e-06,
      "loss": 0.3404,
      "step": 3423
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.300482197373418,
      "learning_rate": 4.079900007289275e-06,
      "loss": 0.1981,
      "step": 3424
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.156933139622203,
      "learning_rate": 4.079343048746565e-06,
      "loss": 0.5173,
      "step": 3425
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.652559447392053,
      "learning_rate": 4.078785959726275e-06,
      "loss": 0.306,
      "step": 3426
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.7179108038513786,
      "learning_rate": 4.07822874027443e-06,
      "loss": 0.2532,
      "step": 3427
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.4483699265143155,
      "learning_rate": 4.0776713904370645e-06,
      "loss": 0.2491,
      "step": 3428
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.424822019938096,
      "learning_rate": 4.077113910260222e-06,
      "loss": 0.345,
      "step": 3429
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.314161488105666,
      "learning_rate": 4.0765562997899625e-06,
      "loss": 0.4061,
      "step": 3430
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.276746522524165,
      "learning_rate": 4.075998559072349e-06,
      "loss": 0.5404,
      "step": 3431
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.364948799204063,
      "learning_rate": 4.075440688153462e-06,
      "loss": 0.3905,
      "step": 3432
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.274590574196186,
      "learning_rate": 4.07488268707939e-06,
      "loss": 0.2254,
      "step": 3433
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.834153519364844,
      "learning_rate": 4.074324555896231e-06,
      "loss": 0.4051,
      "step": 3434
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.684643084447493,
      "learning_rate": 4.073766294650096e-06,
      "loss": 0.2358,
      "step": 3435
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8826854992725934,
      "learning_rate": 4.073207903387105e-06,
      "loss": 0.1521,
      "step": 3436
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.914320830816995,
      "learning_rate": 4.0726493821533904e-06,
      "loss": 0.2243,
      "step": 3437
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.5084339522645505,
      "learning_rate": 4.072090730995094e-06,
      "loss": 0.3183,
      "step": 3438
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.744971232484636,
      "learning_rate": 4.071531949958368e-06,
      "loss": 0.1801,
      "step": 3439
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.821409755455583,
      "learning_rate": 4.070973039089379e-06,
      "loss": 0.3201,
      "step": 3440
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.318000798046844,
      "learning_rate": 4.0704139984342985e-06,
      "loss": 0.0548,
      "step": 3441
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.481227366351618,
      "learning_rate": 4.069854828039313e-06,
      "loss": 0.1784,
      "step": 3442
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.903605758581183,
      "learning_rate": 4.069295527950619e-06,
      "loss": 0.3453,
      "step": 3443
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.782954173046914,
      "learning_rate": 4.0687360982144204e-06,
      "loss": 0.3176,
      "step": 3444
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.318973813536676,
      "learning_rate": 4.068176538876938e-06,
      "loss": 0.3046,
      "step": 3445
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.475961040821014,
      "learning_rate": 4.067616849984397e-06,
      "loss": 0.5805,
      "step": 3446
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.748033890800679,
      "learning_rate": 4.067057031583039e-06,
      "loss": 0.2378,
      "step": 3447
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.954995559632398,
      "learning_rate": 4.066497083719111e-06,
      "loss": 0.3037,
      "step": 3448
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.392084858491689,
      "learning_rate": 4.065937006438874e-06,
      "loss": 0.4588,
      "step": 3449
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.671948791763504,
      "learning_rate": 4.0653767997886e-06,
      "loss": 0.5665,
      "step": 3450
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.2455246909765836,
      "learning_rate": 4.0648164638145684e-06,
      "loss": 0.204,
      "step": 3451
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.320550402692458,
      "learning_rate": 4.064255998563073e-06,
      "loss": 0.4797,
      "step": 3452
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.251335350131432,
      "learning_rate": 4.063695404080415e-06,
      "loss": 0.2544,
      "step": 3453
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.103598392553635,
      "learning_rate": 4.0631346804129105e-06,
      "loss": 0.249,
      "step": 3454
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.066500064590892,
      "learning_rate": 4.062573827606881e-06,
      "loss": 0.4378,
      "step": 3455
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.091128604355378,
      "learning_rate": 4.0620128457086635e-06,
      "loss": 0.2921,
      "step": 3456
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.194813728683406,
      "learning_rate": 4.0614517347646035e-06,
      "loss": 0.5169,
      "step": 3457
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.494179180978702,
      "learning_rate": 4.060890494821055e-06,
      "loss": 0.4063,
      "step": 3458
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.152456225852616,
      "learning_rate": 4.060329125924388e-06,
      "loss": 0.4418,
      "step": 3459
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.004664540687594,
      "learning_rate": 4.059767628120978e-06,
      "loss": 0.1998,
      "step": 3460
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.660311536004401,
      "learning_rate": 4.059206001457213e-06,
      "loss": 0.2838,
      "step": 3461
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.9612986584685244,
      "learning_rate": 4.058644245979493e-06,
      "loss": 0.3491,
      "step": 3462
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.27832894111899,
      "learning_rate": 4.058082361734227e-06,
      "loss": 0.5534,
      "step": 3463
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3646996520981527,
      "learning_rate": 4.057520348767835e-06,
      "loss": 0.2713,
      "step": 3464
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.130843341010224,
      "learning_rate": 4.056958207126749e-06,
      "loss": 0.3193,
      "step": 3465
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.846638321494829,
      "learning_rate": 4.056395936857408e-06,
      "loss": 0.3661,
      "step": 3466
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.794551509109857,
      "learning_rate": 4.055833538006265e-06,
      "loss": 0.4016,
      "step": 3467
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.77048829312485,
      "learning_rate": 4.055271010619784e-06,
      "loss": 0.3126,
      "step": 3468
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.449404173536539,
      "learning_rate": 4.054708354744437e-06,
      "loss": 0.2741,
      "step": 3469
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.724548416456885,
      "learning_rate": 4.054145570426707e-06,
      "loss": 0.3487,
      "step": 3470
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.330331323682189,
      "learning_rate": 4.053582657713091e-06,
      "loss": 0.4276,
      "step": 3471
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.3308721250191318,
      "learning_rate": 4.053019616650092e-06,
      "loss": 0.221,
      "step": 3472
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.656169727358784,
      "learning_rate": 4.052456447284226e-06,
      "loss": 0.4939,
      "step": 3473
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.480045597789725,
      "learning_rate": 4.05189314966202e-06,
      "loss": 0.2372,
      "step": 3474
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.6719027246812055,
      "learning_rate": 4.051329723830011e-06,
      "loss": 0.3284,
      "step": 3475
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.724219500875463,
      "learning_rate": 4.050766169834745e-06,
      "loss": 0.3993,
      "step": 3476
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.664871413850912,
      "learning_rate": 4.050202487722782e-06,
      "loss": 0.4284,
      "step": 3477
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.899362586954651,
      "learning_rate": 4.049638677540688e-06,
      "loss": 0.3453,
      "step": 3478
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.9634368558736695,
      "learning_rate": 4.049074739335046e-06,
      "loss": 0.352,
      "step": 3479
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.068607551632309,
      "learning_rate": 4.0485106731524426e-06,
      "loss": 0.3968,
      "step": 3480
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3319379570457213,
      "learning_rate": 4.04794647903948e-06,
      "loss": 0.0242,
      "step": 3481
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.983955833487037,
      "learning_rate": 4.0473821570427675e-06,
      "loss": 0.3978,
      "step": 3482
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.725525982635452,
      "learning_rate": 4.046817707208927e-06,
      "loss": 0.4448,
      "step": 3483
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.221268737089714,
      "learning_rate": 4.046253129584593e-06,
      "loss": 0.5217,
      "step": 3484
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.53432285647861,
      "learning_rate": 4.0456884242164055e-06,
      "loss": 0.3834,
      "step": 3485
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.921217663001062,
      "learning_rate": 4.0451235911510185e-06,
      "loss": 0.1437,
      "step": 3486
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.840622187146103,
      "learning_rate": 4.044558630435096e-06,
      "loss": 0.3204,
      "step": 3487
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.953766043213206,
      "learning_rate": 4.043993542115312e-06,
      "loss": 0.3992,
      "step": 3488
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.525339209642273,
      "learning_rate": 4.0434283262383514e-06,
      "loss": 0.3758,
      "step": 3489
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.146662889243512,
      "learning_rate": 4.0428629828509094e-06,
      "loss": 0.4347,
      "step": 3490
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.516971163370045,
      "learning_rate": 4.042297511999691e-06,
      "loss": 0.4383,
      "step": 3491
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.1285365825737115,
      "learning_rate": 4.0417319137314145e-06,
      "loss": 0.1983,
      "step": 3492
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.637877286129634,
      "learning_rate": 4.041166188092807e-06,
      "loss": 0.6262,
      "step": 3493
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.550772324265268,
      "learning_rate": 4.040600335130603e-06,
      "loss": 0.4735,
      "step": 3494
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.611379966237507,
      "learning_rate": 4.040034354891554e-06,
      "loss": 0.4857,
      "step": 3495
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7374177691620203,
      "learning_rate": 4.039468247422416e-06,
      "loss": 0.0549,
      "step": 3496
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.736634501705318,
      "learning_rate": 4.038902012769959e-06,
      "loss": 0.3313,
      "step": 3497
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.023327977449651,
      "learning_rate": 4.0383356509809636e-06,
      "loss": 0.2937,
      "step": 3498
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.421667205929566,
      "learning_rate": 4.0377691621022185e-06,
      "loss": 0.3516,
      "step": 3499
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.204264712890094,
      "learning_rate": 4.0372025461805245e-06,
      "loss": 0.1779,
      "step": 3500
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.737698575734867,
      "learning_rate": 4.036635803262692e-06,
      "loss": 0.2743,
      "step": 3501
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.90155496384385,
      "learning_rate": 4.036068933395543e-06,
      "loss": 0.149,
      "step": 3502
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.771484437992958,
      "learning_rate": 4.03550193662591e-06,
      "loss": 0.4127,
      "step": 3503
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.120811748417529,
      "learning_rate": 4.034934813000635e-06,
      "loss": 0.3131,
      "step": 3504
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.252831452522409,
      "learning_rate": 4.0343675625665724e-06,
      "loss": 0.5724,
      "step": 3505
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.1595679187794494,
      "learning_rate": 4.033800185370582e-06,
      "loss": 0.2489,
      "step": 3506
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.644706805463049,
      "learning_rate": 4.0332326814595425e-06,
      "loss": 0.4265,
      "step": 3507
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.7231727517352144,
      "learning_rate": 4.0326650508803345e-06,
      "loss": 0.2634,
      "step": 3508
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.298222012614571,
      "learning_rate": 4.032097293679855e-06,
      "loss": 0.4571,
      "step": 3509
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.9446801668069,
      "learning_rate": 4.031529409905008e-06,
      "loss": 0.3951,
      "step": 3510
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.768075817282496,
      "learning_rate": 4.03096139960271e-06,
      "loss": 0.4859,
      "step": 3511
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.911535102173915,
      "learning_rate": 4.030393262819887e-06,
      "loss": 0.3341,
      "step": 3512
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.248689696454667,
      "learning_rate": 4.029824999603476e-06,
      "loss": 0.1009,
      "step": 3513
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.60596374315893,
      "learning_rate": 4.029256610000424e-06,
      "loss": 0.3347,
      "step": 3514
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.313092744977978,
      "learning_rate": 4.0286880940576875e-06,
      "loss": 0.1376,
      "step": 3515
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.254353125245911,
      "learning_rate": 4.028119451822235e-06,
      "loss": 0.3453,
      "step": 3516
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.042221494338329,
      "learning_rate": 4.027550683341047e-06,
      "loss": 0.3963,
      "step": 3517
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.090827982729387,
      "learning_rate": 4.026981788661111e-06,
      "loss": 0.1426,
      "step": 3518
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.63900571173211,
      "learning_rate": 4.026412767829425e-06,
      "loss": 0.4051,
      "step": 3519
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.269775316362526,
      "learning_rate": 4.025843620893e-06,
      "loss": 0.507,
      "step": 3520
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.4710480820806344,
      "learning_rate": 4.025274347898855e-06,
      "loss": 0.2825,
      "step": 3521
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.659252830329387,
      "learning_rate": 4.0247049488940225e-06,
      "loss": 0.3053,
      "step": 3522
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.122734129038974,
      "learning_rate": 4.0241354239255415e-06,
      "loss": 0.5461,
      "step": 3523
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.9360574715808494,
      "learning_rate": 4.023565773040465e-06,
      "loss": 0.2402,
      "step": 3524
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.938574849900348,
      "learning_rate": 4.022995996285853e-06,
      "loss": 0.4619,
      "step": 3525
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.408620091063064,
      "learning_rate": 4.02242609370878e-06,
      "loss": 0.4783,
      "step": 3526
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.021783409231752,
      "learning_rate": 4.021856065356326e-06,
      "loss": 0.2554,
      "step": 3527
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.298944278279803,
      "learning_rate": 4.021285911275586e-06,
      "loss": 0.2948,
      "step": 3528
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.902763196154152,
      "learning_rate": 4.020715631513661e-06,
      "loss": 0.1341,
      "step": 3529
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.6418152243115,
      "learning_rate": 4.0201452261176675e-06,
      "loss": 0.3919,
      "step": 3530
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.585169074549058,
      "learning_rate": 4.019574695134728e-06,
      "loss": 0.1665,
      "step": 3531
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.981232240650717,
      "learning_rate": 4.0190040386119775e-06,
      "loss": 0.4136,
      "step": 3532
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.208733242822764,
      "learning_rate": 4.01843325659656e-06,
      "loss": 0.4488,
      "step": 3533
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.865038638487144,
      "learning_rate": 4.017862349135632e-06,
      "loss": 0.3954,
      "step": 3534
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.388596617702835,
      "learning_rate": 4.017291316276358e-06,
      "loss": 0.253,
      "step": 3535
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.135658733920124,
      "learning_rate": 4.016720158065914e-06,
      "loss": 0.2007,
      "step": 3536
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2947503543409704,
      "learning_rate": 4.0161488745514875e-06,
      "loss": 0.0969,
      "step": 3537
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.990722097245135,
      "learning_rate": 4.015577465780274e-06,
      "loss": 0.3425,
      "step": 3538
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.172014522454199,
      "learning_rate": 4.015005931799481e-06,
      "loss": 0.2238,
      "step": 3539
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.120920404428767,
      "learning_rate": 4.014434272656324e-06,
      "loss": 0.2376,
      "step": 3540
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.441100010897445,
      "learning_rate": 4.013862488398034e-06,
      "loss": 0.3901,
      "step": 3541
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.195442989795918,
      "learning_rate": 4.013290579071847e-06,
      "loss": 0.3592,
      "step": 3542
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.872260564764381,
      "learning_rate": 4.012718544725011e-06,
      "loss": 0.3279,
      "step": 3543
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.739918177990012,
      "learning_rate": 4.012146385404785e-06,
      "loss": 0.3788,
      "step": 3544
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.782925706364858,
      "learning_rate": 4.01157410115844e-06,
      "loss": 0.0796,
      "step": 3545
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.820589303322385,
      "learning_rate": 4.011001692033252e-06,
      "loss": 0.316,
      "step": 3546
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.278669615844711,
      "learning_rate": 4.010429158076513e-06,
      "loss": 0.3968,
      "step": 3547
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.498614322292342,
      "learning_rate": 4.009856499335522e-06,
      "loss": 0.3198,
      "step": 3548
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.9002288522037025,
      "learning_rate": 4.009283715857589e-06,
      "loss": 0.3607,
      "step": 3549
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.889182393150909,
      "learning_rate": 4.0087108076900356e-06,
      "loss": 0.2423,
      "step": 3550
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.346703246979389,
      "learning_rate": 4.0081377748801916e-06,
      "loss": 0.2334,
      "step": 3551
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.088780190667666,
      "learning_rate": 4.007564617475399e-06,
      "loss": 0.3552,
      "step": 3552
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.885641574617801,
      "learning_rate": 4.0069913355230085e-06,
      "loss": 0.1817,
      "step": 3553
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.2922624065339985,
      "learning_rate": 4.006417929070383e-06,
      "loss": 0.1669,
      "step": 3554
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.470884760143504,
      "learning_rate": 4.005844398164893e-06,
      "loss": 0.4361,
      "step": 3555
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3362332835717443,
      "learning_rate": 4.005270742853921e-06,
      "loss": 0.2298,
      "step": 3556
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.606069145516631,
      "learning_rate": 4.004696963184862e-06,
      "loss": 0.299,
      "step": 3557
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.182713418012491,
      "learning_rate": 4.004123059205116e-06,
      "loss": 0.2146,
      "step": 3558
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.362334965433888,
      "learning_rate": 4.003549030962097e-06,
      "loss": 0.2781,
      "step": 3559
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.6761909835844793,
      "learning_rate": 4.002974878503229e-06,
      "loss": 0.2197,
      "step": 3560
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.8480271514583015,
      "learning_rate": 4.0024006018759455e-06,
      "loss": 0.2782,
      "step": 3561
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.988973477492796,
      "learning_rate": 4.00182620112769e-06,
      "loss": 0.353,
      "step": 3562
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1722619852263674,
      "learning_rate": 4.001251676305916e-06,
      "loss": 0.0764,
      "step": 3563
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.589696025431677,
      "learning_rate": 4.00067702745809e-06,
      "loss": 0.2432,
      "step": 3564
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.469428273297105,
      "learning_rate": 4.000102254631684e-06,
      "loss": 0.5094,
      "step": 3565
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.891011134467451,
      "learning_rate": 3.999527357874186e-06,
      "loss": 0.3795,
      "step": 3566
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.073044932396124,
      "learning_rate": 3.998952337233088e-06,
      "loss": 0.3614,
      "step": 3567
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.075843373069815,
      "learning_rate": 3.998377192755896e-06,
      "loss": 0.2595,
      "step": 3568
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.5549451583854683,
      "learning_rate": 3.997801924490129e-06,
      "loss": 0.2365,
      "step": 3569
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.4957200663521206,
      "learning_rate": 3.997226532483309e-06,
      "loss": 0.2217,
      "step": 3570
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.537320126424226,
      "learning_rate": 3.996651016782974e-06,
      "loss": 0.3292,
      "step": 3571
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.523532242016407,
      "learning_rate": 3.996075377436669e-06,
      "loss": 0.4345,
      "step": 3572
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.221846580114407,
      "learning_rate": 3.9954996144919514e-06,
      "loss": 0.4578,
      "step": 3573
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0530889424483503,
      "learning_rate": 3.9949237279963876e-06,
      "loss": 0.124,
      "step": 3574
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.9573077884983223,
      "learning_rate": 3.994347717997555e-06,
      "loss": 0.1927,
      "step": 3575
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.444191910157606,
      "learning_rate": 3.9937715845430405e-06,
      "loss": 0.2955,
      "step": 3576
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.8634116672513894,
      "learning_rate": 3.99319532768044e-06,
      "loss": 0.2205,
      "step": 3577
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.074193375134128,
      "learning_rate": 3.992618947457363e-06,
      "loss": 0.2867,
      "step": 3578
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.795736072382943,
      "learning_rate": 3.992042443921427e-06,
      "loss": 0.2933,
      "step": 3579
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.914703541411304,
      "learning_rate": 3.991465817120259e-06,
      "loss": 0.4961,
      "step": 3580
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.174041843757733,
      "learning_rate": 3.990889067101498e-06,
      "loss": 0.5977,
      "step": 3581
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.136650898065782,
      "learning_rate": 3.99031219391279e-06,
      "loss": 0.374,
      "step": 3582
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.65091818357623,
      "learning_rate": 3.989735197601796e-06,
      "loss": 0.6856,
      "step": 3583
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8713371567233983,
      "learning_rate": 3.9891580782161844e-06,
      "loss": 0.168,
      "step": 3584
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.773916410573125,
      "learning_rate": 3.988580835803633e-06,
      "loss": 0.3784,
      "step": 3585
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.242614076398955,
      "learning_rate": 3.988003470411831e-06,
      "loss": 0.3338,
      "step": 3586
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.992712090516937,
      "learning_rate": 3.987425982088477e-06,
      "loss": 0.4071,
      "step": 3587
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.8871393970736268,
      "learning_rate": 3.986848370881282e-06,
      "loss": 0.2822,
      "step": 3588
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.851872510220802,
      "learning_rate": 3.986270636837963e-06,
      "loss": 0.2829,
      "step": 3589
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.047377581168256,
      "learning_rate": 3.985692780006251e-06,
      "loss": 0.2196,
      "step": 3590
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.37986981219306,
      "learning_rate": 3.985114800433886e-06,
      "loss": 0.0304,
      "step": 3591
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6121190379150425,
      "learning_rate": 3.984536698168618e-06,
      "loss": 0.123,
      "step": 3592
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3707391332661705,
      "learning_rate": 3.983958473258205e-06,
      "loss": 0.2003,
      "step": 3593
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.658240050302396,
      "learning_rate": 3.983380125750419e-06,
      "loss": 0.3395,
      "step": 3594
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.454177433289474,
      "learning_rate": 3.982801655693039e-06,
      "loss": 0.2544,
      "step": 3595
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.903450098168645,
      "learning_rate": 3.982223063133857e-06,
      "loss": 0.5231,
      "step": 3596
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.587839855513092,
      "learning_rate": 3.981644348120672e-06,
      "loss": 0.1941,
      "step": 3597
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.806426181204499,
      "learning_rate": 3.981065510701295e-06,
      "loss": 0.3462,
      "step": 3598
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.305314578343846,
      "learning_rate": 3.980486550923547e-06,
      "loss": 0.2488,
      "step": 3599
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.174580496875258,
      "learning_rate": 3.979907468835258e-06,
      "loss": 0.0983,
      "step": 3600
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.700840327190412,
      "learning_rate": 3.97932826448427e-06,
      "loss": 0.3434,
      "step": 3601
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.0702216733424,
      "learning_rate": 3.978748937918433e-06,
      "loss": 0.5585,
      "step": 3602
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.952909147993611,
      "learning_rate": 3.9781694891856095e-06,
      "loss": 0.5694,
      "step": 3603
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.941090850198945,
      "learning_rate": 3.97758991833367e-06,
      "loss": 0.1053,
      "step": 3604
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.124989746142513,
      "learning_rate": 3.9770102254104945e-06,
      "loss": 0.2957,
      "step": 3605
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.141896823734735,
      "learning_rate": 3.976430410463976e-06,
      "loss": 0.3753,
      "step": 3606
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8386611368146175,
      "learning_rate": 3.975850473542017e-06,
      "loss": 0.1382,
      "step": 3607
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.494427785977769,
      "learning_rate": 3.975270414692526e-06,
      "loss": 0.4869,
      "step": 3608
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.812841301321538,
      "learning_rate": 3.974690233963426e-06,
      "loss": 0.1118,
      "step": 3609
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.830261563117664,
      "learning_rate": 3.97410993140265e-06,
      "loss": 0.509,
      "step": 3610
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.374738297713571,
      "learning_rate": 3.9735295070581375e-06,
      "loss": 0.3129,
      "step": 3611
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.180937993397005,
      "learning_rate": 3.972948960977843e-06,
      "loss": 0.343,
      "step": 3612
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9698728036050537,
      "learning_rate": 3.972368293209725e-06,
      "loss": 0.1524,
      "step": 3613
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.5905179197763895,
      "learning_rate": 3.971787503801758e-06,
      "loss": 0.4856,
      "step": 3614
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.3129126895791305,
      "learning_rate": 3.971206592801924e-06,
      "loss": 0.4234,
      "step": 3615
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.159095660868645,
      "learning_rate": 3.970625560258213e-06,
      "loss": 0.3247,
      "step": 3616
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.1417660555501845,
      "learning_rate": 3.97004440621863e-06,
      "loss": 0.5819,
      "step": 3617
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1093900091841618,
      "learning_rate": 3.969463130731183e-06,
      "loss": 0.1776,
      "step": 3618
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.315893479987707,
      "learning_rate": 3.968881733843898e-06,
      "loss": 0.4826,
      "step": 3619
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.4140818504636297,
      "learning_rate": 3.9683002156048055e-06,
      "loss": 0.1483,
      "step": 3620
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.984148571629504,
      "learning_rate": 3.9677185760619475e-06,
      "loss": 0.2073,
      "step": 3621
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.160564414514489,
      "learning_rate": 3.967136815263377e-06,
      "loss": 0.1812,
      "step": 3622
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.7636584300234612,
      "learning_rate": 3.966554933257156e-06,
      "loss": 0.2866,
      "step": 3623
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.425743478981851,
      "learning_rate": 3.965972930091356e-06,
      "loss": 0.3855,
      "step": 3624
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.614338782381921,
      "learning_rate": 3.96539080581406e-06,
      "loss": 0.5589,
      "step": 3625
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.6235856402513456,
      "learning_rate": 3.96480856047336e-06,
      "loss": 0.1387,
      "step": 3626
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.730857165176868,
      "learning_rate": 3.964226194117359e-06,
      "loss": 0.131,
      "step": 3627
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.76047980515848,
      "learning_rate": 3.963643706794168e-06,
      "loss": 0.4482,
      "step": 3628
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.3976920860455655,
      "learning_rate": 3.963061098551909e-06,
      "loss": 0.1635,
      "step": 3629
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.522822220411121,
      "learning_rate": 3.962478369438717e-06,
      "loss": 0.23,
      "step": 3630
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.318736296886839,
      "learning_rate": 3.961895519502731e-06,
      "loss": 0.4828,
      "step": 3631
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0599931634765363,
      "learning_rate": 3.961312548792104e-06,
      "loss": 0.2102,
      "step": 3632
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.6421132266207294,
      "learning_rate": 3.960729457355e-06,
      "loss": 0.3453,
      "step": 3633
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.325259620033761,
      "learning_rate": 3.960146245239589e-06,
      "loss": 0.293,
      "step": 3634
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9807810228200973,
      "learning_rate": 3.9595629124940535e-06,
      "loss": 0.3587,
      "step": 3635
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.063471238356893,
      "learning_rate": 3.958979459166587e-06,
      "loss": 0.2182,
      "step": 3636
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.28055594138382,
      "learning_rate": 3.9583958853053896e-06,
      "loss": 0.5032,
      "step": 3637
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.004656465905728,
      "learning_rate": 3.9578121909586744e-06,
      "loss": 0.1562,
      "step": 3638
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.57311309259876,
      "learning_rate": 3.957228376174663e-06,
      "loss": 0.5642,
      "step": 3639
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.5087874092883014,
      "learning_rate": 3.956644441001589e-06,
      "loss": 0.1124,
      "step": 3640
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.775302731329534,
      "learning_rate": 3.956060385487692e-06,
      "loss": 0.5389,
      "step": 3641
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.0839937317546564,
      "learning_rate": 3.955476209681225e-06,
      "loss": 0.1792,
      "step": 3642
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8668816524980527,
      "learning_rate": 3.954891913630449e-06,
      "loss": 0.0521,
      "step": 3643
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.744355426690695,
      "learning_rate": 3.954307497383636e-06,
      "loss": 0.1705,
      "step": 3644
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.322969857134643,
      "learning_rate": 3.953722960989067e-06,
      "loss": 0.4944,
      "step": 3645
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.010315974035983,
      "learning_rate": 3.953138304495035e-06,
      "loss": 0.4474,
      "step": 3646
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.8577562773582788,
      "learning_rate": 3.9525535279498405e-06,
      "loss": 0.1691,
      "step": 3647
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.6615244655384656,
      "learning_rate": 3.951968631401795e-06,
      "loss": 0.1808,
      "step": 3648
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9345984030419208,
      "learning_rate": 3.9513836148992195e-06,
      "loss": 0.0671,
      "step": 3649
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.218244272597617,
      "learning_rate": 3.950798478490447e-06,
      "loss": 0.4678,
      "step": 3650
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9880001005278247,
      "learning_rate": 3.950213222223815e-06,
      "loss": 0.2871,
      "step": 3651
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.0067252618140765,
      "learning_rate": 3.9496278461476775e-06,
      "loss": 0.268,
      "step": 3652
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.929399627898238,
      "learning_rate": 3.949042350310394e-06,
      "loss": 0.2624,
      "step": 3653
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.917245388719424,
      "learning_rate": 3.948456734760336e-06,
      "loss": 0.4552,
      "step": 3654
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9897016908732916,
      "learning_rate": 3.947870999545885e-06,
      "loss": 0.3487,
      "step": 3655
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.1246463050186355,
      "learning_rate": 3.947285144715429e-06,
      "loss": 0.1955,
      "step": 3656
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.386778313314754,
      "learning_rate": 3.946699170317372e-06,
      "loss": 0.3399,
      "step": 3657
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.817307344415654,
      "learning_rate": 3.94611307640012e-06,
      "loss": 0.5289,
      "step": 3658
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.042214991572931,
      "learning_rate": 3.945526863012097e-06,
      "loss": 0.3532,
      "step": 3659
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.738166995166118,
      "learning_rate": 3.9449405302017315e-06,
      "loss": 0.2388,
      "step": 3660
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.371574692759848,
      "learning_rate": 3.944354078017464e-06,
      "loss": 0.4084,
      "step": 3661
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.6030368867101705,
      "learning_rate": 3.943767506507743e-06,
      "loss": 0.3334,
      "step": 3662
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.313109596009611,
      "learning_rate": 3.9431808157210285e-06,
      "loss": 0.464,
      "step": 3663
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.977103051710538,
      "learning_rate": 3.942594005705791e-06,
      "loss": 0.4651,
      "step": 3664
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.13654134113218,
      "learning_rate": 3.94200707651051e-06,
      "loss": 0.6099,
      "step": 3665
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2639887775208525,
      "learning_rate": 3.9414200281836735e-06,
      "loss": 0.0684,
      "step": 3666
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.130931677336291,
      "learning_rate": 3.940832860773782e-06,
      "loss": 0.3632,
      "step": 3667
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.059316091055098,
      "learning_rate": 3.940245574329342e-06,
      "loss": 0.2567,
      "step": 3668
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.492734380875819,
      "learning_rate": 3.939658168898875e-06,
      "loss": 0.1673,
      "step": 3669
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.5288338337277585,
      "learning_rate": 3.939070644530909e-06,
      "loss": 0.2708,
      "step": 3670
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.633259494672537,
      "learning_rate": 3.9384830012739806e-06,
      "loss": 0.1609,
      "step": 3671
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2219972440082634,
      "learning_rate": 3.937895239176639e-06,
      "loss": 0.2072,
      "step": 3672
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.81392286822732,
      "learning_rate": 3.937307358287443e-06,
      "loss": 0.2579,
      "step": 3673
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.000634284877818,
      "learning_rate": 3.93671935865496e-06,
      "loss": 0.3099,
      "step": 3674
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1326270475143834,
      "learning_rate": 3.936131240327766e-06,
      "loss": 0.0482,
      "step": 3675
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.352491780558864,
      "learning_rate": 3.935543003354452e-06,
      "loss": 0.3996,
      "step": 3676
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.020073727413324,
      "learning_rate": 3.934954647783611e-06,
      "loss": 0.1577,
      "step": 3677
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1952611594562272,
      "learning_rate": 3.934366173663854e-06,
      "loss": 0.1569,
      "step": 3678
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9261035549775437,
      "learning_rate": 3.933777581043795e-06,
      "loss": 0.0056,
      "step": 3679
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.08912957141392,
      "learning_rate": 3.933188869972062e-06,
      "loss": 0.2471,
      "step": 3680
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.567029174197511,
      "learning_rate": 3.932600040497292e-06,
      "loss": 0.2554,
      "step": 3681
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.37383301062063,
      "learning_rate": 3.932011092668128e-06,
      "loss": 0.2952,
      "step": 3682
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.408599071179716,
      "learning_rate": 3.93142202653323e-06,
      "loss": 0.2397,
      "step": 3683
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.532105076297745,
      "learning_rate": 3.930832842141262e-06,
      "loss": 0.2275,
      "step": 3684
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.368122043782301,
      "learning_rate": 3.9302435395409e-06,
      "loss": 0.3121,
      "step": 3685
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.5515813568705723,
      "learning_rate": 3.929654118780828e-06,
      "loss": 0.2789,
      "step": 3686
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2652246255301245,
      "learning_rate": 3.929064579909742e-06,
      "loss": 0.2589,
      "step": 3687
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.148622971770795,
      "learning_rate": 3.928474922976346e-06,
      "loss": 0.1682,
      "step": 3688
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.881524782998523,
      "learning_rate": 3.9278851480293565e-06,
      "loss": 0.4756,
      "step": 3689
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.488043269261111,
      "learning_rate": 3.927295255117497e-06,
      "loss": 0.0983,
      "step": 3690
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.281772917797114,
      "learning_rate": 3.9267052442895e-06,
      "loss": 0.1384,
      "step": 3691
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.278547056851671,
      "learning_rate": 3.926115115594111e-06,
      "loss": 0.5672,
      "step": 3692
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.6631894072816324,
      "learning_rate": 3.925524869080083e-06,
      "loss": 0.4326,
      "step": 3693
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.926579822937851,
      "learning_rate": 3.92493450479618e-06,
      "loss": 0.3006,
      "step": 3694
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.790260932186326,
      "learning_rate": 3.9243440227911725e-06,
      "loss": 0.2461,
      "step": 3695
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.153972356641021,
      "learning_rate": 3.923753423113846e-06,
      "loss": 0.3273,
      "step": 3696
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.944677265099221,
      "learning_rate": 3.923162705812992e-06,
      "loss": 0.4691,
      "step": 3697
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.987072393276575,
      "learning_rate": 3.922571870937412e-06,
      "loss": 0.1746,
      "step": 3698
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.421529978633327,
      "learning_rate": 3.921980918535921e-06,
      "loss": 0.4056,
      "step": 3699
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.512516262938848,
      "learning_rate": 3.9213898486573355e-06,
      "loss": 0.2069,
      "step": 3700
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.531811008380078,
      "learning_rate": 3.92079866135049e-06,
      "loss": 0.3845,
      "step": 3701
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.504957224227891,
      "learning_rate": 3.920207356664225e-06,
      "loss": 0.2611,
      "step": 3702
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.9783759319313505,
      "learning_rate": 3.919615934647392e-06,
      "loss": 0.155,
      "step": 3703
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.910426718293424,
      "learning_rate": 3.919024395348849e-06,
      "loss": 0.1906,
      "step": 3704
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.503373893051257,
      "learning_rate": 3.918432738817468e-06,
      "loss": 0.4589,
      "step": 3705
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9088990357687385,
      "learning_rate": 3.917840965102129e-06,
      "loss": 0.2226,
      "step": 3706
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.462870425319588,
      "learning_rate": 3.91724907425172e-06,
      "loss": 0.4914,
      "step": 3707
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1598922299395498,
      "learning_rate": 3.916657066315141e-06,
      "loss": 0.1,
      "step": 3708
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.08087435749273,
      "learning_rate": 3.916064941341301e-06,
      "loss": 0.3363,
      "step": 3709
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.7050390201741923,
      "learning_rate": 3.915472699379118e-06,
      "loss": 0.3564,
      "step": 3710
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.678353614709266,
      "learning_rate": 3.914880340477521e-06,
      "loss": 0.4177,
      "step": 3711
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.492813348071021,
      "learning_rate": 3.914287864685445e-06,
      "loss": 0.4193,
      "step": 3712
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.807770162143156,
      "learning_rate": 3.913695272051842e-06,
      "loss": 0.545,
      "step": 3713
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.545662633679307,
      "learning_rate": 3.913102562625665e-06,
      "loss": 0.0918,
      "step": 3714
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.090736610674807,
      "learning_rate": 3.9125097364558824e-06,
      "loss": 0.2218,
      "step": 3715
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.7970486810390276,
      "learning_rate": 3.911916793591471e-06,
      "loss": 0.1384,
      "step": 3716
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.564380095723674,
      "learning_rate": 3.911323734081416e-06,
      "loss": 0.1678,
      "step": 3717
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9924069373815785,
      "learning_rate": 3.910730557974714e-06,
      "loss": 0.3572,
      "step": 3718
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.06853429354646,
      "learning_rate": 3.91013726532037e-06,
      "loss": 0.3801,
      "step": 3719
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.695947894586772,
      "learning_rate": 3.909543856167398e-06,
      "loss": 0.5536,
      "step": 3720
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.00575807912343,
      "learning_rate": 3.908950330564823e-06,
      "loss": 0.387,
      "step": 3721
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.580341329535757,
      "learning_rate": 3.9083566885616796e-06,
      "loss": 0.2684,
      "step": 3722
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.094618311050671,
      "learning_rate": 3.9077629302070116e-06,
      "loss": 0.5142,
      "step": 3723
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.7517202666439675,
      "learning_rate": 3.907169055549872e-06,
      "loss": 0.2951,
      "step": 3724
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.54367315558692,
      "learning_rate": 3.906575064639324e-06,
      "loss": 0.2188,
      "step": 3725
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.2885088845619403,
      "learning_rate": 3.90598095752444e-06,
      "loss": 0.2679,
      "step": 3726
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.064362453748705,
      "learning_rate": 3.905386734254304e-06,
      "loss": 0.4594,
      "step": 3727
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.648029656637852,
      "learning_rate": 3.904792394878004e-06,
      "loss": 0.7527,
      "step": 3728
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9764508104587293,
      "learning_rate": 3.904197939444646e-06,
      "loss": 0.2974,
      "step": 3729
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.926840754745452,
      "learning_rate": 3.903603368003337e-06,
      "loss": 0.2469,
      "step": 3730
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.451582820913571,
      "learning_rate": 3.903008680603199e-06,
      "loss": 0.3723,
      "step": 3731
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.81424682125572,
      "learning_rate": 3.9024138772933635e-06,
      "loss": 0.6771,
      "step": 3732
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.416272658852293,
      "learning_rate": 3.901818958122968e-06,
      "loss": 0.3895,
      "step": 3733
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.620186385342624,
      "learning_rate": 3.9012239231411645e-06,
      "loss": 0.4079,
      "step": 3734
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.273652232937082,
      "learning_rate": 3.900628772397109e-06,
      "loss": 0.3513,
      "step": 3735
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.839358249892077,
      "learning_rate": 3.900033505939971e-06,
      "loss": 0.538,
      "step": 3736
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.352004983262516,
      "learning_rate": 3.89943812381893e-06,
      "loss": 0.2929,
      "step": 3737
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.9371600948753716,
      "learning_rate": 3.898842626083171e-06,
      "loss": 0.3332,
      "step": 3738
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.6344647016892804,
      "learning_rate": 3.898247012781893e-06,
      "loss": 0.1401,
      "step": 3739
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1975802024371744,
      "learning_rate": 3.8976512839643025e-06,
      "loss": 0.1639,
      "step": 3740
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.147669400702976,
      "learning_rate": 3.8970554396796145e-06,
      "loss": 0.4925,
      "step": 3741
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.817405908446071,
      "learning_rate": 3.896459479977056e-06,
      "loss": 0.2926,
      "step": 3742
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.256995846209692,
      "learning_rate": 3.895863404905861e-06,
      "loss": 0.2613,
      "step": 3743
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.666895098858478,
      "learning_rate": 3.895267214515276e-06,
      "loss": 0.5369,
      "step": 3744
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.772150208452789,
      "learning_rate": 3.894670908854554e-06,
      "loss": 0.5334,
      "step": 3745
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.000751945260792,
      "learning_rate": 3.894074487972959e-06,
      "loss": 0.3296,
      "step": 3746
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.437473044394042,
      "learning_rate": 3.8934779519197645e-06,
      "loss": 0.3365,
      "step": 3747
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.480211101083018,
      "learning_rate": 3.892881300744253e-06,
      "loss": 0.1867,
      "step": 3748
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.51381222911311,
      "learning_rate": 3.892284534495718e-06,
      "loss": 0.2171,
      "step": 3749
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.005217418775944,
      "learning_rate": 3.891687653223461e-06,
      "loss": 0.5215,
      "step": 3750
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.73526591611506,
      "learning_rate": 3.891090656976794e-06,
      "loss": 0.15,
      "step": 3751
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.539576979993799,
      "learning_rate": 3.890493545805035e-06,
      "loss": 0.4914,
      "step": 3752
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.428669515334112,
      "learning_rate": 3.889896319757518e-06,
      "loss": 0.2605,
      "step": 3753
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.3265169502992915,
      "learning_rate": 3.88929897888358e-06,
      "loss": 0.4187,
      "step": 3754
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.225402279207235,
      "learning_rate": 3.8887015232325735e-06,
      "loss": 0.1508,
      "step": 3755
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.336437117041612,
      "learning_rate": 3.888103952853854e-06,
      "loss": 0.259,
      "step": 3756
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.983561533632079,
      "learning_rate": 3.8875062677967925e-06,
      "loss": 0.1976,
      "step": 3757
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.5944266162236653,
      "learning_rate": 3.886908468110765e-06,
      "loss": 0.2672,
      "step": 3758
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.449643264288724,
      "learning_rate": 3.886310553845159e-06,
      "loss": 0.4517,
      "step": 3759
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.1676037421024414,
      "learning_rate": 3.885712525049373e-06,
      "loss": 0.4476,
      "step": 3760
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.4290314788822673,
      "learning_rate": 3.885114381772812e-06,
      "loss": 0.18,
      "step": 3761
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.647260769358969,
      "learning_rate": 3.8845161240648915e-06,
      "loss": 0.3534,
      "step": 3762
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5338790908202682,
      "learning_rate": 3.883917751975036e-06,
      "loss": 0.0593,
      "step": 3763
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.816662915576356,
      "learning_rate": 3.883319265552681e-06,
      "loss": 0.3528,
      "step": 3764
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.4555543333755083,
      "learning_rate": 3.882720664847271e-06,
      "loss": 0.3116,
      "step": 3765
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.8113984373780125,
      "learning_rate": 3.8821219499082586e-06,
      "loss": 0.2648,
      "step": 3766
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.815115023565192,
      "learning_rate": 3.881523120785106e-06,
      "loss": 0.5404,
      "step": 3767
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.017431027157411,
      "learning_rate": 3.880924177527287e-06,
      "loss": 0.1877,
      "step": 3768
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.06733776585238,
      "learning_rate": 3.880325120184283e-06,
      "loss": 0.3426,
      "step": 3769
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.672444419059265,
      "learning_rate": 3.879725948805585e-06,
      "loss": 0.3178,
      "step": 3770
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.4224971883174606,
      "learning_rate": 3.879126663440693e-06,
      "loss": 0.1121,
      "step": 3771
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.526012322228652,
      "learning_rate": 3.8785272641391176e-06,
      "loss": 0.3459,
      "step": 3772
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.8115843580674382,
      "learning_rate": 3.877927750950378e-06,
      "loss": 0.313,
      "step": 3773
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.529859105744814,
      "learning_rate": 3.877328123924004e-06,
      "loss": 0.3371,
      "step": 3774
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.936653510525414,
      "learning_rate": 3.876728383109533e-06,
      "loss": 0.2084,
      "step": 3775
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.775484213146822,
      "learning_rate": 3.876128528556513e-06,
      "loss": 0.3578,
      "step": 3776
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.2874508645973055,
      "learning_rate": 3.875528560314499e-06,
      "loss": 0.3543,
      "step": 3777
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.453773032546963,
      "learning_rate": 3.874928478433061e-06,
      "loss": 0.3159,
      "step": 3778
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5615358264654704,
      "learning_rate": 3.874328282961772e-06,
      "loss": 0.1247,
      "step": 3779
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.3855688049921344,
      "learning_rate": 3.873727973950219e-06,
      "loss": 0.2593,
      "step": 3780
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.67746756520816,
      "learning_rate": 3.873127551447996e-06,
      "loss": 0.3874,
      "step": 3781
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.037234645732008,
      "learning_rate": 3.8725270155047055e-06,
      "loss": 0.2987,
      "step": 3782
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.452582146340652,
      "learning_rate": 3.871926366169963e-06,
      "loss": 0.4565,
      "step": 3783
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.108578813595276,
      "learning_rate": 3.87132560349339e-06,
      "loss": 0.4047,
      "step": 3784
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.28848854698769,
      "learning_rate": 3.87072472752462e-06,
      "loss": 0.4514,
      "step": 3785
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.297377048895478,
      "learning_rate": 3.870123738313292e-06,
      "loss": 0.1681,
      "step": 3786
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.01364290053557,
      "learning_rate": 3.869522635909058e-06,
      "loss": 0.451,
      "step": 3787
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.220359839740346,
      "learning_rate": 3.868921420361579e-06,
      "loss": 0.2725,
      "step": 3788
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.010917765584959,
      "learning_rate": 3.8683200917205235e-06,
      "loss": 0.2995,
      "step": 3789
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.589408269117776,
      "learning_rate": 3.86771865003557e-06,
      "loss": 0.414,
      "step": 3790
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.490394217851714,
      "learning_rate": 3.867117095356407e-06,
      "loss": 0.5099,
      "step": 3791
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.615992991888204,
      "learning_rate": 3.866515427732734e-06,
      "loss": 0.3367,
      "step": 3792
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.729757322399389,
      "learning_rate": 3.865913647214254e-06,
      "loss": 0.3737,
      "step": 3793
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5808402045375325,
      "learning_rate": 3.865311753850687e-06,
      "loss": 0.2484,
      "step": 3794
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6542430591411943,
      "learning_rate": 3.864709747691755e-06,
      "loss": 0.2184,
      "step": 3795
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.046565796092614,
      "learning_rate": 3.864107628787196e-06,
      "loss": 0.5984,
      "step": 3796
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.2817940884046037,
      "learning_rate": 3.863505397186752e-06,
      "loss": 0.2046,
      "step": 3797
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.52513782511823,
      "learning_rate": 3.8629030529401765e-06,
      "loss": 0.3985,
      "step": 3798
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.079468954059213,
      "learning_rate": 3.862300596097233e-06,
      "loss": 0.415,
      "step": 3799
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6002409499476182,
      "learning_rate": 3.861698026707693e-06,
      "loss": 0.2891,
      "step": 3800
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.99609505230759,
      "learning_rate": 3.8610953448213384e-06,
      "loss": 0.4054,
      "step": 3801
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.321828629067082,
      "learning_rate": 3.86049255048796e-06,
      "loss": 0.3331,
      "step": 3802
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.635824575743923,
      "learning_rate": 3.859889643757356e-06,
      "loss": 0.1895,
      "step": 3803
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0463282108272853,
      "learning_rate": 3.859286624679337e-06,
      "loss": 0.0883,
      "step": 3804
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.204606658839396,
      "learning_rate": 3.8586834933037225e-06,
      "loss": 0.3793,
      "step": 3805
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.09333032588095,
      "learning_rate": 3.858080249680337e-06,
      "loss": 0.1653,
      "step": 3806
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.597129132288316,
      "learning_rate": 3.857476893859021e-06,
      "loss": 0.5323,
      "step": 3807
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.791943397054493,
      "learning_rate": 3.85687342588962e-06,
      "loss": 0.2878,
      "step": 3808
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.2645991108841885,
      "learning_rate": 3.856269845821988e-06,
      "loss": 0.2455,
      "step": 3809
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.508133571304252,
      "learning_rate": 3.855666153705991e-06,
      "loss": 0.3559,
      "step": 3810
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.885388402673002,
      "learning_rate": 3.855062349591502e-06,
      "loss": 0.3513,
      "step": 3811
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.425067436454286,
      "learning_rate": 3.854458433528407e-06,
      "loss": 0.3983,
      "step": 3812
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.8331257652635067,
      "learning_rate": 3.853854405566595e-06,
      "loss": 0.2665,
      "step": 3813
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.8199949800442683,
      "learning_rate": 3.853250265755971e-06,
      "loss": 0.0849,
      "step": 3814
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.385357307699593,
      "learning_rate": 3.852646014146444e-06,
      "loss": 0.2489,
      "step": 3815
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.977141974395972,
      "learning_rate": 3.8520416507879344e-06,
      "loss": 0.3747,
      "step": 3816
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.31803795887791,
      "learning_rate": 3.851437175730374e-06,
      "loss": 0.2698,
      "step": 3817
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.4976034569524765,
      "learning_rate": 3.850832589023698e-06,
      "loss": 0.3081,
      "step": 3818
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.053081290962239,
      "learning_rate": 3.850227890717857e-06,
      "loss": 0.1775,
      "step": 3819
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.697100090171944,
      "learning_rate": 3.8496230808628075e-06,
      "loss": 0.5061,
      "step": 3820
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.756150120720348,
      "learning_rate": 3.8490181595085165e-06,
      "loss": 0.4765,
      "step": 3821
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.134491665115111,
      "learning_rate": 3.848413126704958e-06,
      "loss": 0.3974,
      "step": 3822
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.184867899890049,
      "learning_rate": 3.847807982502119e-06,
      "loss": 0.061,
      "step": 3823
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5386435360608473,
      "learning_rate": 3.847202726949991e-06,
      "loss": 0.2765,
      "step": 3824
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.148217408339083,
      "learning_rate": 3.84659736009858e-06,
      "loss": 0.3925,
      "step": 3825
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.100115421541015,
      "learning_rate": 3.845991881997897e-06,
      "loss": 0.3918,
      "step": 3826
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.884687538804943,
      "learning_rate": 3.845386292697962e-06,
      "loss": 0.4044,
      "step": 3827
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.662528315485075,
      "learning_rate": 3.84478059224881e-06,
      "loss": 0.2663,
      "step": 3828
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.2439341105874675,
      "learning_rate": 3.844174780700478e-06,
      "loss": 0.3066,
      "step": 3829
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.061792264328234,
      "learning_rate": 3.843568858103015e-06,
      "loss": 0.2587,
      "step": 3830
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.357864351062785,
      "learning_rate": 3.84296282450648e-06,
      "loss": 0.3974,
      "step": 3831
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6008260687454907,
      "learning_rate": 3.842356679960941e-06,
      "loss": 0.2704,
      "step": 3832
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6883046748234825,
      "learning_rate": 3.841750424516474e-06,
      "loss": 0.2895,
      "step": 3833
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.439505024705157,
      "learning_rate": 3.841144058223165e-06,
      "loss": 0.2233,
      "step": 3834
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.785667341296522,
      "learning_rate": 3.840537581131111e-06,
      "loss": 0.3948,
      "step": 3835
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.5068169699408993,
      "learning_rate": 3.839930993290412e-06,
      "loss": 0.2701,
      "step": 3836
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6565228379460746,
      "learning_rate": 3.839324294751185e-06,
      "loss": 0.1855,
      "step": 3837
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9588428282562336,
      "learning_rate": 3.838717485563551e-06,
      "loss": 0.049,
      "step": 3838
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.864906261854486,
      "learning_rate": 3.838110565777641e-06,
      "loss": 0.2096,
      "step": 3839
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.588626146161524,
      "learning_rate": 3.837503535443598e-06,
      "loss": 0.4623,
      "step": 3840
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.31918566564753,
      "learning_rate": 3.836896394611569e-06,
      "loss": 0.3396,
      "step": 3841
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.755455909116771,
      "learning_rate": 3.836289143331714e-06,
      "loss": 0.4169,
      "step": 3842
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3571374185437035,
      "learning_rate": 3.835681781654203e-06,
      "loss": 0.0859,
      "step": 3843
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.625160367149851,
      "learning_rate": 3.83507430962921e-06,
      "loss": 0.3136,
      "step": 3844
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.7356023494107617,
      "learning_rate": 3.8344667273069245e-06,
      "loss": 0.269,
      "step": 3845
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.646860971852209,
      "learning_rate": 3.83385903473754e-06,
      "loss": 0.1403,
      "step": 3846
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.642403114139395,
      "learning_rate": 3.833251231971261e-06,
      "loss": 0.4036,
      "step": 3847
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6449394236033443,
      "learning_rate": 3.832643319058303e-06,
      "loss": 0.209,
      "step": 3848
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.8366072137565155,
      "learning_rate": 3.832035296048886e-06,
      "loss": 0.5328,
      "step": 3849
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.2298249599169955,
      "learning_rate": 3.831427162993244e-06,
      "loss": 0.4008,
      "step": 3850
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1704336642119033,
      "learning_rate": 3.830818919941618e-06,
      "loss": 0.2439,
      "step": 3851
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.220569055024333,
      "learning_rate": 3.830210566944258e-06,
      "loss": 0.3899,
      "step": 3852
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.228620189024729,
      "learning_rate": 3.8296021040514206e-06,
      "loss": 0.2859,
      "step": 3853
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.095799882077851,
      "learning_rate": 3.828993531313378e-06,
      "loss": 0.2027,
      "step": 3854
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.554824102006452,
      "learning_rate": 3.828384848780405e-06,
      "loss": 0.3351,
      "step": 3855
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.610353921525349,
      "learning_rate": 3.827776056502789e-06,
      "loss": 0.3595,
      "step": 3856
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2236861244132067,
      "learning_rate": 3.827167154530823e-06,
      "loss": 0.0305,
      "step": 3857
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.7856094987639577,
      "learning_rate": 3.826558142914815e-06,
      "loss": 0.3191,
      "step": 3858
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.09208941188376,
      "learning_rate": 3.825949021705077e-06,
      "loss": 0.3474,
      "step": 3859
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.995095865957048,
      "learning_rate": 3.825339790951932e-06,
      "loss": 0.5894,
      "step": 3860
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.763553449193804,
      "learning_rate": 3.824730450705711e-06,
      "loss": 0.0595,
      "step": 3861
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.115060590632845,
      "learning_rate": 3.824121001016754e-06,
      "loss": 0.3628,
      "step": 3862
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.6516350890597065,
      "learning_rate": 3.823511441935412e-06,
      "loss": 0.2821,
      "step": 3863
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.560016750789516,
      "learning_rate": 3.822901773512044e-06,
      "loss": 0.3832,
      "step": 3864
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.376799599907808,
      "learning_rate": 3.822291995797017e-06,
      "loss": 0.4429,
      "step": 3865
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.05861795555519,
      "learning_rate": 3.821682108840708e-06,
      "loss": 0.4451,
      "step": 3866
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.562807219445786,
      "learning_rate": 3.821072112693503e-06,
      "loss": 0.3783,
      "step": 3867
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.691544630814998,
      "learning_rate": 3.820462007405797e-06,
      "loss": 0.4244,
      "step": 3868
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.099281885287403,
      "learning_rate": 3.819851793027994e-06,
      "loss": 0.3373,
      "step": 3869
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.686496590466868,
      "learning_rate": 3.819241469610508e-06,
      "loss": 0.482,
      "step": 3870
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.461954508302775,
      "learning_rate": 3.818631037203758e-06,
      "loss": 0.4441,
      "step": 3871
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.199237548178328,
      "learning_rate": 3.8180204958581765e-06,
      "loss": 0.2923,
      "step": 3872
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.166121014705124,
      "learning_rate": 3.817409845624204e-06,
      "loss": 0.3941,
      "step": 3873
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.725153112883,
      "learning_rate": 3.81679908655229e-06,
      "loss": 0.4031,
      "step": 3874
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.622214672403531,
      "learning_rate": 3.81618821869289e-06,
      "loss": 0.3953,
      "step": 3875
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.794392474182881,
      "learning_rate": 3.8155772420964734e-06,
      "loss": 0.277,
      "step": 3876
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.2716599326297455,
      "learning_rate": 3.814966156813513e-06,
      "loss": 0.2924,
      "step": 3877
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.345486511984712,
      "learning_rate": 3.8143549628944975e-06,
      "loss": 0.2363,
      "step": 3878
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.0288916286817202,
      "learning_rate": 3.8137436603899185e-06,
      "loss": 0.2185,
      "step": 3879
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.190179816209984,
      "learning_rate": 3.81313224935028e-06,
      "loss": 0.1626,
      "step": 3880
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.541662517854159,
      "learning_rate": 3.8125207298260914e-06,
      "loss": 0.4006,
      "step": 3881
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.951989005278172,
      "learning_rate": 3.8119091018678756e-06,
      "loss": 0.3791,
      "step": 3882
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.063266911200157,
      "learning_rate": 3.811297365526162e-06,
      "loss": 0.2346,
      "step": 3883
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.886473491144627,
      "learning_rate": 3.8106855208514894e-06,
      "loss": 0.4461,
      "step": 3884
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.534755118963499,
      "learning_rate": 3.8100735678944046e-06,
      "loss": 0.2305,
      "step": 3885
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.809743951387125,
      "learning_rate": 3.809461506705464e-06,
      "loss": 0.2763,
      "step": 3886
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.1754722925226235,
      "learning_rate": 3.8088493373352336e-06,
      "loss": 0.2475,
      "step": 3887
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9023623585912226,
      "learning_rate": 3.808237059834289e-06,
      "loss": 0.2003,
      "step": 3888
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.122876523124452,
      "learning_rate": 3.8076246742532117e-06,
      "loss": 0.5496,
      "step": 3889
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.4668459255304507,
      "learning_rate": 3.807012180642595e-06,
      "loss": 0.222,
      "step": 3890
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.032424616637112,
      "learning_rate": 3.8063995790530395e-06,
      "loss": 0.3236,
      "step": 3891
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.547936349240214,
      "learning_rate": 3.8057868695351564e-06,
      "loss": 0.3387,
      "step": 3892
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.3639109843220734,
      "learning_rate": 3.8051740521395634e-06,
      "loss": 0.2242,
      "step": 3893
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.552151127135212,
      "learning_rate": 3.8045611269168896e-06,
      "loss": 0.4595,
      "step": 3894
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.400481507146518,
      "learning_rate": 3.8039480939177716e-06,
      "loss": 0.2832,
      "step": 3895
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.917386341131842,
      "learning_rate": 3.8033349531928543e-06,
      "loss": 0.445,
      "step": 3896
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.54432565848614,
      "learning_rate": 3.802721704792793e-06,
      "loss": 0.2632,
      "step": 3897
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9077285474688987,
      "learning_rate": 3.802108348768252e-06,
      "loss": 0.2044,
      "step": 3898
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.53074011702579,
      "learning_rate": 3.801494885169903e-06,
      "loss": 0.4553,
      "step": 3899
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.571318698584371,
      "learning_rate": 3.800881314048428e-06,
      "loss": 0.4014,
      "step": 3900
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.895631098138602,
      "learning_rate": 3.800267635454515e-06,
      "loss": 0.4566,
      "step": 3901
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.035940584623637,
      "learning_rate": 3.7996538494388656e-06,
      "loss": 0.2254,
      "step": 3902
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2830672500434317,
      "learning_rate": 3.799039956052188e-06,
      "loss": 0.2424,
      "step": 3903
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.429129539071369,
      "learning_rate": 3.7984259553451974e-06,
      "loss": 0.577,
      "step": 3904
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.217424894999238,
      "learning_rate": 3.7978118473686204e-06,
      "loss": 0.1844,
      "step": 3905
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.235576200184693,
      "learning_rate": 3.797197632173191e-06,
      "loss": 0.2777,
      "step": 3906
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.830518913375487,
      "learning_rate": 3.7965833098096527e-06,
      "loss": 0.301,
      "step": 3907
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.277312186239868,
      "learning_rate": 3.7959688803287593e-06,
      "loss": 0.3244,
      "step": 3908
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.762072766965544,
      "learning_rate": 3.79535434378127e-06,
      "loss": 0.1082,
      "step": 3909
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.182313927822098,
      "learning_rate": 3.7947397002179563e-06,
      "loss": 0.3273,
      "step": 3910
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.037535817509862,
      "learning_rate": 3.7941249496895954e-06,
      "loss": 0.3075,
      "step": 3911
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.276831154137153,
      "learning_rate": 3.7935100922469757e-06,
      "loss": 0.3298,
      "step": 3912
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.6520651824725485,
      "learning_rate": 3.7928951279408956e-06,
      "loss": 0.4774,
      "step": 3913
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.749653263426172,
      "learning_rate": 3.7922800568221575e-06,
      "loss": 0.4514,
      "step": 3914
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.996796366989836,
      "learning_rate": 3.7916648789415766e-06,
      "loss": 0.4281,
      "step": 3915
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.922104692221765,
      "learning_rate": 3.791049594349977e-06,
      "loss": 0.5649,
      "step": 3916
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.303374387685765,
      "learning_rate": 3.7904342030981887e-06,
      "loss": 0.201,
      "step": 3917
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9640576193053434,
      "learning_rate": 3.7898187052370537e-06,
      "loss": 0.3745,
      "step": 3918
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.177694087321822,
      "learning_rate": 3.7892031008174214e-06,
      "loss": 0.3613,
      "step": 3919
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.887022514425497,
      "learning_rate": 3.788587389890148e-06,
      "loss": 0.3372,
      "step": 3920
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9027193941721023,
      "learning_rate": 3.7879715725061032e-06,
      "loss": 0.2844,
      "step": 3921
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9405522131927264,
      "learning_rate": 3.7873556487161613e-06,
      "loss": 0.3216,
      "step": 3922
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.8600066565714775,
      "learning_rate": 3.786739618571208e-06,
      "loss": 0.2767,
      "step": 3923
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.441279201213196,
      "learning_rate": 3.7861234821221353e-06,
      "loss": 0.2846,
      "step": 3924
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.744902266514368,
      "learning_rate": 3.7855072394198455e-06,
      "loss": 0.2638,
      "step": 3925
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.017105789487479,
      "learning_rate": 3.7848908905152516e-06,
      "loss": 0.339,
      "step": 3926
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.583453730526493,
      "learning_rate": 3.78427443545927e-06,
      "loss": 0.2581,
      "step": 3927
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.450226443321362,
      "learning_rate": 3.7836578743028322e-06,
      "loss": 0.3668,
      "step": 3928
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.5630798584611862,
      "learning_rate": 3.783041207096875e-06,
      "loss": 0.2763,
      "step": 3929
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.202820898107501,
      "learning_rate": 3.782424433892342e-06,
      "loss": 0.3483,
      "step": 3930
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.987912869008164,
      "learning_rate": 3.78180755474019e-06,
      "loss": 0.3503,
      "step": 3931
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.175740451272135,
      "learning_rate": 3.7811905696913825e-06,
      "loss": 0.1437,
      "step": 3932
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.094642427480686,
      "learning_rate": 3.7805734787968916e-06,
      "loss": 0.3744,
      "step": 3933
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.6669405177993015,
      "learning_rate": 3.779956282107698e-06,
      "loss": 0.3677,
      "step": 3934
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.72482535946106,
      "learning_rate": 3.7793389796747914e-06,
      "loss": 0.3188,
      "step": 3935
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2138719490933667,
      "learning_rate": 3.7787215715491716e-06,
      "loss": 0.2933,
      "step": 3936
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.283796707429345,
      "learning_rate": 3.778104057781844e-06,
      "loss": 0.3696,
      "step": 3937
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.7901054032002888,
      "learning_rate": 3.777486438423825e-06,
      "loss": 0.3057,
      "step": 3938
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.597081384541417,
      "learning_rate": 3.77686871352614e-06,
      "loss": 0.1456,
      "step": 3939
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.669621105309501,
      "learning_rate": 3.7762508831398227e-06,
      "loss": 0.3635,
      "step": 3940
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.487446787868056,
      "learning_rate": 3.7756329473159137e-06,
      "loss": 0.1779,
      "step": 3941
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.853301261243177,
      "learning_rate": 3.775014906105465e-06,
      "loss": 0.4145,
      "step": 3942
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.009906871238484,
      "learning_rate": 3.7743967595595356e-06,
      "loss": 0.2926,
      "step": 3943
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.655538702131722,
      "learning_rate": 3.773778507729194e-06,
      "loss": 0.0387,
      "step": 3944
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.11200600014402,
      "learning_rate": 3.7731601506655174e-06,
      "loss": 0.4701,
      "step": 3945
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.69493279587049,
      "learning_rate": 3.772541688419591e-06,
      "loss": 0.4333,
      "step": 3946
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.261994060837244,
      "learning_rate": 3.771923121042509e-06,
      "loss": 0.4006,
      "step": 3947
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.916266655793664,
      "learning_rate": 3.771304448585375e-06,
      "loss": 0.4158,
      "step": 3948
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8231638436222095,
      "learning_rate": 3.7706856710993e-06,
      "loss": 0.3594,
      "step": 3949
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.0508282616365867,
      "learning_rate": 3.7700667886354047e-06,
      "loss": 0.089,
      "step": 3950
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.8037525986877965,
      "learning_rate": 3.7694478012448187e-06,
      "loss": 0.2837,
      "step": 3951
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.309755535243665,
      "learning_rate": 3.7688287089786786e-06,
      "loss": 0.4064,
      "step": 3952
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.117640266969755,
      "learning_rate": 3.768209511888131e-06,
      "loss": 0.5916,
      "step": 3953
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.784450042209126,
      "learning_rate": 3.767590210024332e-06,
      "loss": 0.296,
      "step": 3954
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.060148912556706,
      "learning_rate": 3.766970803438444e-06,
      "loss": 0.4171,
      "step": 3955
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.62357234841756,
      "learning_rate": 3.7663512921816403e-06,
      "loss": 0.3902,
      "step": 3956
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.842391681731419,
      "learning_rate": 3.7657316763051003e-06,
      "loss": 0.2202,
      "step": 3957
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.252345745977204,
      "learning_rate": 3.765111955860016e-06,
      "loss": 0.5608,
      "step": 3958
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.459072507452741,
      "learning_rate": 3.7644921308975836e-06,
      "loss": 0.4718,
      "step": 3959
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.031148460519736,
      "learning_rate": 3.7638722014690114e-06,
      "loss": 0.3679,
      "step": 3960
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.3923472670800905,
      "learning_rate": 3.763252167625514e-06,
      "loss": 0.4107,
      "step": 3961
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.974035223215897,
      "learning_rate": 3.762632029418316e-06,
      "loss": 0.374,
      "step": 3962
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.75719500786711,
      "learning_rate": 3.7620117868986498e-06,
      "loss": 0.3898,
      "step": 3963
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.449226941214141,
      "learning_rate": 3.7613914401177575e-06,
      "loss": 0.5308,
      "step": 3964
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.77113595794797,
      "learning_rate": 3.760770989126889e-06,
      "loss": 0.1901,
      "step": 3965
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.437478247078316,
      "learning_rate": 3.760150433977302e-06,
      "loss": 0.2545,
      "step": 3966
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.853244927801857,
      "learning_rate": 3.7595297747202643e-06,
      "loss": 0.3132,
      "step": 3967
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.973157373736267,
      "learning_rate": 3.758909011407052e-06,
      "loss": 0.2994,
      "step": 3968
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.563323752147033,
      "learning_rate": 3.7582881440889497e-06,
      "loss": 0.5571,
      "step": 3969
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.290592218974564,
      "learning_rate": 3.757667172817249e-06,
      "loss": 0.2377,
      "step": 3970
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.5513203156430024,
      "learning_rate": 3.7570460976432534e-06,
      "loss": 0.2282,
      "step": 3971
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.833990746378398,
      "learning_rate": 3.7564249186182715e-06,
      "loss": 0.3064,
      "step": 3972
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.803446370342577,
      "learning_rate": 3.7558036357936227e-06,
      "loss": 0.3508,
      "step": 3973
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4385228758863966,
      "learning_rate": 3.755182249220635e-06,
      "loss": 0.1123,
      "step": 3974
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.74862208122065,
      "learning_rate": 3.754560758950644e-06,
      "loss": 0.168,
      "step": 3975
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.01246220744637,
      "learning_rate": 3.7539391650349933e-06,
      "loss": 0.2433,
      "step": 3976
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.277558545420366,
      "learning_rate": 3.7533174675250356e-06,
      "loss": 0.2482,
      "step": 3977
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.504991888511777,
      "learning_rate": 3.7526956664721347e-06,
      "loss": 0.2755,
      "step": 3978
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.895501408502185,
      "learning_rate": 3.752073761927659e-06,
      "loss": 0.2861,
      "step": 3979
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.7489405469900055,
      "learning_rate": 3.7514517539429884e-06,
      "loss": 0.3946,
      "step": 3980
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.142613427538114,
      "learning_rate": 3.750829642569509e-06,
      "loss": 0.3462,
      "step": 3981
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.437404145124437,
      "learning_rate": 3.7502074278586153e-06,
      "loss": 0.5009,
      "step": 3982
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.071406105426733,
      "learning_rate": 3.749585109861715e-06,
      "loss": 0.3577,
      "step": 3983
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.584681692986226,
      "learning_rate": 3.74896268863022e-06,
      "loss": 0.4034,
      "step": 3984
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.559413860631105,
      "learning_rate": 3.74834016421555e-06,
      "loss": 0.4207,
      "step": 3985
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2414124672128173,
      "learning_rate": 3.7477175366691363e-06,
      "loss": 0.096,
      "step": 3986
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.11011824531728,
      "learning_rate": 3.747094806042416e-06,
      "loss": 0.3311,
      "step": 3987
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.411541574352496,
      "learning_rate": 3.7464719723868383e-06,
      "loss": 0.3706,
      "step": 3988
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.9503290963453552,
      "learning_rate": 3.7458490357538567e-06,
      "loss": 0.349,
      "step": 3989
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.867037914738536,
      "learning_rate": 3.745225996194936e-06,
      "loss": 0.2597,
      "step": 3990
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.489620597538562,
      "learning_rate": 3.7446028537615485e-06,
      "loss": 0.2519,
      "step": 3991
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.67995292710198,
      "learning_rate": 3.743979608505175e-06,
      "loss": 0.3586,
      "step": 3992
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.102163054284272,
      "learning_rate": 3.7433562604773054e-06,
      "loss": 0.2092,
      "step": 3993
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.14586063084222,
      "learning_rate": 3.7427328097294373e-06,
      "loss": 0.4677,
      "step": 3994
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.078698212274654,
      "learning_rate": 3.742109256313077e-06,
      "loss": 0.1467,
      "step": 3995
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.907393497007393,
      "learning_rate": 3.74148560027974e-06,
      "loss": 0.1803,
      "step": 3996
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.338822577474265,
      "learning_rate": 3.740861841680949e-06,
      "loss": 0.4423,
      "step": 3997
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.8642087468752697,
      "learning_rate": 3.7402379805682366e-06,
      "loss": 0.1912,
      "step": 3998
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.167117950687652,
      "learning_rate": 3.7396140169931433e-06,
      "loss": 0.3123,
      "step": 3999
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.132851160751913,
      "learning_rate": 3.738989951007217e-06,
      "loss": 0.3024,
      "step": 4000
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.095692500347719,
      "learning_rate": 3.738365782662015e-06,
      "loss": 0.292,
      "step": 4001
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6376567896517784,
      "learning_rate": 3.737741512009103e-06,
      "loss": 0.2617,
      "step": 4002
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.881847612772176,
      "learning_rate": 3.7371171391000564e-06,
      "loss": 0.266,
      "step": 4003
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.482395763250964,
      "learning_rate": 3.7364926639864574e-06,
      "loss": 0.0681,
      "step": 4004
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.483010947733964,
      "learning_rate": 3.7358680867198964e-06,
      "loss": 0.2416,
      "step": 4005
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.010393560911372,
      "learning_rate": 3.735243407351973e-06,
      "loss": 0.4153,
      "step": 4006
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5885666870236737,
      "learning_rate": 3.7346186259342946e-06,
      "loss": 0.2741,
      "step": 4007
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9240061792781367,
      "learning_rate": 3.7339937425184796e-06,
      "loss": 0.2338,
      "step": 4008
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.9689189447864655,
      "learning_rate": 3.733368757156151e-06,
      "loss": 0.0715,
      "step": 4009
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.76279100722026,
      "learning_rate": 3.7327436698989427e-06,
      "loss": 0.1404,
      "step": 4010
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.367687319117145,
      "learning_rate": 3.732118480798496e-06,
      "loss": 0.4285,
      "step": 4011
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.775716157857203,
      "learning_rate": 3.731493189906461e-06,
      "loss": 0.1515,
      "step": 4012
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.176032622203043,
      "learning_rate": 3.730867797274497e-06,
      "loss": 0.098,
      "step": 4013
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.4657070753552666,
      "learning_rate": 3.73024230295427e-06,
      "loss": 0.0885,
      "step": 4014
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.8639703822255935,
      "learning_rate": 3.7296167069974546e-06,
      "loss": 0.3297,
      "step": 4015
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5033620861016543,
      "learning_rate": 3.728991009455737e-06,
      "loss": 0.2651,
      "step": 4016
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.662568834253889,
      "learning_rate": 3.728365210380806e-06,
      "loss": 0.3645,
      "step": 4017
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.287595427494699,
      "learning_rate": 3.7277393098243647e-06,
      "loss": 0.5201,
      "step": 4018
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.160314682740059,
      "learning_rate": 3.7271133078381204e-06,
      "loss": 0.4927,
      "step": 4019
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.652958260266686,
      "learning_rate": 3.726487204473791e-06,
      "loss": 0.3261,
      "step": 4020
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.486205231205078,
      "learning_rate": 3.7258609997831017e-06,
      "loss": 0.3064,
      "step": 4021
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.216337219637617,
      "learning_rate": 3.725234693817787e-06,
      "loss": 0.3396,
      "step": 4022
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.314202791348141,
      "learning_rate": 3.7246082866295887e-06,
      "loss": 0.4002,
      "step": 4023
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8134177687169926,
      "learning_rate": 3.723981778270258e-06,
      "loss": 0.1339,
      "step": 4024
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.047613916122432,
      "learning_rate": 3.723355168791553e-06,
      "loss": 0.3044,
      "step": 4025
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.372778194231569,
      "learning_rate": 3.7227284582452428e-06,
      "loss": 0.3068,
      "step": 4026
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9771983316118416,
      "learning_rate": 3.7221016466831005e-06,
      "loss": 0.4058,
      "step": 4027
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.144691884707098,
      "learning_rate": 3.7214747341569136e-06,
      "loss": 0.3067,
      "step": 4028
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6789867047864213,
      "learning_rate": 3.720847720718472e-06,
      "loss": 0.3028,
      "step": 4029
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8133248544210896,
      "learning_rate": 3.720220606419577e-06,
      "loss": 0.0771,
      "step": 4030
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8225820478839667,
      "learning_rate": 3.7195933913120374e-06,
      "loss": 0.3353,
      "step": 4031
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7274066486020088,
      "learning_rate": 3.7189660754476715e-06,
      "loss": 0.2621,
      "step": 4032
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8215992466438977,
      "learning_rate": 3.7183386588783053e-06,
      "loss": 0.2368,
      "step": 4033
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.818357834311553,
      "learning_rate": 3.7177111416557717e-06,
      "loss": 0.5688,
      "step": 4034
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.3427991782277346,
      "learning_rate": 3.7170835238319143e-06,
      "loss": 0.2267,
      "step": 4035
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.238629861486029,
      "learning_rate": 3.7164558054585824e-06,
      "loss": 0.3325,
      "step": 4036
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.583090868705083,
      "learning_rate": 3.715827986587636e-06,
      "loss": 0.1332,
      "step": 4037
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.716865152857654,
      "learning_rate": 3.7152000672709433e-06,
      "loss": 0.4769,
      "step": 4038
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.496736042359182,
      "learning_rate": 3.7145720475603774e-06,
      "loss": 0.2823,
      "step": 4039
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.894179135389166,
      "learning_rate": 3.713943927507824e-06,
      "loss": 0.3846,
      "step": 4040
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.292200863124737,
      "learning_rate": 3.713315707165176e-06,
      "loss": 0.1233,
      "step": 4041
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.598709140106297,
      "learning_rate": 3.7126873865843314e-06,
      "loss": 0.2435,
      "step": 4042
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.733643452430367,
      "learning_rate": 3.7120589658172014e-06,
      "loss": 0.5259,
      "step": 4043
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.750592917460503,
      "learning_rate": 3.711430444915701e-06,
      "loss": 0.2325,
      "step": 4044
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.756056615028903,
      "learning_rate": 3.710801823931758e-06,
      "loss": 0.4492,
      "step": 4045
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.442886099490769,
      "learning_rate": 3.7101731029173038e-06,
      "loss": 0.257,
      "step": 4046
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8171937642987923,
      "learning_rate": 3.709544281924281e-06,
      "loss": 0.3389,
      "step": 4047
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.166744412590996,
      "learning_rate": 3.7089153610046404e-06,
      "loss": 0.3375,
      "step": 4048
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.602391636615193,
      "learning_rate": 3.7082863402103387e-06,
      "loss": 0.406,
      "step": 4049
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8261108042100043,
      "learning_rate": 3.7076572195933436e-06,
      "loss": 0.2215,
      "step": 4050
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9718417019917758,
      "learning_rate": 3.7070279992056302e-06,
      "loss": 0.0764,
      "step": 4051
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.913968836237462,
      "learning_rate": 3.7063986790991805e-06,
      "loss": 0.4026,
      "step": 4052
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.209440818104927,
      "learning_rate": 3.7057692593259875e-06,
      "loss": 0.2356,
      "step": 4053
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.6361416950452945,
      "learning_rate": 3.7051397399380493e-06,
      "loss": 0.5247,
      "step": 4054
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.986398798475423,
      "learning_rate": 3.7045101209873746e-06,
      "loss": 0.2354,
      "step": 4055
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0098358779479732,
      "learning_rate": 3.7038804025259796e-06,
      "loss": 0.2119,
      "step": 4056
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.312401158217113,
      "learning_rate": 3.703250584605887e-06,
      "loss": 0.4938,
      "step": 4057
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.198632060856005,
      "learning_rate": 3.7026206672791304e-06,
      "loss": 0.4593,
      "step": 4058
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.5293586022731036,
      "learning_rate": 3.7019906505977508e-06,
      "loss": 0.1126,
      "step": 4059
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.790344071078243,
      "learning_rate": 3.701360534613797e-06,
      "loss": 0.1795,
      "step": 4060
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.901460434247244,
      "learning_rate": 3.7007303193793255e-06,
      "loss": 0.1822,
      "step": 4061
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.127313991556961,
      "learning_rate": 3.700100004946402e-06,
      "loss": 0.4138,
      "step": 4062
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7607522565138867,
      "learning_rate": 3.6994695913670992e-06,
      "loss": 0.1131,
      "step": 4063
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.658351481571668,
      "learning_rate": 3.6988390786935003e-06,
      "loss": 0.1579,
      "step": 4064
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.898275111396248,
      "learning_rate": 3.698208466977694e-06,
      "loss": 0.3576,
      "step": 4065
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.4263019817651923,
      "learning_rate": 3.6975777562717783e-06,
      "loss": 0.1159,
      "step": 4066
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.584216560299987,
      "learning_rate": 3.69694694662786e-06,
      "loss": 0.0471,
      "step": 4067
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.2566486895710245,
      "learning_rate": 3.6963160380980534e-06,
      "loss": 0.2899,
      "step": 4068
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5134475822070796,
      "learning_rate": 3.6956850307344806e-06,
      "loss": 0.2738,
      "step": 4069
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.29301786837084,
      "learning_rate": 3.6950539245892738e-06,
      "loss": 0.412,
      "step": 4070
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5325127917725485,
      "learning_rate": 3.6944227197145698e-06,
      "loss": 0.0086,
      "step": 4071
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.171926675569348,
      "learning_rate": 3.6937914161625167e-06,
      "loss": 0.4719,
      "step": 4072
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.474179014911505,
      "learning_rate": 3.6931600139852693e-06,
      "loss": 0.2147,
      "step": 4073
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8558184965582587,
      "learning_rate": 3.692528513234992e-06,
      "loss": 0.3068,
      "step": 4074
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.669877689994271,
      "learning_rate": 3.6918969139638556e-06,
      "loss": 0.4509,
      "step": 4075
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.823573382756537,
      "learning_rate": 3.69126521622404e-06,
      "loss": 0.2897,
      "step": 4076
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.769588710928839,
      "learning_rate": 3.690633420067731e-06,
      "loss": 0.3556,
      "step": 4077
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.1338770666430587,
      "learning_rate": 3.690001525547128e-06,
      "loss": 0.2463,
      "step": 4078
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.354119603095363,
      "learning_rate": 3.6893695327144323e-06,
      "loss": 0.2449,
      "step": 4079
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6342362682610037,
      "learning_rate": 3.6887374416218574e-06,
      "loss": 0.3144,
      "step": 4080
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.537417058367153,
      "learning_rate": 3.6881052523216234e-06,
      "loss": 0.3008,
      "step": 4081
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.344223057817349,
      "learning_rate": 3.687472964865957e-06,
      "loss": 0.212,
      "step": 4082
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.330349967443163,
      "learning_rate": 3.6868405793070972e-06,
      "loss": 0.2072,
      "step": 4083
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.2230696933110754,
      "learning_rate": 3.686208095697288e-06,
      "loss": 0.2199,
      "step": 4084
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.244245008004557,
      "learning_rate": 3.685575514088781e-06,
      "loss": 0.1084,
      "step": 4085
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.161586144735892,
      "learning_rate": 3.6849428345338377e-06,
      "loss": 0.5491,
      "step": 4086
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.513216702108144,
      "learning_rate": 3.684310057084726e-06,
      "loss": 0.3283,
      "step": 4087
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.124960631360362,
      "learning_rate": 3.683677181793725e-06,
      "loss": 0.3604,
      "step": 4088
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.634427208851154,
      "learning_rate": 3.683044208713118e-06,
      "loss": 0.7247,
      "step": 4089
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.73644617004746,
      "learning_rate": 3.6824111378951988e-06,
      "loss": 0.1886,
      "step": 4090
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7377812473155316,
      "learning_rate": 3.6817779693922682e-06,
      "loss": 0.2452,
      "step": 4091
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.7382649196330817,
      "learning_rate": 3.6811447032566356e-06,
      "loss": 0.2849,
      "step": 4092
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9922764486120452,
      "learning_rate": 3.680511339540619e-06,
      "loss": 0.1102,
      "step": 4093
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.309919755554874,
      "learning_rate": 3.6798778782965426e-06,
      "loss": 0.3399,
      "step": 4094
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.5841770952194776,
      "learning_rate": 3.679244319576741e-06,
      "loss": 0.1693,
      "step": 4095
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.951748474301394,
      "learning_rate": 3.678610663433555e-06,
      "loss": 0.4018,
      "step": 4096
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.989555418937315,
      "learning_rate": 3.677976909919334e-06,
      "loss": 0.3267,
      "step": 4097
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.211218005163147,
      "learning_rate": 3.6773430590864367e-06,
      "loss": 0.1477,
      "step": 4098
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.265328047155728,
      "learning_rate": 3.6767091109872284e-06,
      "loss": 0.4441,
      "step": 4099
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.8488630933477355,
      "learning_rate": 3.676075065674082e-06,
      "loss": 0.2744,
      "step": 4100
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.4628354155389034,
      "learning_rate": 3.675440923199379e-06,
      "loss": 0.1382,
      "step": 4101
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.045973832868391,
      "learning_rate": 3.6748066836155106e-06,
      "loss": 0.4716,
      "step": 4102
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.3383593834332235,
      "learning_rate": 3.6741723469748737e-06,
      "loss": 0.3987,
      "step": 4103
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.325912093849595,
      "learning_rate": 3.6735379133298743e-06,
      "loss": 0.1286,
      "step": 4104
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.951312513094245,
      "learning_rate": 3.672903382732926e-06,
      "loss": 0.5019,
      "step": 4105
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.102836740545422,
      "learning_rate": 3.67226875523645e-06,
      "loss": 0.2187,
      "step": 4106
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9439866704501116,
      "learning_rate": 3.6716340308928766e-06,
      "loss": 0.2715,
      "step": 4107
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.914961066292653,
      "learning_rate": 3.670999209754644e-06,
      "loss": 0.3927,
      "step": 4108
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.536388998732376,
      "learning_rate": 3.6703642918741988e-06,
      "loss": 0.2473,
      "step": 4109
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.6197663950279244,
      "learning_rate": 3.6697292773039933e-06,
      "loss": 0.3068,
      "step": 4110
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.67293021592896,
      "learning_rate": 3.6690941660964897e-06,
      "loss": 0.2892,
      "step": 4111
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.817232633228814,
      "learning_rate": 3.6684589583041573e-06,
      "loss": 0.4706,
      "step": 4112
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.350737467765941,
      "learning_rate": 3.667823653979475e-06,
      "loss": 0.3174,
      "step": 4113
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.533318414085304,
      "learning_rate": 3.6671882531749283e-06,
      "loss": 0.4562,
      "step": 4114
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.842297524269241,
      "learning_rate": 3.666552755943009e-06,
      "loss": 0.1435,
      "step": 4115
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.412858002699026,
      "learning_rate": 3.6659171623362222e-06,
      "loss": 0.21,
      "step": 4116
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.3164442611921245,
      "learning_rate": 3.665281472407074e-06,
      "loss": 0.2388,
      "step": 4117
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.9051248259901143,
      "learning_rate": 3.6646456862080846e-06,
      "loss": 0.2902,
      "step": 4118
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.708625407999043,
      "learning_rate": 3.6640098037917785e-06,
      "loss": 0.3788,
      "step": 4119
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.690054691137693,
      "learning_rate": 3.6633738252106895e-06,
      "loss": 0.3312,
      "step": 4120
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.6808212236461135,
      "learning_rate": 3.6627377505173584e-06,
      "loss": 0.3217,
      "step": 4121
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.7600230323634887,
      "learning_rate": 3.6621015797643345e-06,
      "loss": 0.1851,
      "step": 4122
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.105983353182932,
      "learning_rate": 3.661465313004176e-06,
      "loss": 0.3978,
      "step": 4123
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.467367590118211,
      "learning_rate": 3.660828950289448e-06,
      "loss": 0.2562,
      "step": 4124
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.481609319295048,
      "learning_rate": 3.6601924916727226e-06,
      "loss": 0.3417,
      "step": 4125
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.952814370611476,
      "learning_rate": 3.6595559372065827e-06,
      "loss": 0.4054,
      "step": 4126
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.938839291830374,
      "learning_rate": 3.658919286943615e-06,
      "loss": 0.5352,
      "step": 4127
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.522122358839695,
      "learning_rate": 3.6582825409364186e-06,
      "loss": 0.2977,
      "step": 4128
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.6558307314069602,
      "learning_rate": 3.6576456992375976e-06,
      "loss": 0.2804,
      "step": 4129
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9017027725908475,
      "learning_rate": 3.6570087618997635e-06,
      "loss": 0.1494,
      "step": 4130
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8837350879846917,
      "learning_rate": 3.656371728975539e-06,
      "loss": 0.3888,
      "step": 4131
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.009494309377723,
      "learning_rate": 3.655734600517551e-06,
      "loss": 0.3064,
      "step": 4132
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.049892755489164,
      "learning_rate": 3.655097376578437e-06,
      "loss": 0.3619,
      "step": 4133
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5891261437357826,
      "learning_rate": 3.6544600572108415e-06,
      "loss": 0.2903,
      "step": 4134
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1280562722550247,
      "learning_rate": 3.6538226424674146e-06,
      "loss": 0.2263,
      "step": 4135
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1899720552252697,
      "learning_rate": 3.653185132400819e-06,
      "loss": 0.1039,
      "step": 4136
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.8211490334683225,
      "learning_rate": 3.6525475270637205e-06,
      "loss": 0.2443,
      "step": 4137
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.095157030582282,
      "learning_rate": 3.6519098265087966e-06,
      "loss": 0.2714,
      "step": 4138
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.003943865551017,
      "learning_rate": 3.65127203078873e-06,
      "loss": 0.3581,
      "step": 4139
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.3320546802336235,
      "learning_rate": 3.650634139956213e-06,
      "loss": 0.3344,
      "step": 4140
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.076273227926804,
      "learning_rate": 3.649996154063945e-06,
      "loss": 0.4481,
      "step": 4141
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.253893306327693,
      "learning_rate": 3.6493580731646314e-06,
      "loss": 0.3376,
      "step": 4142
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.040341446529094,
      "learning_rate": 3.6487198973109907e-06,
      "loss": 0.3206,
      "step": 4143
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1791219660939203,
      "learning_rate": 3.648081626555743e-06,
      "loss": 0.2088,
      "step": 4144
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.1634802828633024,
      "learning_rate": 3.6474432609516197e-06,
      "loss": 0.2239,
      "step": 4145
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.25011986102637,
      "learning_rate": 3.646804800551361e-06,
      "loss": 0.2614,
      "step": 4146
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.940745657080821,
      "learning_rate": 3.6461662454077106e-06,
      "loss": 0.2849,
      "step": 4147
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.436036422585126,
      "learning_rate": 3.6455275955734255e-06,
      "loss": 0.2411,
      "step": 4148
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.1155691351817785,
      "learning_rate": 3.6448888511012657e-06,
      "loss": 0.3298,
      "step": 4149
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.3350749034217975,
      "learning_rate": 3.644250012044003e-06,
      "loss": 0.3869,
      "step": 4150
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.50161760231835,
      "learning_rate": 3.6436110784544144e-06,
      "loss": 0.3054,
      "step": 4151
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.9945931836787105,
      "learning_rate": 3.642972050385284e-06,
      "loss": 0.2755,
      "step": 4152
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.0396312171653035,
      "learning_rate": 3.6423329278894074e-06,
      "loss": 0.2374,
      "step": 4153
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.51643773394796,
      "learning_rate": 3.641693711019585e-06,
      "loss": 0.3921,
      "step": 4154
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.447732606974932,
      "learning_rate": 3.6410543998286253e-06,
      "loss": 0.3139,
      "step": 4155
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.117488146841359,
      "learning_rate": 3.6404149943693456e-06,
      "loss": 0.2081,
      "step": 4156
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.467381714589403,
      "learning_rate": 3.63977549469457e-06,
      "loss": 0.2261,
      "step": 4157
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.290606673317658,
      "learning_rate": 3.6391359008571307e-06,
      "loss": 0.3301,
      "step": 4158
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.5998932743481737,
      "learning_rate": 3.638496212909869e-06,
      "loss": 0.2867,
      "step": 4159
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.930953815920561,
      "learning_rate": 3.637856430905632e-06,
      "loss": 0.1993,
      "step": 4160
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3707239192877676,
      "learning_rate": 3.637216554897275e-06,
      "loss": 0.1809,
      "step": 4161
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.408868329493421,
      "learning_rate": 3.636576584937661e-06,
      "loss": 0.5247,
      "step": 4162
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.704257580545927,
      "learning_rate": 3.635936521079662e-06,
      "loss": 0.368,
      "step": 4163
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.081229615326893,
      "learning_rate": 3.635296363376158e-06,
      "loss": 0.3315,
      "step": 4164
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.552245296398955,
      "learning_rate": 3.634656111880034e-06,
      "loss": 0.3551,
      "step": 4165
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.165627192600827,
      "learning_rate": 3.6340157666441843e-06,
      "loss": 0.2122,
      "step": 4166
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.443813356129511,
      "learning_rate": 3.633375327721512e-06,
      "loss": 0.3995,
      "step": 4167
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.958686674361145,
      "learning_rate": 3.6327347951649267e-06,
      "loss": 0.3641,
      "step": 4168
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.641432101884483,
      "learning_rate": 3.6320941690273463e-06,
      "loss": 0.5628,
      "step": 4169
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.8398225441432556,
      "learning_rate": 3.6314534493616965e-06,
      "loss": 0.1279,
      "step": 4170
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.2076163224148764,
      "learning_rate": 3.6308126362209097e-06,
      "loss": 0.1927,
      "step": 4171
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.322708716088132,
      "learning_rate": 3.6301717296579263e-06,
      "loss": 0.2253,
      "step": 4172
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.092785265640324,
      "learning_rate": 3.629530729725696e-06,
      "loss": 0.4683,
      "step": 4173
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.734871307198582,
      "learning_rate": 3.628889636477175e-06,
      "loss": 0.3292,
      "step": 4174
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.168675752973682,
      "learning_rate": 3.6282484499653263e-06,
      "loss": 0.1461,
      "step": 4175
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.0437425788783115,
      "learning_rate": 3.6276071702431228e-06,
      "loss": 0.2187,
      "step": 4176
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.838157569604393,
      "learning_rate": 3.6269657973635424e-06,
      "loss": 0.4259,
      "step": 4177
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.308640293590356,
      "learning_rate": 3.626324331379573e-06,
      "loss": 0.247,
      "step": 4178
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8954949724236774,
      "learning_rate": 3.62568277234421e-06,
      "loss": 0.3737,
      "step": 4179
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.6039308542501685,
      "learning_rate": 3.625041120310455e-06,
      "loss": 0.4977,
      "step": 4180
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.9762492470121487,
      "learning_rate": 3.6243993753313194e-06,
      "loss": 0.174,
      "step": 4181
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.358349855038549,
      "learning_rate": 3.623757537459818e-06,
      "loss": 0.4449,
      "step": 4182
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.062429265390732,
      "learning_rate": 3.62311560674898e-06,
      "loss": 0.4202,
      "step": 4183
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.726102961604748,
      "learning_rate": 3.6224735832518365e-06,
      "loss": 0.2069,
      "step": 4184
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.081772305447274,
      "learning_rate": 3.6218314670214288e-06,
      "loss": 0.3672,
      "step": 4185
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.233492971476211,
      "learning_rate": 3.6211892581108054e-06,
      "loss": 0.3643,
      "step": 4186
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.428503625804239,
      "learning_rate": 3.6205469565730216e-06,
      "loss": 0.1216,
      "step": 4187
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.799282665125188,
      "learning_rate": 3.619904562461143e-06,
      "loss": 0.4354,
      "step": 4188
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.527906277157118,
      "learning_rate": 3.61926207582824e-06,
      "loss": 0.4354,
      "step": 4189
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.102111443672788,
      "learning_rate": 3.618619496727391e-06,
      "loss": 0.353,
      "step": 4190
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.9665391269098333,
      "learning_rate": 3.617976825211684e-06,
      "loss": 0.3168,
      "step": 4191
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.9594130024341982,
      "learning_rate": 3.6173340613342124e-06,
      "loss": 0.3351,
      "step": 4192
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.791611372339618,
      "learning_rate": 3.616691205148079e-06,
      "loss": 0.3581,
      "step": 4193
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.698797588335175,
      "learning_rate": 3.616048256706393e-06,
      "loss": 0.4574,
      "step": 4194
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.423138375083067,
      "learning_rate": 3.615405216062272e-06,
      "loss": 0.2699,
      "step": 4195
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.673212534724762,
      "learning_rate": 3.61476208326884e-06,
      "loss": 0.2756,
      "step": 4196
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.590923214156035,
      "learning_rate": 3.6141188583792305e-06,
      "loss": 0.6275,
      "step": 4197
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.195625101514778,
      "learning_rate": 3.613475541446583e-06,
      "loss": 0.3256,
      "step": 4198
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.7685056776251535,
      "learning_rate": 3.6128321325240458e-06,
      "loss": 0.5456,
      "step": 4199
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.98315148695062,
      "learning_rate": 3.6121886316647738e-06,
      "loss": 0.2391,
      "step": 4200
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.5160330558594906,
      "learning_rate": 3.6115450389219286e-06,
      "loss": 0.4717,
      "step": 4201
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.276628669235739,
      "learning_rate": 3.6109013543486824e-06,
      "loss": 0.28,
      "step": 4202
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.371643119629829,
      "learning_rate": 3.6102575779982136e-06,
      "loss": 0.5207,
      "step": 4203
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.621004382944956,
      "learning_rate": 3.609613709923706e-06,
      "loss": 0.5748,
      "step": 4204
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.116643021017709,
      "learning_rate": 3.608969750178355e-06,
      "loss": 0.2104,
      "step": 4205
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.7123372653124282,
      "learning_rate": 3.608325698815359e-06,
      "loss": 0.0863,
      "step": 4206
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.676829568106578,
      "learning_rate": 3.6076815558879276e-06,
      "loss": 0.2613,
      "step": 4207
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.632907181716588,
      "learning_rate": 3.607037321449278e-06,
      "loss": 0.4108,
      "step": 4208
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.3938249731387327,
      "learning_rate": 3.606392995552631e-06,
      "loss": 0.2529,
      "step": 4209
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.350751319343075,
      "learning_rate": 3.60574857825122e-06,
      "loss": 0.2127,
      "step": 4210
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.250563595299225,
      "learning_rate": 3.605104069598282e-06,
      "loss": 0.2497,
      "step": 4211
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.773853888129306,
      "learning_rate": 3.6044594696470644e-06,
      "loss": 0.3643,
      "step": 4212
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.2253576407926,
      "learning_rate": 3.60381477845082e-06,
      "loss": 0.4133,
      "step": 4213
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.264099159956622,
      "learning_rate": 3.60316999606281e-06,
      "loss": 0.4287,
      "step": 4214
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8306365592683522,
      "learning_rate": 3.6025251225363045e-06,
      "loss": 0.2612,
      "step": 4215
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.7511353427630962,
      "learning_rate": 3.6018801579245776e-06,
      "loss": 0.1556,
      "step": 4216
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9264289720378134,
      "learning_rate": 3.6012351022809142e-06,
      "loss": 0.1781,
      "step": 4217
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.223066275340934,
      "learning_rate": 3.6005899556586066e-06,
      "loss": 0.1719,
      "step": 4218
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.9033493909587573,
      "learning_rate": 3.5999447181109527e-06,
      "loss": 0.3145,
      "step": 4219
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.307144030600299,
      "learning_rate": 3.5992993896912587e-06,
      "loss": 0.2196,
      "step": 4220
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.8344882961551012,
      "learning_rate": 3.5986539704528383e-06,
      "loss": 0.3174,
      "step": 4221
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.264247224282054,
      "learning_rate": 3.5980084604490133e-06,
      "loss": 0.2791,
      "step": 4222
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.476006747900043,
      "learning_rate": 3.5973628597331135e-06,
      "loss": 0.3328,
      "step": 4223
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.7326132666177547,
      "learning_rate": 3.5967171683584735e-06,
      "loss": 0.2849,
      "step": 4224
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.2708407066965925,
      "learning_rate": 3.596071386378438e-06,
      "loss": 0.3243,
      "step": 4225
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.931881718626119,
      "learning_rate": 3.595425513846359e-06,
      "loss": 0.1843,
      "step": 4226
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.027620966853251,
      "learning_rate": 3.5947795508155936e-06,
      "loss": 0.3676,
      "step": 4227
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.8423526495224847,
      "learning_rate": 3.5941334973395097e-06,
      "loss": 0.2855,
      "step": 4228
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.206178483260814,
      "learning_rate": 3.5934873534714807e-06,
      "loss": 0.3373,
      "step": 4229
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.683386729615154,
      "learning_rate": 3.592841119264887e-06,
      "loss": 0.1992,
      "step": 4230
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.744761385899102,
      "learning_rate": 3.5921947947731184e-06,
      "loss": 0.271,
      "step": 4231
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.351067221813345,
      "learning_rate": 3.5915483800495705e-06,
      "loss": 0.4396,
      "step": 4232
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.951332142976353,
      "learning_rate": 3.590901875147647e-06,
      "loss": 0.184,
      "step": 4233
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.4061126728261453,
      "learning_rate": 3.590255280120759e-06,
      "loss": 0.2189,
      "step": 4234
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.629944261287949,
      "learning_rate": 3.5896085950223247e-06,
      "loss": 0.2267,
      "step": 4235
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.95294203327404,
      "learning_rate": 3.588961819905771e-06,
      "loss": 0.5049,
      "step": 4236
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.2253156095383595,
      "learning_rate": 3.58831495482453e-06,
      "loss": 0.1287,
      "step": 4237
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.446019321804585,
      "learning_rate": 3.5876679998320435e-06,
      "loss": 0.2812,
      "step": 4238
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.517918662048551,
      "learning_rate": 3.587020954981759e-06,
      "loss": 0.4231,
      "step": 4239
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.7213896516378933,
      "learning_rate": 3.586373820327133e-06,
      "loss": 0.2017,
      "step": 4240
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.739133037390339,
      "learning_rate": 3.585726595921628e-06,
      "loss": 0.2963,
      "step": 4241
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.032902877992094,
      "learning_rate": 3.585079281818715e-06,
      "loss": 0.4398,
      "step": 4242
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.798747988526218,
      "learning_rate": 3.5844318780718713e-06,
      "loss": 0.3223,
      "step": 4243
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.2983435740454152,
      "learning_rate": 3.583784384734582e-06,
      "loss": 0.1991,
      "step": 4244
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.1608708841706017,
      "learning_rate": 3.5831368018603407e-06,
      "loss": 0.1174,
      "step": 4245
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.441219950861798,
      "learning_rate": 3.5824891295026474e-06,
      "loss": 0.1875,
      "step": 4246
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.660667719369234,
      "learning_rate": 3.5818413677150093e-06,
      "loss": 0.4553,
      "step": 4247
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.9837188356827897,
      "learning_rate": 3.5811935165509415e-06,
      "loss": 0.0964,
      "step": 4248
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.266813523410169,
      "learning_rate": 3.5805455760639662e-06,
      "loss": 0.3725,
      "step": 4249
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.485532499995145,
      "learning_rate": 3.579897546307613e-06,
      "loss": 0.3704,
      "step": 4250
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.201678160521983,
      "learning_rate": 3.579249427335419e-06,
      "loss": 0.1536,
      "step": 4251
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.417085939901637,
      "learning_rate": 3.578601219200928e-06,
      "loss": 0.3204,
      "step": 4252
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.2474978312867115,
      "learning_rate": 3.5779529219576932e-06,
      "loss": 0.3279,
      "step": 4253
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.506478728916482,
      "learning_rate": 3.577304535659272e-06,
      "loss": 0.2215,
      "step": 4254
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.5867834125529456,
      "learning_rate": 3.5766560603592324e-06,
      "loss": 0.14,
      "step": 4255
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.207791482932693,
      "learning_rate": 3.576007496111148e-06,
      "loss": 0.2951,
      "step": 4256
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.336065994976776,
      "learning_rate": 3.575358842968599e-06,
      "loss": 0.2941,
      "step": 4257
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.100849842349121,
      "learning_rate": 3.5747101009851748e-06,
      "loss": 0.1967,
      "step": 4258
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.49690096413598,
      "learning_rate": 3.574061270214471e-06,
      "loss": 0.6299,
      "step": 4259
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.507954932061144,
      "learning_rate": 3.5734123507100915e-06,
      "loss": 0.2445,
      "step": 4260
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.325989355915776,
      "learning_rate": 3.572763342525646e-06,
      "loss": 0.1811,
      "step": 4261
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.597926524588493,
      "learning_rate": 3.5721142457147518e-06,
      "loss": 0.4174,
      "step": 4262
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3218171978233655,
      "learning_rate": 3.571465060331035e-06,
      "loss": 0.2468,
      "step": 4263
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.995800244882533,
      "learning_rate": 3.570815786428129e-06,
      "loss": 0.2392,
      "step": 4264
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.531660474877529,
      "learning_rate": 3.570166424059672e-06,
      "loss": 0.2731,
      "step": 4265
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6697176982971644,
      "learning_rate": 3.5695169732793124e-06,
      "loss": 0.0445,
      "step": 4266
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.940798346432592,
      "learning_rate": 3.5688674341407033e-06,
      "loss": 0.3522,
      "step": 4267
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.160163144091331,
      "learning_rate": 3.568217806697507e-06,
      "loss": 0.2421,
      "step": 4268
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.58426040444384,
      "learning_rate": 3.567568091003393e-06,
      "loss": 0.5659,
      "step": 4269
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.009857096478862,
      "learning_rate": 3.566918287112038e-06,
      "loss": 0.1607,
      "step": 4270
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.36618993614586,
      "learning_rate": 3.5662683950771248e-06,
      "loss": 0.2627,
      "step": 4271
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.6584215429573455,
      "learning_rate": 3.565618414952343e-06,
      "loss": 0.2066,
      "step": 4272
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.8350461457613605,
      "learning_rate": 3.564968346791393e-06,
      "loss": 0.235,
      "step": 4273
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.075232594400839,
      "learning_rate": 3.5643181906479802e-06,
      "loss": 0.155,
      "step": 4274
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.0083287579639535,
      "learning_rate": 3.5636679465758162e-06,
      "loss": 0.3108,
      "step": 4275
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.76912135945376,
      "learning_rate": 3.5630176146286214e-06,
      "loss": 0.3179,
      "step": 4276
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.294367350419887,
      "learning_rate": 3.5623671948601223e-06,
      "loss": 0.3767,
      "step": 4277
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0224832588903356,
      "learning_rate": 3.5617166873240543e-06,
      "loss": 0.0963,
      "step": 4278
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.8107030716752024,
      "learning_rate": 3.5610660920741598e-06,
      "loss": 0.2982,
      "step": 4279
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.734137289098981,
      "learning_rate": 3.5604154091641865e-06,
      "loss": 0.1292,
      "step": 4280
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.703575203608129,
      "learning_rate": 3.559764638647891e-06,
      "loss": 0.2315,
      "step": 4281
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.8406313112527712,
      "learning_rate": 3.5591137805790355e-06,
      "loss": 0.2948,
      "step": 4282
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.676939327286211,
      "learning_rate": 3.5584628350113933e-06,
      "loss": 0.2982,
      "step": 4283
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.150460085211162,
      "learning_rate": 3.5578118019987408e-06,
      "loss": 0.3331,
      "step": 4284
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.945911262914721,
      "learning_rate": 3.557160681594863e-06,
      "loss": 0.5724,
      "step": 4285
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.505339709358595,
      "learning_rate": 3.556509473853554e-06,
      "loss": 0.4587,
      "step": 4286
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.7333022169959076,
      "learning_rate": 3.5558581788286102e-06,
      "loss": 0.3275,
      "step": 4287
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.174958177942963,
      "learning_rate": 3.5552067965738415e-06,
      "loss": 0.4532,
      "step": 4288
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.19507277094739,
      "learning_rate": 3.55455532714306e-06,
      "loss": 0.2996,
      "step": 4289
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.783582193464245,
      "learning_rate": 3.553903770590088e-06,
      "loss": 0.3818,
      "step": 4290
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.261027478234527,
      "learning_rate": 3.5532521269687542e-06,
      "loss": 0.2739,
      "step": 4291
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.566102549151821,
      "learning_rate": 3.5526003963328914e-06,
      "loss": 0.4312,
      "step": 4292
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.2745395199400384,
      "learning_rate": 3.551948578736346e-06,
      "loss": 0.3516,
      "step": 4293
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.227747327245502,
      "learning_rate": 3.551296674232967e-06,
      "loss": 0.5151,
      "step": 4294
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.96340365158574,
      "learning_rate": 3.55064468287661e-06,
      "loss": 0.3513,
      "step": 4295
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.987683613394473,
      "learning_rate": 3.5499926047211403e-06,
      "loss": 0.1612,
      "step": 4296
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.026517052775575,
      "learning_rate": 3.5493404398204296e-06,
      "loss": 0.4436,
      "step": 4297
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.59298748065801,
      "learning_rate": 3.548688188228356e-06,
      "loss": 0.4709,
      "step": 4298
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.8059449885415475,
      "learning_rate": 3.5480358499988065e-06,
      "loss": 0.2501,
      "step": 4299
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2457321582391025,
      "learning_rate": 3.547383425185674e-06,
      "loss": 0.1012,
      "step": 4300
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.837589616405263,
      "learning_rate": 3.5467309138428563e-06,
      "loss": 0.3939,
      "step": 4301
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.7961757635075686,
      "learning_rate": 3.5460783160242628e-06,
      "loss": 0.3069,
      "step": 4302
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.006841826177933,
      "learning_rate": 3.5454256317838077e-06,
      "loss": 0.3183,
      "step": 4303
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.977357110193397,
      "learning_rate": 3.544772861175413e-06,
      "loss": 0.252,
      "step": 4304
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.722136347603987,
      "learning_rate": 3.544120004253006e-06,
      "loss": 0.447,
      "step": 4305
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.418025481301561,
      "learning_rate": 3.543467061070523e-06,
      "loss": 0.308,
      "step": 4306
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.033948578820753,
      "learning_rate": 3.5428140316819076e-06,
      "loss": 0.2882,
      "step": 4307
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.292400520466944,
      "learning_rate": 3.5421609161411096e-06,
      "loss": 0.3643,
      "step": 4308
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.909362126408001,
      "learning_rate": 3.5415077145020865e-06,
      "loss": 0.4182,
      "step": 4309
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.251634873848706,
      "learning_rate": 3.540854426818802e-06,
      "loss": 0.6111,
      "step": 4310
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.7045271281534875,
      "learning_rate": 3.5402010531452275e-06,
      "loss": 0.0719,
      "step": 4311
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.927987142466689,
      "learning_rate": 3.5395475935353417e-06,
      "loss": 0.2224,
      "step": 4312
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.6255259764948073,
      "learning_rate": 3.538894048043131e-06,
      "loss": 0.0812,
      "step": 4313
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.982485200119896,
      "learning_rate": 3.538240416722587e-06,
      "loss": 0.4161,
      "step": 4314
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.256847556899638,
      "learning_rate": 3.5375866996277107e-06,
      "loss": 0.2551,
      "step": 4315
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.472180882751806,
      "learning_rate": 3.5369328968125073e-06,
      "loss": 0.246,
      "step": 4316
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.7055328721444685,
      "learning_rate": 3.5362790083309917e-06,
      "loss": 0.4434,
      "step": 4317
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.8305819571329955,
      "learning_rate": 3.535625034237186e-06,
      "loss": 0.4769,
      "step": 4318
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.449559489214049,
      "learning_rate": 3.534970974585117e-06,
      "loss": 0.4099,
      "step": 4319
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.822019390297152,
      "learning_rate": 3.53431682942882e-06,
      "loss": 0.349,
      "step": 4320
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.890730247531691,
      "learning_rate": 3.533662598822338e-06,
      "loss": 0.2133,
      "step": 4321
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3708871438570274,
      "learning_rate": 3.533008282819719e-06,
      "loss": 0.1351,
      "step": 4322
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.161931437054984,
      "learning_rate": 3.5323538814750213e-06,
      "loss": 0.2212,
      "step": 4323
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.2944676870691945,
      "learning_rate": 3.5316993948423066e-06,
      "loss": 0.3931,
      "step": 4324
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.8060832045480897,
      "learning_rate": 3.531044822975646e-06,
      "loss": 0.2673,
      "step": 4325
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.6803365783647104,
      "learning_rate": 3.530390165929117e-06,
      "loss": 0.2792,
      "step": 4326
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.25687122164641,
      "learning_rate": 3.529735423756805e-06,
      "loss": 0.3526,
      "step": 4327
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.642037720045009,
      "learning_rate": 3.5290805965128007e-06,
      "loss": 0.2872,
      "step": 4328
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.447367787588808,
      "learning_rate": 3.528425684251202e-06,
      "loss": 0.3497,
      "step": 4329
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.7062326056147175,
      "learning_rate": 3.5277706870261157e-06,
      "loss": 0.3027,
      "step": 4330
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.863465188685267,
      "learning_rate": 3.5271156048916546e-06,
      "loss": 0.3334,
      "step": 4331
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.7670765826607555,
      "learning_rate": 3.526460437901937e-06,
      "loss": 0.3665,
      "step": 4332
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.444828544263069,
      "learning_rate": 3.5258051861110915e-06,
      "loss": 0.2324,
      "step": 4333
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.423273031462882,
      "learning_rate": 3.5251498495732505e-06,
      "loss": 0.1682,
      "step": 4334
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.579304051576486,
      "learning_rate": 3.524494428342554e-06,
      "loss": 0.3855,
      "step": 4335
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.9229123234510133,
      "learning_rate": 3.523838922473152e-06,
      "loss": 0.1371,
      "step": 4336
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.423616031913131,
      "learning_rate": 3.5231833320191973e-06,
      "loss": 0.3003,
      "step": 4337
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.686635739792346,
      "learning_rate": 3.522527657034852e-06,
      "loss": 0.5791,
      "step": 4338
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.729396592548729,
      "learning_rate": 3.521871897574285e-06,
      "loss": 0.4254,
      "step": 4339
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.155133979401558,
      "learning_rate": 3.521216053691672e-06,
      "loss": 0.5556,
      "step": 4340
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.811663464970883,
      "learning_rate": 3.520560125441195e-06,
      "loss": 0.3768,
      "step": 4341
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.298795405743272,
      "learning_rate": 3.519904112877044e-06,
      "loss": 0.2887,
      "step": 4342
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.514957053644864,
      "learning_rate": 3.5192480160534154e-06,
      "loss": 0.2551,
      "step": 4343
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.9669001397443355,
      "learning_rate": 3.518591835024513e-06,
      "loss": 0.2051,
      "step": 4344
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8507936877152495,
      "learning_rate": 3.5179355698445473e-06,
      "loss": 0.2852,
      "step": 4345
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.780878245548417,
      "learning_rate": 3.5172792205677355e-06,
      "loss": 0.2875,
      "step": 4346
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.929225125951385,
      "learning_rate": 3.516622787248301e-06,
      "loss": 0.185,
      "step": 4347
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.742558427397659,
      "learning_rate": 3.5159662699404774e-06,
      "loss": 0.2281,
      "step": 4348
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8084498405696605,
      "learning_rate": 3.5153096686985006e-06,
      "loss": 0.38,
      "step": 4349
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.832724440264584,
      "learning_rate": 3.5146529835766175e-06,
      "loss": 0.2966,
      "step": 4350
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.116464724369258,
      "learning_rate": 3.513996214629079e-06,
      "loss": 0.1479,
      "step": 4351
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.086983875295809,
      "learning_rate": 3.5133393619101448e-06,
      "loss": 0.2489,
      "step": 4352
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.837068343130248,
      "learning_rate": 3.5126824254740805e-06,
      "loss": 0.3734,
      "step": 4353
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.669409523517122,
      "learning_rate": 3.512025405375159e-06,
      "loss": 0.4662,
      "step": 4354
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.0433861160547244,
      "learning_rate": 3.5113683016676602e-06,
      "loss": 0.1756,
      "step": 4355
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.0736717264112468,
      "learning_rate": 3.5107111144058715e-06,
      "loss": 0.2098,
      "step": 4356
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.049783562941118,
      "learning_rate": 3.510053843644085e-06,
      "loss": 0.3381,
      "step": 4357
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.891047808067241,
      "learning_rate": 3.5093964894366015e-06,
      "loss": 0.3051,
      "step": 4358
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.7650894508161286,
      "learning_rate": 3.5087390518377303e-06,
      "loss": 0.3159,
      "step": 4359
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.4443327391579497,
      "learning_rate": 3.5080815309017835e-06,
      "loss": 0.2062,
      "step": 4360
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.896022750788493,
      "learning_rate": 3.507423926683083e-06,
      "loss": 0.2646,
      "step": 4361
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.986204131291998,
      "learning_rate": 3.5067662392359568e-06,
      "loss": 0.2259,
      "step": 4362
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.0275085518106595,
      "learning_rate": 3.5061084686147397e-06,
      "loss": 0.2976,
      "step": 4363
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.670358869390494,
      "learning_rate": 3.5054506148737737e-06,
      "loss": 0.5151,
      "step": 4364
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8686858972082465,
      "learning_rate": 3.5047926780674075e-06,
      "loss": 0.2371,
      "step": 4365
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.793822943491846,
      "learning_rate": 3.504134658249997e-06,
      "loss": 0.122,
      "step": 4366
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.229828568399979,
      "learning_rate": 3.5034765554759027e-06,
      "loss": 0.4439,
      "step": 4367
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.787670535267368,
      "learning_rate": 3.5028183697994955e-06,
      "loss": 0.3171,
      "step": 4368
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.214414737880198,
      "learning_rate": 3.5021601012751516e-06,
      "loss": 0.3614,
      "step": 4369
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.651665214101783,
      "learning_rate": 3.5015017499572533e-06,
      "loss": 0.3083,
      "step": 4370
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.2356697726520856,
      "learning_rate": 3.5008433159001907e-06,
      "loss": 0.1032,
      "step": 4371
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.406957927906856,
      "learning_rate": 3.5001847991583604e-06,
      "loss": 0.3325,
      "step": 4372
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.9187979717939654,
      "learning_rate": 3.499526199786165e-06,
      "loss": 0.2841,
      "step": 4373
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.573799090228754,
      "learning_rate": 3.498867517838016e-06,
      "loss": 0.397,
      "step": 4374
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.957490957024616,
      "learning_rate": 3.4982087533683294e-06,
      "loss": 0.1424,
      "step": 4375
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.251745457669932,
      "learning_rate": 3.4975499064315298e-06,
      "loss": 0.202,
      "step": 4376
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8778324865781535,
      "learning_rate": 3.4968909770820475e-06,
      "loss": 0.3085,
      "step": 4377
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.086201218675915,
      "learning_rate": 3.49623196537432e-06,
      "loss": 0.3225,
      "step": 4378
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.658492052035701,
      "learning_rate": 3.4955728713627922e-06,
      "loss": 0.2491,
      "step": 4379
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.334534458333559,
      "learning_rate": 3.4949136951019146e-06,
      "loss": 0.3981,
      "step": 4380
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.302684125192622,
      "learning_rate": 3.4942544366461456e-06,
      "loss": 0.3808,
      "step": 4381
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.92806458588412,
      "learning_rate": 3.4935950960499486e-06,
      "loss": 0.3044,
      "step": 4382
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.545880518961156,
      "learning_rate": 3.492935673367797e-06,
      "loss": 0.4363,
      "step": 4383
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1616122953576102,
      "learning_rate": 3.4922761686541686e-06,
      "loss": 0.0839,
      "step": 4384
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.749825272700733,
      "learning_rate": 3.491616581963547e-06,
      "loss": 0.4801,
      "step": 4385
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.3956171682568135,
      "learning_rate": 3.490956913350426e-06,
      "loss": 0.4257,
      "step": 4386
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.27936846314115,
      "learning_rate": 3.490297162869302e-06,
      "loss": 0.099,
      "step": 4387
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.060659328065661,
      "learning_rate": 3.4896373305746823e-06,
      "loss": 0.3153,
      "step": 4388
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.2308857186766184,
      "learning_rate": 3.4889774165210784e-06,
      "loss": 0.248,
      "step": 4389
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.134183131574437,
      "learning_rate": 3.4883174207630088e-06,
      "loss": 0.1776,
      "step": 4390
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8009172713355674,
      "learning_rate": 3.487657343355e-06,
      "loss": 0.2831,
      "step": 4391
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.7477024615785735,
      "learning_rate": 3.486997184351582e-06,
      "loss": 0.155,
      "step": 4392
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.783148946121208,
      "learning_rate": 3.4863369438072968e-06,
      "loss": 0.1853,
      "step": 4393
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.129525211109743,
      "learning_rate": 3.4856766217766895e-06,
      "loss": 0.2593,
      "step": 4394
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.1431318008899325,
      "learning_rate": 3.485016218314312e-06,
      "loss": 0.1233,
      "step": 4395
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.335328036503096,
      "learning_rate": 3.484355733474723e-06,
      "loss": 0.4436,
      "step": 4396
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.912051804570991,
      "learning_rate": 3.4836951673124897e-06,
      "loss": 0.1142,
      "step": 4397
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.1480159445194125,
      "learning_rate": 3.4830345198821847e-06,
      "loss": 0.1871,
      "step": 4398
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.291540044013511,
      "learning_rate": 3.4823737912383875e-06,
      "loss": 0.3784,
      "step": 4399
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.040579147303177,
      "learning_rate": 3.481712981435684e-06,
      "loss": 0.4605,
      "step": 4400
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.943340540791445,
      "learning_rate": 3.4810520905286664e-06,
      "loss": 0.3124,
      "step": 4401
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.860967947159777,
      "learning_rate": 3.480391118571935e-06,
      "loss": 0.2399,
      "step": 4402
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.2887897692405454,
      "learning_rate": 3.479730065620096e-06,
      "loss": 0.604,
      "step": 4403
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9823066238965372,
      "learning_rate": 3.479068931727763e-06,
      "loss": 0.0727,
      "step": 4404
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.221286305942506,
      "learning_rate": 3.4784077169495545e-06,
      "loss": 0.2398,
      "step": 4405
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.4907299062448685,
      "learning_rate": 3.4777464213400976e-06,
      "loss": 0.3976,
      "step": 4406
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.456448364990262,
      "learning_rate": 3.477085044954024e-06,
      "loss": 0.4006,
      "step": 4407
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.2965271546860264,
      "learning_rate": 3.476423587845976e-06,
      "loss": 0.1995,
      "step": 4408
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.18435299472433,
      "learning_rate": 3.475762050070597e-06,
      "loss": 0.4978,
      "step": 4409
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2031824920142347,
      "learning_rate": 3.475100431682542e-06,
      "loss": 0.0969,
      "step": 4410
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.6006917141564365,
      "learning_rate": 3.4744387327364693e-06,
      "loss": 0.3697,
      "step": 4411
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.637937093181274,
      "learning_rate": 3.473776953287046e-06,
      "loss": 0.2844,
      "step": 4412
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.63718494336665,
      "learning_rate": 3.4731150933889455e-06,
      "loss": 0.193,
      "step": 4413
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.508059389061264,
      "learning_rate": 3.4724531530968476e-06,
      "loss": 0.2954,
      "step": 4414
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.37563637587726,
      "learning_rate": 3.471791132465437e-06,
      "loss": 0.2272,
      "step": 4415
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.252557746434914,
      "learning_rate": 3.4711290315494072e-06,
      "loss": 0.2463,
      "step": 4416
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.550113480181691,
      "learning_rate": 3.4704668504034578e-06,
      "loss": 0.3517,
      "step": 4417
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.802528773973703,
      "learning_rate": 3.4698045890822963e-06,
      "loss": 0.2044,
      "step": 4418
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.315324790393712,
      "learning_rate": 3.4691422476406335e-06,
      "loss": 0.3037,
      "step": 4419
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.831557218178395,
      "learning_rate": 3.4684798261331898e-06,
      "loss": 0.2652,
      "step": 4420
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.778744240308241,
      "learning_rate": 3.467817324614691e-06,
      "loss": 0.5337,
      "step": 4421
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.8152318166194927,
      "learning_rate": 3.46715474313987e-06,
      "loss": 0.2628,
      "step": 4422
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.888270918391274,
      "learning_rate": 3.4664920817634657e-06,
      "loss": 0.3636,
      "step": 4423
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.167875855230275,
      "learning_rate": 3.4658293405402245e-06,
      "loss": 0.431,
      "step": 4424
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.79228451939911,
      "learning_rate": 3.4651665195248975e-06,
      "loss": 0.3575,
      "step": 4425
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.073219215347603,
      "learning_rate": 3.4645036187722453e-06,
      "loss": 0.3209,
      "step": 4426
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.86412884920326,
      "learning_rate": 3.4638406383370327e-06,
      "loss": 0.3434,
      "step": 4427
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.8049892690599174,
      "learning_rate": 3.4631775782740316e-06,
      "loss": 0.3295,
      "step": 4428
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.0809437703055575,
      "learning_rate": 3.462514438638022e-06,
      "loss": 0.453,
      "step": 4429
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.2670942281887205,
      "learning_rate": 3.461851219483788e-06,
      "loss": 0.2768,
      "step": 4430
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.6488494281652075,
      "learning_rate": 3.461187920866122e-06,
      "loss": 0.4591,
      "step": 4431
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.042395988321581,
      "learning_rate": 3.460524542839822e-06,
      "loss": 0.2878,
      "step": 4432
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.605016625095532,
      "learning_rate": 3.4598610854596935e-06,
      "loss": 0.3065,
      "step": 4433
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3409310650168567,
      "learning_rate": 3.459197548780549e-06,
      "loss": 0.2422,
      "step": 4434
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.392131158701197,
      "learning_rate": 3.458533932857205e-06,
      "loss": 0.2337,
      "step": 4435
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.432925085593234,
      "learning_rate": 3.4578702377444872e-06,
      "loss": 0.2927,
      "step": 4436
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.5774182092899385,
      "learning_rate": 3.4572064634972257e-06,
      "loss": 0.3178,
      "step": 4437
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.317539737737469,
      "learning_rate": 3.4565426101702605e-06,
      "loss": 0.3599,
      "step": 4438
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.8853680472468213,
      "learning_rate": 3.455878677818433e-06,
      "loss": 0.1355,
      "step": 4439
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.838583322348048,
      "learning_rate": 3.455214666496597e-06,
      "loss": 0.3815,
      "step": 4440
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.239452770185887,
      "learning_rate": 3.4545505762596078e-06,
      "loss": 0.4235,
      "step": 4441
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.417923323237768,
      "learning_rate": 3.4538864071623296e-06,
      "loss": 0.1609,
      "step": 4442
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.555187206767811,
      "learning_rate": 3.4532221592596336e-06,
      "loss": 0.4023,
      "step": 4443
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.836360723420115,
      "learning_rate": 3.4525578326063957e-06,
      "loss": 0.3324,
      "step": 4444
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.02759780962836,
      "learning_rate": 3.4518934272575e-06,
      "loss": 0.2643,
      "step": 4445
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.396350612781855,
      "learning_rate": 3.451228943267836e-06,
      "loss": 0.2614,
      "step": 4446
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.838777291204982,
      "learning_rate": 3.450564380692301e-06,
      "loss": 0.387,
      "step": 4447
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5610456240108592,
      "learning_rate": 3.4498997395857968e-06,
      "loss": 0.0412,
      "step": 4448
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.187806148858569,
      "learning_rate": 3.449235020003233e-06,
      "loss": 0.3033,
      "step": 4449
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.3563264912782236,
      "learning_rate": 3.4485702219995266e-06,
      "loss": 0.1038,
      "step": 4450
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.872621854657298,
      "learning_rate": 3.4479053456295986e-06,
      "loss": 0.2848,
      "step": 4451
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.750095327991059,
      "learning_rate": 3.4472403909483775e-06,
      "loss": 0.3792,
      "step": 4452
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.8142553407433875,
      "learning_rate": 3.4465753580108e-06,
      "loss": 0.2727,
      "step": 4453
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.20914791545139,
      "learning_rate": 3.445910246871807e-06,
      "loss": 0.4243,
      "step": 4454
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.181082164705995,
      "learning_rate": 3.445245057586347e-06,
      "loss": 0.1175,
      "step": 4455
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.01026215900132,
      "learning_rate": 3.4445797902093753e-06,
      "loss": 0.4099,
      "step": 4456
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.245985443355488,
      "learning_rate": 3.443914444795852e-06,
      "loss": 0.1607,
      "step": 4457
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.9551049599566945,
      "learning_rate": 3.443249021400744e-06,
      "loss": 0.3923,
      "step": 4458
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.814023087551492,
      "learning_rate": 3.4425835200790276e-06,
      "loss": 0.3986,
      "step": 4459
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.8757637271048293,
      "learning_rate": 3.4419179408856817e-06,
      "loss": 0.0615,
      "step": 4460
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.475847367534337,
      "learning_rate": 3.441252283875694e-06,
      "loss": 0.1719,
      "step": 4461
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.630599113903705,
      "learning_rate": 3.4405865491040563e-06,
      "loss": 0.4041,
      "step": 4462
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.417990983931962,
      "learning_rate": 3.439920736625769e-06,
      "loss": 0.1481,
      "step": 4463
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5210304263163583,
      "learning_rate": 3.43925484649584e-06,
      "loss": 0.3293,
      "step": 4464
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5618816142434717,
      "learning_rate": 3.43858887876928e-06,
      "loss": 0.2658,
      "step": 4465
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.382488140994815,
      "learning_rate": 3.437922833501109e-06,
      "loss": 0.2042,
      "step": 4466
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.4508794781470815,
      "learning_rate": 3.4372567107463508e-06,
      "loss": 0.587,
      "step": 4467
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5903582707914694,
      "learning_rate": 3.4365905105600382e-06,
      "loss": 0.0959,
      "step": 4468
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.580432716234485,
      "learning_rate": 3.435924232997211e-06,
      "loss": 0.3353,
      "step": 4469
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.471231083617953,
      "learning_rate": 3.435257878112911e-06,
      "loss": 0.5016,
      "step": 4470
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.173904783574969,
      "learning_rate": 3.4345914459621908e-06,
      "loss": 0.4049,
      "step": 4471
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.492704157665788,
      "learning_rate": 3.4339249366001073e-06,
      "loss": 0.4191,
      "step": 4472
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.070344594727811,
      "learning_rate": 3.433258350081724e-06,
      "loss": 0.4336,
      "step": 4473
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6225935935269855,
      "learning_rate": 3.4325916864621125e-06,
      "loss": 0.2259,
      "step": 4474
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.881851942662907,
      "learning_rate": 3.4319249457963472e-06,
      "loss": 0.2915,
      "step": 4475
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.778413123851687,
      "learning_rate": 3.431258128139512e-06,
      "loss": 0.309,
      "step": 4476
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.3958498351554818,
      "learning_rate": 3.4305912335466955e-06,
      "loss": 0.1237,
      "step": 4477
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.051752339569713,
      "learning_rate": 3.429924262072994e-06,
      "loss": 0.31,
      "step": 4478
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.0566162584612044,
      "learning_rate": 3.4292572137735097e-06,
      "loss": 0.24,
      "step": 4479
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.520534529082731,
      "learning_rate": 3.42859008870335e-06,
      "loss": 0.1654,
      "step": 4480
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.479132520129585,
      "learning_rate": 3.4279228869176302e-06,
      "loss": 0.1293,
      "step": 4481
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6951228751233334,
      "learning_rate": 3.4272556084714692e-06,
      "loss": 0.1155,
      "step": 4482
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.202523755409608,
      "learning_rate": 3.4265882534199975e-06,
      "loss": 0.4022,
      "step": 4483
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.818733963666171,
      "learning_rate": 3.425920821818347e-06,
      "loss": 0.4529,
      "step": 4484
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.84799013107984,
      "learning_rate": 3.4252533137216577e-06,
      "loss": 0.3634,
      "step": 4485
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.067130107181637,
      "learning_rate": 3.4245857291850754e-06,
      "loss": 0.264,
      "step": 4486
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.967077138650309,
      "learning_rate": 3.4239180682637528e-06,
      "loss": 0.5151,
      "step": 4487
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.353543478945049,
      "learning_rate": 3.4232503310128497e-06,
      "loss": 0.3401,
      "step": 4488
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.05929567053522,
      "learning_rate": 3.422582517487531e-06,
      "loss": 0.3059,
      "step": 4489
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.903320363995885,
      "learning_rate": 3.4219146277429678e-06,
      "loss": 0.3845,
      "step": 4490
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.573723003099975,
      "learning_rate": 3.421246661834338e-06,
      "loss": 0.3528,
      "step": 4491
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8987790538471,
      "learning_rate": 3.4205786198168243e-06,
      "loss": 0.4208,
      "step": 4492
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.145200504333146,
      "learning_rate": 3.41991050174562e-06,
      "loss": 0.2033,
      "step": 4493
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.142782664379731,
      "learning_rate": 3.4192423076759197e-06,
      "loss": 0.3085,
      "step": 4494
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.119045349662358,
      "learning_rate": 3.4185740376629273e-06,
      "loss": 0.2149,
      "step": 4495
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.843704830629133,
      "learning_rate": 3.41790569176185e-06,
      "loss": 0.2567,
      "step": 4496
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.368452334902323,
      "learning_rate": 3.4172372700279055e-06,
      "loss": 0.3196,
      "step": 4497
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.084110970505712,
      "learning_rate": 3.416568772516315e-06,
      "loss": 0.5132,
      "step": 4498
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.682765001291204,
      "learning_rate": 3.415900199282306e-06,
      "loss": 0.3656,
      "step": 4499
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.384352669040547,
      "learning_rate": 3.415231550381113e-06,
      "loss": 0.4893,
      "step": 4500
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.49643201587685,
      "learning_rate": 3.414562825867976e-06,
      "loss": 0.3545,
      "step": 4501
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.011423950438376,
      "learning_rate": 3.4138940257981425e-06,
      "loss": 0.3742,
      "step": 4502
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.102137519724797,
      "learning_rate": 3.413225150226865e-06,
      "loss": 0.1492,
      "step": 4503
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.978881568131904,
      "learning_rate": 3.4125561992094033e-06,
      "loss": 0.3425,
      "step": 4504
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5044553083748666,
      "learning_rate": 3.411887172801022e-06,
      "loss": 0.2553,
      "step": 4505
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.927202615510398,
      "learning_rate": 3.411218071056993e-06,
      "loss": 0.3064,
      "step": 4506
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8682856794190275,
      "learning_rate": 3.410548894032595e-06,
      "loss": 0.3319,
      "step": 4507
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.180937877035494,
      "learning_rate": 3.4098796417831115e-06,
      "loss": 0.3742,
      "step": 4508
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.019916195680005,
      "learning_rate": 3.4092103143638335e-06,
      "loss": 0.2751,
      "step": 4509
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.374969704488578,
      "learning_rate": 3.408540911830056e-06,
      "loss": 0.5802,
      "step": 4510
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6037391692352148,
      "learning_rate": 3.4078714342370828e-06,
      "loss": 0.223,
      "step": 4511
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.9802091350272475,
      "learning_rate": 3.4072018816402223e-06,
      "loss": 0.2497,
      "step": 4512
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.75369442809561,
      "learning_rate": 3.406532254094791e-06,
      "loss": 0.11,
      "step": 4513
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.896804142164942,
      "learning_rate": 3.4058625516561093e-06,
      "loss": 0.3454,
      "step": 4514
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.143505338479911,
      "learning_rate": 3.405192774379505e-06,
      "loss": 0.4016,
      "step": 4515
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.654575984213122,
      "learning_rate": 3.4045229223203107e-06,
      "loss": 0.3651,
      "step": 4516
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8045128375211292,
      "learning_rate": 3.403852995533867e-06,
      "loss": 0.3152,
      "step": 4517
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8164153866955062,
      "learning_rate": 3.4031829940755213e-06,
      "loss": 0.0154,
      "step": 4518
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.152515037424124,
      "learning_rate": 3.4025129180006244e-06,
      "loss": 0.2976,
      "step": 4519
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.2180273250691815,
      "learning_rate": 3.401842767364535e-06,
      "loss": 0.3286,
      "step": 4520
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.356212703306528,
      "learning_rate": 3.401172542222617e-06,
      "loss": 0.3311,
      "step": 4521
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.963276439146577,
      "learning_rate": 3.4005022426302427e-06,
      "loss": 0.3963,
      "step": 4522
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.056337786265482,
      "learning_rate": 3.399831868642787e-06,
      "loss": 0.3981,
      "step": 4523
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.908498524607385,
      "learning_rate": 3.399161420315635e-06,
      "loss": 0.1919,
      "step": 4524
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.479074607739294,
      "learning_rate": 3.398490897704174e-06,
      "loss": 0.2532,
      "step": 4525
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.369633225323017,
      "learning_rate": 3.3978203008638007e-06,
      "loss": 0.391,
      "step": 4526
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.339903730556575,
      "learning_rate": 3.397149629849915e-06,
      "loss": 0.1726,
      "step": 4527
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2024085827093947,
      "learning_rate": 3.3964788847179266e-06,
      "loss": 0.0296,
      "step": 4528
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.6517509398235246,
      "learning_rate": 3.395808065523247e-06,
      "loss": 0.3041,
      "step": 4529
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8375374161636238,
      "learning_rate": 3.395137172321298e-06,
      "loss": 0.2948,
      "step": 4530
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.9937702569998166,
      "learning_rate": 3.3944662051675037e-06,
      "loss": 0.1386,
      "step": 4531
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8014263773849604,
      "learning_rate": 3.393795164117296e-06,
      "loss": 0.2379,
      "step": 4532
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.8250130115106127,
      "learning_rate": 3.393124049226115e-06,
      "loss": 0.2455,
      "step": 4533
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.700336399271132,
      "learning_rate": 3.3924528605494044e-06,
      "loss": 0.5979,
      "step": 4534
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.272221774905596,
      "learning_rate": 3.3917815981426126e-06,
      "loss": 0.4504,
      "step": 4535
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.5673997054841955,
      "learning_rate": 3.391110262061198e-06,
      "loss": 0.5455,
      "step": 4536
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.08988117978627,
      "learning_rate": 3.390438852360622e-06,
      "loss": 0.2768,
      "step": 4537
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.988277157204314,
      "learning_rate": 3.389767369096354e-06,
      "loss": 0.3868,
      "step": 4538
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.862372197052691,
      "learning_rate": 3.389095812323868e-06,
      "loss": 0.2171,
      "step": 4539
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.935193459110184,
      "learning_rate": 3.3884241820986453e-06,
      "loss": 0.5011,
      "step": 4540
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.026317862800136,
      "learning_rate": 3.3877524784761725e-06,
      "loss": 0.2508,
      "step": 4541
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.254196583637085,
      "learning_rate": 3.3870807015119415e-06,
      "loss": 0.24,
      "step": 4542
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.328179424716486,
      "learning_rate": 3.3864088512614527e-06,
      "loss": 0.2021,
      "step": 4543
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.377497970340581,
      "learning_rate": 3.38573692778021e-06,
      "loss": 0.4528,
      "step": 4544
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.179224055582999,
      "learning_rate": 3.3850649311237256e-06,
      "loss": 0.3662,
      "step": 4545
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.686067428332378,
      "learning_rate": 3.3843928613475156e-06,
      "loss": 0.2639,
      "step": 4546
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.2224493418141456,
      "learning_rate": 3.3837207185071033e-06,
      "loss": 0.1271,
      "step": 4547
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.378976654184847,
      "learning_rate": 3.383048502658017e-06,
      "loss": 0.4227,
      "step": 4548
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.378551524849373,
      "learning_rate": 3.382376213855794e-06,
      "loss": 0.5187,
      "step": 4549
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.513723587483803,
      "learning_rate": 3.381703852155974e-06,
      "loss": 0.3868,
      "step": 4550
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.251645024035851,
      "learning_rate": 3.3810314176141047e-06,
      "loss": 0.2347,
      "step": 4551
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.587637793993155,
      "learning_rate": 3.380358910285738e-06,
      "loss": 0.592,
      "step": 4552
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.698829569825452,
      "learning_rate": 3.3796863302264354e-06,
      "loss": 0.3051,
      "step": 4553
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.436711686626426,
      "learning_rate": 3.3790136774917604e-06,
      "loss": 0.3118,
      "step": 4554
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.4552888968073034,
      "learning_rate": 3.3783409521372856e-06,
      "loss": 0.2098,
      "step": 4555
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.5724406659635686,
      "learning_rate": 3.377668154218587e-06,
      "loss": 0.1135,
      "step": 4556
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.334412834222784,
      "learning_rate": 3.376995283791249e-06,
      "loss": 0.3569,
      "step": 4557
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.025201520206651,
      "learning_rate": 3.37632234091086e-06,
      "loss": 0.3085,
      "step": 4558
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.471866298899471,
      "learning_rate": 3.3756493256330153e-06,
      "loss": 0.4144,
      "step": 4559
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.5580642604147212,
      "learning_rate": 3.3749762380133165e-06,
      "loss": 0.2151,
      "step": 4560
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.681260611693176,
      "learning_rate": 3.3743030781073716e-06,
      "loss": 0.4466,
      "step": 4561
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.403187757478537,
      "learning_rate": 3.3736298459707916e-06,
      "loss": 0.4462,
      "step": 4562
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.354470335202082,
      "learning_rate": 3.3729565416591974e-06,
      "loss": 0.3286,
      "step": 4563
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.943551834749159,
      "learning_rate": 3.372283165228214e-06,
      "loss": 0.0755,
      "step": 4564
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.523509052436471,
      "learning_rate": 3.371609716733472e-06,
      "loss": 0.3978,
      "step": 4565
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.992338722272154,
      "learning_rate": 3.370936196230608e-06,
      "loss": 0.1568,
      "step": 4566
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.390465626220885,
      "learning_rate": 3.3702626037752657e-06,
      "loss": 0.4796,
      "step": 4567
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.584228931825612,
      "learning_rate": 3.369588939423093e-06,
      "loss": 0.4342,
      "step": 4568
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.907610340013593,
      "learning_rate": 3.368915203229747e-06,
      "loss": 0.4942,
      "step": 4569
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.099546304247292,
      "learning_rate": 3.368241395250886e-06,
      "loss": 0.4196,
      "step": 4570
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.794088766256053,
      "learning_rate": 3.3675675155421783e-06,
      "loss": 0.3459,
      "step": 4571
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4464379970725862,
      "learning_rate": 3.366893564159295e-06,
      "loss": 0.2658,
      "step": 4572
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.198990048048132,
      "learning_rate": 3.366219541157916e-06,
      "loss": 0.4025,
      "step": 4573
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2417830277699173,
      "learning_rate": 3.3655454465937253e-06,
      "loss": 0.1128,
      "step": 4574
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.1135784982753325,
      "learning_rate": 3.3648712805224137e-06,
      "loss": 0.238,
      "step": 4575
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.221851309093853,
      "learning_rate": 3.364197042999677e-06,
      "loss": 0.2867,
      "step": 4576
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.374183922577403,
      "learning_rate": 3.363522734081217e-06,
      "loss": 0.3489,
      "step": 4577
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.143845183078808,
      "learning_rate": 3.3628483538227423e-06,
      "loss": 0.2433,
      "step": 4578
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.487572171508534,
      "learning_rate": 3.3621739022799683e-06,
      "loss": 0.2667,
      "step": 4579
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.900884738157544,
      "learning_rate": 3.3614993795086127e-06,
      "loss": 0.2061,
      "step": 4580
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.430306723317722,
      "learning_rate": 3.3608247855644014e-06,
      "loss": 0.368,
      "step": 4581
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7491397342274535,
      "learning_rate": 3.360150120503067e-06,
      "loss": 0.2024,
      "step": 4582
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.5770237833384226,
      "learning_rate": 3.3594753843803473e-06,
      "loss": 0.2552,
      "step": 4583
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2614075162196796,
      "learning_rate": 3.358800577251985e-06,
      "loss": 0.2135,
      "step": 4584
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.018778771058904,
      "learning_rate": 3.3581256991737292e-06,
      "loss": 0.3607,
      "step": 4585
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.009018874640202,
      "learning_rate": 3.3574507502013354e-06,
      "loss": 0.3587,
      "step": 4586
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4434502358123584,
      "learning_rate": 3.356775730390564e-06,
      "loss": 0.2896,
      "step": 4587
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.428137415883407,
      "learning_rate": 3.3561006397971828e-06,
      "loss": 0.5673,
      "step": 4588
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.148558147242833,
      "learning_rate": 3.3554254784769642e-06,
      "loss": 0.3465,
      "step": 4589
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.047957632793691,
      "learning_rate": 3.3547502464856863e-06,
      "loss": 0.1955,
      "step": 4590
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7156317371268854,
      "learning_rate": 3.3540749438791343e-06,
      "loss": 0.2227,
      "step": 4591
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.854069168952488,
      "learning_rate": 3.353399570713096e-06,
      "loss": 0.292,
      "step": 4592
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.486769024773444,
      "learning_rate": 3.3527241270433703e-06,
      "loss": 0.354,
      "step": 4593
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.094131522010618,
      "learning_rate": 3.352048612925758e-06,
      "loss": 0.2149,
      "step": 4594
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7391771319415597,
      "learning_rate": 3.3513730284160666e-06,
      "loss": 0.3107,
      "step": 4595
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.18865148543444,
      "learning_rate": 3.35069737357011e-06,
      "loss": 0.1302,
      "step": 4596
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.997771563831143,
      "learning_rate": 3.3500216484437054e-06,
      "loss": 0.276,
      "step": 4597
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.35355811497651,
      "learning_rate": 3.3493458530926816e-06,
      "loss": 0.0904,
      "step": 4598
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.599906702408555,
      "learning_rate": 3.348669987572867e-06,
      "loss": 0.2439,
      "step": 4599
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.432027108041385,
      "learning_rate": 3.3479940519400987e-06,
      "loss": 0.1936,
      "step": 4600
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7544878878053396,
      "learning_rate": 3.347318046250219e-06,
      "loss": 0.2531,
      "step": 4601
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.3349841203639197,
      "learning_rate": 3.3466419705590763e-06,
      "loss": 0.0827,
      "step": 4602
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.8107279078482135,
      "learning_rate": 3.345965824922526e-06,
      "loss": 0.4131,
      "step": 4603
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.405651182591588,
      "learning_rate": 3.345289609396426e-06,
      "loss": 0.5465,
      "step": 4604
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.568254832322562,
      "learning_rate": 3.344613324036643e-06,
      "loss": 0.1178,
      "step": 4605
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.654028346712977,
      "learning_rate": 3.3439369688990477e-06,
      "loss": 0.3265,
      "step": 4606
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.540122752644783,
      "learning_rate": 3.3432605440395173e-06,
      "loss": 0.4255,
      "step": 4607
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.210199294129901,
      "learning_rate": 3.342584049513936e-06,
      "loss": 0.4717,
      "step": 4608
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.6128081448795095,
      "learning_rate": 3.3419074853781914e-06,
      "loss": 0.3092,
      "step": 4609
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.66617234259947,
      "learning_rate": 3.3412308516881775e-06,
      "loss": 0.2895,
      "step": 4610
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.449480498877227,
      "learning_rate": 3.340554148499795e-06,
      "loss": 0.248,
      "step": 4611
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.849639472614852,
      "learning_rate": 3.3398773758689497e-06,
      "loss": 0.5469,
      "step": 4612
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.047692188204782,
      "learning_rate": 3.339200533851554e-06,
      "loss": 0.2583,
      "step": 4613
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.112794605039049,
      "learning_rate": 3.338523622503524e-06,
      "loss": 0.2604,
      "step": 4614
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.596785309684564,
      "learning_rate": 3.3378466418807835e-06,
      "loss": 0.2156,
      "step": 4615
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.590228265720066,
      "learning_rate": 3.337169592039261e-06,
      "loss": 0.3161,
      "step": 4616
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5945985001606358,
      "learning_rate": 3.3364924730348913e-06,
      "loss": 0.026,
      "step": 4617
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.822425669533337,
      "learning_rate": 3.3358152849236147e-06,
      "loss": 0.3148,
      "step": 4618
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.298718133145983,
      "learning_rate": 3.335138027761377e-06,
      "loss": 0.3718,
      "step": 4619
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.95577341187716,
      "learning_rate": 3.33446070160413e-06,
      "loss": 0.4327,
      "step": 4620
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.4925677546979217,
      "learning_rate": 3.3337833065078305e-06,
      "loss": 0.1129,
      "step": 4621
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.139241397604077,
      "learning_rate": 3.3331058425284424e-06,
      "loss": 0.2645,
      "step": 4622
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.584659719233678,
      "learning_rate": 3.3324283097219344e-06,
      "loss": 0.1418,
      "step": 4623
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.661334024847383,
      "learning_rate": 3.33175070814428e-06,
      "loss": 0.3439,
      "step": 4624
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.751701425280382,
      "learning_rate": 3.3310730378514605e-06,
      "loss": 0.4711,
      "step": 4625
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.487028418678971,
      "learning_rate": 3.330395298899461e-06,
      "loss": 0.2556,
      "step": 4626
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.1148682737489795,
      "learning_rate": 3.3297174913442724e-06,
      "loss": 0.3268,
      "step": 4627
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.8987018277382472,
      "learning_rate": 3.3290396152418936e-06,
      "loss": 0.3143,
      "step": 4628
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.987865380727173,
      "learning_rate": 3.328361670648326e-06,
      "loss": 0.4328,
      "step": 4629
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.324350453253901,
      "learning_rate": 3.327683657619578e-06,
      "loss": 0.3685,
      "step": 4630
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.627870427152122,
      "learning_rate": 3.327005576211665e-06,
      "loss": 0.3376,
      "step": 4631
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.623466135018657,
      "learning_rate": 3.326327426480605e-06,
      "loss": 0.1958,
      "step": 4632
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.992634040859221,
      "learning_rate": 3.325649208482425e-06,
      "loss": 0.2501,
      "step": 4633
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.591305533947781,
      "learning_rate": 3.324970922273155e-06,
      "loss": 0.4146,
      "step": 4634
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.114841353845202,
      "learning_rate": 3.3242925679088315e-06,
      "loss": 0.327,
      "step": 4635
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.785054478360327,
      "learning_rate": 3.3236141454454985e-06,
      "loss": 0.2047,
      "step": 4636
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7453984416205124,
      "learning_rate": 3.322935654939201e-06,
      "loss": 0.3356,
      "step": 4637
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.6250974282315926,
      "learning_rate": 3.322257096445996e-06,
      "loss": 0.288,
      "step": 4638
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.196326396213644,
      "learning_rate": 3.3215784700219404e-06,
      "loss": 0.5497,
      "step": 4639
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2246760944601,
      "learning_rate": 3.3208997757230997e-06,
      "loss": 0.0133,
      "step": 4640
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.496292944504812,
      "learning_rate": 3.320221013605544e-06,
      "loss": 0.2712,
      "step": 4641
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.9200280894671486,
      "learning_rate": 3.3195421837253494e-06,
      "loss": 0.4026,
      "step": 4642
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.104005736304836,
      "learning_rate": 3.3188632861385984e-06,
      "loss": 0.1538,
      "step": 4643
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.4841301661639603,
      "learning_rate": 3.3181843209013758e-06,
      "loss": 0.2341,
      "step": 4644
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.505787249667623,
      "learning_rate": 3.3175052880697774e-06,
      "loss": 0.2989,
      "step": 4645
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.706929452442551,
      "learning_rate": 3.3168261876999e-06,
      "loss": 0.3775,
      "step": 4646
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.216422798698861,
      "learning_rate": 3.3161470198478464e-06,
      "loss": 0.4495,
      "step": 4647
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.6154754311722455,
      "learning_rate": 3.3154677845697286e-06,
      "loss": 0.2295,
      "step": 4648
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.957409332160341,
      "learning_rate": 3.3147884819216596e-06,
      "loss": 0.3635,
      "step": 4649
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.53833178273209,
      "learning_rate": 3.314109111959762e-06,
      "loss": 0.5345,
      "step": 4650
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.8001415806707763,
      "learning_rate": 3.31342967474016e-06,
      "loss": 0.2276,
      "step": 4651
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.3523379260774746,
      "learning_rate": 3.312750170318987e-06,
      "loss": 0.1939,
      "step": 4652
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.079134548334216,
      "learning_rate": 3.3120705987523786e-06,
      "loss": 0.1648,
      "step": 4653
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.649869208828925,
      "learning_rate": 3.3113909600964796e-06,
      "loss": 0.3137,
      "step": 4654
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.787606025412062,
      "learning_rate": 3.310711254407437e-06,
      "loss": 0.4619,
      "step": 4655
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.2475410165681575,
      "learning_rate": 3.310031481741406e-06,
      "loss": 0.2568,
      "step": 4656
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.428045437623484,
      "learning_rate": 3.3093516421545447e-06,
      "loss": 0.2687,
      "step": 4657
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.441736223144872,
      "learning_rate": 3.308671735703019e-06,
      "loss": 0.4919,
      "step": 4658
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.556581638359455,
      "learning_rate": 3.307991762442999e-06,
      "loss": 0.3733,
      "step": 4659
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.019212719100541,
      "learning_rate": 3.3073117224306612e-06,
      "loss": 0.3554,
      "step": 4660
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.545344495283543,
      "learning_rate": 3.306631615722187e-06,
      "loss": 0.359,
      "step": 4661
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.579992405213842,
      "learning_rate": 3.3059514423737626e-06,
      "loss": 0.4296,
      "step": 4662
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.474148713728616,
      "learning_rate": 3.3052712024415823e-06,
      "loss": 0.484,
      "step": 4663
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.49190621638783,
      "learning_rate": 3.304590895981843e-06,
      "loss": 0.2563,
      "step": 4664
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2056234190642834,
      "learning_rate": 3.3039105230507483e-06,
      "loss": 0.1929,
      "step": 4665
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.269866393270177,
      "learning_rate": 3.303230083704508e-06,
      "loss": 0.4673,
      "step": 4666
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.3325076942805016,
      "learning_rate": 3.302549577999336e-06,
      "loss": 0.2819,
      "step": 4667
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.631146466805592,
      "learning_rate": 3.3018690059914527e-06,
      "loss": 0.3725,
      "step": 4668
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.107163959884205,
      "learning_rate": 3.301188367737083e-06,
      "loss": 0.5641,
      "step": 4669
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.430250836357703,
      "learning_rate": 3.3005076632924592e-06,
      "loss": 0.2094,
      "step": 4670
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4931665594491903,
      "learning_rate": 3.2998268927138166e-06,
      "loss": 0.0098,
      "step": 4671
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.079860248263399,
      "learning_rate": 3.299146056057397e-06,
      "loss": 0.3961,
      "step": 4672
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.2292329862597837,
      "learning_rate": 3.2984651533794487e-06,
      "loss": 0.2658,
      "step": 4673
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.703031316519472,
      "learning_rate": 3.297784184736224e-06,
      "loss": 0.1935,
      "step": 4674
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.7241069148292985,
      "learning_rate": 3.297103150183981e-06,
      "loss": 0.2773,
      "step": 4675
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.195335060538634,
      "learning_rate": 3.2964220497789846e-06,
      "loss": 0.4437,
      "step": 4676
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.0935789011119805,
      "learning_rate": 3.2957408835775022e-06,
      "loss": 0.2977,
      "step": 4677
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.48235197378499,
      "learning_rate": 3.295059651635809e-06,
      "loss": 0.4322,
      "step": 4678
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.166639849673011,
      "learning_rate": 3.2943783540101866e-06,
      "loss": 0.2902,
      "step": 4679
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.9283937806319855,
      "learning_rate": 3.293696990756918e-06,
      "loss": 0.5696,
      "step": 4680
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.5182062359214257,
      "learning_rate": 3.2930155619322966e-06,
      "loss": 0.1339,
      "step": 4681
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0554022022744776,
      "learning_rate": 3.2923340675926163e-06,
      "loss": 0.2202,
      "step": 4682
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.684381448905774,
      "learning_rate": 3.29165250779418e-06,
      "loss": 0.0301,
      "step": 4683
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.487367784248532,
      "learning_rate": 3.2909708825932957e-06,
      "loss": 0.2157,
      "step": 4684
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4639188844112128,
      "learning_rate": 3.290289192046274e-06,
      "loss": 0.3028,
      "step": 4685
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.266803309376317,
      "learning_rate": 3.289607436209435e-06,
      "loss": 0.3468,
      "step": 4686
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.199601817188491,
      "learning_rate": 3.288925615139099e-06,
      "loss": 0.3162,
      "step": 4687
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.1826390951908956,
      "learning_rate": 3.2882437288915982e-06,
      "loss": 0.1815,
      "step": 4688
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.623391281672761,
      "learning_rate": 3.287561777523264e-06,
      "loss": 0.4593,
      "step": 4689
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.648036970296752,
      "learning_rate": 3.286879761090438e-06,
      "loss": 0.3348,
      "step": 4690
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.320149151336515,
      "learning_rate": 3.286197679649464e-06,
      "loss": 0.493,
      "step": 4691
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.7626827414116963,
      "learning_rate": 3.285515533256691e-06,
      "loss": 0.2121,
      "step": 4692
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.0154628584071395,
      "learning_rate": 3.2848333219684762e-06,
      "loss": 0.2302,
      "step": 4693
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.160003587206618,
      "learning_rate": 3.284151045841181e-06,
      "loss": 0.2494,
      "step": 4694
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.400125985012193,
      "learning_rate": 3.283468704931171e-06,
      "loss": 0.3808,
      "step": 4695
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.4448782850511996,
      "learning_rate": 3.2827862992948167e-06,
      "loss": 0.1301,
      "step": 4696
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.309710837272856,
      "learning_rate": 3.282103828988496e-06,
      "loss": 0.5099,
      "step": 4697
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.3710538869345115,
      "learning_rate": 3.281421294068592e-06,
      "loss": 0.3903,
      "step": 4698
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.986608254689501,
      "learning_rate": 3.280738694591492e-06,
      "loss": 0.3671,
      "step": 4699
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.9783442919799143,
      "learning_rate": 3.2800560306135887e-06,
      "loss": 0.1331,
      "step": 4700
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4495973185578854,
      "learning_rate": 3.2793733021912798e-06,
      "loss": 0.0038,
      "step": 4701
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5530669294255297,
      "learning_rate": 3.2786905093809707e-06,
      "loss": 0.318,
      "step": 4702
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.6462441016486897,
      "learning_rate": 3.278007652239069e-06,
      "loss": 0.307,
      "step": 4703
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.1972207429186135,
      "learning_rate": 3.2773247308219897e-06,
      "loss": 0.2774,
      "step": 4704
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.380474568961962,
      "learning_rate": 3.2766417451861522e-06,
      "loss": 0.3044,
      "step": 4705
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5916604803899137,
      "learning_rate": 3.2759586953879807e-06,
      "loss": 0.2705,
      "step": 4706
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.922849434920132,
      "learning_rate": 3.2752755814839066e-06,
      "loss": 0.196,
      "step": 4707
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9356047166725765,
      "learning_rate": 3.274592403530365e-06,
      "loss": 0.2301,
      "step": 4708
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.5627397317761,
      "learning_rate": 3.2739091615837965e-06,
      "loss": 0.5373,
      "step": 4709
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4006361237040994,
      "learning_rate": 3.273225855700648e-06,
      "loss": 0.2224,
      "step": 4710
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.493900788151862,
      "learning_rate": 3.272542485937369e-06,
      "loss": 0.312,
      "step": 4711
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.068989766293849,
      "learning_rate": 3.271859052350418e-06,
      "loss": 0.199,
      "step": 4712
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.126554539732517,
      "learning_rate": 3.2711755549962558e-06,
      "loss": 0.3382,
      "step": 4713
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.968641348887342,
      "learning_rate": 3.270491993931351e-06,
      "loss": 0.5632,
      "step": 4714
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.097368069499608,
      "learning_rate": 3.269808369212175e-06,
      "loss": 0.3619,
      "step": 4715
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.484198307022185,
      "learning_rate": 3.2691246808952053e-06,
      "loss": 0.5088,
      "step": 4716
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.1135905466166505,
      "learning_rate": 3.268440929036925e-06,
      "loss": 0.1861,
      "step": 4717
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.8573875934728648,
      "learning_rate": 3.2677571136938234e-06,
      "loss": 0.304,
      "step": 4718
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.6513994845588975,
      "learning_rate": 3.267073234922393e-06,
      "loss": 0.304,
      "step": 4719
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.940395748882283,
      "learning_rate": 3.266389292779133e-06,
      "loss": 0.2625,
      "step": 4720
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.710221701836649,
      "learning_rate": 3.2657052873205462e-06,
      "loss": 0.3288,
      "step": 4721
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.586266122228208,
      "learning_rate": 3.2650212186031427e-06,
      "loss": 0.4696,
      "step": 4722
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.819952819435908,
      "learning_rate": 3.2643370866834373e-06,
      "loss": 0.4078,
      "step": 4723
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.608102417167921,
      "learning_rate": 3.2636528916179487e-06,
      "loss": 0.3876,
      "step": 4724
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.2883425297642646,
      "learning_rate": 3.262968633463203e-06,
      "loss": 0.1974,
      "step": 4725
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.525563098117967,
      "learning_rate": 3.2622843122757286e-06,
      "loss": 0.232,
      "step": 4726
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.5019898057228795,
      "learning_rate": 3.261599928112061e-06,
      "loss": 0.1976,
      "step": 4727
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.057422410242129,
      "learning_rate": 3.260915481028743e-06,
      "loss": 0.1039,
      "step": 4728
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.262342430155024,
      "learning_rate": 3.2602309710823176e-06,
      "loss": 0.2112,
      "step": 4729
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.180612055714284,
      "learning_rate": 3.2595463983293364e-06,
      "loss": 0.3171,
      "step": 4730
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.8559995152244806,
      "learning_rate": 3.2588617628263563e-06,
      "loss": 0.1727,
      "step": 4731
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.1476261301039825,
      "learning_rate": 3.258177064629937e-06,
      "loss": 0.2206,
      "step": 4732
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.313311630949394,
      "learning_rate": 3.2574923037966466e-06,
      "loss": 0.1856,
      "step": 4733
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.428382265982045,
      "learning_rate": 3.2568074803830553e-06,
      "loss": 0.3089,
      "step": 4734
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.23760077602546,
      "learning_rate": 3.25612259444574e-06,
      "loss": 0.309,
      "step": 4735
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.731800480020281,
      "learning_rate": 3.255437646041284e-06,
      "loss": 0.2736,
      "step": 4736
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.531725338390099,
      "learning_rate": 3.2547526352262725e-06,
      "loss": 0.2407,
      "step": 4737
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.1336003694790953,
      "learning_rate": 3.2540675620572994e-06,
      "loss": 0.1676,
      "step": 4738
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.103943089971184,
      "learning_rate": 3.2533824265909614e-06,
      "loss": 0.0547,
      "step": 4739
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.00209303829483,
      "learning_rate": 3.2526972288838597e-06,
      "loss": 0.1606,
      "step": 4740
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.906800490067263,
      "learning_rate": 3.252011968992604e-06,
      "loss": 0.3535,
      "step": 4741
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.454970293375335,
      "learning_rate": 3.251326646973807e-06,
      "loss": 0.2765,
      "step": 4742
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.243107553867994,
      "learning_rate": 3.2506412628840854e-06,
      "loss": 0.1252,
      "step": 4743
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.978744012825483,
      "learning_rate": 3.2499558167800627e-06,
      "loss": 0.1914,
      "step": 4744
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.859651208655054,
      "learning_rate": 3.2492703087183676e-06,
      "loss": 0.363,
      "step": 4745
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.01006744637955,
      "learning_rate": 3.2485847387556326e-06,
      "loss": 0.362,
      "step": 4746
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.267239541762313,
      "learning_rate": 3.2478991069484967e-06,
      "loss": 0.4819,
      "step": 4747
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.4058899288888185,
      "learning_rate": 3.247213413353604e-06,
      "loss": 0.4237,
      "step": 4748
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.129340041132839,
      "learning_rate": 3.2465276580276013e-06,
      "loss": 0.263,
      "step": 4749
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.994604033307342,
      "learning_rate": 3.245841841027144e-06,
      "loss": 0.6261,
      "step": 4750
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.240879025807172,
      "learning_rate": 3.245155962408891e-06,
      "loss": 0.209,
      "step": 4751
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.218478005721386,
      "learning_rate": 3.2444700222295056e-06,
      "loss": 0.3676,
      "step": 4752
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.347606094292481,
      "learning_rate": 3.2437840205456567e-06,
      "loss": 0.1695,
      "step": 4753
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4433401725030044,
      "learning_rate": 3.2430979574140183e-06,
      "loss": 0.2644,
      "step": 4754
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.348416231098496,
      "learning_rate": 3.2424118328912703e-06,
      "loss": 0.3042,
      "step": 4755
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.130701188117499,
      "learning_rate": 3.2417256470340962e-06,
      "loss": 0.3362,
      "step": 4756
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.0630102109377235,
      "learning_rate": 3.241039399899186e-06,
      "loss": 0.1363,
      "step": 4757
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.491757126647936,
      "learning_rate": 3.2403530915432337e-06,
      "loss": 0.1189,
      "step": 4758
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.887107146265746,
      "learning_rate": 3.239666722022938e-06,
      "loss": 0.3729,
      "step": 4759
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.487580381673162,
      "learning_rate": 3.2389802913950045e-06,
      "loss": 0.1206,
      "step": 4760
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.662295735727135,
      "learning_rate": 3.2382937997161427e-06,
      "loss": 0.286,
      "step": 4761
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.265526382877191,
      "learning_rate": 3.2376072470430664e-06,
      "loss": 0.4612,
      "step": 4762
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.133797740055912,
      "learning_rate": 3.2369206334324955e-06,
      "loss": 0.4298,
      "step": 4763
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.770944662913861,
      "learning_rate": 3.236233958941155e-06,
      "loss": 0.4381,
      "step": 4764
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.252866795092926,
      "learning_rate": 3.2355472236257745e-06,
      "loss": 0.2223,
      "step": 4765
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.562307975744711,
      "learning_rate": 3.2348604275430894e-06,
      "loss": 0.4507,
      "step": 4766
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.8728396375205287,
      "learning_rate": 3.234173570749837e-06,
      "loss": 0.3518,
      "step": 4767
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.6027064951691665,
      "learning_rate": 3.233486653302764e-06,
      "loss": 0.2272,
      "step": 4768
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.439931614516756,
      "learning_rate": 3.232799675258621e-06,
      "loss": 0.3608,
      "step": 4769
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.869020281090106,
      "learning_rate": 3.2321126366741608e-06,
      "loss": 0.3504,
      "step": 4770
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.811791833184031,
      "learning_rate": 3.2314255376061443e-06,
      "loss": 0.4692,
      "step": 4771
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5372095881336922,
      "learning_rate": 3.2307383781113348e-06,
      "loss": 0.3095,
      "step": 4772
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.649454735197049,
      "learning_rate": 3.2300511582465034e-06,
      "loss": 0.1475,
      "step": 4773
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.189919762348924,
      "learning_rate": 3.2293638780684254e-06,
      "loss": 0.236,
      "step": 4774
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.743925114408006,
      "learning_rate": 3.2286765376338792e-06,
      "loss": 0.2292,
      "step": 4775
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.983392366196884,
      "learning_rate": 3.2279891369996503e-06,
      "loss": 0.4116,
      "step": 4776
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.057805345970182,
      "learning_rate": 3.227301676222527e-06,
      "loss": 0.2268,
      "step": 4777
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.505336521471886,
      "learning_rate": 3.2266141553593054e-06,
      "loss": 0.4805,
      "step": 4778
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.2272834281367455,
      "learning_rate": 3.2259265744667854e-06,
      "loss": 0.2748,
      "step": 4779
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.969033216430592,
      "learning_rate": 3.225238933601771e-06,
      "loss": 0.1304,
      "step": 4780
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.74087849849601,
      "learning_rate": 3.224551232821071e-06,
      "loss": 0.3073,
      "step": 4781
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.511206085455183,
      "learning_rate": 3.2238634721815e-06,
      "loss": 0.0871,
      "step": 4782
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.9989264639308875,
      "learning_rate": 3.2231756517398775e-06,
      "loss": 0.3228,
      "step": 4783
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.031986051042493,
      "learning_rate": 3.2224877715530294e-06,
      "loss": 0.2591,
      "step": 4784
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.9651776574933,
      "learning_rate": 3.221799831677783e-06,
      "loss": 0.0709,
      "step": 4785
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.330574105784113,
      "learning_rate": 3.2211118321709746e-06,
      "loss": 0.1508,
      "step": 4786
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.9625447439714403,
      "learning_rate": 3.22042377308944e-06,
      "loss": 0.14,
      "step": 4787
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.5714304448861656,
      "learning_rate": 3.2197356544900266e-06,
      "loss": 0.2209,
      "step": 4788
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.672681053574256,
      "learning_rate": 3.219047476429582e-06,
      "loss": 0.6178,
      "step": 4789
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.858979043185417,
      "learning_rate": 3.2183592389649604e-06,
      "loss": 0.4055,
      "step": 4790
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.196146570595499,
      "learning_rate": 3.21767094215302e-06,
      "loss": 0.2568,
      "step": 4791
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.677858173963849,
      "learning_rate": 3.216982586050624e-06,
      "loss": 0.336,
      "step": 4792
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.4935497150911217,
      "learning_rate": 3.2162941707146432e-06,
      "loss": 0.1854,
      "step": 4793
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.094792883730139,
      "learning_rate": 3.21560569620195e-06,
      "loss": 0.3436,
      "step": 4794
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.0162008899087205,
      "learning_rate": 3.214917162569422e-06,
      "loss": 0.2237,
      "step": 4795
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.420520507529095,
      "learning_rate": 3.214228569873944e-06,
      "loss": 0.2883,
      "step": 4796
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.410645269947167,
      "learning_rate": 3.2135399181724013e-06,
      "loss": 0.4672,
      "step": 4797
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8177937659073615,
      "learning_rate": 3.21285120752169e-06,
      "loss": 0.1532,
      "step": 4798
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.632028873460921,
      "learning_rate": 3.212162437978707e-06,
      "loss": 0.3303,
      "step": 4799
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.2493572185698065,
      "learning_rate": 3.2114736096003553e-06,
      "loss": 0.0616,
      "step": 4800
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3212402037825197,
      "learning_rate": 3.2107847224435413e-06,
      "loss": 0.2356,
      "step": 4801
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.364660341503804,
      "learning_rate": 3.210095776565178e-06,
      "loss": 0.455,
      "step": 4802
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.261572236915278,
      "learning_rate": 3.209406772022184e-06,
      "loss": 0.1609,
      "step": 4803
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.356912330478083,
      "learning_rate": 3.2087177088714806e-06,
      "loss": 0.1902,
      "step": 4804
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.6706419648933752,
      "learning_rate": 3.2080285871699945e-06,
      "loss": 0.1575,
      "step": 4805
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.160458545462576,
      "learning_rate": 3.2073394069746573e-06,
      "loss": 0.5307,
      "step": 4806
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.229642143225473,
      "learning_rate": 3.206650168342406e-06,
      "loss": 0.3037,
      "step": 4807
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1296806587703747,
      "learning_rate": 3.205960871330183e-06,
      "loss": 0.1262,
      "step": 4808
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.226596535412189,
      "learning_rate": 3.2052715159949344e-06,
      "loss": 0.1787,
      "step": 4809
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.1826264256526615,
      "learning_rate": 3.20458210239361e-06,
      "loss": 0.4782,
      "step": 4810
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.724727824481679,
      "learning_rate": 3.203892630583167e-06,
      "loss": 0.2562,
      "step": 4811
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.033233608830763,
      "learning_rate": 3.2032031006205655e-06,
      "loss": 0.2978,
      "step": 4812
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.2056214759542256,
      "learning_rate": 3.2025135125627722e-06,
      "loss": 0.1499,
      "step": 4813
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.310898385280305,
      "learning_rate": 3.2018238664667566e-06,
      "loss": 0.228,
      "step": 4814
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.2104740386829227,
      "learning_rate": 3.201134162389494e-06,
      "loss": 0.1566,
      "step": 4815
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.838843752766329,
      "learning_rate": 3.2004444003879635e-06,
      "loss": 0.2951,
      "step": 4816
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.258881611200773,
      "learning_rate": 3.199754580519151e-06,
      "loss": 0.239,
      "step": 4817
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.115235647973391,
      "learning_rate": 3.1990647028400467e-06,
      "loss": 0.3834,
      "step": 4818
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.62088897806271,
      "learning_rate": 3.1983747674076437e-06,
      "loss": 0.2988,
      "step": 4819
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.412848122781393,
      "learning_rate": 3.1976847742789406e-06,
      "loss": 0.4908,
      "step": 4820
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.922429239038783,
      "learning_rate": 3.1969947235109426e-06,
      "loss": 0.0943,
      "step": 4821
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.45336984404086,
      "learning_rate": 3.1963046151606573e-06,
      "loss": 0.4507,
      "step": 4822
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.341394459955303,
      "learning_rate": 3.195614449285099e-06,
      "loss": 0.3362,
      "step": 4823
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.043126242624003,
      "learning_rate": 3.1949242259412855e-06,
      "loss": 0.4144,
      "step": 4824
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.760392695268398,
      "learning_rate": 3.1942339451862385e-06,
      "loss": 0.4598,
      "step": 4825
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.946441670922658,
      "learning_rate": 3.1935436070769876e-06,
      "loss": 0.4868,
      "step": 4826
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.791453157727164,
      "learning_rate": 3.1928532116705635e-06,
      "loss": 0.4768,
      "step": 4827
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8215956847850405,
      "learning_rate": 3.1921627590240047e-06,
      "loss": 0.2498,
      "step": 4828
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.306368714881234,
      "learning_rate": 3.1914722491943516e-06,
      "loss": 0.4468,
      "step": 4829
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.136323355282104,
      "learning_rate": 3.190781682238651e-06,
      "loss": 0.3576,
      "step": 4830
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.015975279241194,
      "learning_rate": 3.1900910582139556e-06,
      "loss": 0.356,
      "step": 4831
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.585770561763247,
      "learning_rate": 3.189400377177319e-06,
      "loss": 0.4343,
      "step": 4832
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.126783765894817,
      "learning_rate": 3.1887096391858043e-06,
      "loss": 0.3721,
      "step": 4833
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.020739571899408,
      "learning_rate": 3.1880188442964755e-06,
      "loss": 0.1583,
      "step": 4834
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.161147304634649,
      "learning_rate": 3.1873279925664023e-06,
      "loss": 0.3622,
      "step": 4835
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.24697999129568,
      "learning_rate": 3.1866370840526613e-06,
      "loss": 0.5402,
      "step": 4836
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.961453976643534,
      "learning_rate": 3.1859461188123297e-06,
      "loss": 0.275,
      "step": 4837
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.4359489510998835,
      "learning_rate": 3.1852550969024937e-06,
      "loss": 0.213,
      "step": 4838
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8985791119885307,
      "learning_rate": 3.184564018380241e-06,
      "loss": 0.354,
      "step": 4839
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.955863534385494,
      "learning_rate": 3.1838728833026645e-06,
      "loss": 0.3661,
      "step": 4840
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.810213719097269,
      "learning_rate": 3.1831816917268644e-06,
      "loss": 0.3862,
      "step": 4841
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.187525726550299,
      "learning_rate": 3.1824904437099413e-06,
      "loss": 0.4617,
      "step": 4842
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.651290450643386,
      "learning_rate": 3.181799139309005e-06,
      "loss": 0.2881,
      "step": 4843
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.371930116326899,
      "learning_rate": 3.181107778581165e-06,
      "loss": 0.2534,
      "step": 4844
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.564102281041971,
      "learning_rate": 3.180416361583541e-06,
      "loss": 0.3323,
      "step": 4845
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.7040359884426275,
      "learning_rate": 3.179724888373253e-06,
      "loss": 0.5009,
      "step": 4846
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.566575097818915,
      "learning_rate": 3.1790333590074263e-06,
      "loss": 0.482,
      "step": 4847
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.704184346910678,
      "learning_rate": 3.1783417735431933e-06,
      "loss": 0.3564,
      "step": 4848
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.07088866490669,
      "learning_rate": 3.1776501320376883e-06,
      "loss": 0.1277,
      "step": 4849
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.700579966807758,
      "learning_rate": 3.176958434548052e-06,
      "loss": 0.2571,
      "step": 4850
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.4255961249776656,
      "learning_rate": 3.176266681131429e-06,
      "loss": 0.2155,
      "step": 4851
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.083385086631037,
      "learning_rate": 3.1755748718449676e-06,
      "loss": 0.4749,
      "step": 4852
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.3177186062090644,
      "learning_rate": 3.174883006745824e-06,
      "loss": 0.2422,
      "step": 4853
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.5368322432026633,
      "learning_rate": 3.174191085891153e-06,
      "loss": 0.2965,
      "step": 4854
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.946251759644638,
      "learning_rate": 3.1734991093381214e-06,
      "loss": 0.3971,
      "step": 4855
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.803513283425842,
      "learning_rate": 3.1728070771438953e-06,
      "loss": 0.2886,
      "step": 4856
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8968888177957353,
      "learning_rate": 3.172114989365646e-06,
      "loss": 0.2897,
      "step": 4857
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.1398446668204016,
      "learning_rate": 3.171422846060552e-06,
      "loss": 0.2159,
      "step": 4858
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8426164203203528,
      "learning_rate": 3.1707306472857947e-06,
      "loss": 0.1542,
      "step": 4859
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.561249256539306,
      "learning_rate": 3.1700383930985603e-06,
      "loss": 0.4061,
      "step": 4860
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.18307577104287,
      "learning_rate": 3.169346083556038e-06,
      "loss": 0.4691,
      "step": 4861
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.41272666835684,
      "learning_rate": 3.168653718715423e-06,
      "loss": 0.3488,
      "step": 4862
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.820589401129377,
      "learning_rate": 3.167961298633917e-06,
      "loss": 0.3269,
      "step": 4863
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.110743527063059,
      "learning_rate": 3.167268823368724e-06,
      "loss": 0.3876,
      "step": 4864
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.247283624901222,
      "learning_rate": 3.1665762929770515e-06,
      "loss": 0.3263,
      "step": 4865
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.227807950498134,
      "learning_rate": 3.165883707516114e-06,
      "loss": 0.4916,
      "step": 4866
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.235896717602455,
      "learning_rate": 3.1651910670431285e-06,
      "loss": 0.4412,
      "step": 4867
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.0240122413158135,
      "learning_rate": 3.1644983716153187e-06,
      "loss": 0.5167,
      "step": 4868
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.031425822301446,
      "learning_rate": 3.1638056212899115e-06,
      "loss": 0.2868,
      "step": 4869
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.80309427864316,
      "learning_rate": 3.1631128161241388e-06,
      "loss": 0.3917,
      "step": 4870
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.830663344681829,
      "learning_rate": 3.1624199561752363e-06,
      "loss": 0.5009,
      "step": 4871
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.465021031552495,
      "learning_rate": 3.1617270415004435e-06,
      "loss": 0.2088,
      "step": 4872
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.796060331716556,
      "learning_rate": 3.161034072157007e-06,
      "loss": 0.1214,
      "step": 4873
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.439327513398572,
      "learning_rate": 3.1603410482021773e-06,
      "loss": 0.3817,
      "step": 4874
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9422362563447257,
      "learning_rate": 3.1596479696932076e-06,
      "loss": 0.3476,
      "step": 4875
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.4260586924780725,
      "learning_rate": 3.1589548366873565e-06,
      "loss": 0.1186,
      "step": 4876
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.314188883231265,
      "learning_rate": 3.1582616492418868e-06,
      "loss": 0.3725,
      "step": 4877
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.118498977953253,
      "learning_rate": 3.1575684074140665e-06,
      "loss": 0.4801,
      "step": 4878
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.3083232662721134,
      "learning_rate": 3.15687511126117e-06,
      "loss": 0.2316,
      "step": 4879
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.734100139724577,
      "learning_rate": 3.1561817608404717e-06,
      "loss": 0.3049,
      "step": 4880
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.462274750813631,
      "learning_rate": 3.155488356209253e-06,
      "loss": 0.3178,
      "step": 4881
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.457629829409487,
      "learning_rate": 3.1547948974248e-06,
      "loss": 0.3801,
      "step": 4882
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8113402540703922,
      "learning_rate": 3.154101384544403e-06,
      "loss": 0.3624,
      "step": 4883
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8927096456140444,
      "learning_rate": 3.153407817625357e-06,
      "loss": 0.3513,
      "step": 4884
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2398455239830817,
      "learning_rate": 3.1527141967249596e-06,
      "loss": 0.1044,
      "step": 4885
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.831894627912729,
      "learning_rate": 3.152020521900516e-06,
      "loss": 0.3296,
      "step": 4886
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.9792198221773294,
      "learning_rate": 3.1513267932093327e-06,
      "loss": 0.1483,
      "step": 4887
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.538441928868677,
      "learning_rate": 3.150633010708724e-06,
      "loss": 0.3459,
      "step": 4888
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.8536388119986014,
      "learning_rate": 3.149939174456006e-06,
      "loss": 0.3091,
      "step": 4889
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.8769888472018863,
      "learning_rate": 3.1492452845085e-06,
      "loss": 0.0859,
      "step": 4890
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.704168703139494,
      "learning_rate": 3.1485513409235314e-06,
      "loss": 0.3572,
      "step": 4891
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.774075720263906,
      "learning_rate": 3.1478573437584294e-06,
      "loss": 0.7259,
      "step": 4892
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.090087807667036,
      "learning_rate": 3.147163293070531e-06,
      "loss": 0.2282,
      "step": 4893
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.950235844804193,
      "learning_rate": 3.1464691889171744e-06,
      "loss": 0.4286,
      "step": 4894
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.432747795958101,
      "learning_rate": 3.1457750313557028e-06,
      "loss": 0.4163,
      "step": 4895
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.51917312633719,
      "learning_rate": 3.1450808204434646e-06,
      "loss": 0.3484,
      "step": 4896
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.621701696878572,
      "learning_rate": 3.1443865562378103e-06,
      "loss": 0.178,
      "step": 4897
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.430103041043835,
      "learning_rate": 3.143692238796099e-06,
      "loss": 0.38,
      "step": 4898
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.4349410438481724,
      "learning_rate": 3.142997868175691e-06,
      "loss": 0.3082,
      "step": 4899
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.0233676557585487,
      "learning_rate": 3.142303444433951e-06,
      "loss": 0.1534,
      "step": 4900
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7274290007130364,
      "learning_rate": 3.141608967628249e-06,
      "loss": 0.0609,
      "step": 4901
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.928322107788959,
      "learning_rate": 3.14091443781596e-06,
      "loss": 0.3767,
      "step": 4902
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.57115128135993,
      "learning_rate": 3.140219855054463e-06,
      "loss": 0.1522,
      "step": 4903
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.485023253837687,
      "learning_rate": 3.1395252194011407e-06,
      "loss": 0.5535,
      "step": 4904
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.869377297678959,
      "learning_rate": 3.1388305309133793e-06,
      "loss": 0.4002,
      "step": 4905
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.192096608416151,
      "learning_rate": 3.138135789648572e-06,
      "loss": 0.281,
      "step": 4906
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.099283968641256,
      "learning_rate": 3.137440995664114e-06,
      "loss": 0.5579,
      "step": 4907
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.5552036982318835,
      "learning_rate": 3.136746149017406e-06,
      "loss": 0.4763,
      "step": 4908
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.5443707736060466,
      "learning_rate": 3.1360512497658533e-06,
      "loss": 0.1975,
      "step": 4909
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.790856373704132,
      "learning_rate": 3.135356297966865e-06,
      "loss": 0.1275,
      "step": 4910
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.4174279365940645,
      "learning_rate": 3.134661293677854e-06,
      "loss": 0.4698,
      "step": 4911
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.5161205903766835,
      "learning_rate": 3.1339662369562387e-06,
      "loss": 0.3126,
      "step": 4912
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.2719821420638717,
      "learning_rate": 3.133271127859441e-06,
      "loss": 0.2113,
      "step": 4913
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9883698591658194,
      "learning_rate": 3.1325759664448873e-06,
      "loss": 0.3982,
      "step": 4914
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.4306908955738695,
      "learning_rate": 3.131880752770009e-06,
      "loss": 0.3491,
      "step": 4915
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.5282157503790694,
      "learning_rate": 3.13118548689224e-06,
      "loss": 0.1908,
      "step": 4916
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.7776344923764227,
      "learning_rate": 3.130490168869021e-06,
      "loss": 0.2592,
      "step": 4917
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6943418085513926,
      "learning_rate": 3.129794798757796e-06,
      "loss": 0.2886,
      "step": 4918
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4323683231767124,
      "learning_rate": 3.1290993766160116e-06,
      "loss": 0.2193,
      "step": 4919
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.396484397029752,
      "learning_rate": 3.1284039025011216e-06,
      "loss": 0.2909,
      "step": 4920
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.809480035492748,
      "learning_rate": 3.127708376470582e-06,
      "loss": 0.1864,
      "step": 4921
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.829514227560115,
      "learning_rate": 3.127012798581853e-06,
      "loss": 0.4564,
      "step": 4922
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.4214729835558915,
      "learning_rate": 3.1263171688924014e-06,
      "loss": 0.2992,
      "step": 4923
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.391302252503503,
      "learning_rate": 3.125621487459696e-06,
      "loss": 0.5373,
      "step": 4924
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.74019458296164,
      "learning_rate": 3.1249257543412102e-06,
      "loss": 0.1829,
      "step": 4925
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.490029005702957,
      "learning_rate": 3.1242299695944234e-06,
      "loss": 0.3056,
      "step": 4926
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.2030299391766786,
      "learning_rate": 3.1235341332768153e-06,
      "loss": 0.288,
      "step": 4927
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2057760843654735,
      "learning_rate": 3.1228382454458757e-06,
      "loss": 0.1149,
      "step": 4928
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.045823420159244,
      "learning_rate": 3.122142306159093e-06,
      "loss": 0.2211,
      "step": 4929
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4511148585928493,
      "learning_rate": 3.121446315473964e-06,
      "loss": 0.2437,
      "step": 4930
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1038924263117558,
      "learning_rate": 3.1207502734479866e-06,
      "loss": 0.2077,
      "step": 4931
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.7305554306795616,
      "learning_rate": 3.1200541801386646e-06,
      "loss": 0.3141,
      "step": 4932
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.192935205630842,
      "learning_rate": 3.119358035603507e-06,
      "loss": 0.3835,
      "step": 4933
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.768628555974112,
      "learning_rate": 3.1186618399000247e-06,
      "loss": 0.1684,
      "step": 4934
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.068238780691385,
      "learning_rate": 3.1179655930857338e-06,
      "loss": 0.4363,
      "step": 4935
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.789805949270492,
      "learning_rate": 3.117269295218156e-06,
      "loss": 0.2945,
      "step": 4936
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.9625536903835425,
      "learning_rate": 3.116572946354815e-06,
      "loss": 0.1881,
      "step": 4937
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.997661369874759,
      "learning_rate": 3.1158765465532404e-06,
      "loss": 0.4385,
      "step": 4938
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.253124086789594,
      "learning_rate": 3.1151800958709648e-06,
      "loss": 0.3742,
      "step": 4939
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.329248783609188,
      "learning_rate": 3.1144835943655256e-06,
      "loss": 0.3319,
      "step": 4940
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.0653900746526865,
      "learning_rate": 3.113787042094465e-06,
      "loss": 0.198,
      "step": 4941
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8682650803708747,
      "learning_rate": 3.113090439115327e-06,
      "loss": 0.0699,
      "step": 4942
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.578470875969412,
      "learning_rate": 3.112393785485664e-06,
      "loss": 0.402,
      "step": 4943
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.506260634375419,
      "learning_rate": 3.111697081263028e-06,
      "loss": 0.1921,
      "step": 4944
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.990301183062897,
      "learning_rate": 3.1110003265049787e-06,
      "loss": 0.3205,
      "step": 4945
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.214477310614006,
      "learning_rate": 3.1103035212690787e-06,
      "loss": 0.2275,
      "step": 4946
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.30948986614763,
      "learning_rate": 3.1096066656128925e-06,
      "loss": 0.2114,
      "step": 4947
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.699057995691039,
      "learning_rate": 3.1089097595939932e-06,
      "loss": 0.4609,
      "step": 4948
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4417399100585833,
      "learning_rate": 3.108212803269954e-06,
      "loss": 0.2219,
      "step": 4949
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.4968242989384475,
      "learning_rate": 3.107515796698356e-06,
      "loss": 0.2899,
      "step": 4950
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.633108128881131,
      "learning_rate": 3.106818739936781e-06,
      "loss": 0.3844,
      "step": 4951
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.188875083247561,
      "learning_rate": 3.106121633042816e-06,
      "loss": 0.2485,
      "step": 4952
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.130254734056694,
      "learning_rate": 3.105424476074054e-06,
      "loss": 0.2152,
      "step": 4953
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.981930075136838,
      "learning_rate": 3.1047272690880904e-06,
      "loss": 0.2958,
      "step": 4954
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.424215844398684,
      "learning_rate": 3.104030012142524e-06,
      "loss": 0.2462,
      "step": 4955
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.409363123165759,
      "learning_rate": 3.10333270529496e-06,
      "loss": 0.3866,
      "step": 4956
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4387966646295216,
      "learning_rate": 3.1026353486030054e-06,
      "loss": 0.2079,
      "step": 4957
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.738063970570216,
      "learning_rate": 3.1019379421242734e-06,
      "loss": 0.2901,
      "step": 4958
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.870080582082895,
      "learning_rate": 3.1012404859163785e-06,
      "loss": 0.2534,
      "step": 4959
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.2698646925650525,
      "learning_rate": 3.1005429800369436e-06,
      "loss": 0.3294,
      "step": 4960
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6410435737323477,
      "learning_rate": 3.0998454245435923e-06,
      "loss": 0.1718,
      "step": 4961
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.594273886026298,
      "learning_rate": 3.0991478194939515e-06,
      "loss": 0.5425,
      "step": 4962
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.239426986767386,
      "learning_rate": 3.098450164945656e-06,
      "loss": 0.4242,
      "step": 4963
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6521626697771077,
      "learning_rate": 3.097752460956342e-06,
      "loss": 0.2793,
      "step": 4964
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.172477799128823,
      "learning_rate": 3.0970547075836503e-06,
      "loss": 0.2022,
      "step": 4965
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.5711321098720195,
      "learning_rate": 3.096356904885226e-06,
      "loss": 0.3767,
      "step": 4966
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.681681679261479,
      "learning_rate": 3.0956590529187173e-06,
      "loss": 0.1373,
      "step": 4967
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.076388907333688,
      "learning_rate": 3.0949611517417787e-06,
      "loss": 0.1994,
      "step": 4968
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.59549111086661,
      "learning_rate": 3.0942632014120667e-06,
      "loss": 0.4862,
      "step": 4969
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.282253089633197,
      "learning_rate": 3.0935652019872425e-06,
      "loss": 0.1891,
      "step": 4970
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.1345894415909092,
      "learning_rate": 3.0928671535249718e-06,
      "loss": 0.2268,
      "step": 4971
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.723155182566518,
      "learning_rate": 3.092169056082922e-06,
      "loss": 0.2894,
      "step": 4972
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.6816575985243203,
      "learning_rate": 3.091470909718769e-06,
      "loss": 0.1682,
      "step": 4973
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.336574212179533,
      "learning_rate": 3.0907727144901894e-06,
      "loss": 0.3413,
      "step": 4974
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.206914821295698,
      "learning_rate": 3.0900744704548645e-06,
      "loss": 0.2468,
      "step": 4975
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.3721580464959695,
      "learning_rate": 3.08937617767048e-06,
      "loss": 0.3875,
      "step": 4976
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.111964416008556,
      "learning_rate": 3.0886778361947246e-06,
      "loss": 0.4285,
      "step": 4977
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.796547348850406,
      "learning_rate": 3.087979446085293e-06,
      "loss": 0.5079,
      "step": 4978
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.521601059242475,
      "learning_rate": 3.087281007399882e-06,
      "loss": 0.4411,
      "step": 4979
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.751472211525078,
      "learning_rate": 3.086582520196194e-06,
      "loss": 0.145,
      "step": 4980
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.470163697986655,
      "learning_rate": 3.085883984531934e-06,
      "loss": 0.0618,
      "step": 4981
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.140139860922409,
      "learning_rate": 3.0851854004648107e-06,
      "loss": 0.5612,
      "step": 4982
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.573384206766277,
      "learning_rate": 3.084486768052539e-06,
      "loss": 0.4721,
      "step": 4983
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.932066491324931,
      "learning_rate": 3.0837880873528363e-06,
      "loss": 0.5106,
      "step": 4984
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4765981131586607,
      "learning_rate": 3.0830893584234245e-06,
      "loss": 0.1337,
      "step": 4985
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.482076959411211,
      "learning_rate": 3.0823905813220283e-06,
      "loss": 0.2634,
      "step": 4986
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.115938648766443,
      "learning_rate": 3.081691756106377e-06,
      "loss": 0.3026,
      "step": 4987
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.167314762972917,
      "learning_rate": 3.080992882834205e-06,
      "loss": 0.5489,
      "step": 4988
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4929499745201116,
      "learning_rate": 3.0802939615632498e-06,
      "loss": 0.0554,
      "step": 4989
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.684734365099613,
      "learning_rate": 3.0795949923512525e-06,
      "loss": 0.356,
      "step": 4990
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.086086558703498,
      "learning_rate": 3.0788959752559583e-06,
      "loss": 0.2842,
      "step": 4991
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.4063781299936795,
      "learning_rate": 3.0781969103351163e-06,
      "loss": 0.1013,
      "step": 4992
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6900222745273927,
      "learning_rate": 3.077497797646481e-06,
      "loss": 0.3231,
      "step": 4993
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.6230052649297235,
      "learning_rate": 3.0767986372478088e-06,
      "loss": 0.4945,
      "step": 4994
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2760497545193306,
      "learning_rate": 3.0760994291968615e-06,
      "loss": 0.0525,
      "step": 4995
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.6944709577598673,
      "learning_rate": 3.075400173551404e-06,
      "loss": 0.3252,
      "step": 4996
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.263686555958294,
      "learning_rate": 3.0747008703692034e-06,
      "loss": 0.3062,
      "step": 4997
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.923107099267787,
      "learning_rate": 3.074001519708036e-06,
      "loss": 0.5925,
      "step": 4998
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.787879101761248,
      "learning_rate": 3.073302121625677e-06,
      "loss": 0.1697,
      "step": 4999
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.2109002189663265,
      "learning_rate": 3.0726026761799077e-06,
      "loss": 0.2106,
      "step": 5000
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.9830355189398374,
      "learning_rate": 3.071903183428513e-06,
      "loss": 0.1779,
      "step": 5001
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.342607761299718,
      "learning_rate": 3.071203643429279e-06,
      "loss": 0.5422,
      "step": 5002
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.349615717130405,
      "learning_rate": 3.0705040562400025e-06,
      "loss": 0.382,
      "step": 5003
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.644106739164708,
      "learning_rate": 3.0698044219184773e-06,
      "loss": 0.3041,
      "step": 5004
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2145929428475575,
      "learning_rate": 3.0691047405225048e-06,
      "loss": 0.1097,
      "step": 5005
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.575203548748124,
      "learning_rate": 3.068405012109888e-06,
      "loss": 0.2748,
      "step": 5006
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.469665106530607,
      "learning_rate": 3.0677052367384364e-06,
      "loss": 0.5403,
      "step": 5007
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.62628954463138,
      "learning_rate": 3.0670054144659613e-06,
      "loss": 0.1585,
      "step": 5008
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.48096579271516,
      "learning_rate": 3.066305545350279e-06,
      "loss": 0.1974,
      "step": 5009
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.328537431391629,
      "learning_rate": 3.0656056294492086e-06,
      "loss": 0.3817,
      "step": 5010
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.503669800817744,
      "learning_rate": 3.064905666820574e-06,
      "loss": 0.5395,
      "step": 5011
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.570926184175203,
      "learning_rate": 3.064205657522203e-06,
      "loss": 0.2766,
      "step": 5012
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.03475746544935,
      "learning_rate": 3.063505601611927e-06,
      "loss": 0.4807,
      "step": 5013
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.249240756430706,
      "learning_rate": 3.062805499147581e-06,
      "loss": 0.3561,
      "step": 5014
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.38367125337341,
      "learning_rate": 3.0621053501870037e-06,
      "loss": 0.2532,
      "step": 5015
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.57037210693007,
      "learning_rate": 3.0614051547880374e-06,
      "loss": 0.3337,
      "step": 5016
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.6021719862159234,
      "learning_rate": 3.06070491300853e-06,
      "loss": 0.1983,
      "step": 5017
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.873432903084344,
      "learning_rate": 3.0600046249063326e-06,
      "loss": 0.2308,
      "step": 5018
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.938232534957135,
      "learning_rate": 3.0593042905392984e-06,
      "loss": 0.2151,
      "step": 5019
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.7370992210860345,
      "learning_rate": 3.058603909965286e-06,
      "loss": 0.5619,
      "step": 5020
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.85263908826624,
      "learning_rate": 3.057903483242156e-06,
      "loss": 0.2797,
      "step": 5021
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2486460287677077,
      "learning_rate": 3.057203010427776e-06,
      "loss": 0.0508,
      "step": 5022
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.063805165502391,
      "learning_rate": 3.0565024915800157e-06,
      "loss": 0.3146,
      "step": 5023
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.5549290906327307,
      "learning_rate": 3.0558019267567477e-06,
      "loss": 0.0279,
      "step": 5024
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.909454995385955,
      "learning_rate": 3.0551013160158487e-06,
      "loss": 0.1907,
      "step": 5025
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.077147875449621,
      "learning_rate": 3.0544006594152015e-06,
      "loss": 0.073,
      "step": 5026
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.055163088810523,
      "learning_rate": 3.0536999570126892e-06,
      "loss": 0.2373,
      "step": 5027
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.8953775134325577,
      "learning_rate": 3.052999208866202e-06,
      "loss": 0.2267,
      "step": 5028
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.8990391385268186,
      "learning_rate": 3.052298415033631e-06,
      "loss": 0.1228,
      "step": 5029
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.796313846653311,
      "learning_rate": 3.051597575572872e-06,
      "loss": 0.2311,
      "step": 5030
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.689748909753795,
      "learning_rate": 3.050896690541826e-06,
      "loss": 0.3834,
      "step": 5031
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.73774816710167,
      "learning_rate": 3.0501957599983965e-06,
      "loss": 0.4102,
      "step": 5032
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.589826728323691,
      "learning_rate": 3.049494784000491e-06,
      "loss": 0.4687,
      "step": 5033
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.0656633144181615,
      "learning_rate": 3.0487937626060203e-06,
      "loss": 0.3455,
      "step": 5034
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.190830174467479,
      "learning_rate": 3.0480926958728996e-06,
      "loss": 0.3697,
      "step": 5035
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.329347682125806,
      "learning_rate": 3.047391583859047e-06,
      "loss": 0.341,
      "step": 5036
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.01710376423225,
      "learning_rate": 3.0466904266223856e-06,
      "loss": 0.2903,
      "step": 5037
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.55027342988467,
      "learning_rate": 3.0459892242208423e-06,
      "loss": 0.1428,
      "step": 5038
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.4099981886169237,
      "learning_rate": 3.0452879767123457e-06,
      "loss": 0.0944,
      "step": 5039
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.780123958360437,
      "learning_rate": 3.0445866841548296e-06,
      "loss": 0.5027,
      "step": 5040
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.918441209239242,
      "learning_rate": 3.0438853466062314e-06,
      "loss": 0.2151,
      "step": 5041
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.331495167892945,
      "learning_rate": 3.0431839641244925e-06,
      "loss": 0.3468,
      "step": 5042
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.727158861147262,
      "learning_rate": 3.0424825367675582e-06,
      "loss": 0.4291,
      "step": 5043
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.4294555979338264,
      "learning_rate": 3.041781064593376e-06,
      "loss": 0.1835,
      "step": 5044
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.199215093847567,
      "learning_rate": 3.041079547659898e-06,
      "loss": 0.3095,
      "step": 5045
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.185526053332576,
      "learning_rate": 3.0403779860250808e-06,
      "loss": 0.1208,
      "step": 5046
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.14780066157603,
      "learning_rate": 3.039676379746884e-06,
      "loss": 0.4143,
      "step": 5047
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.28771869442166,
      "learning_rate": 3.0389747288832706e-06,
      "loss": 0.3102,
      "step": 5048
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.206943505471962,
      "learning_rate": 3.0382730334922067e-06,
      "loss": 0.1514,
      "step": 5049
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.360284343652712,
      "learning_rate": 3.037571293631665e-06,
      "loss": 0.2358,
      "step": 5050
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.318167856922191,
      "learning_rate": 3.036869509359618e-06,
      "loss": 0.2286,
      "step": 5051
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.408889687285139,
      "learning_rate": 3.036167680734044e-06,
      "loss": 0.3801,
      "step": 5052
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.6711942258784656,
      "learning_rate": 3.0354658078129256e-06,
      "loss": 0.2588,
      "step": 5053
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.595331567587207,
      "learning_rate": 3.0347638906542464e-06,
      "loss": 0.4805,
      "step": 5054
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.354847034388392,
      "learning_rate": 3.0340619293159972e-06,
      "loss": 0.4288,
      "step": 5055
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.9076281986418566,
      "learning_rate": 3.03335992385617e-06,
      "loss": 0.2986,
      "step": 5056
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.272351238413334,
      "learning_rate": 3.03265787433276e-06,
      "loss": 0.442,
      "step": 5057
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.807603471512081,
      "learning_rate": 3.0319557808037687e-06,
      "loss": 0.3441,
      "step": 5058
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.255711954439796,
      "learning_rate": 3.0312536433271976e-06,
      "loss": 0.2598,
      "step": 5059
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.881278108503005,
      "learning_rate": 3.0305514619610565e-06,
      "loss": 0.3287,
      "step": 5060
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.8369652361373516,
      "learning_rate": 3.0298492367633542e-06,
      "loss": 0.2826,
      "step": 5061
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.5601400210861325,
      "learning_rate": 3.0291469677921056e-06,
      "loss": 0.363,
      "step": 5062
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.212816284361954,
      "learning_rate": 3.0284446551053285e-06,
      "loss": 0.3329,
      "step": 5063
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.3329658979433434,
      "learning_rate": 3.027742298761045e-06,
      "loss": 0.1402,
      "step": 5064
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.330371088570464,
      "learning_rate": 3.0270398988172806e-06,
      "loss": 0.2775,
      "step": 5065
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.753468995270949,
      "learning_rate": 3.0263374553320633e-06,
      "loss": 0.4267,
      "step": 5066
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.4170024683881217,
      "learning_rate": 3.0256349683634264e-06,
      "loss": 0.2672,
      "step": 5067
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.814036708629305,
      "learning_rate": 3.0249324379694045e-06,
      "loss": 0.2843,
      "step": 5068
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.846020846555507,
      "learning_rate": 3.0242298642080386e-06,
      "loss": 0.2214,
      "step": 5069
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.687823090333912,
      "learning_rate": 3.0235272471373724e-06,
      "loss": 0.5702,
      "step": 5070
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.668341603377335,
      "learning_rate": 3.022824586815451e-06,
      "loss": 0.2029,
      "step": 5071
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.947652417525209,
      "learning_rate": 3.0221218833003254e-06,
      "loss": 0.4798,
      "step": 5072
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.5926763254290117,
      "learning_rate": 3.021419136650049e-06,
      "loss": 0.3205,
      "step": 5073
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.6439067710678135,
      "learning_rate": 3.020716346922681e-06,
      "loss": 0.4998,
      "step": 5074
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.890769332160241,
      "learning_rate": 3.020013514176281e-06,
      "loss": 0.2306,
      "step": 5075
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9822121999301334,
      "learning_rate": 3.0193106384689144e-06,
      "loss": 0.0721,
      "step": 5076
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.188012852530459,
      "learning_rate": 3.0186077198586482e-06,
      "loss": 0.3622,
      "step": 5077
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.30918263351565,
      "learning_rate": 3.017904758403555e-06,
      "loss": 0.2164,
      "step": 5078
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.9690201297712213,
      "learning_rate": 3.0172017541617095e-06,
      "loss": 0.3161,
      "step": 5079
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.262784493875218,
      "learning_rate": 3.0164987071911918e-06,
      "loss": 0.2988,
      "step": 5080
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.9886413693724005,
      "learning_rate": 3.0157956175500823e-06,
      "loss": 0.4169,
      "step": 5081
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.168756806112075,
      "learning_rate": 3.0150924852964674e-06,
      "loss": 0.4699,
      "step": 5082
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.906288124488619,
      "learning_rate": 3.014389310488437e-06,
      "loss": 0.3648,
      "step": 5083
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.122759660625205,
      "learning_rate": 3.013686093184084e-06,
      "loss": 0.3957,
      "step": 5084
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.185134383592378,
      "learning_rate": 3.0129828334415046e-06,
      "loss": 0.429,
      "step": 5085
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.484789988841989,
      "learning_rate": 3.0122795313187985e-06,
      "loss": 0.4187,
      "step": 5086
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.297458036127003,
      "learning_rate": 3.0115761868740684e-06,
      "loss": 0.2932,
      "step": 5087
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.711453148740165,
      "learning_rate": 3.010872800165422e-06,
      "loss": 0.3364,
      "step": 5088
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.512354828062086,
      "learning_rate": 3.01016937125097e-06,
      "loss": 0.2052,
      "step": 5089
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.702871893341287,
      "learning_rate": 3.0094659001888262e-06,
      "loss": 0.4211,
      "step": 5090
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.702537860350029,
      "learning_rate": 3.0087623870371067e-06,
      "loss": 0.2686,
      "step": 5091
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.963415875205847,
      "learning_rate": 3.0080588318539328e-06,
      "loss": 0.3876,
      "step": 5092
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.522880965086896,
      "learning_rate": 3.00735523469743e-06,
      "loss": 0.1286,
      "step": 5093
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.479627623157386,
      "learning_rate": 3.0066515956257246e-06,
      "loss": 0.5179,
      "step": 5094
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.074790406799178,
      "learning_rate": 3.0059479146969487e-06,
      "loss": 0.2021,
      "step": 5095
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.928735837829921,
      "learning_rate": 3.0052441919692364e-06,
      "loss": 0.3724,
      "step": 5096
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.645264204373484,
      "learning_rate": 3.0045404275007256e-06,
      "loss": 0.3111,
      "step": 5097
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.036437339904618,
      "learning_rate": 3.003836621349559e-06,
      "loss": 0.3131,
      "step": 5098
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.94162799862385,
      "learning_rate": 3.0031327735738808e-06,
      "loss": 0.1479,
      "step": 5099
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.144050860543681,
      "learning_rate": 3.0024288842318393e-06,
      "loss": 0.3614,
      "step": 5100
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.469293001448648,
      "learning_rate": 3.0017249533815873e-06,
      "loss": 0.3509,
      "step": 5101
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.576771798227218,
      "learning_rate": 3.001020981081278e-06,
      "loss": 0.2801,
      "step": 5102
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.421874812476753,
      "learning_rate": 3.0003169673890733e-06,
      "loss": 0.3133,
      "step": 5103
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0022785078335787,
      "learning_rate": 2.999612912363133e-06,
      "loss": 0.1985,
      "step": 5104
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.007453836826411,
      "learning_rate": 2.998908816061623e-06,
      "loss": 0.2799,
      "step": 5105
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5463509400399778,
      "learning_rate": 2.9982046785427132e-06,
      "loss": 0.1417,
      "step": 5106
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.851560902227128,
      "learning_rate": 2.9975004998645745e-06,
      "loss": 0.3635,
      "step": 5107
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.468538131031678,
      "learning_rate": 2.9967962800853846e-06,
      "loss": 0.2687,
      "step": 5108
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.999055781610225,
      "learning_rate": 2.9960920192633217e-06,
      "loss": 0.406,
      "step": 5109
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.2663200548875317,
      "learning_rate": 2.995387717456568e-06,
      "loss": 0.1123,
      "step": 5110
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.895538457786936,
      "learning_rate": 2.99468337472331e-06,
      "loss": 0.4013,
      "step": 5111
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.134484187462316,
      "learning_rate": 2.9939789911217366e-06,
      "loss": 0.2962,
      "step": 5112
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.89365576511777,
      "learning_rate": 2.993274566710041e-06,
      "loss": 0.324,
      "step": 5113
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.398139118222913,
      "learning_rate": 2.992570101546419e-06,
      "loss": 0.4704,
      "step": 5114
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.752357546431005,
      "learning_rate": 2.991865595689071e-06,
      "loss": 0.1962,
      "step": 5115
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0136165569630013,
      "learning_rate": 2.9911610491961977e-06,
      "loss": 0.1592,
      "step": 5116
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.099018692105976,
      "learning_rate": 2.990456462126007e-06,
      "loss": 0.366,
      "step": 5117
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.07114053419646,
      "learning_rate": 2.989751834536708e-06,
      "loss": 0.3924,
      "step": 5118
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.274710197022785,
      "learning_rate": 2.989047166486514e-06,
      "loss": 0.5371,
      "step": 5119
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.265563672509059,
      "learning_rate": 2.9883424580336407e-06,
      "loss": 0.4519,
      "step": 5120
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.136424303486428,
      "learning_rate": 2.987637709236307e-06,
      "loss": 0.3665,
      "step": 5121
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.450449645035891,
      "learning_rate": 2.9869329201527363e-06,
      "loss": 0.1302,
      "step": 5122
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.759648470738097,
      "learning_rate": 2.986228090841156e-06,
      "loss": 0.1734,
      "step": 5123
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.8044666869480617,
      "learning_rate": 2.9855232213597945e-06,
      "loss": 0.151,
      "step": 5124
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.291256810057828,
      "learning_rate": 2.984818311766885e-06,
      "loss": 0.3675,
      "step": 5125
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.560863569668388,
      "learning_rate": 2.984113362120663e-06,
      "loss": 0.4254,
      "step": 5126
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.5094835815731775,
      "learning_rate": 2.9834083724793685e-06,
      "loss": 0.2739,
      "step": 5127
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.141966003702786,
      "learning_rate": 2.9827033429012442e-06,
      "loss": 0.3225,
      "step": 5128
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.536097279822168,
      "learning_rate": 2.981998273444537e-06,
      "loss": 0.1631,
      "step": 5129
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.563560318659316,
      "learning_rate": 2.9812931641674948e-06,
      "loss": 0.2782,
      "step": 5130
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.197688904646076,
      "learning_rate": 2.980588015128372e-06,
      "loss": 0.4406,
      "step": 5131
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.325022602357946,
      "learning_rate": 2.9798828263854222e-06,
      "loss": 0.2067,
      "step": 5132
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.361643832887742,
      "learning_rate": 2.979177597996907e-06,
      "loss": 0.5262,
      "step": 5133
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.7796150377703555,
      "learning_rate": 2.9784723300210883e-06,
      "loss": 0.3026,
      "step": 5134
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.8829098096274337,
      "learning_rate": 2.9777670225162303e-06,
      "loss": 0.2664,
      "step": 5135
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.651150059821297,
      "learning_rate": 2.977061675540604e-06,
      "loss": 0.4517,
      "step": 5136
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.356553576605324,
      "learning_rate": 2.9763562891524805e-06,
      "loss": 0.3527,
      "step": 5137
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.8969660094539433,
      "learning_rate": 2.9756508634101366e-06,
      "loss": 0.351,
      "step": 5138
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.138908074793277,
      "learning_rate": 2.9749453983718502e-06,
      "loss": 0.5938,
      "step": 5139
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.727873193661517,
      "learning_rate": 2.9742398940959027e-06,
      "loss": 0.5583,
      "step": 5140
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5782264902301235,
      "learning_rate": 2.9735343506405812e-06,
      "loss": 0.1218,
      "step": 5141
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.003297251130122,
      "learning_rate": 2.972828768064173e-06,
      "loss": 0.2094,
      "step": 5142
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.639837900450093,
      "learning_rate": 2.97212314642497e-06,
      "loss": 0.2831,
      "step": 5143
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.897353522308479,
      "learning_rate": 2.971417485781267e-06,
      "loss": 0.2614,
      "step": 5144
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.663569696811518,
      "learning_rate": 2.970711786191363e-06,
      "loss": 0.4614,
      "step": 5145
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.21594668327156,
      "learning_rate": 2.970006047713559e-06,
      "loss": 0.3584,
      "step": 5146
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.335219806842325,
      "learning_rate": 2.969300270406159e-06,
      "loss": 0.3279,
      "step": 5147
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.334266406488415,
      "learning_rate": 2.9685944543274725e-06,
      "loss": 0.3502,
      "step": 5148
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2180607257540923,
      "learning_rate": 2.9678885995358087e-06,
      "loss": 0.0209,
      "step": 5149
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.627741567105319,
      "learning_rate": 2.967182706089484e-06,
      "loss": 0.1492,
      "step": 5150
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.853943514925339,
      "learning_rate": 2.9664767740468135e-06,
      "loss": 0.4305,
      "step": 5151
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.301748828095196,
      "learning_rate": 2.9657708034661194e-06,
      "loss": 0.2971,
      "step": 5152
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.8416905473869285,
      "learning_rate": 2.9650647944057247e-06,
      "loss": 0.4605,
      "step": 5153
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.779900007078968,
      "learning_rate": 2.9643587469239567e-06,
      "loss": 0.1988,
      "step": 5154
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.955640626998099,
      "learning_rate": 2.963652661079146e-06,
      "loss": 0.2993,
      "step": 5155
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.924489514399324,
      "learning_rate": 2.962946536929626e-06,
      "loss": 0.2931,
      "step": 5156
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.016316614398952,
      "learning_rate": 2.962240374533732e-06,
      "loss": 0.2669,
      "step": 5157
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.397221832831223,
      "learning_rate": 2.961534173949805e-06,
      "loss": 0.3837,
      "step": 5158
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.3654744429487975,
      "learning_rate": 2.960827935236187e-06,
      "loss": 0.2866,
      "step": 5159
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.079046835365893,
      "learning_rate": 2.9601216584512244e-06,
      "loss": 0.1613,
      "step": 5160
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.339233987315818,
      "learning_rate": 2.959415343653267e-06,
      "loss": 0.3477,
      "step": 5161
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.0358150757976503,
      "learning_rate": 2.958708990900665e-06,
      "loss": 0.1539,
      "step": 5162
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.655807291165759,
      "learning_rate": 2.958002600251776e-06,
      "loss": 0.0952,
      "step": 5163
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9648217387735842,
      "learning_rate": 2.9572961717649574e-06,
      "loss": 0.1705,
      "step": 5164
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.23454777195111,
      "learning_rate": 2.956589705498571e-06,
      "loss": 0.2229,
      "step": 5165
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.393694846202282,
      "learning_rate": 2.955883201510982e-06,
      "loss": 0.2406,
      "step": 5166
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.6717311056559785,
      "learning_rate": 2.955176659860557e-06,
      "loss": 0.4959,
      "step": 5167
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.483950288390628,
      "learning_rate": 2.9544700806056685e-06,
      "loss": 0.2185,
      "step": 5168
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.69465190889923,
      "learning_rate": 2.9537634638046907e-06,
      "loss": 0.2745,
      "step": 5169
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.402909743700769,
      "learning_rate": 2.9530568095159994e-06,
      "loss": 0.3845,
      "step": 5170
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.201539104890341,
      "learning_rate": 2.9523501177979767e-06,
      "loss": 0.2774,
      "step": 5171
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.7569593933117815,
      "learning_rate": 2.951643388709004e-06,
      "loss": 0.5518,
      "step": 5172
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.28254672103804,
      "learning_rate": 2.9509366223074688e-06,
      "loss": 0.3751,
      "step": 5173
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.368208369005648,
      "learning_rate": 2.9502298186517615e-06,
      "loss": 0.2928,
      "step": 5174
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.4481160391485,
      "learning_rate": 2.949522977800274e-06,
      "loss": 0.2513,
      "step": 5175
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.8341522142310023,
      "learning_rate": 2.9488160998114013e-06,
      "loss": 0.2292,
      "step": 5176
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.952329778549979,
      "learning_rate": 2.948109184743543e-06,
      "loss": 0.4448,
      "step": 5177
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.942793381183645,
      "learning_rate": 2.9474022326551003e-06,
      "loss": 0.2199,
      "step": 5178
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.922203676052118,
      "learning_rate": 2.9466952436044794e-06,
      "loss": 0.374,
      "step": 5179
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.036884743935448,
      "learning_rate": 2.9459882176500874e-06,
      "loss": 0.2831,
      "step": 5180
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.468685755523063,
      "learning_rate": 2.9452811548503356e-06,
      "loss": 0.4824,
      "step": 5181
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.4809837370303045,
      "learning_rate": 2.9445740552636375e-06,
      "loss": 0.412,
      "step": 5182
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.63096239472261,
      "learning_rate": 2.9438669189484105e-06,
      "loss": 0.3217,
      "step": 5183
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.4503355202036863,
      "learning_rate": 2.9431597459630752e-06,
      "loss": 0.1405,
      "step": 5184
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.362639420262293,
      "learning_rate": 2.9424525363660546e-06,
      "loss": 0.2852,
      "step": 5185
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.551317657104393,
      "learning_rate": 2.941745290215774e-06,
      "loss": 0.2723,
      "step": 5186
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.061948139336184,
      "learning_rate": 2.9410380075706637e-06,
      "loss": 0.3789,
      "step": 5187
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.696108233423711,
      "learning_rate": 2.9403306884891547e-06,
      "loss": 0.3762,
      "step": 5188
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.5961698928090415,
      "learning_rate": 2.9396233330296843e-06,
      "loss": 0.3123,
      "step": 5189
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.76073579434785,
      "learning_rate": 2.938915941250689e-06,
      "loss": 0.1242,
      "step": 5190
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.887893138820894,
      "learning_rate": 2.9382085132106105e-06,
      "loss": 0.2428,
      "step": 5191
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1403009916082754,
      "learning_rate": 2.9375010489678925e-06,
      "loss": 0.2124,
      "step": 5192
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.3613976915254,
      "learning_rate": 2.9367935485809827e-06,
      "loss": 0.4045,
      "step": 5193
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.9953984719165105,
      "learning_rate": 2.9360860121083324e-06,
      "loss": 0.1465,
      "step": 5194
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.7049790318773836,
      "learning_rate": 2.935378439608394e-06,
      "loss": 0.2078,
      "step": 5195
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6850102058401983,
      "learning_rate": 2.9346708311396226e-06,
      "loss": 0.3239,
      "step": 5196
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.449314936475415,
      "learning_rate": 2.933963186760478e-06,
      "loss": 0.1242,
      "step": 5197
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.637898599326144,
      "learning_rate": 2.933255506529423e-06,
      "loss": 0.1144,
      "step": 5198
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.126988726530453,
      "learning_rate": 2.932547790504922e-06,
      "loss": 0.336,
      "step": 5199
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.5224461801258142,
      "learning_rate": 2.931840038745444e-06,
      "loss": 0.1521,
      "step": 5200
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.2398703372351836,
      "learning_rate": 2.931132251309459e-06,
      "loss": 0.1754,
      "step": 5201
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.9228996235022953,
      "learning_rate": 2.9304244282554393e-06,
      "loss": 0.1493,
      "step": 5202
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.381480915811077,
      "learning_rate": 2.929716569641865e-06,
      "loss": 0.2951,
      "step": 5203
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.401887672529384,
      "learning_rate": 2.929008675527215e-06,
      "loss": 0.4219,
      "step": 5204
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.864135573833964,
      "learning_rate": 2.928300745969971e-06,
      "loss": 0.3254,
      "step": 5205
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.796163265185676,
      "learning_rate": 2.927592781028619e-06,
      "loss": 0.2248,
      "step": 5206
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.4459948042069293,
      "learning_rate": 2.9268847807616475e-06,
      "loss": 0.2549,
      "step": 5207
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.567207304358708,
      "learning_rate": 2.9261767452275493e-06,
      "loss": 0.2371,
      "step": 5208
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.815938195972413,
      "learning_rate": 2.925468674484818e-06,
      "loss": 0.3364,
      "step": 5209
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6225336432934467,
      "learning_rate": 2.92476056859195e-06,
      "loss": 0.2251,
      "step": 5210
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.6763182434594324,
      "learning_rate": 2.9240524276074463e-06,
      "loss": 0.1219,
      "step": 5211
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.269817411094885,
      "learning_rate": 2.9233442515898095e-06,
      "loss": 0.3907,
      "step": 5212
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.554040712650505,
      "learning_rate": 2.9226360405975475e-06,
      "loss": 0.3292,
      "step": 5213
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.194011617641844,
      "learning_rate": 2.9219277946891673e-06,
      "loss": 0.3729,
      "step": 5214
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.4727318172592945,
      "learning_rate": 2.9212195139231812e-06,
      "loss": 0.2652,
      "step": 5215
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1436579968177116,
      "learning_rate": 2.9205111983581043e-06,
      "loss": 0.2275,
      "step": 5216
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.066124492644694,
      "learning_rate": 2.919802848052453e-06,
      "loss": 0.3892,
      "step": 5217
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.715970586008878,
      "learning_rate": 2.9190944630647494e-06,
      "loss": 0.311,
      "step": 5218
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.507650031093725,
      "learning_rate": 2.918386043453516e-06,
      "loss": 0.4293,
      "step": 5219
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.096535402827932,
      "learning_rate": 2.917677589277279e-06,
      "loss": 0.3727,
      "step": 5220
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.7022868043144666,
      "learning_rate": 2.9169691005945666e-06,
      "loss": 0.135,
      "step": 5221
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.255138417425894,
      "learning_rate": 2.916260577463912e-06,
      "loss": 0.3287,
      "step": 5222
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.155235557025148,
      "learning_rate": 2.915552019943849e-06,
      "loss": 0.3907,
      "step": 5223
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.083045116526055,
      "learning_rate": 2.914843428092916e-06,
      "loss": 0.3601,
      "step": 5224
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.0877189016327202,
      "learning_rate": 2.9141348019696525e-06,
      "loss": 0.1558,
      "step": 5225
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.315324740811498,
      "learning_rate": 2.9134261416326017e-06,
      "loss": 0.2064,
      "step": 5226
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.053532488255491,
      "learning_rate": 2.9127174471403097e-06,
      "loss": 0.2819,
      "step": 5227
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.0949746903746975,
      "learning_rate": 2.912008718551327e-06,
      "loss": 0.6025,
      "step": 5228
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.209628636167298,
      "learning_rate": 2.9112999559242035e-06,
      "loss": 0.3106,
      "step": 5229
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.4406732005806395,
      "learning_rate": 2.910591159317493e-06,
      "loss": 0.2749,
      "step": 5230
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.576249270958081,
      "learning_rate": 2.9098823287897554e-06,
      "loss": 0.3793,
      "step": 5231
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.171002843771981,
      "learning_rate": 2.9091734643995483e-06,
      "loss": 0.3506,
      "step": 5232
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.05595664492432,
      "learning_rate": 2.908464566205436e-06,
      "loss": 0.3386,
      "step": 5233
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.012859035330841,
      "learning_rate": 2.907755634265984e-06,
      "loss": 0.2665,
      "step": 5234
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.1296279166593375,
      "learning_rate": 2.9070466686397603e-06,
      "loss": 0.2336,
      "step": 5235
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.331353937025013,
      "learning_rate": 2.9063376693853367e-06,
      "loss": 0.2233,
      "step": 5236
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.4908603071058115,
      "learning_rate": 2.9056286365612867e-06,
      "loss": 0.3266,
      "step": 5237
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.8148583693140647,
      "learning_rate": 2.9049195702261883e-06,
      "loss": 0.2007,
      "step": 5238
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.585405189462412,
      "learning_rate": 2.9042104704386196e-06,
      "loss": 0.3797,
      "step": 5239
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.836522720709617,
      "learning_rate": 2.9035013372571633e-06,
      "loss": 0.3664,
      "step": 5240
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.136457385431846,
      "learning_rate": 2.9027921707404056e-06,
      "loss": 0.3784,
      "step": 5241
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.119147871663276,
      "learning_rate": 2.902082970946933e-06,
      "loss": 0.2209,
      "step": 5242
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.311161459011665,
      "learning_rate": 2.9013737379353368e-06,
      "loss": 0.1806,
      "step": 5243
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.3430766427713743,
      "learning_rate": 2.90066447176421e-06,
      "loss": 0.1424,
      "step": 5244
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.335463853042418,
      "learning_rate": 2.899955172492149e-06,
      "loss": 0.4196,
      "step": 5245
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.083528158673192,
      "learning_rate": 2.8992458401777524e-06,
      "loss": 0.233,
      "step": 5246
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.330116212962648,
      "learning_rate": 2.898536474879622e-06,
      "loss": 0.1837,
      "step": 5247
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.304643006531079,
      "learning_rate": 2.897827076656362e-06,
      "loss": 0.2326,
      "step": 5248
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.465066283654208,
      "learning_rate": 2.8971176455665788e-06,
      "loss": 0.2071,
      "step": 5249
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.383222555367005,
      "learning_rate": 2.896408181668884e-06,
      "loss": 0.2697,
      "step": 5250
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.184221541603763,
      "learning_rate": 2.8956986850218876e-06,
      "loss": 0.11,
      "step": 5251
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.668639022744886,
      "learning_rate": 2.8949891556842056e-06,
      "loss": 0.2882,
      "step": 5252
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.725225050083634,
      "learning_rate": 2.8942795937144574e-06,
      "loss": 0.1137,
      "step": 5253
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.550103520387627,
      "learning_rate": 2.8935699991712614e-06,
      "loss": 0.3144,
      "step": 5254
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.680958348124925,
      "learning_rate": 2.8928603721132416e-06,
      "loss": 0.1346,
      "step": 5255
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.513883344336451,
      "learning_rate": 2.8921507125990246e-06,
      "loss": 0.243,
      "step": 5256
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1401991764458317,
      "learning_rate": 2.8914410206872373e-06,
      "loss": 0.0915,
      "step": 5257
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.870265073821483,
      "learning_rate": 2.8907312964365126e-06,
      "loss": 0.2714,
      "step": 5258
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.404182759634184,
      "learning_rate": 2.890021539905483e-06,
      "loss": 0.5369,
      "step": 5259
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.0991474187646153,
      "learning_rate": 2.8893117511527874e-06,
      "loss": 0.1969,
      "step": 5260
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.876558005838939,
      "learning_rate": 2.888601930237063e-06,
      "loss": 0.45,
      "step": 5261
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.799168449950827,
      "learning_rate": 2.8878920772169517e-06,
      "loss": 0.1849,
      "step": 5262
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.974253892084128,
      "learning_rate": 2.8871821921510995e-06,
      "loss": 0.4381,
      "step": 5263
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.924651436471515,
      "learning_rate": 2.886472275098152e-06,
      "loss": 0.24,
      "step": 5264
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5241212879989874,
      "learning_rate": 2.8857623261167613e-06,
      "loss": 0.2396,
      "step": 5265
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.190349685705682,
      "learning_rate": 2.885052345265578e-06,
      "loss": 0.4177,
      "step": 5266
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.7550169545879575,
      "learning_rate": 2.884342332603257e-06,
      "loss": 0.1036,
      "step": 5267
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.2547245206179936,
      "learning_rate": 2.8836322881884572e-06,
      "loss": 0.3908,
      "step": 5268
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.0624510048617415,
      "learning_rate": 2.8829222120798384e-06,
      "loss": 0.386,
      "step": 5269
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.1494169063296855,
      "learning_rate": 2.8822121043360645e-06,
      "loss": 0.2893,
      "step": 5270
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.957034582724144,
      "learning_rate": 2.8815019650158e-06,
      "loss": 0.356,
      "step": 5271
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.187432963888149,
      "learning_rate": 2.880791794177713e-06,
      "loss": 0.3339,
      "step": 5272
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.463367603767126,
      "learning_rate": 2.8800815918804755e-06,
      "loss": 0.3175,
      "step": 5273
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.068163829436656,
      "learning_rate": 2.8793713581827605e-06,
      "loss": 0.217,
      "step": 5274
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.216099996241864,
      "learning_rate": 2.8786610931432433e-06,
      "loss": 0.2995,
      "step": 5275
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.70597700140972,
      "learning_rate": 2.877950796820604e-06,
      "loss": 0.3479,
      "step": 5276
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.285395133350757,
      "learning_rate": 2.8772404692735217e-06,
      "loss": 0.3058,
      "step": 5277
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.447030882128726,
      "learning_rate": 2.8765301105606812e-06,
      "loss": 0.2539,
      "step": 5278
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.61418839939288,
      "learning_rate": 2.87581972074077e-06,
      "loss": 0.3857,
      "step": 5279
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.0905297433580663,
      "learning_rate": 2.8751092998724763e-06,
      "loss": 0.2569,
      "step": 5280
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.9405129899554643,
      "learning_rate": 2.8743988480144906e-06,
      "loss": 0.3486,
      "step": 5281
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.231926943340344,
      "learning_rate": 2.873688365225508e-06,
      "loss": 0.4892,
      "step": 5282
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.738083905033917,
      "learning_rate": 2.8729778515642235e-06,
      "loss": 0.5209,
      "step": 5283
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.229415500113866,
      "learning_rate": 2.872267307089339e-06,
      "loss": 0.4605,
      "step": 5284
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.508732129829379,
      "learning_rate": 2.8715567318595542e-06,
      "loss": 0.3561,
      "step": 5285
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.2402152379722997,
      "learning_rate": 2.870846125933574e-06,
      "loss": 0.2564,
      "step": 5286
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.081638303901507,
      "learning_rate": 2.870135489370105e-06,
      "loss": 0.3522,
      "step": 5287
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.748314134558888,
      "learning_rate": 2.8694248222278557e-06,
      "loss": 0.2977,
      "step": 5288
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.7664152315468327,
      "learning_rate": 2.86871412456554e-06,
      "loss": 0.1827,
      "step": 5289
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.961542868704998,
      "learning_rate": 2.8680033964418708e-06,
      "loss": 0.4061,
      "step": 5290
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5735170078342255,
      "learning_rate": 2.867292637915565e-06,
      "loss": 0.1883,
      "step": 5291
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.550913958471067,
      "learning_rate": 2.866581849045341e-06,
      "loss": 0.2147,
      "step": 5292
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.590292945386645,
      "learning_rate": 2.8658710298899235e-06,
      "loss": 0.1026,
      "step": 5293
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.7834168858419743,
      "learning_rate": 2.865160180508035e-06,
      "loss": 0.3794,
      "step": 5294
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.961123121132617,
      "learning_rate": 2.8644493009584027e-06,
      "loss": 0.2385,
      "step": 5295
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.315202088615803,
      "learning_rate": 2.8637383912997563e-06,
      "loss": 0.3811,
      "step": 5296
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.25135557055097,
      "learning_rate": 2.863027451590826e-06,
      "loss": 0.3757,
      "step": 5297
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7969272312499573,
      "learning_rate": 2.862316481890348e-06,
      "loss": 0.0586,
      "step": 5298
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.118574230170169,
      "learning_rate": 2.8616054822570593e-06,
      "loss": 0.4144,
      "step": 5299
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.093345932375062,
      "learning_rate": 2.8608944527496985e-06,
      "loss": 0.2102,
      "step": 5300
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.4787883962317965,
      "learning_rate": 2.8601833934270066e-06,
      "loss": 0.2511,
      "step": 5301
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.443291108761145,
      "learning_rate": 2.8594723043477284e-06,
      "loss": 0.1612,
      "step": 5302
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.732314740521956,
      "learning_rate": 2.8587611855706115e-06,
      "loss": 0.3197,
      "step": 5303
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.779088729618239,
      "learning_rate": 2.8580500371544045e-06,
      "loss": 0.3835,
      "step": 5304
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.913519268299445,
      "learning_rate": 2.8573388591578588e-06,
      "loss": 0.3851,
      "step": 5305
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.640109170667809,
      "learning_rate": 2.8566276516397283e-06,
      "loss": 0.2935,
      "step": 5306
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.0412339353562325,
      "learning_rate": 2.8559164146587694e-06,
      "loss": 0.2351,
      "step": 5307
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.741435508530229,
      "learning_rate": 2.8552051482737414e-06,
      "loss": 0.2105,
      "step": 5308
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.4387453374594275,
      "learning_rate": 2.8544938525434065e-06,
      "loss": 0.3235,
      "step": 5309
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.39948684787703,
      "learning_rate": 2.853782527526526e-06,
      "loss": 0.2035,
      "step": 5310
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.8162061205712865,
      "learning_rate": 2.853071173281868e-06,
      "loss": 0.1991,
      "step": 5311
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.4752970779343957,
      "learning_rate": 2.852359789868201e-06,
      "loss": 0.3288,
      "step": 5312
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.986869671128744,
      "learning_rate": 2.851648377344296e-06,
      "loss": 0.2181,
      "step": 5313
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.3723419148941853,
      "learning_rate": 2.850936935768926e-06,
      "loss": 0.2301,
      "step": 5314
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.963810308948555,
      "learning_rate": 2.8502254652008664e-06,
      "loss": 0.3647,
      "step": 5315
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.541111865168709,
      "learning_rate": 2.8495139656988963e-06,
      "loss": 0.1142,
      "step": 5316
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.499228172559946,
      "learning_rate": 2.8488024373217954e-06,
      "loss": 0.3269,
      "step": 5317
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.794613722335876,
      "learning_rate": 2.848090880128348e-06,
      "loss": 0.3147,
      "step": 5318
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.847153479185968,
      "learning_rate": 2.8473792941773383e-06,
      "loss": 0.523,
      "step": 5319
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.439238758830093,
      "learning_rate": 2.8466676795275545e-06,
      "loss": 0.4246,
      "step": 5320
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.683459293878377,
      "learning_rate": 2.8459560362377864e-06,
      "loss": 0.1619,
      "step": 5321
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.011690382087715,
      "learning_rate": 2.8452443643668266e-06,
      "loss": 0.2715,
      "step": 5322
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.089752893456432,
      "learning_rate": 2.8445326639734704e-06,
      "loss": 0.4574,
      "step": 5323
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.94160479233305,
      "learning_rate": 2.843820935116515e-06,
      "loss": 0.2117,
      "step": 5324
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.324341715706382,
      "learning_rate": 2.8431091778547592e-06,
      "loss": 0.3939,
      "step": 5325
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.9476979709466953,
      "learning_rate": 2.842397392247005e-06,
      "loss": 0.4428,
      "step": 5326
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.8736779267678165,
      "learning_rate": 2.8416855783520568e-06,
      "loss": 0.2201,
      "step": 5327
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.709846058071825,
      "learning_rate": 2.840973736228722e-06,
      "loss": 0.371,
      "step": 5328
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.49536562817875,
      "learning_rate": 2.840261865935809e-06,
      "loss": 0.3186,
      "step": 5329
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.4104493982681503,
      "learning_rate": 2.839549967532128e-06,
      "loss": 0.1743,
      "step": 5330
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.407194214863511,
      "learning_rate": 2.838838041076494e-06,
      "loss": 0.2645,
      "step": 5331
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.385313379085445,
      "learning_rate": 2.8381260866277217e-06,
      "loss": 0.4525,
      "step": 5332
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.452906126635241,
      "learning_rate": 2.837414104244631e-06,
      "loss": 0.2626,
      "step": 5333
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.599041703254827,
      "learning_rate": 2.83670209398604e-06,
      "loss": 0.3102,
      "step": 5334
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.890431022069335,
      "learning_rate": 2.8359900559107735e-06,
      "loss": 0.3721,
      "step": 5335
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.7433115755780486,
      "learning_rate": 2.8352779900776557e-06,
      "loss": 0.3169,
      "step": 5336
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.7744444045572845,
      "learning_rate": 2.8345658965455137e-06,
      "loss": 0.2291,
      "step": 5337
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.8272048586526606,
      "learning_rate": 2.8338537753731784e-06,
      "loss": 0.1531,
      "step": 5338
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.206935257914981,
      "learning_rate": 2.8331416266194806e-06,
      "loss": 0.3267,
      "step": 5339
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.185525870890649,
      "learning_rate": 2.8324294503432543e-06,
      "loss": 0.1253,
      "step": 5340
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.583110930682741,
      "learning_rate": 2.8317172466033378e-06,
      "loss": 0.2305,
      "step": 5341
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.932019934978707,
      "learning_rate": 2.8310050154585673e-06,
      "loss": 0.2521,
      "step": 5342
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.369313225254441,
      "learning_rate": 2.830292756967786e-06,
      "loss": 0.2897,
      "step": 5343
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.919933631697959,
      "learning_rate": 2.8295804711898374e-06,
      "loss": 0.163,
      "step": 5344
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.1266579482753722,
      "learning_rate": 2.8288681581835648e-06,
      "loss": 0.2061,
      "step": 5345
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.243426419064427,
      "learning_rate": 2.8281558180078176e-06,
      "loss": 0.5128,
      "step": 5346
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.788791364952855,
      "learning_rate": 2.8274434507214456e-06,
      "loss": 0.15,
      "step": 5347
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.8189222500232476,
      "learning_rate": 2.826731056383302e-06,
      "loss": 0.1688,
      "step": 5348
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.817954455469976,
      "learning_rate": 2.826018635052239e-06,
      "loss": 0.2757,
      "step": 5349
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.2027607742473725,
      "learning_rate": 2.825306186787115e-06,
      "loss": 0.5584,
      "step": 5350
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.053656586019003,
      "learning_rate": 2.82459371164679e-06,
      "loss": 0.2444,
      "step": 5351
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.956019077254517,
      "learning_rate": 2.8238812096901223e-06,
      "loss": 0.0794,
      "step": 5352
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.958610293133119,
      "learning_rate": 2.823168680975978e-06,
      "loss": 0.3339,
      "step": 5353
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.039972059902961,
      "learning_rate": 2.822456125563221e-06,
      "loss": 0.329,
      "step": 5354
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3894673691669883,
      "learning_rate": 2.8217435435107203e-06,
      "loss": 0.2364,
      "step": 5355
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.694233875672846,
      "learning_rate": 2.8210309348773457e-06,
      "loss": 0.2934,
      "step": 5356
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3759637834485825,
      "learning_rate": 2.8203182997219684e-06,
      "loss": 0.1983,
      "step": 5357
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.6847255355015327,
      "learning_rate": 2.8196056381034642e-06,
      "loss": 0.1838,
      "step": 5358
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.538407532542956,
      "learning_rate": 2.818892950080709e-06,
      "loss": 0.2493,
      "step": 5359
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.595225394966203,
      "learning_rate": 2.818180235712582e-06,
      "loss": 0.2866,
      "step": 5360
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.5952460148281173,
      "learning_rate": 2.817467495057964e-06,
      "loss": 0.0846,
      "step": 5361
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.542131831329889,
      "learning_rate": 2.816754728175737e-06,
      "loss": 0.1454,
      "step": 5362
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1156340539106795,
      "learning_rate": 2.8160419351247883e-06,
      "loss": 0.2206,
      "step": 5363
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.699636752251585,
      "learning_rate": 2.8153291159640033e-06,
      "loss": 0.3195,
      "step": 5364
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.3885618200611995,
      "learning_rate": 2.8146162707522737e-06,
      "loss": 0.3613,
      "step": 5365
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2426938613207454,
      "learning_rate": 2.8139033995484903e-06,
      "loss": 0.2638,
      "step": 5366
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.5597151899617443,
      "learning_rate": 2.8131905024115463e-06,
      "loss": 0.232,
      "step": 5367
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.6496447215830097,
      "learning_rate": 2.812477579400339e-06,
      "loss": 0.2283,
      "step": 5368
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.9172580754968536,
      "learning_rate": 2.8117646305737663e-06,
      "loss": 0.3167,
      "step": 5369
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.399277388163713,
      "learning_rate": 2.811051655990728e-06,
      "loss": 0.6155,
      "step": 5370
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.868128332899077,
      "learning_rate": 2.810338655710127e-06,
      "loss": 0.2786,
      "step": 5371
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.011146077053296,
      "learning_rate": 2.8096256297908673e-06,
      "loss": 0.4428,
      "step": 5372
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.028601649929284,
      "learning_rate": 2.8089125782918562e-06,
      "loss": 0.2749,
      "step": 5373
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1613146796980915,
      "learning_rate": 2.8081995012720025e-06,
      "loss": 0.059,
      "step": 5374
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.905216404446313,
      "learning_rate": 2.807486398790218e-06,
      "loss": 0.4609,
      "step": 5375
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.484588497167329,
      "learning_rate": 2.806773270905414e-06,
      "loss": 0.3033,
      "step": 5376
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8808295492759943,
      "learning_rate": 2.8060601176765055e-06,
      "loss": 0.0729,
      "step": 5377
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.4016005444859365,
      "learning_rate": 2.8053469391624116e-06,
      "loss": 0.2766,
      "step": 5378
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.299238287890895,
      "learning_rate": 2.8046337354220506e-06,
      "loss": 0.4576,
      "step": 5379
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.371708383535898,
      "learning_rate": 2.803920506514344e-06,
      "loss": 0.5131,
      "step": 5380
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0927288682039515,
      "learning_rate": 2.803207252498215e-06,
      "loss": 0.2014,
      "step": 5381
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.19011139849042,
      "learning_rate": 2.802493973432589e-06,
      "loss": 0.1948,
      "step": 5382
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3420741080101144,
      "learning_rate": 2.801780669376394e-06,
      "loss": 0.2717,
      "step": 5383
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.881469511327119,
      "learning_rate": 2.80106734038856e-06,
      "loss": 0.2138,
      "step": 5384
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3489266097369064,
      "learning_rate": 2.800353986528019e-06,
      "loss": 0.255,
      "step": 5385
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.221291531694487,
      "learning_rate": 2.7996406078537036e-06,
      "loss": 0.437,
      "step": 5386
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7012633961950483,
      "learning_rate": 2.79892720442455e-06,
      "loss": 0.2342,
      "step": 5387
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.576841315363568,
      "learning_rate": 2.798213776299496e-06,
      "loss": 0.1427,
      "step": 5388
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.206776096865992,
      "learning_rate": 2.7975003235374825e-06,
      "loss": 0.1788,
      "step": 5389
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.232858908694417,
      "learning_rate": 2.7967868461974506e-06,
      "loss": 0.3142,
      "step": 5390
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.9213144083047173,
      "learning_rate": 2.796073344338345e-06,
      "loss": 0.2262,
      "step": 5391
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.623284004936178,
      "learning_rate": 2.7953598180191104e-06,
      "loss": 0.0988,
      "step": 5392
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.226085767526318,
      "learning_rate": 2.794646267298696e-06,
      "loss": 0.3658,
      "step": 5393
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.860740881359353,
      "learning_rate": 2.793932692236052e-06,
      "loss": 0.2596,
      "step": 5394
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.876403166729081,
      "learning_rate": 2.79321909289013e-06,
      "loss": 0.2419,
      "step": 5395
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.9612941319497503,
      "learning_rate": 2.792505469319884e-06,
      "loss": 0.2516,
      "step": 5396
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.016602692878717,
      "learning_rate": 2.7917918215842693e-06,
      "loss": 0.2835,
      "step": 5397
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.085333944983378,
      "learning_rate": 2.7910781497422457e-06,
      "loss": 0.4463,
      "step": 5398
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.5119236406002985,
      "learning_rate": 2.790364453852773e-06,
      "loss": 0.4143,
      "step": 5399
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.0117999417892864,
      "learning_rate": 2.7896507339748126e-06,
      "loss": 0.2456,
      "step": 5400
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.7740996779339704,
      "learning_rate": 2.788936990167328e-06,
      "loss": 0.0638,
      "step": 5401
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.033882100927613,
      "learning_rate": 2.788223222489285e-06,
      "loss": 0.5341,
      "step": 5402
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.292508638273151,
      "learning_rate": 2.7875094309996536e-06,
      "loss": 0.4038,
      "step": 5403
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.84909926115097,
      "learning_rate": 2.786795615757403e-06,
      "loss": 0.3455,
      "step": 5404
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.945750517425135,
      "learning_rate": 2.786081776821504e-06,
      "loss": 0.3507,
      "step": 5405
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.4490822078541927,
      "learning_rate": 2.7853679142509317e-06,
      "loss": 0.2622,
      "step": 5406
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.9763502128667207,
      "learning_rate": 2.78465402810466e-06,
      "loss": 0.1146,
      "step": 5407
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.346804166123277,
      "learning_rate": 2.78394011844167e-06,
      "loss": 0.3966,
      "step": 5408
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.640968783590945,
      "learning_rate": 2.7832261853209386e-06,
      "loss": 0.3212,
      "step": 5409
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.4441219151963405,
      "learning_rate": 2.782512228801449e-06,
      "loss": 0.2993,
      "step": 5410
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.605671192537662,
      "learning_rate": 2.781798248942183e-06,
      "loss": 0.1847,
      "step": 5411
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.689562206709543,
      "learning_rate": 2.7810842458021273e-06,
      "loss": 0.1814,
      "step": 5412
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.050997178359798,
      "learning_rate": 2.7803702194402697e-06,
      "loss": 0.328,
      "step": 5413
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.864670666474286,
      "learning_rate": 2.7796561699156e-06,
      "loss": 0.5449,
      "step": 5414
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.848756931596075,
      "learning_rate": 2.7789420972871077e-06,
      "loss": 0.4806,
      "step": 5415
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.279901817554039,
      "learning_rate": 2.778228001613787e-06,
      "loss": 0.3361,
      "step": 5416
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.362998126229517,
      "learning_rate": 2.7775138829546326e-06,
      "loss": 0.3732,
      "step": 5417
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.4502772599589417,
      "learning_rate": 2.776799741368642e-06,
      "loss": 0.1782,
      "step": 5418
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.9856273506644166,
      "learning_rate": 2.7760855769148144e-06,
      "loss": 0.3751,
      "step": 5419
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.238133776501825,
      "learning_rate": 2.77537138965215e-06,
      "loss": 0.3704,
      "step": 5420
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3542440672381217,
      "learning_rate": 2.7746571796396506e-06,
      "loss": 0.2709,
      "step": 5421
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.0446774798898755,
      "learning_rate": 2.7739429469363216e-06,
      "loss": 0.1519,
      "step": 5422
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.935367344246504,
      "learning_rate": 2.77322869160117e-06,
      "loss": 0.3086,
      "step": 5423
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.788787874534199,
      "learning_rate": 2.7725144136932035e-06,
      "loss": 0.3046,
      "step": 5424
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3389129318524531,
      "learning_rate": 2.7718001132714323e-06,
      "loss": 0.0049,
      "step": 5425
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.16595055035606,
      "learning_rate": 2.7710857903948672e-06,
      "loss": 0.3826,
      "step": 5426
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.437670793999537,
      "learning_rate": 2.770371445122524e-06,
      "loss": 0.4192,
      "step": 5427
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0129571724082864,
      "learning_rate": 2.769657077513418e-06,
      "loss": 0.0886,
      "step": 5428
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4527899674076465,
      "learning_rate": 2.7689426876265658e-06,
      "loss": 0.1442,
      "step": 5429
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.261524548836526,
      "learning_rate": 2.7682282755209873e-06,
      "loss": 0.3588,
      "step": 5430
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.017360559904428,
      "learning_rate": 2.7675138412557035e-06,
      "loss": 0.3853,
      "step": 5431
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.2870674313226202,
      "learning_rate": 2.766799384889738e-06,
      "loss": 0.1974,
      "step": 5432
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.095037144751365,
      "learning_rate": 2.7660849064821157e-06,
      "loss": 0.1191,
      "step": 5433
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8224960358475224,
      "learning_rate": 2.7653704060918622e-06,
      "loss": 0.1887,
      "step": 5434
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.98898944764236,
      "learning_rate": 2.7646558837780067e-06,
      "loss": 0.0749,
      "step": 5435
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.389408879685675,
      "learning_rate": 2.763941339599581e-06,
      "loss": 0.3643,
      "step": 5436
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.523625562153334,
      "learning_rate": 2.7632267736156143e-06,
      "loss": 0.2977,
      "step": 5437
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.730117972415884,
      "learning_rate": 2.762512185885143e-06,
      "loss": 0.2071,
      "step": 5438
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.387772811932842,
      "learning_rate": 2.761797576467202e-06,
      "loss": 0.4312,
      "step": 5439
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.643912953989562,
      "learning_rate": 2.7610829454208284e-06,
      "loss": 0.3356,
      "step": 5440
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.477611832673612,
      "learning_rate": 2.7603682928050617e-06,
      "loss": 0.3736,
      "step": 5441
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.5783543275999596,
      "learning_rate": 2.759653618678943e-06,
      "loss": 0.2793,
      "step": 5442
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.552496086645674,
      "learning_rate": 2.7589389231015163e-06,
      "loss": 0.2173,
      "step": 5443
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.362448673380242,
      "learning_rate": 2.758224206131824e-06,
      "loss": 0.3565,
      "step": 5444
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.3333220238987256,
      "learning_rate": 2.757509467828915e-06,
      "loss": 0.2579,
      "step": 5445
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.5147543957488794,
      "learning_rate": 2.756794708251836e-06,
      "loss": 0.2566,
      "step": 5446
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.7324558341718435,
      "learning_rate": 2.756079927459636e-06,
      "loss": 0.2094,
      "step": 5447
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.376746227317071,
      "learning_rate": 2.7553651255113694e-06,
      "loss": 0.2396,
      "step": 5448
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.091078747101839,
      "learning_rate": 2.7546503024660875e-06,
      "loss": 0.2376,
      "step": 5449
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.722134265324946,
      "learning_rate": 2.753935458382846e-06,
      "loss": 0.2249,
      "step": 5450
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.8109197336904383,
      "learning_rate": 2.753220593320702e-06,
      "loss": 0.2728,
      "step": 5451
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.45368266193704,
      "learning_rate": 2.752505707338714e-06,
      "loss": 0.3743,
      "step": 5452
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.7414494003696626,
      "learning_rate": 2.7517908004959426e-06,
      "loss": 0.2757,
      "step": 5453
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.0607235204039815,
      "learning_rate": 2.7510758728514497e-06,
      "loss": 0.4806,
      "step": 5454
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.66253484160587,
      "learning_rate": 2.750360924464299e-06,
      "loss": 0.3003,
      "step": 5455
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.606261292622137,
      "learning_rate": 2.7496459553935568e-06,
      "loss": 0.5439,
      "step": 5456
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.105872585558025,
      "learning_rate": 2.748930965698289e-06,
      "loss": 0.2583,
      "step": 5457
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.399186970664956,
      "learning_rate": 2.7482159554375655e-06,
      "loss": 0.344,
      "step": 5458
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.664596649683045,
      "learning_rate": 2.7475009246704566e-06,
      "loss": 0.4468,
      "step": 5459
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.245927963943239,
      "learning_rate": 2.746785873456035e-06,
      "loss": 0.404,
      "step": 5460
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.810677029053138,
      "learning_rate": 2.7460708018533744e-06,
      "loss": 0.3483,
      "step": 5461
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.785582339604259,
      "learning_rate": 2.74535570992155e-06,
      "loss": 0.1265,
      "step": 5462
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.246139612845579,
      "learning_rate": 2.74464059771964e-06,
      "loss": 0.4061,
      "step": 5463
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.593862851830096,
      "learning_rate": 2.7439254653067234e-06,
      "loss": 0.3042,
      "step": 5464
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.4700804497939317,
      "learning_rate": 2.7432103127418806e-06,
      "loss": 0.1205,
      "step": 5465
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.214938460663912,
      "learning_rate": 2.742495140084195e-06,
      "loss": 0.1587,
      "step": 5466
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7700872732912196,
      "learning_rate": 2.741779947392749e-06,
      "loss": 0.3348,
      "step": 5467
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7806222428989633,
      "learning_rate": 2.741064734726629e-06,
      "loss": 0.163,
      "step": 5468
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.17401140403697,
      "learning_rate": 2.7403495021449227e-06,
      "loss": 0.2486,
      "step": 5469
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.5236994255797827,
      "learning_rate": 2.739634249706719e-06,
      "loss": 0.141,
      "step": 5470
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.135528333461807,
      "learning_rate": 2.738918977471109e-06,
      "loss": 0.4327,
      "step": 5471
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.201349759676737,
      "learning_rate": 2.7382036854971834e-06,
      "loss": 0.4881,
      "step": 5472
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.9981369412795416,
      "learning_rate": 2.737488373844037e-06,
      "loss": 0.3407,
      "step": 5473
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.239026134339954,
      "learning_rate": 2.736773042570766e-06,
      "loss": 0.5361,
      "step": 5474
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.870659767823704,
      "learning_rate": 2.736057691736468e-06,
      "loss": 0.5332,
      "step": 5475
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.497632823784296,
      "learning_rate": 2.73534232140024e-06,
      "loss": 0.4626,
      "step": 5476
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.009874645208944,
      "learning_rate": 2.7346269316211826e-06,
      "loss": 0.2945,
      "step": 5477
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.456960663620133,
      "learning_rate": 2.733911522458399e-06,
      "loss": 0.2669,
      "step": 5478
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.9888540711179457,
      "learning_rate": 2.7331960939709925e-06,
      "loss": 0.1722,
      "step": 5479
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.407188643119457,
      "learning_rate": 2.732480646218068e-06,
      "loss": 0.4105,
      "step": 5480
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.6914591066564415,
      "learning_rate": 2.7317651792587324e-06,
      "loss": 0.1761,
      "step": 5481
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.674286038878745,
      "learning_rate": 2.7310496931520936e-06,
      "loss": 0.2255,
      "step": 5482
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.575753427809119,
      "learning_rate": 2.7303341879572615e-06,
      "loss": 0.3716,
      "step": 5483
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.9772997575522635,
      "learning_rate": 2.729618663733349e-06,
      "loss": 0.1818,
      "step": 5484
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.351102564404861,
      "learning_rate": 2.728903120539468e-06,
      "loss": 0.3935,
      "step": 5485
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.3399969527793885,
      "learning_rate": 2.728187558434733e-06,
      "loss": 0.2326,
      "step": 5486
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.396383969325271,
      "learning_rate": 2.7274719774782602e-06,
      "loss": 0.1647,
      "step": 5487
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.467420990877768,
      "learning_rate": 2.726756377729169e-06,
      "loss": 0.4876,
      "step": 5488
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.969862234819394,
      "learning_rate": 2.726040759246577e-06,
      "loss": 0.3058,
      "step": 5489
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.6498881979450846,
      "learning_rate": 2.725325122089606e-06,
      "loss": 0.2943,
      "step": 5490
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.020850429018855,
      "learning_rate": 2.724609466317378e-06,
      "loss": 0.2397,
      "step": 5491
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.348889742446438,
      "learning_rate": 2.7238937919890167e-06,
      "loss": 0.3461,
      "step": 5492
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.224488716652352,
      "learning_rate": 2.7231780991636488e-06,
      "loss": 0.2583,
      "step": 5493
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.4018513602682146,
      "learning_rate": 2.7224623879003996e-06,
      "loss": 0.0833,
      "step": 5494
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.260471142333894,
      "learning_rate": 2.7217466582583994e-06,
      "loss": 0.177,
      "step": 5495
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.11471394059707,
      "learning_rate": 2.7210309102967774e-06,
      "loss": 0.3496,
      "step": 5496
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.573557216789675,
      "learning_rate": 2.7203151440746646e-06,
      "loss": 0.3642,
      "step": 5497
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.927651313125404,
      "learning_rate": 2.7195993596511948e-06,
      "loss": 0.251,
      "step": 5498
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.584809774195559,
      "learning_rate": 2.718883557085503e-06,
      "loss": 0.2193,
      "step": 5499
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.537729229275442,
      "learning_rate": 2.7181677364367255e-06,
      "loss": 0.2059,
      "step": 5500
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.05838432699296,
      "learning_rate": 2.7174518977639984e-06,
      "loss": 0.4117,
      "step": 5501
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.525913986171137,
      "learning_rate": 2.716736041126461e-06,
      "loss": 0.2858,
      "step": 5502
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.55141964785059,
      "learning_rate": 2.7160201665832555e-06,
      "loss": 0.3424,
      "step": 5503
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7325339551243495,
      "learning_rate": 2.7153042741935226e-06,
      "loss": 0.1501,
      "step": 5504
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.068140639245917,
      "learning_rate": 2.714588364016406e-06,
      "loss": 0.3034,
      "step": 5505
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.894945156085572,
      "learning_rate": 2.7138724361110515e-06,
      "loss": 0.3583,
      "step": 5506
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.172437837489053,
      "learning_rate": 2.713156490536603e-06,
      "loss": 0.3769,
      "step": 5507
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.1957292942694258,
      "learning_rate": 2.7124405273522118e-06,
      "loss": 0.252,
      "step": 5508
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.478056399884968,
      "learning_rate": 2.711724546617026e-06,
      "loss": 0.3371,
      "step": 5509
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.1360031487032236,
      "learning_rate": 2.711008548390196e-06,
      "loss": 0.1864,
      "step": 5510
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.392856367417845,
      "learning_rate": 2.710292532730874e-06,
      "loss": 0.4675,
      "step": 5511
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.516310708531004,
      "learning_rate": 2.7095764996982138e-06,
      "loss": 0.2551,
      "step": 5512
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.189321186596893,
      "learning_rate": 2.708860449351371e-06,
      "loss": 0.2898,
      "step": 5513
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.34122278023624,
      "learning_rate": 2.7081443817495017e-06,
      "loss": 0.3454,
      "step": 5514
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.880056036747218,
      "learning_rate": 2.7074282969517646e-06,
      "loss": 0.232,
      "step": 5515
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.055988881799539,
      "learning_rate": 2.706712195017318e-06,
      "loss": 0.2832,
      "step": 5516
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.8478524110732573,
      "learning_rate": 2.705996076005323e-06,
      "loss": 0.0872,
      "step": 5517
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.629880396862908,
      "learning_rate": 2.7052799399749425e-06,
      "loss": 0.4743,
      "step": 5518
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.089564367670974,
      "learning_rate": 2.7045637869853403e-06,
      "loss": 0.5235,
      "step": 5519
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.6871356327724936,
      "learning_rate": 2.7038476170956805e-06,
      "loss": 0.2076,
      "step": 5520
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.334659032670532,
      "learning_rate": 2.7031314303651295e-06,
      "loss": 0.0913,
      "step": 5521
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.575062304432275,
      "learning_rate": 2.702415226852856e-06,
      "loss": 0.1699,
      "step": 5522
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.930611430109337,
      "learning_rate": 2.7016990066180294e-06,
      "loss": 0.3981,
      "step": 5523
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.400708277211389,
      "learning_rate": 2.7009827697198193e-06,
      "loss": 0.3673,
      "step": 5524
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.0762512803829276,
      "learning_rate": 2.700266516217398e-06,
      "loss": 0.148,
      "step": 5525
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.0141048948092015,
      "learning_rate": 2.699550246169939e-06,
      "loss": 0.0865,
      "step": 5526
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.393491348104203,
      "learning_rate": 2.6988339596366168e-06,
      "loss": 0.333,
      "step": 5527
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.973520020791531,
      "learning_rate": 2.6981176566766084e-06,
      "loss": 0.7298,
      "step": 5528
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.83241717538489,
      "learning_rate": 2.6974013373490905e-06,
      "loss": 0.327,
      "step": 5529
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.962132163640942,
      "learning_rate": 2.696685001713241e-06,
      "loss": 0.1942,
      "step": 5530
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.246986389717912,
      "learning_rate": 2.6959686498282417e-06,
      "loss": 0.3287,
      "step": 5531
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.660053455934782,
      "learning_rate": 2.6952522817532726e-06,
      "loss": 0.3882,
      "step": 5532
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7805493000429666,
      "learning_rate": 2.6945358975475177e-06,
      "loss": 0.3493,
      "step": 5533
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.0998549365687555,
      "learning_rate": 2.6938194972701614e-06,
      "loss": 0.1583,
      "step": 5534
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.810531669787238,
      "learning_rate": 2.693103080980387e-06,
      "loss": 0.3519,
      "step": 5535
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.06755321737745,
      "learning_rate": 2.692386648737384e-06,
      "loss": 0.2559,
      "step": 5536
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.975438327157379,
      "learning_rate": 2.6916702006003387e-06,
      "loss": 0.2849,
      "step": 5537
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.020169612870699,
      "learning_rate": 2.6909537366284417e-06,
      "loss": 0.1556,
      "step": 5538
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.030495337193037,
      "learning_rate": 2.690237256880883e-06,
      "loss": 0.2668,
      "step": 5539
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.236862193178683,
      "learning_rate": 2.6895207614168546e-06,
      "loss": 0.2521,
      "step": 5540
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.984383043530392,
      "learning_rate": 2.68880425029555e-06,
      "loss": 0.4043,
      "step": 5541
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.8471559256624053,
      "learning_rate": 2.688087723576164e-06,
      "loss": 0.3018,
      "step": 5542
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.598101381391336,
      "learning_rate": 2.6873711813178926e-06,
      "loss": 0.4075,
      "step": 5543
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.716283336567035,
      "learning_rate": 2.686654623579933e-06,
      "loss": 0.1952,
      "step": 5544
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.183080717164125,
      "learning_rate": 2.685938050421483e-06,
      "loss": 0.2613,
      "step": 5545
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.435327438018332,
      "learning_rate": 2.685221461901743e-06,
      "loss": 0.488,
      "step": 5546
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.492460995736797,
      "learning_rate": 2.6845048580799137e-06,
      "loss": 0.1381,
      "step": 5547
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.8677288925828575,
      "learning_rate": 2.683788239015198e-06,
      "loss": 0.1769,
      "step": 5548
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.7426056992062713,
      "learning_rate": 2.683071604766799e-06,
      "loss": 0.2693,
      "step": 5549
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.6685233318717794,
      "learning_rate": 2.6823549553939217e-06,
      "loss": 0.1661,
      "step": 5550
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.251380157407079,
      "learning_rate": 2.6816382909557713e-06,
      "loss": 0.243,
      "step": 5551
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.439662833666999,
      "learning_rate": 2.6809216115115564e-06,
      "loss": 0.3877,
      "step": 5552
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.321719453661745,
      "learning_rate": 2.680204917120485e-06,
      "loss": 0.2703,
      "step": 5553
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.221446853503203,
      "learning_rate": 2.6794882078417665e-06,
      "loss": 0.2969,
      "step": 5554
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.856839054720304,
      "learning_rate": 2.678771483734613e-06,
      "loss": 0.4433,
      "step": 5555
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.605334286572431,
      "learning_rate": 2.6780547448582358e-06,
      "loss": 0.2743,
      "step": 5556
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.297409830949562,
      "learning_rate": 2.6773379912718476e-06,
      "loss": 0.4891,
      "step": 5557
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.478431889854753,
      "learning_rate": 2.676621223034665e-06,
      "loss": 0.2744,
      "step": 5558
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.104245558959532,
      "learning_rate": 2.675904440205902e-06,
      "loss": 0.3044,
      "step": 5559
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.071613140160937,
      "learning_rate": 2.675187642844778e-06,
      "loss": 0.55,
      "step": 5560
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0586740148569949,
      "learning_rate": 2.674470831010509e-06,
      "loss": 0.0145,
      "step": 5561
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0982266249396857,
      "learning_rate": 2.6737540047623146e-06,
      "loss": 0.073,
      "step": 5562
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.1862222852149973,
      "learning_rate": 2.6730371641594173e-06,
      "loss": 0.2535,
      "step": 5563
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.219151260957121,
      "learning_rate": 2.6723203092610366e-06,
      "loss": 0.3677,
      "step": 5564
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.359415313027921,
      "learning_rate": 2.671603440126398e-06,
      "loss": 0.1517,
      "step": 5565
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.324046116344892,
      "learning_rate": 2.6708865568147246e-06,
      "loss": 0.3933,
      "step": 5566
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.0726767617776214,
      "learning_rate": 2.6701696593852403e-06,
      "loss": 0.1389,
      "step": 5567
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.006172648281495,
      "learning_rate": 2.669452747897174e-06,
      "loss": 0.2103,
      "step": 5568
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.594372658905089,
      "learning_rate": 2.6687358224097527e-06,
      "loss": 0.3103,
      "step": 5569
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.1282783795892115,
      "learning_rate": 2.6680188829822044e-06,
      "loss": 0.149,
      "step": 5570
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.211624458436529,
      "learning_rate": 2.6673019296737605e-06,
      "loss": 0.2628,
      "step": 5571
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.450280744606703,
      "learning_rate": 2.6665849625436506e-06,
      "loss": 0.3325,
      "step": 5572
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.8981546491672443,
      "learning_rate": 2.665867981651108e-06,
      "loss": 0.2018,
      "step": 5573
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.347848773012019,
      "learning_rate": 2.665150987055366e-06,
      "loss": 0.3231,
      "step": 5574
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.368405904427422,
      "learning_rate": 2.6644339788156598e-06,
      "loss": 0.4456,
      "step": 5575
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.099265133140624,
      "learning_rate": 2.6637169569912243e-06,
      "loss": 0.3622,
      "step": 5576
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.524994960259014,
      "learning_rate": 2.6629999216412956e-06,
      "loss": 0.169,
      "step": 5577
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.341867921933985,
      "learning_rate": 2.6622828728251126e-06,
      "loss": 0.2676,
      "step": 5578
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.905067686059973,
      "learning_rate": 2.661565810601915e-06,
      "loss": 0.4431,
      "step": 5579
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.00044905760312,
      "learning_rate": 2.660848735030942e-06,
      "loss": 0.206,
      "step": 5580
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.623123645493358,
      "learning_rate": 2.6601316461714356e-06,
      "loss": 0.2677,
      "step": 5581
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.4862930472171065,
      "learning_rate": 2.6594145440826363e-06,
      "loss": 0.2538,
      "step": 5582
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.637886053410032,
      "learning_rate": 2.65869742882379e-06,
      "loss": 0.425,
      "step": 5583
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.2275259679402133,
      "learning_rate": 2.65798030045414e-06,
      "loss": 0.2441,
      "step": 5584
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.199722907235807,
      "learning_rate": 2.6572631590329325e-06,
      "loss": 0.2237,
      "step": 5585
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.978934798414051,
      "learning_rate": 2.656546004619413e-06,
      "loss": 0.4906,
      "step": 5586
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.324965825375047,
      "learning_rate": 2.65582883727283e-06,
      "loss": 0.2599,
      "step": 5587
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.724484051242074,
      "learning_rate": 2.6551116570524326e-06,
      "loss": 0.4011,
      "step": 5588
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.218975470231098,
      "learning_rate": 2.654394464017471e-06,
      "loss": 0.362,
      "step": 5589
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.5477613038231577,
      "learning_rate": 2.6536772582271962e-06,
      "loss": 0.2543,
      "step": 5590
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.074397057114063,
      "learning_rate": 2.6529600397408593e-06,
      "loss": 0.3229,
      "step": 5591
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.8075314839586936,
      "learning_rate": 2.6522428086177137e-06,
      "loss": 0.152,
      "step": 5592
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7808099978225433,
      "learning_rate": 2.651525564917013e-06,
      "loss": 0.1992,
      "step": 5593
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.946965560600043,
      "learning_rate": 2.650808308698014e-06,
      "loss": 0.1743,
      "step": 5594
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8252252302744076,
      "learning_rate": 2.650091040019972e-06,
      "loss": 0.2447,
      "step": 5595
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.363875875487605,
      "learning_rate": 2.649373758942144e-06,
      "loss": 0.2264,
      "step": 5596
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.7830262666945957,
      "learning_rate": 2.6486564655237875e-06,
      "loss": 0.3363,
      "step": 5597
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.682041461080941,
      "learning_rate": 2.6479391598241634e-06,
      "loss": 0.1587,
      "step": 5598
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.2726603893412594,
      "learning_rate": 2.6472218419025307e-06,
      "loss": 0.2273,
      "step": 5599
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.597950498520313,
      "learning_rate": 2.646504511818152e-06,
      "loss": 0.1439,
      "step": 5600
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.419957990554808,
      "learning_rate": 2.6457871696302887e-06,
      "loss": 0.1465,
      "step": 5601
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.002835523288373,
      "learning_rate": 2.645069815398203e-06,
      "loss": 0.1087,
      "step": 5602
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.215921641674614,
      "learning_rate": 2.6443524491811615e-06,
      "loss": 0.4481,
      "step": 5603
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.553256880455985,
      "learning_rate": 2.6436350710384284e-06,
      "loss": 0.1675,
      "step": 5604
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.443663495052928,
      "learning_rate": 2.64291768102927e-06,
      "loss": 0.3858,
      "step": 5605
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.315953680446186,
      "learning_rate": 2.6422002792129535e-06,
      "loss": 0.5837,
      "step": 5606
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.563089812807981,
      "learning_rate": 2.6414828656487455e-06,
      "loss": 0.4076,
      "step": 5607
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.877238283989366,
      "learning_rate": 2.6407654403959186e-06,
      "loss": 0.3324,
      "step": 5608
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.547750917317659,
      "learning_rate": 2.640048003513741e-06,
      "loss": 0.5164,
      "step": 5609
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.360987335243661,
      "learning_rate": 2.6393305550614834e-06,
      "loss": 0.3165,
      "step": 5610
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.787220391543232,
      "learning_rate": 2.6386130950984187e-06,
      "loss": 0.1945,
      "step": 5611
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.24412855411115,
      "learning_rate": 2.6378956236838195e-06,
      "loss": 0.4839,
      "step": 5612
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.098284101981154,
      "learning_rate": 2.6371781408769597e-06,
      "loss": 0.3825,
      "step": 5613
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.7130769412909457,
      "learning_rate": 2.636460646737115e-06,
      "loss": 0.1719,
      "step": 5614
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.458020519229663,
      "learning_rate": 2.635743141323561e-06,
      "loss": 0.3662,
      "step": 5615
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.4079986736252796,
      "learning_rate": 2.635025624695573e-06,
      "loss": 0.1654,
      "step": 5616
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.5942803007872675,
      "learning_rate": 2.6343080969124303e-06,
      "loss": 0.4382,
      "step": 5617
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.3211170011729414,
      "learning_rate": 2.633590558033412e-06,
      "loss": 0.1957,
      "step": 5618
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.116615659717387,
      "learning_rate": 2.632873008117796e-06,
      "loss": 0.3559,
      "step": 5619
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.984773394923949,
      "learning_rate": 2.6321554472248638e-06,
      "loss": 0.1662,
      "step": 5620
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.5480193485178457,
      "learning_rate": 2.6314378754138956e-06,
      "loss": 0.2278,
      "step": 5621
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.131007850961135,
      "learning_rate": 2.6307202927441756e-06,
      "loss": 0.362,
      "step": 5622
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.768741704469491,
      "learning_rate": 2.6300026992749854e-06,
      "loss": 0.2061,
      "step": 5623
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.5970196129140035,
      "learning_rate": 2.62928509506561e-06,
      "loss": 0.4715,
      "step": 5624
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.255458805911085,
      "learning_rate": 2.6285674801753336e-06,
      "loss": 0.2944,
      "step": 5625
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.287852129049745,
      "learning_rate": 2.6278498546634422e-06,
      "loss": 0.2469,
      "step": 5626
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.4075879933642406,
      "learning_rate": 2.6271322185892224e-06,
      "loss": 0.156,
      "step": 5627
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.27674761086556,
      "learning_rate": 2.6264145720119625e-06,
      "loss": 0.3258,
      "step": 5628
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.1515911416069056,
      "learning_rate": 2.6256969149909508e-06,
      "loss": 0.2481,
      "step": 5629
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.997962562493336,
      "learning_rate": 2.6249792475854756e-06,
      "loss": 0.1285,
      "step": 5630
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.305824550230581,
      "learning_rate": 2.624261569854827e-06,
      "loss": 0.4325,
      "step": 5631
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.886290198185201,
      "learning_rate": 2.6235438818582974e-06,
      "loss": 0.3955,
      "step": 5632
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.018832921024684,
      "learning_rate": 2.6228261836551786e-06,
      "loss": 0.4211,
      "step": 5633
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.1086567225414914,
      "learning_rate": 2.6221084753047627e-06,
      "loss": 0.1406,
      "step": 5634
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.474302978328723,
      "learning_rate": 2.6213907568663423e-06,
      "loss": 0.3284,
      "step": 5635
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.213232390039483,
      "learning_rate": 2.620673028399213e-06,
      "loss": 0.2581,
      "step": 5636
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.180328392423449,
      "learning_rate": 2.6199552899626697e-06,
      "loss": 0.0091,
      "step": 5637
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.796786277827634,
      "learning_rate": 2.619237541616009e-06,
      "loss": 0.3292,
      "step": 5638
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.028170913122398,
      "learning_rate": 2.6185197834185273e-06,
      "loss": 0.1598,
      "step": 5639
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.271337704824229,
      "learning_rate": 2.617802015429521e-06,
      "loss": 0.2008,
      "step": 5640
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.229784728292231,
      "learning_rate": 2.6170842377082904e-06,
      "loss": 0.2801,
      "step": 5641
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.835027953397691,
      "learning_rate": 2.616366450314134e-06,
      "loss": 0.3306,
      "step": 5642
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.569349605349419,
      "learning_rate": 2.6156486533063525e-06,
      "loss": 0.2454,
      "step": 5643
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.975935404013659,
      "learning_rate": 2.614930846744246e-06,
      "loss": 0.5405,
      "step": 5644
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.316647778254065,
      "learning_rate": 2.614213030687116e-06,
      "loss": 0.3171,
      "step": 5645
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.102703631497988,
      "learning_rate": 2.6134952051942656e-06,
      "loss": 0.2371,
      "step": 5646
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.705163366935671,
      "learning_rate": 2.6127773703249977e-06,
      "loss": 0.2394,
      "step": 5647
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7658966983123623,
      "learning_rate": 2.612059526138617e-06,
      "loss": 0.1616,
      "step": 5648
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.942990722569167,
      "learning_rate": 2.6113416726944275e-06,
      "loss": 0.4896,
      "step": 5649
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.124068351527358,
      "learning_rate": 2.610623810051734e-06,
      "loss": 0.2196,
      "step": 5650
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.142824954359236,
      "learning_rate": 2.6099059382698443e-06,
      "loss": 0.2972,
      "step": 5651
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.741965184195762,
      "learning_rate": 2.6091880574080646e-06,
      "loss": 0.1055,
      "step": 5652
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.236371608736326,
      "learning_rate": 2.6084701675257034e-06,
      "loss": 0.2826,
      "step": 5653
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7977568367045036,
      "learning_rate": 2.6077522686820677e-06,
      "loss": 0.1446,
      "step": 5654
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6925712985595234,
      "learning_rate": 2.607034360936469e-06,
      "loss": 0.3206,
      "step": 5655
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8329136641285753,
      "learning_rate": 2.606316444348216e-06,
      "loss": 0.2744,
      "step": 5656
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6637285552040386,
      "learning_rate": 2.6055985189766193e-06,
      "loss": 0.2751,
      "step": 5657
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.5493634679308417,
      "learning_rate": 2.6048805848809906e-06,
      "loss": 0.155,
      "step": 5658
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.177362416151829,
      "learning_rate": 2.6041626421206423e-06,
      "loss": 0.4955,
      "step": 5659
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.7644745675419515,
      "learning_rate": 2.6034446907548872e-06,
      "loss": 0.2731,
      "step": 5660
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.113893875907853,
      "learning_rate": 2.6027267308430395e-06,
      "loss": 0.6091,
      "step": 5661
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.101440908775792,
      "learning_rate": 2.602008762444412e-06,
      "loss": 0.3182,
      "step": 5662
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.409676963935518,
      "learning_rate": 2.6012907856183204e-06,
      "loss": 0.4715,
      "step": 5663
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.427760416429342,
      "learning_rate": 2.6005728004240816e-06,
      "loss": 0.3163,
      "step": 5664
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.613273063572505,
      "learning_rate": 2.599854806921011e-06,
      "loss": 0.4761,
      "step": 5665
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.021428864292128,
      "learning_rate": 2.5991368051684255e-06,
      "loss": 0.216,
      "step": 5666
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.3664076565856567,
      "learning_rate": 2.5984187952256433e-06,
      "loss": 0.1662,
      "step": 5667
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.9964663263959985,
      "learning_rate": 2.597700777151983e-06,
      "loss": 0.2296,
      "step": 5668
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.752113793717304,
      "learning_rate": 2.596982751006763e-06,
      "loss": 0.3358,
      "step": 5669
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.341557202653214,
      "learning_rate": 2.5962647168493038e-06,
      "loss": 0.3704,
      "step": 5670
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.127215062357721,
      "learning_rate": 2.5955466747389262e-06,
      "loss": 0.3418,
      "step": 5671
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.152217909841184,
      "learning_rate": 2.5948286247349504e-06,
      "loss": 0.151,
      "step": 5672
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.543712832826098,
      "learning_rate": 2.5941105668966977e-06,
      "loss": 0.3037,
      "step": 5673
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.145542843816784,
      "learning_rate": 2.593392501283492e-06,
      "loss": 0.3905,
      "step": 5674
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.153886057129538,
      "learning_rate": 2.5926744279546564e-06,
      "loss": 0.3912,
      "step": 5675
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.832668366075767,
      "learning_rate": 2.5919563469695137e-06,
      "loss": 0.2502,
      "step": 5676
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.298226218305639,
      "learning_rate": 2.5912382583873875e-06,
      "loss": 0.5257,
      "step": 5677
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8225720560040464,
      "learning_rate": 2.590520162267604e-06,
      "loss": 0.1824,
      "step": 5678
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.783641426174991,
      "learning_rate": 2.5898020586694896e-06,
      "loss": 0.254,
      "step": 5679
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.6898011533174,
      "learning_rate": 2.589083947652368e-06,
      "loss": 0.318,
      "step": 5680
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.051466281434948,
      "learning_rate": 2.5883658292755687e-06,
      "loss": 0.2232,
      "step": 5681
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.469107689574806,
      "learning_rate": 2.5876477035984164e-06,
      "loss": 0.5104,
      "step": 5682
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.031441427091058,
      "learning_rate": 2.5869295706802412e-06,
      "loss": 0.4053,
      "step": 5683
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.464111848598243,
      "learning_rate": 2.5862114305803714e-06,
      "loss": 0.1481,
      "step": 5684
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.510971547897624,
      "learning_rate": 2.585493283358136e-06,
      "loss": 0.4155,
      "step": 5685
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5050651093557805,
      "learning_rate": 2.584775129072864e-06,
      "loss": 0.1486,
      "step": 5686
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.532483922943913,
      "learning_rate": 2.584056967783887e-06,
      "loss": 0.4242,
      "step": 5687
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.8378813384146806,
      "learning_rate": 2.5833387995505343e-06,
      "loss": 0.0823,
      "step": 5688
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6533026712970966,
      "learning_rate": 2.58262062443214e-06,
      "loss": 0.1142,
      "step": 5689
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.8874903016363125,
      "learning_rate": 2.5819024424880345e-06,
      "loss": 0.2316,
      "step": 5690
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.534045746823456,
      "learning_rate": 2.5811842537775507e-06,
      "loss": 0.4458,
      "step": 5691
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.8397162743309075,
      "learning_rate": 2.580466058360021e-06,
      "loss": 0.3324,
      "step": 5692
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.334991727719048,
      "learning_rate": 2.57974785629478e-06,
      "loss": 0.4523,
      "step": 5693
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.7930580027524896,
      "learning_rate": 2.5790296476411636e-06,
      "loss": 0.1506,
      "step": 5694
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4117202840705767,
      "learning_rate": 2.5783114324585048e-06,
      "loss": 0.2522,
      "step": 5695
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.22961260580243,
      "learning_rate": 2.5775932108061393e-06,
      "loss": 0.3922,
      "step": 5696
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.71549836774674,
      "learning_rate": 2.5768749827434014e-06,
      "loss": 0.3624,
      "step": 5697
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.962562360465091,
      "learning_rate": 2.576156748329631e-06,
      "loss": 0.2448,
      "step": 5698
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.377424855848098,
      "learning_rate": 2.5754385076241633e-06,
      "loss": 0.3657,
      "step": 5699
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6435509080601896,
      "learning_rate": 2.5747202606863363e-06,
      "loss": 0.1296,
      "step": 5700
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.092015735300875,
      "learning_rate": 2.5740020075754873e-06,
      "loss": 0.2835,
      "step": 5701
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.276574706484719,
      "learning_rate": 2.5732837483509545e-06,
      "loss": 0.1166,
      "step": 5702
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.377620327695549,
      "learning_rate": 2.5725654830720787e-06,
      "loss": 0.3083,
      "step": 5703
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.02132732674732,
      "learning_rate": 2.571847211798198e-06,
      "loss": 0.3306,
      "step": 5704
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.00577149546578,
      "learning_rate": 2.571128934588653e-06,
      "loss": 0.3775,
      "step": 5705
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.7988084342412543,
      "learning_rate": 2.570410651502785e-06,
      "loss": 0.291,
      "step": 5706
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6038880482720894,
      "learning_rate": 2.569692362599932e-06,
      "loss": 0.3497,
      "step": 5707
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.7122622287476363,
      "learning_rate": 2.568974067939439e-06,
      "loss": 0.2956,
      "step": 5708
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.134162079526592,
      "learning_rate": 2.5682557675806472e-06,
      "loss": 0.2851,
      "step": 5709
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.2185696403775355,
      "learning_rate": 2.5675374615828987e-06,
      "loss": 0.3767,
      "step": 5710
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.975396591726601,
      "learning_rate": 2.566819150005535e-06,
      "loss": 0.233,
      "step": 5711
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.114223464315067,
      "learning_rate": 2.566100832907901e-06,
      "loss": 0.3012,
      "step": 5712
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.814214540953058,
      "learning_rate": 2.565382510349341e-06,
      "loss": 0.4962,
      "step": 5713
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.214470611182316,
      "learning_rate": 2.564664182389198e-06,
      "loss": 0.3604,
      "step": 5714
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.6963569686422835,
      "learning_rate": 2.5639458490868174e-06,
      "loss": 0.2915,
      "step": 5715
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.0028636746323825,
      "learning_rate": 2.5632275105015442e-06,
      "loss": 0.1821,
      "step": 5716
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.325635524824483,
      "learning_rate": 2.562509166692724e-06,
      "loss": 0.2271,
      "step": 5717
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.8748661646645384,
      "learning_rate": 2.5617908177197034e-06,
      "loss": 0.3098,
      "step": 5718
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.225443062696645,
      "learning_rate": 2.561072463641828e-06,
      "loss": 0.3306,
      "step": 5719
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.633689832466919,
      "learning_rate": 2.5603541045184456e-06,
      "loss": 0.1304,
      "step": 5720
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.879445256469047,
      "learning_rate": 2.559635740408902e-06,
      "loss": 0.3376,
      "step": 5721
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5977472311909224,
      "learning_rate": 2.5589173713725456e-06,
      "loss": 0.173,
      "step": 5722
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.395550430038391,
      "learning_rate": 2.558198997468726e-06,
      "loss": 0.433,
      "step": 5723
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.078691644107608,
      "learning_rate": 2.5574806187567903e-06,
      "loss": 0.2453,
      "step": 5724
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6852306847443144,
      "learning_rate": 2.556762235296088e-06,
      "loss": 0.2015,
      "step": 5725
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.0034552997344752,
      "learning_rate": 2.5560438471459676e-06,
      "loss": 0.1935,
      "step": 5726
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5898835144325973,
      "learning_rate": 2.5553254543657786e-06,
      "loss": 0.2609,
      "step": 5727
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.17279057931156,
      "learning_rate": 2.5546070570148728e-06,
      "loss": 0.3692,
      "step": 5728
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.09018770618885,
      "learning_rate": 2.5538886551526e-06,
      "loss": 0.35,
      "step": 5729
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.668395220921897,
      "learning_rate": 2.5531702488383102e-06,
      "loss": 0.2966,
      "step": 5730
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.267926125978914,
      "learning_rate": 2.552451838131355e-06,
      "loss": 0.3792,
      "step": 5731
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.343725328425566,
      "learning_rate": 2.5517334230910856e-06,
      "loss": 0.4018,
      "step": 5732
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.478856268670843,
      "learning_rate": 2.5510150037768555e-06,
      "loss": 0.2444,
      "step": 5733
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.465274065940834,
      "learning_rate": 2.5502965802480157e-06,
      "loss": 0.4001,
      "step": 5734
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.604816378121286,
      "learning_rate": 2.5495781525639186e-06,
      "loss": 0.2113,
      "step": 5735
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.667158037748396,
      "learning_rate": 2.548859720783918e-06,
      "loss": 0.3122,
      "step": 5736
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.6768591601140526,
      "learning_rate": 2.548141284967366e-06,
      "loss": 0.1468,
      "step": 5737
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2496826976604862,
      "learning_rate": 2.5474228451736177e-06,
      "loss": 0.1199,
      "step": 5738
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.14376002345193,
      "learning_rate": 2.5467044014620267e-06,
      "loss": 0.486,
      "step": 5739
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.9644663107977336,
      "learning_rate": 2.545985953891946e-06,
      "loss": 0.3256,
      "step": 5740
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.3296666249110745,
      "learning_rate": 2.5452675025227325e-06,
      "loss": 0.4039,
      "step": 5741
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.9282618463272123,
      "learning_rate": 2.5445490474137393e-06,
      "loss": 0.1285,
      "step": 5742
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5370514227884104,
      "learning_rate": 2.543830588624322e-06,
      "loss": 0.1192,
      "step": 5743
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.397618957825567,
      "learning_rate": 2.543112126213837e-06,
      "loss": 0.3115,
      "step": 5744
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.060567867632584,
      "learning_rate": 2.542393660241639e-06,
      "loss": 0.2917,
      "step": 5745
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4609341542012966,
      "learning_rate": 2.5416751907670846e-06,
      "loss": 0.2224,
      "step": 5746
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5619432187364994,
      "learning_rate": 2.5409567178495303e-06,
      "loss": 0.2361,
      "step": 5747
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.815324165436471,
      "learning_rate": 2.5402382415483327e-06,
      "loss": 0.2349,
      "step": 5748
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.93630041536183,
      "learning_rate": 2.5395197619228496e-06,
      "loss": 0.3847,
      "step": 5749
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.9500328764869965,
      "learning_rate": 2.538801279032436e-06,
      "loss": 0.1562,
      "step": 5750
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.303505101167118,
      "learning_rate": 2.5380827929364527e-06,
      "loss": 0.2627,
      "step": 5751
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.40487510415097,
      "learning_rate": 2.5373643036942547e-06,
      "loss": 0.3256,
      "step": 5752
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.0618827241391005,
      "learning_rate": 2.5366458113652015e-06,
      "loss": 0.2626,
      "step": 5753
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.311391174470655,
      "learning_rate": 2.5359273160086505e-06,
      "loss": 0.053,
      "step": 5754
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.085672859065956,
      "learning_rate": 2.5352088176839613e-06,
      "loss": 0.2888,
      "step": 5755
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.8127427762157002,
      "learning_rate": 2.5344903164504924e-06,
      "loss": 0.3495,
      "step": 5756
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.659831184159649,
      "learning_rate": 2.533771812367602e-06,
      "loss": 0.544,
      "step": 5757
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.689079617593253,
      "learning_rate": 2.5330533054946506e-06,
      "loss": 0.2768,
      "step": 5758
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3066986585059013,
      "learning_rate": 2.5323347958909965e-06,
      "loss": 0.2622,
      "step": 5759
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.28195461777116,
      "learning_rate": 2.531616283616001e-06,
      "loss": 0.2752,
      "step": 5760
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.443585944581287,
      "learning_rate": 2.5308977687290228e-06,
      "loss": 0.3769,
      "step": 5761
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.2955621103845805,
      "learning_rate": 2.530179251289422e-06,
      "loss": 0.2501,
      "step": 5762
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.234685594452846,
      "learning_rate": 2.52946073135656e-06,
      "loss": 0.3455,
      "step": 5763
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.693922498405195,
      "learning_rate": 2.5287422089897964e-06,
      "loss": 0.5652,
      "step": 5764
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.731438754619773,
      "learning_rate": 2.528023684248493e-06,
      "loss": 0.4123,
      "step": 5765
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.754545761255187,
      "learning_rate": 2.527305157192009e-06,
      "loss": 0.4772,
      "step": 5766
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.845412719899518,
      "learning_rate": 2.526586627879708e-06,
      "loss": 0.4604,
      "step": 5767
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.601039676090113,
      "learning_rate": 2.5258680963709497e-06,
      "loss": 0.3716,
      "step": 5768
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.783251731957733,
      "learning_rate": 2.525149562725096e-06,
      "loss": 0.2754,
      "step": 5769
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.695316510091078,
      "learning_rate": 2.524431027001509e-06,
      "loss": 0.3452,
      "step": 5770
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.620945182427314,
      "learning_rate": 2.5237124892595503e-06,
      "loss": 0.2556,
      "step": 5771
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.438977301193632,
      "learning_rate": 2.5229939495585816e-06,
      "loss": 0.3037,
      "step": 5772
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6347141932935374,
      "learning_rate": 2.5222754079579657e-06,
      "loss": 0.187,
      "step": 5773
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.580641394309725,
      "learning_rate": 2.5215568645170644e-06,
      "loss": 0.514,
      "step": 5774
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.588188942525081,
      "learning_rate": 2.5208383192952416e-06,
      "loss": 0.296,
      "step": 5775
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3359496179730956,
      "learning_rate": 2.5201197723518587e-06,
      "loss": 0.2083,
      "step": 5776
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3833325693178224,
      "learning_rate": 2.5194012237462782e-06,
      "loss": 0.1077,
      "step": 5777
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.163179185327584,
      "learning_rate": 2.518682673537864e-06,
      "loss": 0.4175,
      "step": 5778
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.22877047505847,
      "learning_rate": 2.5179641217859797e-06,
      "loss": 0.3581,
      "step": 5779
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.079409701269256,
      "learning_rate": 2.517245568549987e-06,
      "loss": 0.3477,
      "step": 5780
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.124148594006243,
      "learning_rate": 2.5165270138892508e-06,
      "loss": 0.3493,
      "step": 5781
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.439390744358861,
      "learning_rate": 2.5158084578631327e-06,
      "loss": 0.2822,
      "step": 5782
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.43516279367167,
      "learning_rate": 2.515089900530997e-06,
      "loss": 0.1656,
      "step": 5783
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.012891712692361,
      "learning_rate": 2.5143713419522087e-06,
      "loss": 0.2917,
      "step": 5784
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5020325651942925,
      "learning_rate": 2.5136527821861306e-06,
      "loss": 0.1252,
      "step": 5785
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.4556130116567294,
      "learning_rate": 2.5129342212921265e-06,
      "loss": 0.286,
      "step": 5786
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.1037462174715733,
      "learning_rate": 2.5122156593295598e-06,
      "loss": 0.1732,
      "step": 5787
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5453170550859108,
      "learning_rate": 2.5114970963577953e-06,
      "loss": 0.2558,
      "step": 5788
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.594896682367196,
      "learning_rate": 2.510778532436197e-06,
      "loss": 0.2574,
      "step": 5789
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.215355818470069,
      "learning_rate": 2.51005996762413e-06,
      "loss": 0.2188,
      "step": 5790
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.664985737875613,
      "learning_rate": 2.5093414019809577e-06,
      "loss": 0.2926,
      "step": 5791
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.914651155055497,
      "learning_rate": 2.508622835566044e-06,
      "loss": 0.3304,
      "step": 5792
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5508945756846297,
      "learning_rate": 2.5079042684387537e-06,
      "loss": 0.2569,
      "step": 5793
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.189677617946642,
      "learning_rate": 2.507185700658452e-06,
      "loss": 0.3536,
      "step": 5794
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.078241129172782,
      "learning_rate": 2.5064671322845034e-06,
      "loss": 0.3995,
      "step": 5795
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.612790529811262,
      "learning_rate": 2.505748563376272e-06,
      "loss": 0.4539,
      "step": 5796
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.8423437976802433,
      "learning_rate": 2.505029993993121e-06,
      "loss": 0.3056,
      "step": 5797
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.805501505513689,
      "learning_rate": 2.504311424194417e-06,
      "loss": 0.5709,
      "step": 5798
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.1108060740016645,
      "learning_rate": 2.5035928540395253e-06,
      "loss": 0.2777,
      "step": 5799
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.249777472159915,
      "learning_rate": 2.5028742835878093e-06,
      "loss": 0.3619,
      "step": 5800
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.066560817954737,
      "learning_rate": 2.5021557128986344e-06,
      "loss": 0.2829,
      "step": 5801
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.6430470328692084,
      "learning_rate": 2.5014371420313637e-06,
      "loss": 0.2493,
      "step": 5802
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.8429404862819956,
      "learning_rate": 2.5007185710453647e-06,
      "loss": 0.137,
      "step": 5803
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.370424622569346,
      "learning_rate": 2.5e-06,
      "loss": 0.3939,
      "step": 5804
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.404150876639908,
      "learning_rate": 2.4992814289546357e-06,
      "loss": 0.336,
      "step": 5805
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.217748494197521,
      "learning_rate": 2.498562857968637e-06,
      "loss": 0.4294,
      "step": 5806
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7959547883291345,
      "learning_rate": 2.4978442871013664e-06,
      "loss": 0.2452,
      "step": 5807
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.863388854269682,
      "learning_rate": 2.497125716412192e-06,
      "loss": 0.1014,
      "step": 5808
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.3534296222137336,
      "learning_rate": 2.496407145960475e-06,
      "loss": 0.2998,
      "step": 5809
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8553239458748747,
      "learning_rate": 2.495688575805583e-06,
      "loss": 0.1224,
      "step": 5810
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.206653123674837,
      "learning_rate": 2.4949700060068793e-06,
      "loss": 0.2073,
      "step": 5811
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.416613587650445,
      "learning_rate": 2.4942514366237292e-06,
      "loss": 0.2679,
      "step": 5812
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.031576345460209,
      "learning_rate": 2.493532867715498e-06,
      "loss": 0.4163,
      "step": 5813
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.424340230834137,
      "learning_rate": 2.4928142993415486e-06,
      "loss": 0.5204,
      "step": 5814
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.054821411524885,
      "learning_rate": 2.4920957315612467e-06,
      "loss": 0.4996,
      "step": 5815
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.0108157146802395,
      "learning_rate": 2.4913771644339567e-06,
      "loss": 0.4093,
      "step": 5816
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.508643751661064,
      "learning_rate": 2.4906585980190427e-06,
      "loss": 0.1191,
      "step": 5817
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.666993903061976,
      "learning_rate": 2.489940032375871e-06,
      "loss": 0.4668,
      "step": 5818
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.0615026864925445,
      "learning_rate": 2.489221467563803e-06,
      "loss": 0.2981,
      "step": 5819
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.826746006411625,
      "learning_rate": 2.488502903642205e-06,
      "loss": 0.2215,
      "step": 5820
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.193225365888707,
      "learning_rate": 2.487784340670441e-06,
      "loss": 0.609,
      "step": 5821
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.463880660063583,
      "learning_rate": 2.4870657787078744e-06,
      "loss": 0.5157,
      "step": 5822
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.954834944472275,
      "learning_rate": 2.486347217813871e-06,
      "loss": 0.443,
      "step": 5823
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.428630151895993,
      "learning_rate": 2.485628658047792e-06,
      "loss": 0.5229,
      "step": 5824
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.007866587284195,
      "learning_rate": 2.484910099469003e-06,
      "loss": 0.2564,
      "step": 5825
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.1049862907441454,
      "learning_rate": 2.4841915421368686e-06,
      "loss": 0.5065,
      "step": 5826
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.021705342008418,
      "learning_rate": 2.4834729861107505e-06,
      "loss": 0.3404,
      "step": 5827
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.226266619728756,
      "learning_rate": 2.482754431450013e-06,
      "loss": 0.258,
      "step": 5828
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.6464973201982143,
      "learning_rate": 2.4820358782140215e-06,
      "loss": 0.3107,
      "step": 5829
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.509221666933185,
      "learning_rate": 2.481317326462136e-06,
      "loss": 0.1511,
      "step": 5830
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.517788520006733,
      "learning_rate": 2.480598776253722e-06,
      "loss": 0.2504,
      "step": 5831
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.432599564875584,
      "learning_rate": 2.479880227648142e-06,
      "loss": 0.3681,
      "step": 5832
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.654101541864977,
      "learning_rate": 2.4791616807047588e-06,
      "loss": 0.4044,
      "step": 5833
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.080548262678054,
      "learning_rate": 2.478443135482936e-06,
      "loss": 0.3686,
      "step": 5834
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.083724278559385,
      "learning_rate": 2.4777245920420347e-06,
      "loss": 0.2882,
      "step": 5835
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.006269359676949,
      "learning_rate": 2.4770060504414192e-06,
      "loss": 0.4077,
      "step": 5836
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.199010831778123,
      "learning_rate": 2.4762875107404505e-06,
      "loss": 0.106,
      "step": 5837
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.430086929912683,
      "learning_rate": 2.4755689729984915e-06,
      "loss": 0.3381,
      "step": 5838
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8587291281904403,
      "learning_rate": 2.474850437274905e-06,
      "loss": 0.2371,
      "step": 5839
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.751564111833204,
      "learning_rate": 2.474131903629051e-06,
      "loss": 0.2649,
      "step": 5840
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.185816905442697,
      "learning_rate": 2.4734133721202928e-06,
      "loss": 0.3591,
      "step": 5841
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2871372476933436,
      "learning_rate": 2.4726948428079913e-06,
      "loss": 0.1265,
      "step": 5842
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8856195171928514,
      "learning_rate": 2.4719763157515075e-06,
      "loss": 0.3481,
      "step": 5843
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.849833987532878,
      "learning_rate": 2.4712577910102045e-06,
      "loss": 0.363,
      "step": 5844
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.4293403476798865,
      "learning_rate": 2.4705392686434406e-06,
      "loss": 0.4285,
      "step": 5845
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7636147008151757,
      "learning_rate": 2.469820748710579e-06,
      "loss": 0.2756,
      "step": 5846
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.580498755658191,
      "learning_rate": 2.4691022312709776e-06,
      "loss": 0.4199,
      "step": 5847
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.236237922431302,
      "learning_rate": 2.468383716383999e-06,
      "loss": 0.3421,
      "step": 5848
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.1044454833114483,
      "learning_rate": 2.467665204109004e-06,
      "loss": 0.2161,
      "step": 5849
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.924155423010109,
      "learning_rate": 2.46694669450535e-06,
      "loss": 0.5894,
      "step": 5850
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.756879411049171,
      "learning_rate": 2.4662281876323987e-06,
      "loss": 0.4524,
      "step": 5851
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9029634396128574,
      "learning_rate": 2.4655096835495084e-06,
      "loss": 0.1786,
      "step": 5852
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.4797156738370774,
      "learning_rate": 2.4647911823160387e-06,
      "loss": 0.296,
      "step": 5853
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.435463166914168,
      "learning_rate": 2.4640726839913504e-06,
      "loss": 0.4813,
      "step": 5854
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.190585685020022,
      "learning_rate": 2.4633541886347994e-06,
      "loss": 0.4101,
      "step": 5855
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8914387534767205,
      "learning_rate": 2.462635696305746e-06,
      "loss": 0.2802,
      "step": 5856
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.845700349792296,
      "learning_rate": 2.461917207063548e-06,
      "loss": 0.1951,
      "step": 5857
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.638808487643313,
      "learning_rate": 2.4611987209675635e-06,
      "loss": 0.5424,
      "step": 5858
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8973397400063603,
      "learning_rate": 2.460480238077152e-06,
      "loss": 0.2794,
      "step": 5859
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.193625313140622,
      "learning_rate": 2.4597617584516677e-06,
      "loss": 0.3674,
      "step": 5860
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.346007977354848,
      "learning_rate": 2.4590432821504705e-06,
      "loss": 0.2044,
      "step": 5861
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.123336078733298,
      "learning_rate": 2.4583248092329158e-06,
      "loss": 0.4464,
      "step": 5862
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.577622293537648,
      "learning_rate": 2.4576063397583615e-06,
      "loss": 0.2597,
      "step": 5863
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.384197406895584,
      "learning_rate": 2.4568878737861644e-06,
      "loss": 0.4541,
      "step": 5864
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.470809327674683,
      "learning_rate": 2.4561694113756786e-06,
      "loss": 0.2739,
      "step": 5865
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.870595518141755,
      "learning_rate": 2.455450952586261e-06,
      "loss": 0.1746,
      "step": 5866
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.7683064639080106,
      "learning_rate": 2.4547324974772684e-06,
      "loss": 0.0917,
      "step": 5867
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.5904119651103708,
      "learning_rate": 2.454014046108054e-06,
      "loss": 0.317,
      "step": 5868
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.718721152062177,
      "learning_rate": 2.4532955985379746e-06,
      "loss": 0.1005,
      "step": 5869
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7147169445502555,
      "learning_rate": 2.452577154826383e-06,
      "loss": 0.1768,
      "step": 5870
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.1816438163676315,
      "learning_rate": 2.451858715032634e-06,
      "loss": 0.3797,
      "step": 5871
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.160519623734339,
      "learning_rate": 2.451140279216083e-06,
      "loss": 0.3267,
      "step": 5872
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.120360303043361,
      "learning_rate": 2.450421847436082e-06,
      "loss": 0.4853,
      "step": 5873
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.280360518420043,
      "learning_rate": 2.4497034197519855e-06,
      "loss": 0.2908,
      "step": 5874
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.48356885955288,
      "learning_rate": 2.4489849962231453e-06,
      "loss": 0.4594,
      "step": 5875
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.8055555103345564,
      "learning_rate": 2.4482665769089144e-06,
      "loss": 0.2635,
      "step": 5876
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.1572349868605265,
      "learning_rate": 2.447548161868646e-06,
      "loss": 0.287,
      "step": 5877
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.09309383135796,
      "learning_rate": 2.44682975116169e-06,
      "loss": 0.2168,
      "step": 5878
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.8237652530991433,
      "learning_rate": 2.4461113448474012e-06,
      "loss": 0.1586,
      "step": 5879
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.390040023787688,
      "learning_rate": 2.4453929429851276e-06,
      "loss": 0.2535,
      "step": 5880
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.046838674219428,
      "learning_rate": 2.4446745456342214e-06,
      "loss": 0.324,
      "step": 5881
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.262158470054922,
      "learning_rate": 2.4439561528540332e-06,
      "loss": 0.2512,
      "step": 5882
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.583900792352416,
      "learning_rate": 2.4432377647039127e-06,
      "loss": 0.5198,
      "step": 5883
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.685141258636088,
      "learning_rate": 2.442519381243211e-06,
      "loss": 0.3524,
      "step": 5884
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.535771704881475,
      "learning_rate": 2.441801002531275e-06,
      "loss": 0.3096,
      "step": 5885
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.159202075501785,
      "learning_rate": 2.4410826286274544e-06,
      "loss": 0.6479,
      "step": 5886
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.5946257304211238,
      "learning_rate": 2.440364259591099e-06,
      "loss": 0.1472,
      "step": 5887
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.439853875234242,
      "learning_rate": 2.4396458954815556e-06,
      "loss": 0.2534,
      "step": 5888
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.665835725663125,
      "learning_rate": 2.4389275363581734e-06,
      "loss": 0.2546,
      "step": 5889
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9849570923007778,
      "learning_rate": 2.438209182280298e-06,
      "loss": 0.4173,
      "step": 5890
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.2877810054524867,
      "learning_rate": 2.4374908333072765e-06,
      "loss": 0.274,
      "step": 5891
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.085667552993291,
      "learning_rate": 2.4367724894984566e-06,
      "loss": 0.3261,
      "step": 5892
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.323867689650524,
      "learning_rate": 2.436054150913183e-06,
      "loss": 0.42,
      "step": 5893
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.040413845567395,
      "learning_rate": 2.435335817610803e-06,
      "loss": 0.1217,
      "step": 5894
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.267424155355662,
      "learning_rate": 2.43461748965066e-06,
      "loss": 0.2094,
      "step": 5895
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.526676190715663,
      "learning_rate": 2.4338991670920993e-06,
      "loss": 0.4496,
      "step": 5896
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.383839483145744,
      "learning_rate": 2.4331808499944655e-06,
      "loss": 0.122,
      "step": 5897
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.17065150077593,
      "learning_rate": 2.432462538417102e-06,
      "loss": 0.2819,
      "step": 5898
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.881315203006964,
      "learning_rate": 2.4317442324193536e-06,
      "loss": 0.2906,
      "step": 5899
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.0645658012967902,
      "learning_rate": 2.4310259320605613e-06,
      "loss": 0.176,
      "step": 5900
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.53369404765228,
      "learning_rate": 2.430307637400068e-06,
      "loss": 0.3478,
      "step": 5901
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7699192523345775,
      "learning_rate": 2.4295893484972167e-06,
      "loss": 0.3211,
      "step": 5902
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.959181621555041,
      "learning_rate": 2.4288710654113473e-06,
      "loss": 0.3352,
      "step": 5903
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.948846923051538,
      "learning_rate": 2.4281527882018032e-06,
      "loss": 0.1511,
      "step": 5904
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.000902806494907,
      "learning_rate": 2.427434516927922e-06,
      "loss": 0.4358,
      "step": 5905
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.354743306728322,
      "learning_rate": 2.426716251649046e-06,
      "loss": 0.237,
      "step": 5906
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.718403262240324,
      "learning_rate": 2.425997992424514e-06,
      "loss": 0.2501,
      "step": 5907
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.053558400124433,
      "learning_rate": 2.4252797393136645e-06,
      "loss": 0.2359,
      "step": 5908
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.169887849956771,
      "learning_rate": 2.4245614923758367e-06,
      "loss": 0.1725,
      "step": 5909
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.016509198874105,
      "learning_rate": 2.4238432516703697e-06,
      "loss": 0.3262,
      "step": 5910
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.7422254890503863,
      "learning_rate": 2.4231250172565986e-06,
      "loss": 0.2859,
      "step": 5911
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9542159628460904,
      "learning_rate": 2.4224067891938624e-06,
      "loss": 0.2481,
      "step": 5912
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.635233951833206,
      "learning_rate": 2.421688567541496e-06,
      "loss": 0.3395,
      "step": 5913
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.378127932422877,
      "learning_rate": 2.4209703523588364e-06,
      "loss": 0.2534,
      "step": 5914
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.50131538044865,
      "learning_rate": 2.4202521437052202e-06,
      "loss": 0.5206,
      "step": 5915
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.587611730772254,
      "learning_rate": 2.4195339416399796e-06,
      "loss": 0.2588,
      "step": 5916
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.66289442846309,
      "learning_rate": 2.4188157462224505e-06,
      "loss": 0.2867,
      "step": 5917
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.03870187619544,
      "learning_rate": 2.4180975575119663e-06,
      "loss": 0.1476,
      "step": 5918
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.197960913489302,
      "learning_rate": 2.4173793755678604e-06,
      "loss": 0.4377,
      "step": 5919
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.442488905212661,
      "learning_rate": 2.416661200449466e-06,
      "loss": 0.4185,
      "step": 5920
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2221966311390413,
      "learning_rate": 2.415943032216114e-06,
      "loss": 0.1878,
      "step": 5921
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6821156702976747,
      "learning_rate": 2.415224870927137e-06,
      "loss": 0.2801,
      "step": 5922
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.141561575121208,
      "learning_rate": 2.414506716641865e-06,
      "loss": 0.541,
      "step": 5923
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.151914851429514,
      "learning_rate": 2.413788569419629e-06,
      "loss": 0.2507,
      "step": 5924
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.373975524431338,
      "learning_rate": 2.4130704293197596e-06,
      "loss": 0.0865,
      "step": 5925
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.5905522815613757,
      "learning_rate": 2.412352296401584e-06,
      "loss": 0.2815,
      "step": 5926
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.78086479804609,
      "learning_rate": 2.4116341707244325e-06,
      "loss": 0.1659,
      "step": 5927
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.754493697261165,
      "learning_rate": 2.4109160523476323e-06,
      "loss": 0.2234,
      "step": 5928
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.973132641657082,
      "learning_rate": 2.410197941330511e-06,
      "loss": 0.4039,
      "step": 5929
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.240438414293221,
      "learning_rate": 2.4094798377323968e-06,
      "loss": 0.474,
      "step": 5930
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.153251327826301,
      "learning_rate": 2.408761741612613e-06,
      "loss": 0.3304,
      "step": 5931
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.491024462496775,
      "learning_rate": 2.4080436530304876e-06,
      "loss": 0.0942,
      "step": 5932
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.277367092800305,
      "learning_rate": 2.407325572045344e-06,
      "loss": 0.2171,
      "step": 5933
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2043647288671515,
      "learning_rate": 2.406607498716508e-06,
      "loss": 0.2274,
      "step": 5934
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.16685910194716,
      "learning_rate": 2.405889433103303e-06,
      "loss": 0.2248,
      "step": 5935
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.7129596122809096,
      "learning_rate": 2.405171375265051e-06,
      "loss": 0.2639,
      "step": 5936
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.7551091647625725,
      "learning_rate": 2.404453325261075e-06,
      "loss": 0.3418,
      "step": 5937
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0952423795658777,
      "learning_rate": 2.4037352831506966e-06,
      "loss": 0.1574,
      "step": 5938
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.103886619912476,
      "learning_rate": 2.403017248993237e-06,
      "loss": 0.2815,
      "step": 5939
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.791852369344596,
      "learning_rate": 2.4022992228480183e-06,
      "loss": 0.3429,
      "step": 5940
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.580055731751162,
      "learning_rate": 2.4015812047743575e-06,
      "loss": 0.3808,
      "step": 5941
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.137260437606928,
      "learning_rate": 2.4008631948315757e-06,
      "loss": 0.5602,
      "step": 5942
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.727446478169106,
      "learning_rate": 2.4001451930789897e-06,
      "loss": 0.3342,
      "step": 5943
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.18453726792126,
      "learning_rate": 2.399427199575919e-06,
      "loss": 0.4909,
      "step": 5944
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.209999087726688,
      "learning_rate": 2.3987092143816804e-06,
      "loss": 0.5397,
      "step": 5945
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.013614553220636,
      "learning_rate": 2.3979912375555893e-06,
      "loss": 0.3876,
      "step": 5946
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.630861335141664,
      "learning_rate": 2.397273269156962e-06,
      "loss": 0.3159,
      "step": 5947
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.987301356060629,
      "learning_rate": 2.3965553092451136e-06,
      "loss": 0.2844,
      "step": 5948
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.9983786416272045,
      "learning_rate": 2.395837357879358e-06,
      "loss": 0.366,
      "step": 5949
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.9712634544054435,
      "learning_rate": 2.3951194151190106e-06,
      "loss": 0.2936,
      "step": 5950
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.929945662014371,
      "learning_rate": 2.3944014810233816e-06,
      "loss": 0.2271,
      "step": 5951
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.503033299200959,
      "learning_rate": 2.3936835556517845e-06,
      "loss": 0.2221,
      "step": 5952
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.8243514055716443,
      "learning_rate": 2.392965639063532e-06,
      "loss": 0.1333,
      "step": 5953
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.880999069566754,
      "learning_rate": 2.392247731317932e-06,
      "loss": 0.3459,
      "step": 5954
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.469490168361898,
      "learning_rate": 2.391529832474298e-06,
      "loss": 0.2753,
      "step": 5955
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6967994959420727,
      "learning_rate": 2.3908119425919362e-06,
      "loss": 0.1716,
      "step": 5956
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.491494031597143,
      "learning_rate": 2.390094061730156e-06,
      "loss": 0.4859,
      "step": 5957
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.5406786451511545,
      "learning_rate": 2.389376189948267e-06,
      "loss": 0.3572,
      "step": 5958
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.94476630619909,
      "learning_rate": 2.3886583273055737e-06,
      "loss": 0.4183,
      "step": 5959
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.941897122961172,
      "learning_rate": 2.3879404738613844e-06,
      "loss": 0.2394,
      "step": 5960
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.058584699012688,
      "learning_rate": 2.3872226296750027e-06,
      "loss": 0.3743,
      "step": 5961
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.856876883647403,
      "learning_rate": 2.3865047948057348e-06,
      "loss": 0.345,
      "step": 5962
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.8956608001383195,
      "learning_rate": 2.385786969312885e-06,
      "loss": 0.1521,
      "step": 5963
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.658110618601851,
      "learning_rate": 2.3850691532557545e-06,
      "loss": 0.3153,
      "step": 5964
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.3551715001815765,
      "learning_rate": 2.3843513466936487e-06,
      "loss": 0.4197,
      "step": 5965
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.216152916865577,
      "learning_rate": 2.3836335496858667e-06,
      "loss": 0.7045,
      "step": 5966
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.9568854298501765,
      "learning_rate": 2.38291576229171e-06,
      "loss": 0.279,
      "step": 5967
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.393975379456991,
      "learning_rate": 2.3821979845704794e-06,
      "loss": 0.2308,
      "step": 5968
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.4384928185624406,
      "learning_rate": 2.381480216581474e-06,
      "loss": 0.1837,
      "step": 5969
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.8882276892025596,
      "learning_rate": 2.3807624583839923e-06,
      "loss": 0.2519,
      "step": 5970
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.42041731636728,
      "learning_rate": 2.3800447100373307e-06,
      "loss": 0.3914,
      "step": 5971
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.427535687592862,
      "learning_rate": 2.3793269716007873e-06,
      "loss": 0.241,
      "step": 5972
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.791759968652485,
      "learning_rate": 2.3786092431336586e-06,
      "loss": 0.3984,
      "step": 5973
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7012007605469184,
      "learning_rate": 2.377891524695238e-06,
      "loss": 0.1617,
      "step": 5974
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.966733347417958,
      "learning_rate": 2.3771738163448226e-06,
      "loss": 0.1649,
      "step": 5975
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.653311722028565,
      "learning_rate": 2.376456118141703e-06,
      "loss": 0.4369,
      "step": 5976
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.14174036188502,
      "learning_rate": 2.3757384301451732e-06,
      "loss": 0.2121,
      "step": 5977
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.041655699341887,
      "learning_rate": 2.3750207524145257e-06,
      "loss": 0.3083,
      "step": 5978
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.424535625706364,
      "learning_rate": 2.37430308500905e-06,
      "loss": 0.1928,
      "step": 5979
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.507868803756544,
      "learning_rate": 2.3735854279880387e-06,
      "loss": 0.2231,
      "step": 5980
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.091434004844798,
      "learning_rate": 2.3728677814107785e-06,
      "loss": 0.2462,
      "step": 5981
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.7739425427068043,
      "learning_rate": 2.3721501453365586e-06,
      "loss": 0.2772,
      "step": 5982
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.956522927213076,
      "learning_rate": 2.3714325198246672e-06,
      "loss": 0.3277,
      "step": 5983
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.782366149799863,
      "learning_rate": 2.3707149049343907e-06,
      "loss": 0.2797,
      "step": 5984
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.0959410125271942,
      "learning_rate": 2.369997300725016e-06,
      "loss": 0.2157,
      "step": 5985
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.5980996231407127,
      "learning_rate": 2.3692797072558256e-06,
      "loss": 0.3035,
      "step": 5986
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.6632665018137067,
      "learning_rate": 2.3685621245861048e-06,
      "loss": 0.3656,
      "step": 5987
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.916118154504279,
      "learning_rate": 2.3678445527751375e-06,
      "loss": 0.3479,
      "step": 5988
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.162076910014902,
      "learning_rate": 2.3671269918822047e-06,
      "loss": 0.1815,
      "step": 5989
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.386777901918464,
      "learning_rate": 2.3664094419665884e-06,
      "loss": 0.3808,
      "step": 5990
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.25524612479767,
      "learning_rate": 2.36569190308757e-06,
      "loss": 0.274,
      "step": 5991
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0139762841100852,
      "learning_rate": 2.3649743753044276e-06,
      "loss": 0.1044,
      "step": 5992
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.775897897323823,
      "learning_rate": 2.3642568586764405e-06,
      "loss": 0.3758,
      "step": 5993
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.827210789104615,
      "learning_rate": 2.3635393532628854e-06,
      "loss": 0.1695,
      "step": 5994
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.960294004947789,
      "learning_rate": 2.3628218591230403e-06,
      "loss": 0.2695,
      "step": 5995
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.621622766800731,
      "learning_rate": 2.3621043763161817e-06,
      "loss": 0.2768,
      "step": 5996
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.105513776588294,
      "learning_rate": 2.361386904901582e-06,
      "loss": 0.4839,
      "step": 5997
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.725579540184023,
      "learning_rate": 2.3606694449385174e-06,
      "loss": 0.5764,
      "step": 5998
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.81215821685003,
      "learning_rate": 2.35995199648626e-06,
      "loss": 0.3638,
      "step": 5999
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.106722656286227,
      "learning_rate": 2.3592345596040818e-06,
      "loss": 0.3364,
      "step": 6000
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.708449713029526,
      "learning_rate": 2.358517134351255e-06,
      "loss": 0.5362,
      "step": 6001
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.9473293626958457,
      "learning_rate": 2.3577997207870473e-06,
      "loss": 0.1127,
      "step": 6002
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.582938807974665,
      "learning_rate": 2.3570823189707314e-06,
      "loss": 0.409,
      "step": 6003
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6710377252144077,
      "learning_rate": 2.3563649289615724e-06,
      "loss": 0.2045,
      "step": 6004
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.4782996315196697,
      "learning_rate": 2.355647550818839e-06,
      "loss": 0.2194,
      "step": 6005
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.331358538972659,
      "learning_rate": 2.3549301846017978e-06,
      "loss": 0.1367,
      "step": 6006
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.9762061292006123,
      "learning_rate": 2.354212830369712e-06,
      "loss": 0.2206,
      "step": 6007
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.696830469451779,
      "learning_rate": 2.3534954881818492e-06,
      "loss": 0.1739,
      "step": 6008
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.080429953020883,
      "learning_rate": 2.3527781580974697e-06,
      "loss": 0.2194,
      "step": 6009
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.529863679397945,
      "learning_rate": 2.352060840175837e-06,
      "loss": 0.0922,
      "step": 6010
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.34659344008791,
      "learning_rate": 2.3513435344762138e-06,
      "loss": 0.4189,
      "step": 6011
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.7756329740922085,
      "learning_rate": 2.350626241057857e-06,
      "loss": 0.2954,
      "step": 6012
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8761317273993099,
      "learning_rate": 2.349908959980029e-06,
      "loss": 0.0107,
      "step": 6013
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.5074920107701693,
      "learning_rate": 2.3491916913019866e-06,
      "loss": 0.1498,
      "step": 6014
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.549971788130569,
      "learning_rate": 2.348474435082987e-06,
      "loss": 0.1236,
      "step": 6015
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.220293612542654,
      "learning_rate": 2.347757191382287e-06,
      "loss": 0.2045,
      "step": 6016
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.1660898313354675,
      "learning_rate": 2.347039960259141e-06,
      "loss": 0.314,
      "step": 6017
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.443387043802675,
      "learning_rate": 2.346322741772805e-06,
      "loss": 0.2504,
      "step": 6018
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.679036276648902,
      "learning_rate": 2.3456055359825293e-06,
      "loss": 0.3479,
      "step": 6019
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.062619900247903,
      "learning_rate": 2.3448883429475673e-06,
      "loss": 0.1915,
      "step": 6020
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.26745483823196,
      "learning_rate": 2.3441711627271703e-06,
      "loss": 0.4157,
      "step": 6021
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.237547551766405,
      "learning_rate": 2.3434539953805873e-06,
      "loss": 0.2218,
      "step": 6022
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.320119103042298,
      "learning_rate": 2.3427368409670692e-06,
      "loss": 0.2719,
      "step": 6023
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.274234614025658,
      "learning_rate": 2.342019699545861e-06,
      "loss": 0.5734,
      "step": 6024
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.4714363564739035,
      "learning_rate": 2.3413025711762104e-06,
      "loss": 0.3348,
      "step": 6025
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.743762266997144,
      "learning_rate": 2.340585455917364e-06,
      "loss": 0.2748,
      "step": 6026
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.740222842132052,
      "learning_rate": 2.3398683538285656e-06,
      "loss": 0.1318,
      "step": 6027
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.6917693035978973,
      "learning_rate": 2.339151264969058e-06,
      "loss": 0.1444,
      "step": 6028
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.970438664415516,
      "learning_rate": 2.3384341893980857e-06,
      "loss": 0.1886,
      "step": 6029
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.717935544635934,
      "learning_rate": 2.3377171271748874e-06,
      "loss": 0.0166,
      "step": 6030
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.933031557035584,
      "learning_rate": 2.3370000783587052e-06,
      "loss": 0.3717,
      "step": 6031
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.419342472204932,
      "learning_rate": 2.3362830430087765e-06,
      "loss": 0.3168,
      "step": 6032
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.76065482241379,
      "learning_rate": 2.3355660211843402e-06,
      "loss": 0.1956,
      "step": 6033
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.7635350324751906,
      "learning_rate": 2.3348490129446344e-06,
      "loss": 0.3178,
      "step": 6034
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7219376689609165,
      "learning_rate": 2.334132018348892e-06,
      "loss": 0.1746,
      "step": 6035
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.703770420079324,
      "learning_rate": 2.33341503745635e-06,
      "loss": 0.2645,
      "step": 6036
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.817604905230456,
      "learning_rate": 2.33269807032624e-06,
      "loss": 0.3177,
      "step": 6037
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.6332047075695106,
      "learning_rate": 2.331981117017795e-06,
      "loss": 0.2524,
      "step": 6038
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.262233118979678,
      "learning_rate": 2.331264177590248e-06,
      "loss": 0.3327,
      "step": 6039
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.869404532681201,
      "learning_rate": 2.3305472521028263e-06,
      "loss": 0.3417,
      "step": 6040
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.16104208807127,
      "learning_rate": 2.32983034061476e-06,
      "loss": 0.2084,
      "step": 6041
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.051431880804744,
      "learning_rate": 2.329113443185276e-06,
      "loss": 0.33,
      "step": 6042
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5365432985793404,
      "learning_rate": 2.328396559873602e-06,
      "loss": 0.1282,
      "step": 6043
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.423773018789782,
      "learning_rate": 2.327679690738964e-06,
      "loss": 0.3995,
      "step": 6044
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.39174922353733,
      "learning_rate": 2.326962835840584e-06,
      "loss": 0.1897,
      "step": 6045
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.097913584951836,
      "learning_rate": 2.3262459952376863e-06,
      "loss": 0.427,
      "step": 6046
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.27376379655601,
      "learning_rate": 2.325529168989492e-06,
      "loss": 0.2271,
      "step": 6047
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.8535357713035547,
      "learning_rate": 2.3248123571552226e-06,
      "loss": 0.3716,
      "step": 6048
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.037884376376567,
      "learning_rate": 2.3240955597940985e-06,
      "loss": 0.3372,
      "step": 6049
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.2503580466408035,
      "learning_rate": 2.3233787769653357e-06,
      "loss": 0.2696,
      "step": 6050
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.736490757347629,
      "learning_rate": 2.3226620087281528e-06,
      "loss": 0.4758,
      "step": 6051
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.4558018656060145,
      "learning_rate": 2.321945255141765e-06,
      "loss": 0.3463,
      "step": 6052
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.392908462958059,
      "learning_rate": 2.3212285162653873e-06,
      "loss": 0.4964,
      "step": 6053
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.8628972936266277,
      "learning_rate": 2.320511792158234e-06,
      "loss": 0.2616,
      "step": 6054
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.301862287357437,
      "learning_rate": 2.3197950828795153e-06,
      "loss": 0.3344,
      "step": 6055
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.475748042094122,
      "learning_rate": 2.3190783884884444e-06,
      "loss": 0.2123,
      "step": 6056
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.086619663422267,
      "learning_rate": 2.318361709044229e-06,
      "loss": 0.2352,
      "step": 6057
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.797692692366145,
      "learning_rate": 2.3176450446060787e-06,
      "loss": 0.2226,
      "step": 6058
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2473828513695144,
      "learning_rate": 2.316928395233202e-06,
      "loss": 0.2109,
      "step": 6059
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.335005901215294,
      "learning_rate": 2.3162117609848027e-06,
      "loss": 0.4136,
      "step": 6060
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.697661451673527,
      "learning_rate": 2.315495141920087e-06,
      "loss": 0.2964,
      "step": 6061
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.943489703140045,
      "learning_rate": 2.3147785380982576e-06,
      "loss": 0.4144,
      "step": 6062
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.421876235205083,
      "learning_rate": 2.3140619495785175e-06,
      "loss": 0.3449,
      "step": 6063
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.786064896204798,
      "learning_rate": 2.3133453764200684e-06,
      "loss": 0.2943,
      "step": 6064
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.56292151416855,
      "learning_rate": 2.3126288186821083e-06,
      "loss": 0.3029,
      "step": 6065
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.47605916734642195,
      "learning_rate": 2.311912276423837e-06,
      "loss": 0.0033,
      "step": 6066
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.153659569503696,
      "learning_rate": 2.3111957497044508e-06,
      "loss": 0.373,
      "step": 6067
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.020280109417581,
      "learning_rate": 2.3104792385831458e-06,
      "loss": 0.3833,
      "step": 6068
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.9082828129954827,
      "learning_rate": 2.309762743119118e-06,
      "loss": 0.1768,
      "step": 6069
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.3984779228475213,
      "learning_rate": 2.309046263371559e-06,
      "loss": 0.1736,
      "step": 6070
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.811537428263458,
      "learning_rate": 2.3083297993996613e-06,
      "loss": 0.4418,
      "step": 6071
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.785215215943229,
      "learning_rate": 2.3076133512626166e-06,
      "loss": 0.1887,
      "step": 6072
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.045901254977352,
      "learning_rate": 2.306896919019613e-06,
      "loss": 0.2101,
      "step": 6073
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.419517590647231,
      "learning_rate": 2.3061805027298403e-06,
      "loss": 0.3092,
      "step": 6074
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.181708433059207,
      "learning_rate": 2.3054641024524827e-06,
      "loss": 0.1722,
      "step": 6075
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.7729567821669843,
      "learning_rate": 2.304747718246728e-06,
      "loss": 0.217,
      "step": 6076
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.9170547842751926,
      "learning_rate": 2.304031350171759e-06,
      "loss": 0.134,
      "step": 6077
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.9150928830944993,
      "learning_rate": 2.3033149982867596e-06,
      "loss": 0.2476,
      "step": 6078
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.744964000905609,
      "learning_rate": 2.302598662650911e-06,
      "loss": 0.2718,
      "step": 6079
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.717379824444088,
      "learning_rate": 2.3018823433233925e-06,
      "loss": 0.4799,
      "step": 6080
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.192050396289763,
      "learning_rate": 2.3011660403633836e-06,
      "loss": 0.2327,
      "step": 6081
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.453884928294412,
      "learning_rate": 2.3004497538300618e-06,
      "loss": 0.1723,
      "step": 6082
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.242609589800027,
      "learning_rate": 2.2997334837826023e-06,
      "loss": 0.1549,
      "step": 6083
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.168154587930148,
      "learning_rate": 2.2990172302801824e-06,
      "loss": 0.3585,
      "step": 6084
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.79433869440843,
      "learning_rate": 2.298300993381972e-06,
      "loss": 0.4671,
      "step": 6085
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.559456880441315,
      "learning_rate": 2.2975847731471444e-06,
      "loss": 0.2188,
      "step": 6086
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.10621230066585,
      "learning_rate": 2.296868569634871e-06,
      "loss": 0.3681,
      "step": 6087
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.285749852585626,
      "learning_rate": 2.2961523829043203e-06,
      "loss": 0.3556,
      "step": 6088
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.3130648440382986,
      "learning_rate": 2.2954362130146613e-06,
      "loss": 0.1987,
      "step": 6089
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.880898936247226,
      "learning_rate": 2.294720060025058e-06,
      "loss": 0.6263,
      "step": 6090
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.492131827354366,
      "learning_rate": 2.2940039239946775e-06,
      "loss": 0.4213,
      "step": 6091
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.887045069354183,
      "learning_rate": 2.293287804982683e-06,
      "loss": 0.3297,
      "step": 6092
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.641853951686769,
      "learning_rate": 2.2925717030482362e-06,
      "loss": 0.3777,
      "step": 6093
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.18768159860267,
      "learning_rate": 2.291855618250499e-06,
      "loss": 0.2176,
      "step": 6094
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.8056614746750776,
      "learning_rate": 2.29113955064863e-06,
      "loss": 0.1773,
      "step": 6095
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.8012284866860835,
      "learning_rate": 2.2904235003017867e-06,
      "loss": 0.1695,
      "step": 6096
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1657237134122247,
      "learning_rate": 2.2897074672691265e-06,
      "loss": 0.0635,
      "step": 6097
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.4816127543230455,
      "learning_rate": 2.2889914516098045e-06,
      "loss": 0.2686,
      "step": 6098
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.365852590452632,
      "learning_rate": 2.2882754533829753e-06,
      "loss": 0.2413,
      "step": 6099
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.694496331446915,
      "learning_rate": 2.2875594726477886e-06,
      "loss": 0.4125,
      "step": 6100
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.119212467920366,
      "learning_rate": 2.286843509463397e-06,
      "loss": 0.3238,
      "step": 6101
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.252753373575381,
      "learning_rate": 2.28612756388895e-06,
      "loss": 0.3723,
      "step": 6102
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.235128325741988,
      "learning_rate": 2.2854116359835947e-06,
      "loss": 0.1919,
      "step": 6103
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.0620009722554435,
      "learning_rate": 2.284695725806479e-06,
      "loss": 0.2959,
      "step": 6104
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.148764615363517,
      "learning_rate": 2.283979833416746e-06,
      "loss": 0.429,
      "step": 6105
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.053042184039074,
      "learning_rate": 2.2832639588735393e-06,
      "loss": 0.1289,
      "step": 6106
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.0363814016421005,
      "learning_rate": 2.282548102236003e-06,
      "loss": 0.2323,
      "step": 6107
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.9131881184451895,
      "learning_rate": 2.281832263563276e-06,
      "loss": 0.354,
      "step": 6108
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.1495319048662305,
      "learning_rate": 2.2811164429144968e-06,
      "loss": 0.2533,
      "step": 6109
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.102502741971985,
      "learning_rate": 2.2804006403488056e-06,
      "loss": 0.5185,
      "step": 6110
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.53574876159999,
      "learning_rate": 2.279684855925336e-06,
      "loss": 0.3317,
      "step": 6111
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.313069375828019,
      "learning_rate": 2.278969089703224e-06,
      "loss": 0.2909,
      "step": 6112
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.633261744126608,
      "learning_rate": 2.278253341741601e-06,
      "loss": 0.0041,
      "step": 6113
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.606871109372734,
      "learning_rate": 2.2775376120996e-06,
      "loss": 0.2595,
      "step": 6114
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.587909217418479,
      "learning_rate": 2.2768219008363525e-06,
      "loss": 0.4233,
      "step": 6115
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.8132495727754465,
      "learning_rate": 2.2761062080109837e-06,
      "loss": 0.2641,
      "step": 6116
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2255236112966688,
      "learning_rate": 2.2753905336826227e-06,
      "loss": 0.2805,
      "step": 6117
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.073390137193674,
      "learning_rate": 2.2746748779103942e-06,
      "loss": 0.3679,
      "step": 6118
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.942888057783215,
      "learning_rate": 2.2739592407534232e-06,
      "loss": 0.5422,
      "step": 6119
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.4410827800582644,
      "learning_rate": 2.273243622270832e-06,
      "loss": 0.0851,
      "step": 6120
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.439239747177113,
      "learning_rate": 2.27252802252174e-06,
      "loss": 0.481,
      "step": 6121
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.750996958161855,
      "learning_rate": 2.271812441565268e-06,
      "loss": 0.2538,
      "step": 6122
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.543720926944283,
      "learning_rate": 2.271096879460533e-06,
      "loss": 0.5086,
      "step": 6123
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5760416898100895,
      "learning_rate": 2.2703813362666514e-06,
      "loss": 0.068,
      "step": 6124
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.73553842496498,
      "learning_rate": 2.2696658120427393e-06,
      "loss": 0.34,
      "step": 6125
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.554541999709323,
      "learning_rate": 2.2689503068479072e-06,
      "loss": 0.4725,
      "step": 6126
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.665107612047309,
      "learning_rate": 2.268234820741269e-06,
      "loss": 0.4725,
      "step": 6127
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.504950565774146,
      "learning_rate": 2.2675193537819326e-06,
      "loss": 0.2217,
      "step": 6128
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.829374234802461,
      "learning_rate": 2.2668039060290075e-06,
      "loss": 0.2493,
      "step": 6129
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.2105543365771765,
      "learning_rate": 2.2660884775416016e-06,
      "loss": 0.225,
      "step": 6130
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.809343000371298,
      "learning_rate": 2.2653730683788178e-06,
      "loss": 0.2954,
      "step": 6131
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.969087132190936,
      "learning_rate": 2.2646576785997613e-06,
      "loss": 0.2486,
      "step": 6132
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.867134131997059,
      "learning_rate": 2.263942308263533e-06,
      "loss": 0.3063,
      "step": 6133
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.16815865764904,
      "learning_rate": 2.2632269574292336e-06,
      "loss": 0.2707,
      "step": 6134
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.8545737062688272,
      "learning_rate": 2.2625116261559637e-06,
      "loss": 0.2831,
      "step": 6135
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.873251047439242,
      "learning_rate": 2.2617963145028175e-06,
      "loss": 0.3427,
      "step": 6136
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.2144692354707045,
      "learning_rate": 2.2610810225288927e-06,
      "loss": 0.3587,
      "step": 6137
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.062635504840198,
      "learning_rate": 2.2603657502932814e-06,
      "loss": 0.2128,
      "step": 6138
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.984282663497506,
      "learning_rate": 2.2596504978550777e-06,
      "loss": 0.2878,
      "step": 6139
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.5558988873993447,
      "learning_rate": 2.258935265273372e-06,
      "loss": 0.1384,
      "step": 6140
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.106993394832651,
      "learning_rate": 2.258220052607252e-06,
      "loss": 0.2913,
      "step": 6141
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.306698625362802,
      "learning_rate": 2.2575048599158064e-06,
      "loss": 0.0972,
      "step": 6142
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.772294178752315,
      "learning_rate": 2.25678968725812e-06,
      "loss": 0.1394,
      "step": 6143
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.003311040226902,
      "learning_rate": 2.256074534693277e-06,
      "loss": 0.3484,
      "step": 6144
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.002069508201448,
      "learning_rate": 2.255359402280361e-06,
      "loss": 0.4936,
      "step": 6145
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.731537962151003,
      "learning_rate": 2.254644290078451e-06,
      "loss": 0.1645,
      "step": 6146
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.301279415771619,
      "learning_rate": 2.2539291981466273e-06,
      "loss": 0.2852,
      "step": 6147
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.708189503819798,
      "learning_rate": 2.253214126543966e-06,
      "loss": 0.4333,
      "step": 6148
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.675580556195086,
      "learning_rate": 2.252499075329544e-06,
      "loss": 0.3108,
      "step": 6149
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.9281971672190528,
      "learning_rate": 2.2517840445624357e-06,
      "loss": 0.3003,
      "step": 6150
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.098208437443793,
      "learning_rate": 2.251069034301712e-06,
      "loss": 0.2946,
      "step": 6151
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.327723514609966,
      "learning_rate": 2.250354044606444e-06,
      "loss": 0.3062,
      "step": 6152
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.8848899463938893,
      "learning_rate": 2.2496390755357013e-06,
      "loss": 0.2004,
      "step": 6153
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.564025153955028,
      "learning_rate": 2.2489241271485507e-06,
      "loss": 0.2234,
      "step": 6154
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.2470198891761335,
      "learning_rate": 2.2482091995040582e-06,
      "loss": 0.338,
      "step": 6155
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.365013478511924,
      "learning_rate": 2.247494292661287e-06,
      "loss": 0.3854,
      "step": 6156
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.053689846372204,
      "learning_rate": 2.2467794066792983e-06,
      "loss": 0.1191,
      "step": 6157
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.603324013843815,
      "learning_rate": 2.2460645416171545e-06,
      "loss": 0.3106,
      "step": 6158
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.324489408035275,
      "learning_rate": 2.245349697533913e-06,
      "loss": 0.3704,
      "step": 6159
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3722686361330876,
      "learning_rate": 2.244634874488632e-06,
      "loss": 0.1207,
      "step": 6160
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4677156843925725,
      "learning_rate": 2.2439200725403644e-06,
      "loss": 0.245,
      "step": 6161
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.339166555759312,
      "learning_rate": 2.243205291748165e-06,
      "loss": 0.3998,
      "step": 6162
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4820798702949856,
      "learning_rate": 2.242490532171086e-06,
      "loss": 0.2797,
      "step": 6163
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4828293697182775,
      "learning_rate": 2.241775793868176e-06,
      "loss": 0.2352,
      "step": 6164
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.676251422041181,
      "learning_rate": 2.2410610768984854e-06,
      "loss": 0.3273,
      "step": 6165
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.236877612134774,
      "learning_rate": 2.2403463813210576e-06,
      "loss": 0.1122,
      "step": 6166
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.406556889522573,
      "learning_rate": 2.239631707194939e-06,
      "loss": 0.2494,
      "step": 6167
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.837455558266182,
      "learning_rate": 2.238917054579173e-06,
      "loss": 0.2858,
      "step": 6168
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.258898516513891,
      "learning_rate": 2.2382024235327986e-06,
      "loss": 0.3634,
      "step": 6169
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.998496391005162,
      "learning_rate": 2.2374878141148583e-06,
      "loss": 0.5098,
      "step": 6170
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.0226734548560454,
      "learning_rate": 2.2367732263843865e-06,
      "loss": 0.1887,
      "step": 6171
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.046665057929935,
      "learning_rate": 2.23605866040042e-06,
      "loss": 0.016,
      "step": 6172
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.227971987605703,
      "learning_rate": 2.2353441162219937e-06,
      "loss": 0.3422,
      "step": 6173
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.7202140984858874,
      "learning_rate": 2.234629593908138e-06,
      "loss": 0.1555,
      "step": 6174
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.17352913623334,
      "learning_rate": 2.233915093517886e-06,
      "loss": 0.2726,
      "step": 6175
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.047981337851216,
      "learning_rate": 2.233200615110263e-06,
      "loss": 0.3118,
      "step": 6176
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9621440381795472,
      "learning_rate": 2.232486158744297e-06,
      "loss": 0.0792,
      "step": 6177
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.8560830202227665,
      "learning_rate": 2.231771724479014e-06,
      "loss": 0.268,
      "step": 6178
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.313190041558848,
      "learning_rate": 2.231057312373435e-06,
      "loss": 0.2854,
      "step": 6179
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.009126947925171,
      "learning_rate": 2.2303429224865837e-06,
      "loss": 0.2512,
      "step": 6180
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3092196482632565,
      "learning_rate": 2.2296285548774764e-06,
      "loss": 0.2043,
      "step": 6181
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.301542803960705,
      "learning_rate": 2.228914209605133e-06,
      "loss": 0.2745,
      "step": 6182
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.303140707653817,
      "learning_rate": 2.228199886728569e-06,
      "loss": 0.3598,
      "step": 6183
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3094377125490064,
      "learning_rate": 2.227485586306797e-06,
      "loss": 0.1707,
      "step": 6184
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.2856773921245646,
      "learning_rate": 2.226771308398831e-06,
      "loss": 0.1474,
      "step": 6185
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.7714376101263785,
      "learning_rate": 2.2260570530636792e-06,
      "loss": 0.3226,
      "step": 6186
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8328849029965102,
      "learning_rate": 2.2253428203603503e-06,
      "loss": 0.0747,
      "step": 6187
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.1404372588951914,
      "learning_rate": 2.2246286103478515e-06,
      "loss": 0.1835,
      "step": 6188
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4919021055444084,
      "learning_rate": 2.223914423085186e-06,
      "loss": 0.2125,
      "step": 6189
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.957116670891337,
      "learning_rate": 2.2232002586313576e-06,
      "loss": 0.3727,
      "step": 6190
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.032532361384196,
      "learning_rate": 2.222486117045368e-06,
      "loss": 0.2906,
      "step": 6191
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.898502918369513,
      "learning_rate": 2.221771998386214e-06,
      "loss": 0.1943,
      "step": 6192
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.913181359235777,
      "learning_rate": 2.2210579027128936e-06,
      "loss": 0.2265,
      "step": 6193
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.577364984763552,
      "learning_rate": 2.220343830084401e-06,
      "loss": 0.2306,
      "step": 6194
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.519193065028082,
      "learning_rate": 2.21962978055973e-06,
      "loss": 0.5194,
      "step": 6195
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8562686725569058,
      "learning_rate": 2.2189157541978735e-06,
      "loss": 0.0742,
      "step": 6196
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.113626369768566,
      "learning_rate": 2.2182017510578175e-06,
      "loss": 0.4248,
      "step": 6197
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.753052347397892,
      "learning_rate": 2.2174877711985527e-06,
      "loss": 0.3126,
      "step": 6198
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.586193417159442,
      "learning_rate": 2.2167738146790623e-06,
      "loss": 0.2437,
      "step": 6199
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.263744853450838,
      "learning_rate": 2.2160598815583306e-06,
      "loss": 0.3818,
      "step": 6200
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.866840697910093,
      "learning_rate": 2.2153459718953403e-06,
      "loss": 0.3053,
      "step": 6201
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.3144966467943338,
      "learning_rate": 2.2146320857490695e-06,
      "loss": 0.2453,
      "step": 6202
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.159010976490989,
      "learning_rate": 2.2139182231784968e-06,
      "loss": 0.3381,
      "step": 6203
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.157753799901266,
      "learning_rate": 2.2132043842425975e-06,
      "loss": 0.3443,
      "step": 6204
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.0161492599673325,
      "learning_rate": 2.212490569000346e-06,
      "loss": 0.3389,
      "step": 6205
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.6337179640964905,
      "learning_rate": 2.2117767775107156e-06,
      "loss": 0.3958,
      "step": 6206
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.047730732180072,
      "learning_rate": 2.2110630098326726e-06,
      "loss": 0.2688,
      "step": 6207
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.0812435055976595,
      "learning_rate": 2.210349266025189e-06,
      "loss": 0.4606,
      "step": 6208
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.071968238338452,
      "learning_rate": 2.209635546147228e-06,
      "loss": 0.597,
      "step": 6209
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.144915362628559,
      "learning_rate": 2.2089218502577542e-06,
      "loss": 0.129,
      "step": 6210
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4347507011157994,
      "learning_rate": 2.2082081784157315e-06,
      "loss": 0.1662,
      "step": 6211
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.352335737318611,
      "learning_rate": 2.2074945306801165e-06,
      "loss": 0.1849,
      "step": 6212
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.7657635666544795,
      "learning_rate": 2.2067809071098713e-06,
      "loss": 0.3543,
      "step": 6213
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.255471498223544,
      "learning_rate": 2.2060673077639488e-06,
      "loss": 0.3247,
      "step": 6214
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.376874316055062,
      "learning_rate": 2.2053537327013043e-06,
      "loss": 0.4858,
      "step": 6215
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.0139295363872405,
      "learning_rate": 2.20464018198089e-06,
      "loss": 0.2178,
      "step": 6216
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.7577911619887376,
      "learning_rate": 2.2039266556616556e-06,
      "loss": 0.3781,
      "step": 6217
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.1340427131619544,
      "learning_rate": 2.2032131538025507e-06,
      "loss": 0.1784,
      "step": 6218
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.78550162890688,
      "learning_rate": 2.2024996764625183e-06,
      "loss": 0.5065,
      "step": 6219
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.865307063220718,
      "learning_rate": 2.2017862237005042e-06,
      "loss": 0.1575,
      "step": 6220
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.459807195769542,
      "learning_rate": 2.201072795575451e-06,
      "loss": 0.3529,
      "step": 6221
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.846934200685975,
      "learning_rate": 2.200359392146297e-06,
      "loss": 0.1745,
      "step": 6222
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.853489189533381,
      "learning_rate": 2.1996460134719825e-06,
      "loss": 0.4667,
      "step": 6223
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4340067514471455,
      "learning_rate": 2.1989326596114403e-06,
      "loss": 0.1207,
      "step": 6224
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.119720219113461,
      "learning_rate": 2.198219330623606e-06,
      "loss": 0.265,
      "step": 6225
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.626591652219807,
      "learning_rate": 2.197506026567412e-06,
      "loss": 0.1592,
      "step": 6226
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.7126967133865705,
      "learning_rate": 2.1967927475017856e-06,
      "loss": 0.4162,
      "step": 6227
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.380574619080912,
      "learning_rate": 2.196079493485657e-06,
      "loss": 0.197,
      "step": 6228
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.786716510919124,
      "learning_rate": 2.1953662645779503e-06,
      "loss": 0.2424,
      "step": 6229
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.484936369406902,
      "learning_rate": 2.194653060837589e-06,
      "loss": 0.4946,
      "step": 6230
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.003187794915971,
      "learning_rate": 2.193939882323495e-06,
      "loss": 0.3112,
      "step": 6231
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.8855906747085522,
      "learning_rate": 2.193226729094587e-06,
      "loss": 0.2181,
      "step": 6232
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.216188967725288,
      "learning_rate": 2.1925136012097824e-06,
      "loss": 0.2723,
      "step": 6233
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.628813191416657,
      "learning_rate": 2.191800498727998e-06,
      "loss": 0.2386,
      "step": 6234
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.761455637600543,
      "learning_rate": 2.191087421708144e-06,
      "loss": 0.1359,
      "step": 6235
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.714308765941496,
      "learning_rate": 2.1903743702091335e-06,
      "loss": 0.3348,
      "step": 6236
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.793162193721143,
      "learning_rate": 2.1896613442898737e-06,
      "loss": 0.2697,
      "step": 6237
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.303936213423527,
      "learning_rate": 2.1889483440092724e-06,
      "loss": 0.4648,
      "step": 6238
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.388850175386015,
      "learning_rate": 2.188235369426235e-06,
      "loss": 0.093,
      "step": 6239
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.859612908302548,
      "learning_rate": 2.1875224205996615e-06,
      "loss": 0.3894,
      "step": 6240
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.813233272758551,
      "learning_rate": 2.1868094975884545e-06,
      "loss": 0.2326,
      "step": 6241
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.625429515397173,
      "learning_rate": 2.1860966004515106e-06,
      "loss": 0.2848,
      "step": 6242
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.4003554416801727,
      "learning_rate": 2.1853837292477267e-06,
      "loss": 0.1761,
      "step": 6243
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.4549422705721198,
      "learning_rate": 2.1846708840359975e-06,
      "loss": 0.148,
      "step": 6244
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.86661944281575,
      "learning_rate": 2.1839580648752126e-06,
      "loss": 0.3933,
      "step": 6245
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.5550331138490785,
      "learning_rate": 2.183245271824264e-06,
      "loss": 0.3109,
      "step": 6246
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.565744244033101,
      "learning_rate": 2.182532504942037e-06,
      "loss": 0.5504,
      "step": 6247
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.221164087428385,
      "learning_rate": 2.1818197642874183e-06,
      "loss": 0.2188,
      "step": 6248
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.81736907499264,
      "learning_rate": 2.181107049919292e-06,
      "loss": 0.0227,
      "step": 6249
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.115612282640217,
      "learning_rate": 2.180394361896536e-06,
      "loss": 0.4351,
      "step": 6250
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.3785449524058175,
      "learning_rate": 2.179681700278032e-06,
      "loss": 0.1399,
      "step": 6251
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.354441165454706,
      "learning_rate": 2.178969065122655e-06,
      "loss": 0.5366,
      "step": 6252
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.406111775460352,
      "learning_rate": 2.1782564564892797e-06,
      "loss": 0.4975,
      "step": 6253
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.9586375103884017,
      "learning_rate": 2.1775438744367798e-06,
      "loss": 0.1828,
      "step": 6254
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.789633301820669,
      "learning_rate": 2.176831319024023e-06,
      "loss": 0.3116,
      "step": 6255
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4967966050921286,
      "learning_rate": 2.1761187903098785e-06,
      "loss": 0.24,
      "step": 6256
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7481953328988884,
      "learning_rate": 2.1754062883532114e-06,
      "loss": 0.1323,
      "step": 6257
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.282621770008579,
      "learning_rate": 2.174693813212885e-06,
      "loss": 0.2342,
      "step": 6258
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.8778004992015274,
      "learning_rate": 2.173981364947762e-06,
      "loss": 0.2788,
      "step": 6259
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.011117240368957,
      "learning_rate": 2.1732689436166994e-06,
      "loss": 0.1976,
      "step": 6260
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.141859259593418,
      "learning_rate": 2.172556549278555e-06,
      "loss": 0.2203,
      "step": 6261
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9854504812433507,
      "learning_rate": 2.171844181992183e-06,
      "loss": 0.302,
      "step": 6262
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.976271889057248,
      "learning_rate": 2.1711318418164356e-06,
      "loss": 0.3648,
      "step": 6263
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.176907932365967,
      "learning_rate": 2.1704195288101643e-06,
      "loss": 0.0474,
      "step": 6264
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.663001346807217,
      "learning_rate": 2.1697072430322142e-06,
      "loss": 0.4541,
      "step": 6265
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.124052087492653,
      "learning_rate": 2.168994984541433e-06,
      "loss": 0.4958,
      "step": 6266
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.674452143244945,
      "learning_rate": 2.1682827533966635e-06,
      "loss": 0.4876,
      "step": 6267
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4183782643507934,
      "learning_rate": 2.1675705496567457e-06,
      "loss": 0.1998,
      "step": 6268
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.1526123344754815,
      "learning_rate": 2.166858373380521e-06,
      "loss": 0.1894,
      "step": 6269
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.98373072403967,
      "learning_rate": 2.166146224626823e-06,
      "loss": 0.3047,
      "step": 6270
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7681556079887675,
      "learning_rate": 2.1654341034544867e-06,
      "loss": 0.2719,
      "step": 6271
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.245685182651288,
      "learning_rate": 2.164722009922345e-06,
      "loss": 0.3447,
      "step": 6272
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.1278583377482425,
      "learning_rate": 2.164009944089227e-06,
      "loss": 0.3279,
      "step": 6273
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3831431239133694,
      "learning_rate": 2.163297906013961e-06,
      "loss": 0.1514,
      "step": 6274
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.618478529027858,
      "learning_rate": 2.1625858957553705e-06,
      "loss": 0.1552,
      "step": 6275
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9234826132211973,
      "learning_rate": 2.1618739133722787e-06,
      "loss": 0.0719,
      "step": 6276
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.242871173561612,
      "learning_rate": 2.1611619589235065e-06,
      "loss": 0.499,
      "step": 6277
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.80982434545676,
      "learning_rate": 2.1604500324678725e-06,
      "loss": 0.2533,
      "step": 6278
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3725269006596483,
      "learning_rate": 2.1597381340641923e-06,
      "loss": 0.3039,
      "step": 6279
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.6349723987105618,
      "learning_rate": 2.1590262637712786e-06,
      "loss": 0.2275,
      "step": 6280
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.702765971126637,
      "learning_rate": 2.158314421647943e-06,
      "loss": 0.4161,
      "step": 6281
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.848102670888579,
      "learning_rate": 2.157602607752996e-06,
      "loss": 0.4334,
      "step": 6282
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3298629071844723,
      "learning_rate": 2.1568908221452416e-06,
      "loss": 0.2507,
      "step": 6283
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4325413569060736,
      "learning_rate": 2.1561790648834864e-06,
      "loss": 0.2616,
      "step": 6284
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.379304524159604,
      "learning_rate": 2.1554673360265304e-06,
      "loss": 0.3251,
      "step": 6285
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.442815339139983,
      "learning_rate": 2.1547556356331734e-06,
      "loss": 0.0623,
      "step": 6286
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.581982614428395,
      "learning_rate": 2.1540439637622144e-06,
      "loss": 0.4456,
      "step": 6287
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.9250989989007214,
      "learning_rate": 2.153332320472446e-06,
      "loss": 0.1472,
      "step": 6288
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.000352945943562,
      "learning_rate": 2.152620705822663e-06,
      "loss": 0.4692,
      "step": 6289
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.2965410530863246,
      "learning_rate": 2.151909119871653e-06,
      "loss": 0.1778,
      "step": 6290
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.3565774568153883,
      "learning_rate": 2.151197562678205e-06,
      "loss": 0.195,
      "step": 6291
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.80187899776255,
      "learning_rate": 2.150486034301105e-06,
      "loss": 0.1754,
      "step": 6292
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3533806855035286,
      "learning_rate": 2.149774534799134e-06,
      "loss": 0.0509,
      "step": 6293
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.38947586155559,
      "learning_rate": 2.1490630642310754e-06,
      "loss": 0.2879,
      "step": 6294
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.524276026684496,
      "learning_rate": 2.148351622655705e-06,
      "loss": 0.3019,
      "step": 6295
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.75286740970991,
      "learning_rate": 2.147640210131799e-06,
      "loss": 0.3249,
      "step": 6296
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.115812482806369,
      "learning_rate": 2.1469288267181322e-06,
      "loss": 0.24,
      "step": 6297
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9858394414036566,
      "learning_rate": 2.146217472473474e-06,
      "loss": 0.2161,
      "step": 6298
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.285723301907709,
      "learning_rate": 2.1455061474565952e-06,
      "loss": 0.3916,
      "step": 6299
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.445429364196582,
      "learning_rate": 2.144794851726259e-06,
      "loss": 0.3852,
      "step": 6300
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.449222756952777,
      "learning_rate": 2.144083585341231e-06,
      "loss": 0.4485,
      "step": 6301
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.17728401494013,
      "learning_rate": 2.1433723483602725e-06,
      "loss": 0.3519,
      "step": 6302
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4047663711736216,
      "learning_rate": 2.1426611408421417e-06,
      "loss": 0.2805,
      "step": 6303
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4939783127042787,
      "learning_rate": 2.1419499628455968e-06,
      "loss": 0.276,
      "step": 6304
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.4644049341067475,
      "learning_rate": 2.141238814429389e-06,
      "loss": 0.364,
      "step": 6305
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.180840319822861,
      "learning_rate": 2.1405276956522716e-06,
      "loss": 0.3778,
      "step": 6306
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.821329131897939,
      "learning_rate": 2.139816606572994e-06,
      "loss": 0.118,
      "step": 6307
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.416823167530231,
      "learning_rate": 2.1391055472503024e-06,
      "loss": 0.2443,
      "step": 6308
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.960845615089512,
      "learning_rate": 2.138394517742942e-06,
      "loss": 0.5044,
      "step": 6309
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.6665611603852435,
      "learning_rate": 2.1376835181096527e-06,
      "loss": 0.0696,
      "step": 6310
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.5809670232251847,
      "learning_rate": 2.1369725484091745e-06,
      "loss": 0.2746,
      "step": 6311
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.79165339875506,
      "learning_rate": 2.1362616087002454e-06,
      "loss": 0.3508,
      "step": 6312
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.871608690246611,
      "learning_rate": 2.1355506990415977e-06,
      "loss": 0.2025,
      "step": 6313
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.1664368099362425,
      "learning_rate": 2.134839819491965e-06,
      "loss": 0.3048,
      "step": 6314
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7587788805989955,
      "learning_rate": 2.134128970110077e-06,
      "loss": 0.253,
      "step": 6315
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2293474700906573,
      "learning_rate": 2.133418150954659e-06,
      "loss": 0.1288,
      "step": 6316
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.56787431169331,
      "learning_rate": 2.1327073620844362e-06,
      "loss": 0.4225,
      "step": 6317
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.875133997905079,
      "learning_rate": 2.13199660355813e-06,
      "loss": 0.3696,
      "step": 6318
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.564225185550136,
      "learning_rate": 2.1312858754344605e-06,
      "loss": 0.4737,
      "step": 6319
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.779814835421954,
      "learning_rate": 2.130575177772145e-06,
      "loss": 0.2601,
      "step": 6320
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.478583509201235,
      "learning_rate": 2.129864510629896e-06,
      "loss": 0.1327,
      "step": 6321
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.625063653635908,
      "learning_rate": 2.1291538740664273e-06,
      "loss": 0.2039,
      "step": 6322
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4696602015720286,
      "learning_rate": 2.1284432681404466e-06,
      "loss": 0.2268,
      "step": 6323
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.686762001302785,
      "learning_rate": 2.127732692910661e-06,
      "loss": 0.5215,
      "step": 6324
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.363741517292609,
      "learning_rate": 2.1270221484357773e-06,
      "loss": 0.4242,
      "step": 6325
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.307153145916411,
      "learning_rate": 2.1263116347744934e-06,
      "loss": 0.2387,
      "step": 6326
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.162443193793711,
      "learning_rate": 2.1256011519855106e-06,
      "loss": 0.3769,
      "step": 6327
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.21756670098228,
      "learning_rate": 2.1248907001275245e-06,
      "loss": 0.2373,
      "step": 6328
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4577600665281705,
      "learning_rate": 2.12418027925923e-06,
      "loss": 0.1936,
      "step": 6329
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.896339732140739,
      "learning_rate": 2.123469889439319e-06,
      "loss": 0.1879,
      "step": 6330
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.402568834530567,
      "learning_rate": 2.1227595307264787e-06,
      "loss": 0.1975,
      "step": 6331
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.7327002353697254,
      "learning_rate": 2.1220492031793974e-06,
      "loss": 0.1408,
      "step": 6332
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.501913971997804,
      "learning_rate": 2.121338906856757e-06,
      "loss": 0.3627,
      "step": 6333
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.74882816487407,
      "learning_rate": 2.12062864181724e-06,
      "loss": 0.3749,
      "step": 6334
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.2115128724999815,
      "learning_rate": 2.1199184081195253e-06,
      "loss": 0.3875,
      "step": 6335
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.365892099835821,
      "learning_rate": 2.1192082058222873e-06,
      "loss": 0.5126,
      "step": 6336
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.095578985698966,
      "learning_rate": 2.118498034984201e-06,
      "loss": 0.382,
      "step": 6337
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7054334619353506,
      "learning_rate": 2.1177878956639363e-06,
      "loss": 0.212,
      "step": 6338
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.864592046062179,
      "learning_rate": 2.1170777879201616e-06,
      "loss": 0.3846,
      "step": 6339
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.013172084378442,
      "learning_rate": 2.116367711811544e-06,
      "loss": 0.3583,
      "step": 6340
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.003357227057462,
      "learning_rate": 2.1156576673967437e-06,
      "loss": 0.2697,
      "step": 6341
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.333767201649507,
      "learning_rate": 2.1149476547344234e-06,
      "loss": 0.2539,
      "step": 6342
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.807065361266682,
      "learning_rate": 2.1142376738832396e-06,
      "loss": 0.1996,
      "step": 6343
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.9020332476652255,
      "learning_rate": 2.1135277249018478e-06,
      "loss": 0.2215,
      "step": 6344
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.8322095551068087,
      "learning_rate": 2.1128178078489013e-06,
      "loss": 0.3259,
      "step": 6345
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.111583048933859,
      "learning_rate": 2.1121079227830487e-06,
      "loss": 0.1035,
      "step": 6346
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.037062474662022,
      "learning_rate": 2.111398069762938e-06,
      "loss": 0.3579,
      "step": 6347
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.491648445082496,
      "learning_rate": 2.1106882488472135e-06,
      "loss": 0.2725,
      "step": 6348
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.824846461734468,
      "learning_rate": 2.1099784600945165e-06,
      "loss": 0.5491,
      "step": 6349
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.5960979630734076,
      "learning_rate": 2.1092687035634886e-06,
      "loss": 0.1293,
      "step": 6350
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.250696763222136,
      "learning_rate": 2.1085589793127635e-06,
      "loss": 0.1053,
      "step": 6351
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.909983766682349,
      "learning_rate": 2.1078492874009766e-06,
      "loss": 0.4284,
      "step": 6352
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.981574933475329,
      "learning_rate": 2.1071396278867588e-06,
      "loss": 0.2534,
      "step": 6353
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.4707784350467508,
      "learning_rate": 2.106430000828739e-06,
      "loss": 0.1913,
      "step": 6354
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.7435435855601185,
      "learning_rate": 2.1057204062855443e-06,
      "loss": 0.2719,
      "step": 6355
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.038170609520437,
      "learning_rate": 2.1050108443157948e-06,
      "loss": 0.3252,
      "step": 6356
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.492105010787709,
      "learning_rate": 2.1043013149781132e-06,
      "loss": 0.465,
      "step": 6357
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7526497696781296,
      "learning_rate": 2.1035918183311173e-06,
      "loss": 0.2445,
      "step": 6358
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.176028728072308,
      "learning_rate": 2.102882354433421e-06,
      "loss": 0.2843,
      "step": 6359
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.418410080559542,
      "learning_rate": 2.1021729233436393e-06,
      "loss": 0.2289,
      "step": 6360
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.696530913494172,
      "learning_rate": 2.101463525120379e-06,
      "loss": 0.3876,
      "step": 6361
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.185704208247117,
      "learning_rate": 2.100754159822248e-06,
      "loss": 0.4499,
      "step": 6362
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0998407267666184,
      "learning_rate": 2.100044827507852e-06,
      "loss": 0.2547,
      "step": 6363
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.247589520874369,
      "learning_rate": 2.0993355282357903e-06,
      "loss": 0.2661,
      "step": 6364
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.0696276989955864,
      "learning_rate": 2.0986262620646645e-06,
      "loss": 0.2254,
      "step": 6365
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.5912044835322643,
      "learning_rate": 2.097917029053068e-06,
      "loss": 0.2709,
      "step": 6366
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.293724672152894,
      "learning_rate": 2.097207829259595e-06,
      "loss": 0.2401,
      "step": 6367
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.332277799071495,
      "learning_rate": 2.096498662742837e-06,
      "loss": 0.2223,
      "step": 6368
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.913398832442126,
      "learning_rate": 2.0957895295613808e-06,
      "loss": 0.0777,
      "step": 6369
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.7550377018103998,
      "learning_rate": 2.095080429773813e-06,
      "loss": 0.288,
      "step": 6370
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.8964288842885075,
      "learning_rate": 2.0943713634387137e-06,
      "loss": 0.4212,
      "step": 6371
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.956177988327319,
      "learning_rate": 2.0936623306146638e-06,
      "loss": 0.261,
      "step": 6372
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.187860036811762,
      "learning_rate": 2.09295333136024e-06,
      "loss": 0.1603,
      "step": 6373
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.3348720506417933,
      "learning_rate": 2.0922443657340163e-06,
      "loss": 0.222,
      "step": 6374
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.080857977823227,
      "learning_rate": 2.091535433794565e-06,
      "loss": 0.273,
      "step": 6375
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.932126775156374,
      "learning_rate": 2.0908265356004525e-06,
      "loss": 0.2558,
      "step": 6376
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.802460694976878,
      "learning_rate": 2.0901176712102454e-06,
      "loss": 0.3975,
      "step": 6377
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.704935539146389,
      "learning_rate": 2.0894088406825075e-06,
      "loss": 0.5052,
      "step": 6378
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9476034428013924,
      "learning_rate": 2.0887000440757978e-06,
      "loss": 0.1797,
      "step": 6379
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1352966004871843,
      "learning_rate": 2.0879912814486744e-06,
      "loss": 0.0713,
      "step": 6380
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.570798318901218,
      "learning_rate": 2.0872825528596907e-06,
      "loss": 0.2227,
      "step": 6381
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.6343458381550455,
      "learning_rate": 2.0865738583673987e-06,
      "loss": 0.4212,
      "step": 6382
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.251686052121278,
      "learning_rate": 2.0858651980303487e-06,
      "loss": 0.2346,
      "step": 6383
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.055041768766293,
      "learning_rate": 2.0851565719070847e-06,
      "loss": 0.3897,
      "step": 6384
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.9870114344230707,
      "learning_rate": 2.0844479800561522e-06,
      "loss": 0.1851,
      "step": 6385
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2124582458462143,
      "learning_rate": 2.083739422536089e-06,
      "loss": 0.1581,
      "step": 6386
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.1066252362471705,
      "learning_rate": 2.0830308994054338e-06,
      "loss": 0.3783,
      "step": 6387
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.868722183881534,
      "learning_rate": 2.082322410722722e-06,
      "loss": 0.5376,
      "step": 6388
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.702672118775028,
      "learning_rate": 2.081613956546485e-06,
      "loss": 0.235,
      "step": 6389
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.90932163244422,
      "learning_rate": 2.080905536935252e-06,
      "loss": 0.4299,
      "step": 6390
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.424565178306873,
      "learning_rate": 2.0801971519475477e-06,
      "loss": 0.3665,
      "step": 6391
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.8668846739362697,
      "learning_rate": 2.0794888016418965e-06,
      "loss": 0.2662,
      "step": 6392
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.376363629206426,
      "learning_rate": 2.0787804860768196e-06,
      "loss": 0.3535,
      "step": 6393
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.417769023237083,
      "learning_rate": 2.0780722053108335e-06,
      "loss": 0.208,
      "step": 6394
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.846370445963258,
      "learning_rate": 2.077363959402453e-06,
      "loss": 0.4258,
      "step": 6395
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.5336688618903866,
      "learning_rate": 2.076655748410191e-06,
      "loss": 0.231,
      "step": 6396
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.899191577711942,
      "learning_rate": 2.0759475723925545e-06,
      "loss": 0.3397,
      "step": 6397
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.7349760902590368,
      "learning_rate": 2.0752394314080514e-06,
      "loss": 0.267,
      "step": 6398
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5192438710167324,
      "learning_rate": 2.0745313255151835e-06,
      "loss": 0.0638,
      "step": 6399
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.536990862601319,
      "learning_rate": 2.073823254772451e-06,
      "loss": 0.5625,
      "step": 6400
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.0926735418014415,
      "learning_rate": 2.073115219238353e-06,
      "loss": 0.2874,
      "step": 6401
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1323914432028532,
      "learning_rate": 2.0724072189713816e-06,
      "loss": 0.0999,
      "step": 6402
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.984345172586333,
      "learning_rate": 2.0716992540300303e-06,
      "loss": 0.3885,
      "step": 6403
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.535695665307415,
      "learning_rate": 2.070991324472786e-06,
      "loss": 0.2991,
      "step": 6404
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.5188689025356337,
      "learning_rate": 2.070283430358135e-06,
      "loss": 0.25,
      "step": 6405
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.390029495618389,
      "learning_rate": 2.0695755717445615e-06,
      "loss": 0.4318,
      "step": 6406
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.374736303316772,
      "learning_rate": 2.0688677486905425e-06,
      "loss": 0.3663,
      "step": 6407
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.189019990043161,
      "learning_rate": 2.0681599612545574e-06,
      "loss": 0.2575,
      "step": 6408
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.655107327417136,
      "learning_rate": 2.0674522094950788e-06,
      "loss": 0.418,
      "step": 6409
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.6851504598501363,
      "learning_rate": 2.0667444934705776e-06,
      "loss": 0.1225,
      "step": 6410
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.680108417861485,
      "learning_rate": 2.066036813239523e-06,
      "loss": 0.3335,
      "step": 6411
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9105253616963913,
      "learning_rate": 2.065329168860378e-06,
      "loss": 0.1476,
      "step": 6412
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.050747024154646,
      "learning_rate": 2.0646215603916074e-06,
      "loss": 0.2264,
      "step": 6413
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.791844003358931,
      "learning_rate": 2.063913987891668e-06,
      "loss": 0.2121,
      "step": 6414
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.7835818789698847,
      "learning_rate": 2.063206451419017e-06,
      "loss": 0.1533,
      "step": 6415
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.176094176810443,
      "learning_rate": 2.0624989510321087e-06,
      "loss": 0.1621,
      "step": 6416
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.896005154531211,
      "learning_rate": 2.06179148678939e-06,
      "loss": 0.2698,
      "step": 6417
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.046774175464978,
      "learning_rate": 2.0610840587493125e-06,
      "loss": 0.3135,
      "step": 6418
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.349068948104855,
      "learning_rate": 2.060376666970317e-06,
      "loss": 0.2246,
      "step": 6419
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.648094759218998,
      "learning_rate": 2.0596693115108453e-06,
      "loss": 0.3632,
      "step": 6420
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.241004689356351,
      "learning_rate": 2.058961992429337e-06,
      "loss": 0.3991,
      "step": 6421
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.5928470717837295,
      "learning_rate": 2.0582547097842265e-06,
      "loss": 0.1638,
      "step": 6422
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.717131178478955,
      "learning_rate": 2.057547463633947e-06,
      "loss": 0.5441,
      "step": 6423
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.474037419310963,
      "learning_rate": 2.0568402540369256e-06,
      "loss": 0.3241,
      "step": 6424
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.4423462154573614,
      "learning_rate": 2.05613308105159e-06,
      "loss": 0.2413,
      "step": 6425
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.478340101590781,
      "learning_rate": 2.0554259447363634e-06,
      "loss": 0.4464,
      "step": 6426
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5233507692951367,
      "learning_rate": 2.054718845149665e-06,
      "loss": 0.0857,
      "step": 6427
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9901046362713781,
      "learning_rate": 2.0540117823499135e-06,
      "loss": 0.0829,
      "step": 6428
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.228668379278687,
      "learning_rate": 2.053304756395521e-06,
      "loss": 0.4245,
      "step": 6429
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.672573457282159,
      "learning_rate": 2.0525977673449e-06,
      "loss": 0.4169,
      "step": 6430
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.6032333901927265,
      "learning_rate": 2.051890815256458e-06,
      "loss": 0.1389,
      "step": 6431
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9312228065865513,
      "learning_rate": 2.051183900188599e-06,
      "loss": 0.2596,
      "step": 6432
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.611577873253786,
      "learning_rate": 2.0504770221997265e-06,
      "loss": 0.2386,
      "step": 6433
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.555586123246969,
      "learning_rate": 2.0497701813482394e-06,
      "loss": 0.2909,
      "step": 6434
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.6391534580204725,
      "learning_rate": 2.049063377692531e-06,
      "loss": 0.2753,
      "step": 6435
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5866333622605775,
      "learning_rate": 2.0483566112909965e-06,
      "loss": 0.1141,
      "step": 6436
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.593666126606081,
      "learning_rate": 2.047649882202024e-06,
      "loss": 0.4088,
      "step": 6437
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.58318312475444,
      "learning_rate": 2.0469431904840006e-06,
      "loss": 0.2904,
      "step": 6438
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.960688699386551,
      "learning_rate": 2.04623653619531e-06,
      "loss": 0.4294,
      "step": 6439
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.729303016983094,
      "learning_rate": 2.045529919394332e-06,
      "loss": 0.4647,
      "step": 6440
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.912844716374547,
      "learning_rate": 2.0448233401394435e-06,
      "loss": 0.6105,
      "step": 6441
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.557978010074395,
      "learning_rate": 2.044116798489019e-06,
      "loss": 0.2364,
      "step": 6442
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.2482435036246438,
      "learning_rate": 2.043410294501429e-06,
      "loss": 0.1143,
      "step": 6443
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.5229505987197784,
      "learning_rate": 2.0427038282350434e-06,
      "loss": 0.4025,
      "step": 6444
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.5728095491734675,
      "learning_rate": 2.041997399748225e-06,
      "loss": 0.0899,
      "step": 6445
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.4351936177510085,
      "learning_rate": 2.0412910090993355e-06,
      "loss": 0.4303,
      "step": 6446
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.578706379184465,
      "learning_rate": 2.040584656346734e-06,
      "loss": 0.4084,
      "step": 6447
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.329052416448495,
      "learning_rate": 2.0398783415487755e-06,
      "loss": 0.2385,
      "step": 6448
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.24957483579186,
      "learning_rate": 2.039172064763814e-06,
      "loss": 0.3651,
      "step": 6449
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.19297694659527,
      "learning_rate": 2.038465826050196e-06,
      "loss": 0.2145,
      "step": 6450
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.802441878498951,
      "learning_rate": 2.037759625466269e-06,
      "loss": 0.2218,
      "step": 6451
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.8607469251464965,
      "learning_rate": 2.0370534630703747e-06,
      "loss": 0.2172,
      "step": 6452
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.568825767008543,
      "learning_rate": 2.0363473389208543e-06,
      "loss": 0.4525,
      "step": 6453
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.6599672635087104,
      "learning_rate": 2.0356412530760437e-06,
      "loss": 0.2032,
      "step": 6454
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.580662013577509,
      "learning_rate": 2.0349352055942757e-06,
      "loss": 0.4313,
      "step": 6455
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.815539718073553,
      "learning_rate": 2.034229196533882e-06,
      "loss": 0.2524,
      "step": 6456
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.01975290988113,
      "learning_rate": 2.033523225953187e-06,
      "loss": 0.3273,
      "step": 6457
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.535988399162076,
      "learning_rate": 2.0328172939105166e-06,
      "loss": 0.2917,
      "step": 6458
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.8708235844166756,
      "learning_rate": 2.0321114004641917e-06,
      "loss": 0.1541,
      "step": 6459
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.119748478874862,
      "learning_rate": 2.0314055456725283e-06,
      "loss": 0.4197,
      "step": 6460
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.814314453246753,
      "learning_rate": 2.0306997295938413e-06,
      "loss": 0.1847,
      "step": 6461
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.355528886509316,
      "learning_rate": 2.0299939522864417e-06,
      "loss": 0.1389,
      "step": 6462
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.68885189615938,
      "learning_rate": 2.0292882138086375e-06,
      "loss": 0.2074,
      "step": 6463
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1612393093734696,
      "learning_rate": 2.028582514218734e-06,
      "loss": 0.0956,
      "step": 6464
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.047115514551518,
      "learning_rate": 2.027876853575031e-06,
      "loss": 0.3823,
      "step": 6465
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.743101052716219,
      "learning_rate": 2.027171231935828e-06,
      "loss": 0.1291,
      "step": 6466
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.482098560560313,
      "learning_rate": 2.0264656493594196e-06,
      "loss": 0.316,
      "step": 6467
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.224593764629576,
      "learning_rate": 2.0257601059040973e-06,
      "loss": 0.4043,
      "step": 6468
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.714594008117766,
      "learning_rate": 2.025054601628151e-06,
      "loss": 0.1725,
      "step": 6469
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.0339079992194575,
      "learning_rate": 2.024349136589864e-06,
      "loss": 0.5077,
      "step": 6470
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.886757683984291,
      "learning_rate": 2.0236437108475204e-06,
      "loss": 0.2952,
      "step": 6471
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.543829674256235,
      "learning_rate": 2.0229383244593966e-06,
      "loss": 0.3479,
      "step": 6472
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.62037569240344,
      "learning_rate": 2.0222329774837697e-06,
      "loss": 0.2939,
      "step": 6473
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.227141962826757,
      "learning_rate": 2.0215276699789134e-06,
      "loss": 0.4981,
      "step": 6474
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.9147358687608724,
      "learning_rate": 2.0208224020030937e-06,
      "loss": 0.2749,
      "step": 6475
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.1768681748607355,
      "learning_rate": 2.020117173614578e-06,
      "loss": 0.3225,
      "step": 6476
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.224832367821796,
      "learning_rate": 2.019411984871629e-06,
      "loss": 0.3975,
      "step": 6477
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.452361312440855,
      "learning_rate": 2.0187068358325052e-06,
      "loss": 0.2264,
      "step": 6478
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.7591483123288105,
      "learning_rate": 2.018001726555464e-06,
      "loss": 0.2628,
      "step": 6479
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.5225153870814383,
      "learning_rate": 2.017296657098756e-06,
      "loss": 0.2538,
      "step": 6480
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.4405583381516225,
      "learning_rate": 2.016591627520632e-06,
      "loss": 0.2119,
      "step": 6481
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.201792360612763,
      "learning_rate": 2.015886637879338e-06,
      "loss": 0.2951,
      "step": 6482
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.635499341820522,
      "learning_rate": 2.0151816882331156e-06,
      "loss": 0.0543,
      "step": 6483
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.06103589905169,
      "learning_rate": 2.0144767786402068e-06,
      "loss": 0.2454,
      "step": 6484
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0002777850457862,
      "learning_rate": 2.0137719091588447e-06,
      "loss": 0.1655,
      "step": 6485
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.019673052402382,
      "learning_rate": 2.0130670798472637e-06,
      "loss": 0.2152,
      "step": 6486
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.516279861825802,
      "learning_rate": 2.012362290763694e-06,
      "loss": 0.4521,
      "step": 6487
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.05009059350463,
      "learning_rate": 2.01165754196636e-06,
      "loss": 0.5911,
      "step": 6488
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43856497839908914,
      "learning_rate": 2.0109528335134877e-06,
      "loss": 0.0037,
      "step": 6489
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.1542367138400005,
      "learning_rate": 2.0102481654632927e-06,
      "loss": 0.2108,
      "step": 6490
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.888867104466177,
      "learning_rate": 2.0095435378739934e-06,
      "loss": 0.3416,
      "step": 6491
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.4469925329255626,
      "learning_rate": 2.008838950803803e-06,
      "loss": 0.0716,
      "step": 6492
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.259512787146332,
      "learning_rate": 2.00813440431093e-06,
      "loss": 0.3076,
      "step": 6493
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.413032686028815,
      "learning_rate": 2.007429898453582e-06,
      "loss": 0.2301,
      "step": 6494
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.6009554799709744,
      "learning_rate": 2.0067254332899597e-06,
      "loss": 0.2474,
      "step": 6495
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.739850241955693,
      "learning_rate": 2.006021008878264e-06,
      "loss": 0.4589,
      "step": 6496
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7583187570856156,
      "learning_rate": 2.005316625276691e-06,
      "loss": 0.3257,
      "step": 6497
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.423872330714675,
      "learning_rate": 2.0046122825434323e-06,
      "loss": 0.2997,
      "step": 6498
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.025151198063702,
      "learning_rate": 2.0039079807366795e-06,
      "loss": 0.107,
      "step": 6499
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.367539005040155,
      "learning_rate": 2.0032037199146158e-06,
      "loss": 0.3223,
      "step": 6500
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.101695744273602,
      "learning_rate": 2.0024995001354255e-06,
      "loss": 0.2959,
      "step": 6501
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0090857188076523,
      "learning_rate": 2.0017953214572876e-06,
      "loss": 0.1919,
      "step": 6502
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.5867291541755435,
      "learning_rate": 2.0010911839383775e-06,
      "loss": 0.2439,
      "step": 6503
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.5001499451128093,
      "learning_rate": 2.000387087636868e-06,
      "loss": 0.3253,
      "step": 6504
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.6271282664805606,
      "learning_rate": 1.999683032610928e-06,
      "loss": 0.138,
      "step": 6505
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.080908652997182,
      "learning_rate": 1.9989790189187217e-06,
      "loss": 0.3232,
      "step": 6506
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.353023391558771,
      "learning_rate": 1.998275046618414e-06,
      "loss": 0.1312,
      "step": 6507
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.909998765590669,
      "learning_rate": 1.997571115768161e-06,
      "loss": 0.4761,
      "step": 6508
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.715007308858282,
      "learning_rate": 1.9968672264261205e-06,
      "loss": 0.4737,
      "step": 6509
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.049392567689247,
      "learning_rate": 1.996163378650442e-06,
      "loss": 0.3394,
      "step": 6510
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.02578066125097,
      "learning_rate": 1.9954595724992748e-06,
      "loss": 0.3961,
      "step": 6511
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.5880001656849863,
      "learning_rate": 1.994755808030765e-06,
      "loss": 0.1407,
      "step": 6512
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.297219742967665,
      "learning_rate": 1.994052085303052e-06,
      "loss": 0.307,
      "step": 6513
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2203256431428926,
      "learning_rate": 1.9933484043742758e-06,
      "loss": 0.2429,
      "step": 6514
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.424671229797253,
      "learning_rate": 1.992644765302571e-06,
      "loss": 0.1221,
      "step": 6515
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.468043902324729,
      "learning_rate": 1.9919411681460676e-06,
      "loss": 0.2391,
      "step": 6516
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.48265434356953,
      "learning_rate": 1.9912376129628945e-06,
      "loss": 0.4636,
      "step": 6517
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.980888223085135,
      "learning_rate": 1.9905340998111746e-06,
      "loss": 0.1684,
      "step": 6518
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.99773433490196,
      "learning_rate": 1.9898306287490297e-06,
      "loss": 0.1648,
      "step": 6519
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.2097730760338523,
      "learning_rate": 1.9891271998345784e-06,
      "loss": 0.011,
      "step": 6520
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.040530400203807,
      "learning_rate": 1.988423813125932e-06,
      "loss": 0.2569,
      "step": 6521
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.162967309829603,
      "learning_rate": 1.9877204686812027e-06,
      "loss": 0.1175,
      "step": 6522
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.411244354293716,
      "learning_rate": 1.987017166558496e-06,
      "loss": 0.4851,
      "step": 6523
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.545144810166877,
      "learning_rate": 1.986313906815916e-06,
      "loss": 0.4158,
      "step": 6524
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.994786630855001,
      "learning_rate": 1.9856106895115637e-06,
      "loss": 0.3508,
      "step": 6525
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.626788286885362,
      "learning_rate": 1.984907514703533e-06,
      "loss": 0.137,
      "step": 6526
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.506302383561484,
      "learning_rate": 1.984204382449919e-06,
      "loss": 0.3412,
      "step": 6527
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.9882112179100515,
      "learning_rate": 1.9835012928088095e-06,
      "loss": 0.5193,
      "step": 6528
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.071348385927807,
      "learning_rate": 1.9827982458382905e-06,
      "loss": 0.2764,
      "step": 6529
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.285514079113418,
      "learning_rate": 1.9820952415964463e-06,
      "loss": 0.2399,
      "step": 6530
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.747281167879266,
      "learning_rate": 1.9813922801413526e-06,
      "loss": 0.6284,
      "step": 6531
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.1276066068525585,
      "learning_rate": 1.980689361531087e-06,
      "loss": 0.3598,
      "step": 6532
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.8155508901744475,
      "learning_rate": 1.9799864858237193e-06,
      "loss": 0.1649,
      "step": 6533
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.526042096719405,
      "learning_rate": 1.979283653077319e-06,
      "loss": 0.1781,
      "step": 6534
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.669742975589513,
      "learning_rate": 1.9785808633499514e-06,
      "loss": 0.2547,
      "step": 6535
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.863536277575287,
      "learning_rate": 1.9778781166996754e-06,
      "loss": 0.3715,
      "step": 6536
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.000697673157929,
      "learning_rate": 1.9771754131845502e-06,
      "loss": 0.2749,
      "step": 6537
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.942631220744654,
      "learning_rate": 1.976472752862629e-06,
      "loss": 0.365,
      "step": 6538
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.0407532340008916,
      "learning_rate": 1.9757701357919614e-06,
      "loss": 0.2376,
      "step": 6539
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.3256441870707962,
      "learning_rate": 1.9750675620305963e-06,
      "loss": 0.2339,
      "step": 6540
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.9441050995046245,
      "learning_rate": 1.974365031636575e-06,
      "loss": 0.2875,
      "step": 6541
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.143687531738628,
      "learning_rate": 1.9736625446679375e-06,
      "loss": 0.2586,
      "step": 6542
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.71705865885681,
      "learning_rate": 1.9729601011827203e-06,
      "loss": 0.3325,
      "step": 6543
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.234765955282488,
      "learning_rate": 1.9722577012389547e-06,
      "loss": 0.5186,
      "step": 6544
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.9996351008896767,
      "learning_rate": 1.9715553448946727e-06,
      "loss": 0.1856,
      "step": 6545
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.9883690111949988,
      "learning_rate": 1.9708530322078953e-06,
      "loss": 0.183,
      "step": 6546
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.875946264337515,
      "learning_rate": 1.970150763236647e-06,
      "loss": 0.423,
      "step": 6547
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.6774903674935615,
      "learning_rate": 1.9694485380389443e-06,
      "loss": 0.3645,
      "step": 6548
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.242003280063907,
      "learning_rate": 1.968746356672802e-06,
      "loss": 0.3298,
      "step": 6549
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.178456225896117,
      "learning_rate": 1.968044219196233e-06,
      "loss": 0.5764,
      "step": 6550
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.63888750487161,
      "learning_rate": 1.9673421256672405e-06,
      "loss": 0.3468,
      "step": 6551
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.480554775287508,
      "learning_rate": 1.9666400761438313e-06,
      "loss": 0.2503,
      "step": 6552
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.083759436386386,
      "learning_rate": 1.9659380706840036e-06,
      "loss": 0.357,
      "step": 6553
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.8407619132027575,
      "learning_rate": 1.9652361093457536e-06,
      "loss": 0.2256,
      "step": 6554
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6925286161064987,
      "learning_rate": 1.964534192187076e-06,
      "loss": 0.0746,
      "step": 6555
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.591596399102881,
      "learning_rate": 1.9638323192659567e-06,
      "loss": 0.2376,
      "step": 6556
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.2533129786250665,
      "learning_rate": 1.9631304906403825e-06,
      "loss": 0.4001,
      "step": 6557
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.05401251859786,
      "learning_rate": 1.962428706368336e-06,
      "loss": 0.2388,
      "step": 6558
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.6458905239945842,
      "learning_rate": 1.9617269665077933e-06,
      "loss": 0.1745,
      "step": 6559
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.939365181785482,
      "learning_rate": 1.961025271116731e-06,
      "loss": 0.3208,
      "step": 6560
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.9771133759860993,
      "learning_rate": 1.960323620253117e-06,
      "loss": 0.2357,
      "step": 6561
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.80319617653678,
      "learning_rate": 1.9596220139749196e-06,
      "loss": 0.2455,
      "step": 6562
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.361476954958274,
      "learning_rate": 1.958920452340103e-06,
      "loss": 0.3513,
      "step": 6563
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.329972117033355,
      "learning_rate": 1.958218935406625e-06,
      "loss": 0.3932,
      "step": 6564
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.8173624067154535,
      "learning_rate": 1.9575174632324434e-06,
      "loss": 0.3018,
      "step": 6565
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.737554642591215,
      "learning_rate": 1.9568160358755084e-06,
      "loss": 0.4382,
      "step": 6566
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.762210608348067,
      "learning_rate": 1.956114653393769e-06,
      "loss": 0.4267,
      "step": 6567
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.69212455005904,
      "learning_rate": 1.9554133158451717e-06,
      "loss": 0.4442,
      "step": 6568
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.2449089259313224,
      "learning_rate": 1.954712023287655e-06,
      "loss": 0.1916,
      "step": 6569
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.155275866527683,
      "learning_rate": 1.9540107757791594e-06,
      "loss": 0.194,
      "step": 6570
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.273198857335606,
      "learning_rate": 1.953309573377615e-06,
      "loss": 0.4486,
      "step": 6571
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.540389241664962,
      "learning_rate": 1.9526084161409533e-06,
      "loss": 0.2754,
      "step": 6572
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.7551355377470172,
      "learning_rate": 1.9519073041271012e-06,
      "loss": 0.2922,
      "step": 6573
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1137181343218296,
      "learning_rate": 1.95120623739398e-06,
      "loss": 0.1233,
      "step": 6574
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8074193386905064,
      "learning_rate": 1.95050521599951e-06,
      "loss": 0.0446,
      "step": 6575
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.024783564590054,
      "learning_rate": 1.9498042400016044e-06,
      "loss": 0.219,
      "step": 6576
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.3134049303760738,
      "learning_rate": 1.9491033094581742e-06,
      "loss": 0.1703,
      "step": 6577
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.718890824856773,
      "learning_rate": 1.948402424427129e-06,
      "loss": 0.4092,
      "step": 6578
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.581108262739516,
      "learning_rate": 1.94770158496637e-06,
      "loss": 0.4271,
      "step": 6579
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.333941365078548,
      "learning_rate": 1.9470007911337996e-06,
      "loss": 0.2379,
      "step": 6580
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.112086576754967,
      "learning_rate": 1.946300042987311e-06,
      "loss": 0.223,
      "step": 6581
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.6626155948903407,
      "learning_rate": 1.945599340584799e-06,
      "loss": 0.2916,
      "step": 6582
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.46281506205001,
      "learning_rate": 1.9448986839841517e-06,
      "loss": 0.1343,
      "step": 6583
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.8561172171719558,
      "learning_rate": 1.944198073243253e-06,
      "loss": 0.2407,
      "step": 6584
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.189811299680405,
      "learning_rate": 1.9434975084199856e-06,
      "loss": 0.3782,
      "step": 6585
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.4641119910123948,
      "learning_rate": 1.9427969895722244e-06,
      "loss": 0.1242,
      "step": 6586
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.184240954740195,
      "learning_rate": 1.9420965167578447e-06,
      "loss": 0.4402,
      "step": 6587
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.738570741173528,
      "learning_rate": 1.941396090034716e-06,
      "loss": 0.4322,
      "step": 6588
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.457782798082285,
      "learning_rate": 1.9406957094607024e-06,
      "loss": 0.2206,
      "step": 6589
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.269579231948112,
      "learning_rate": 1.9399953750936686e-06,
      "loss": 0.2114,
      "step": 6590
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.609717561505509,
      "learning_rate": 1.9392950869914702e-06,
      "loss": 0.3005,
      "step": 6591
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.780390025031608,
      "learning_rate": 1.938594845211963e-06,
      "loss": 0.3615,
      "step": 6592
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8806622283474983,
      "learning_rate": 1.9378946498129975e-06,
      "loss": 0.1358,
      "step": 6593
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.0369797146676385,
      "learning_rate": 1.9371945008524203e-06,
      "loss": 0.2819,
      "step": 6594
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.439709746267444,
      "learning_rate": 1.936494398388073e-06,
      "loss": 0.146,
      "step": 6595
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.544055673080535,
      "learning_rate": 1.935794342477798e-06,
      "loss": 0.3678,
      "step": 6596
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.602987068977395,
      "learning_rate": 1.9350943331794266e-06,
      "loss": 0.267,
      "step": 6597
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.498512735513037,
      "learning_rate": 1.9343943705507922e-06,
      "loss": 0.3515,
      "step": 6598
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.559187607546861,
      "learning_rate": 1.933694454649722e-06,
      "loss": 0.3853,
      "step": 6599
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.848110963011794,
      "learning_rate": 1.932994585534039e-06,
      "loss": 0.182,
      "step": 6600
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.2361041376677995,
      "learning_rate": 1.9322947632615645e-06,
      "loss": 0.1949,
      "step": 6601
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.774145009095844,
      "learning_rate": 1.9315949878901126e-06,
      "loss": 0.4059,
      "step": 6602
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.294408867540662,
      "learning_rate": 1.9308952594774965e-06,
      "loss": 0.3317,
      "step": 6603
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.925819827542834,
      "learning_rate": 1.930195578081523e-06,
      "loss": 0.2307,
      "step": 6604
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.815589616967836,
      "learning_rate": 1.929495943759998e-06,
      "loss": 0.5393,
      "step": 6605
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.638867723571405,
      "learning_rate": 1.9287963565707215e-06,
      "loss": 0.2627,
      "step": 6606
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.094645026314438,
      "learning_rate": 1.928096816571488e-06,
      "loss": 0.3697,
      "step": 6607
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.562206338068896,
      "learning_rate": 1.927397323820093e-06,
      "loss": 0.2864,
      "step": 6608
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.491098044986951,
      "learning_rate": 1.9266978783743236e-06,
      "loss": 0.3514,
      "step": 6609
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.162236103303515,
      "learning_rate": 1.925998480291964e-06,
      "loss": 0.2109,
      "step": 6610
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.630912096066466,
      "learning_rate": 1.925299129630797e-06,
      "loss": 0.3971,
      "step": 6611
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.02527931135007,
      "learning_rate": 1.9245998264485973e-06,
      "loss": 0.2289,
      "step": 6612
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.481802238353764,
      "learning_rate": 1.9239005708031398e-06,
      "loss": 0.2315,
      "step": 6613
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.9403174640151093,
      "learning_rate": 1.9232013627521917e-06,
      "loss": 0.295,
      "step": 6614
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.0992231053299495,
      "learning_rate": 1.9225022023535193e-06,
      "loss": 0.3288,
      "step": 6615
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.895840891498203,
      "learning_rate": 1.9218030896648845e-06,
      "loss": 0.4595,
      "step": 6616
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.184516542739213,
      "learning_rate": 1.921104024744042e-06,
      "loss": 0.3142,
      "step": 6617
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.952712573560934,
      "learning_rate": 1.920405007648749e-06,
      "loss": 0.4721,
      "step": 6618
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.684449588407301,
      "learning_rate": 1.919706038436751e-06,
      "loss": 0.0993,
      "step": 6619
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.1862914821217325,
      "learning_rate": 1.9190071171657953e-06,
      "loss": 0.4112,
      "step": 6620
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.613348731352494,
      "learning_rate": 1.9183082438936234e-06,
      "loss": 0.1401,
      "step": 6621
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.47896969697792,
      "learning_rate": 1.9176094186779725e-06,
      "loss": 0.3225,
      "step": 6622
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.595585088326933,
      "learning_rate": 1.9169106415765768e-06,
      "loss": 0.4497,
      "step": 6623
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.559461547387424,
      "learning_rate": 1.916211912647164e-06,
      "loss": 0.3262,
      "step": 6624
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.220493097404983,
      "learning_rate": 1.9155132319474614e-06,
      "loss": 0.0969,
      "step": 6625
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.1898620957081243,
      "learning_rate": 1.91481459953519e-06,
      "loss": 0.2368,
      "step": 6626
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.878826688914308,
      "learning_rate": 1.914116015468067e-06,
      "loss": 0.397,
      "step": 6627
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.424388389464396,
      "learning_rate": 1.9134174798038073e-06,
      "loss": 0.2796,
      "step": 6628
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.2845823339003433,
      "learning_rate": 1.9127189926001184e-06,
      "loss": 0.1216,
      "step": 6629
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.1187902696098195,
      "learning_rate": 1.9120205539147075e-06,
      "loss": 0.3887,
      "step": 6630
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.0041365228827015,
      "learning_rate": 1.9113221638052758e-06,
      "loss": 0.3233,
      "step": 6631
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9041893764803413,
      "learning_rate": 1.9106238223295206e-06,
      "loss": 0.1472,
      "step": 6632
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7639374522042344,
      "learning_rate": 1.9099255295451367e-06,
      "loss": 0.2388,
      "step": 6633
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.5756132384561345,
      "learning_rate": 1.9092272855098114e-06,
      "loss": 0.3329,
      "step": 6634
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.941349477101968,
      "learning_rate": 1.9085290902812313e-06,
      "loss": 0.3858,
      "step": 6635
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.4833509880768485,
      "learning_rate": 1.9078309439170782e-06,
      "loss": 0.2993,
      "step": 6636
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.6830690966338224,
      "learning_rate": 1.9071328464750295e-06,
      "loss": 0.1688,
      "step": 6637
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.425037913876512,
      "learning_rate": 1.906434798012758e-06,
      "loss": 0.3828,
      "step": 6638
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.315762394085271,
      "learning_rate": 1.9057367985879341e-06,
      "loss": 0.2632,
      "step": 6639
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.6429182590402505,
      "learning_rate": 1.9050388482582217e-06,
      "loss": 0.2718,
      "step": 6640
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.439744321284153,
      "learning_rate": 1.9043409470812829e-06,
      "loss": 0.3758,
      "step": 6641
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.337926607903135,
      "learning_rate": 1.9036430951147745e-06,
      "loss": 0.3939,
      "step": 6642
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.6529307595308818,
      "learning_rate": 1.9029452924163497e-06,
      "loss": 0.1329,
      "step": 6643
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.3155225074126093,
      "learning_rate": 1.9022475390436588e-06,
      "loss": 0.1956,
      "step": 6644
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.686942362216954,
      "learning_rate": 1.9015498350543443e-06,
      "loss": 0.3704,
      "step": 6645
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.075302717269803,
      "learning_rate": 1.9008521805060493e-06,
      "loss": 0.4042,
      "step": 6646
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.156438035898714,
      "learning_rate": 1.9001545754564088e-06,
      "loss": 0.3428,
      "step": 6647
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.551605898572187,
      "learning_rate": 1.8994570199630564e-06,
      "loss": 0.2504,
      "step": 6648
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.876157067831479,
      "learning_rate": 1.8987595140836219e-06,
      "loss": 0.1557,
      "step": 6649
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.652866107293154,
      "learning_rate": 1.8980620578757274e-06,
      "loss": 0.3343,
      "step": 6650
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.6435429130765415,
      "learning_rate": 1.8973646513969955e-06,
      "loss": 0.4646,
      "step": 6651
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.8808924135903533,
      "learning_rate": 1.8966672947050405e-06,
      "loss": 0.3304,
      "step": 6652
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.460454723847402,
      "learning_rate": 1.8959699878574758e-06,
      "loss": 0.5451,
      "step": 6653
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.855999794718239,
      "learning_rate": 1.8952727309119106e-06,
      "loss": 0.299,
      "step": 6654
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.8990183636424733,
      "learning_rate": 1.8945755239259462e-06,
      "loss": 0.3277,
      "step": 6655
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.7562939085104,
      "learning_rate": 1.8938783669571846e-06,
      "loss": 0.0786,
      "step": 6656
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.6858796220655945,
      "learning_rate": 1.89318126006322e-06,
      "loss": 0.4108,
      "step": 6657
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.6694345173352274,
      "learning_rate": 1.8924842033016444e-06,
      "loss": 0.1501,
      "step": 6658
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9968106671119945,
      "learning_rate": 1.8917871967300467e-06,
      "loss": 0.082,
      "step": 6659
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.000779723945013,
      "learning_rate": 1.8910902404060078e-06,
      "loss": 0.2332,
      "step": 6660
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.506868419241136,
      "learning_rate": 1.8903933343871083e-06,
      "loss": 0.2947,
      "step": 6661
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.1224308267291723,
      "learning_rate": 1.8896964787309223e-06,
      "loss": 0.2169,
      "step": 6662
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.256352763191381,
      "learning_rate": 1.8889996734950211e-06,
      "loss": 0.3628,
      "step": 6663
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.8223759992231288,
      "learning_rate": 1.8883029187369724e-06,
      "loss": 0.1691,
      "step": 6664
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9016110703805977,
      "learning_rate": 1.8876062145143366e-06,
      "loss": 0.1677,
      "step": 6665
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.625090156556914,
      "learning_rate": 1.8869095608846734e-06,
      "loss": 0.4166,
      "step": 6666
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.575748353432107,
      "learning_rate": 1.886212957905536e-06,
      "loss": 0.3367,
      "step": 6667
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.427283364368232,
      "learning_rate": 1.8855164056344746e-06,
      "loss": 0.3725,
      "step": 6668
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.9936824463858613,
      "learning_rate": 1.8848199041290363e-06,
      "loss": 0.2174,
      "step": 6669
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.131506258753758,
      "learning_rate": 1.8841234534467607e-06,
      "loss": 0.1864,
      "step": 6670
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.3320898680082935,
      "learning_rate": 1.8834270536451859e-06,
      "loss": 0.3465,
      "step": 6671
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0927127272673456,
      "learning_rate": 1.8827307047818446e-06,
      "loss": 0.1701,
      "step": 6672
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.244822014603915,
      "learning_rate": 1.8820344069142662e-06,
      "loss": 0.238,
      "step": 6673
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.566435286441504,
      "learning_rate": 1.8813381600999766e-06,
      "loss": 0.1668,
      "step": 6674
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.472622354590215,
      "learning_rate": 1.880641964396494e-06,
      "loss": 0.2592,
      "step": 6675
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.968229896718876,
      "learning_rate": 1.8799458198613358e-06,
      "loss": 0.2642,
      "step": 6676
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.046703511929616,
      "learning_rate": 1.8792497265520142e-06,
      "loss": 0.4116,
      "step": 6677
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.371543209424767,
      "learning_rate": 1.8785536845260365e-06,
      "loss": 0.3389,
      "step": 6678
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.010938200839801,
      "learning_rate": 1.8778576938409076e-06,
      "loss": 0.2741,
      "step": 6679
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.419449882652013,
      "learning_rate": 1.877161754554125e-06,
      "loss": 0.4047,
      "step": 6680
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.665882374202842,
      "learning_rate": 1.8764658667231845e-06,
      "loss": 0.3391,
      "step": 6681
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.057865066092741,
      "learning_rate": 1.8757700304055776e-06,
      "loss": 0.2827,
      "step": 6682
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.814646993356579,
      "learning_rate": 1.8750742456587897e-06,
      "loss": 0.3767,
      "step": 6683
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.55724339848168,
      "learning_rate": 1.874378512540305e-06,
      "loss": 0.2367,
      "step": 6684
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7287262885630463,
      "learning_rate": 1.873682831107599e-06,
      "loss": 0.0695,
      "step": 6685
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.604393346698425,
      "learning_rate": 1.872987201418147e-06,
      "loss": 0.271,
      "step": 6686
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.78292345149719,
      "learning_rate": 1.8722916235294192e-06,
      "loss": 0.1585,
      "step": 6687
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.7655536423862848,
      "learning_rate": 1.871596097498879e-06,
      "loss": 0.1238,
      "step": 6688
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.643884161918867,
      "learning_rate": 1.8709006233839895e-06,
      "loss": 0.2356,
      "step": 6689
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.694970464695572,
      "learning_rate": 1.870205201242205e-06,
      "loss": 0.5604,
      "step": 6690
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.7094248949031465,
      "learning_rate": 1.8695098311309795e-06,
      "loss": 0.3321,
      "step": 6691
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.01448110140948,
      "learning_rate": 1.8688145131077606e-06,
      "loss": 0.2068,
      "step": 6692
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.5806664693405383,
      "learning_rate": 1.8681192472299919e-06,
      "loss": 0.1269,
      "step": 6693
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.526177154864554,
      "learning_rate": 1.8674240335551142e-06,
      "loss": 0.2658,
      "step": 6694
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.582511245200103,
      "learning_rate": 1.8667288721405602e-06,
      "loss": 0.2082,
      "step": 6695
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.302375779150455,
      "learning_rate": 1.866033763043762e-06,
      "loss": 0.3841,
      "step": 6696
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.068269367927885,
      "learning_rate": 1.8653387063221467e-06,
      "loss": 0.3483,
      "step": 6697
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.292717495707297,
      "learning_rate": 1.8646437020331353e-06,
      "loss": 0.2653,
      "step": 6698
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.080685020846566,
      "learning_rate": 1.8639487502341475e-06,
      "loss": 0.0709,
      "step": 6699
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.690674956007914,
      "learning_rate": 1.8632538509825945e-06,
      "loss": 0.341,
      "step": 6700
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.5882365823691815,
      "learning_rate": 1.8625590043358866e-06,
      "loss": 0.3729,
      "step": 6701
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.94572787871183,
      "learning_rate": 1.861864210351429e-06,
      "loss": 0.1598,
      "step": 6702
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.5012617009018685,
      "learning_rate": 1.8611694690866212e-06,
      "loss": 0.3088,
      "step": 6703
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.0745720312860207,
      "learning_rate": 1.8604747805988608e-06,
      "loss": 0.1756,
      "step": 6704
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.981953489128452,
      "learning_rate": 1.8597801449455377e-06,
      "loss": 0.2586,
      "step": 6705
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.449219026124584,
      "learning_rate": 1.8590855621840398e-06,
      "loss": 0.4478,
      "step": 6706
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.996965376451565,
      "learning_rate": 1.8583910323717517e-06,
      "loss": 0.4907,
      "step": 6707
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.435805729239966,
      "learning_rate": 1.8576965555660498e-06,
      "loss": 0.1598,
      "step": 6708
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.3285481447981935,
      "learning_rate": 1.8570021318243108e-06,
      "loss": 0.2635,
      "step": 6709
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.64580011385688,
      "learning_rate": 1.856307761203902e-06,
      "loss": 0.364,
      "step": 6710
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.832008316366094,
      "learning_rate": 1.8556134437621903e-06,
      "loss": 0.4868,
      "step": 6711
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.1711020670502394,
      "learning_rate": 1.8549191795565369e-06,
      "loss": 0.344,
      "step": 6712
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.265462156463436,
      "learning_rate": 1.8542249686442978e-06,
      "loss": 0.2426,
      "step": 6713
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.611592928175389,
      "learning_rate": 1.8535308110828269e-06,
      "loss": 0.2182,
      "step": 6714
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.479917692696548,
      "learning_rate": 1.8528367069294694e-06,
      "loss": 0.4827,
      "step": 6715
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.943490472097874,
      "learning_rate": 1.852142656241571e-06,
      "loss": 0.1462,
      "step": 6716
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.803389399384215,
      "learning_rate": 1.8514486590764701e-06,
      "loss": 0.487,
      "step": 6717
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.426569155553823,
      "learning_rate": 1.850754715491501e-06,
      "loss": 0.2663,
      "step": 6718
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.551163620439118,
      "learning_rate": 1.8500608255439942e-06,
      "loss": 0.2382,
      "step": 6719
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.234227383400931,
      "learning_rate": 1.8493669892912762e-06,
      "loss": 0.4256,
      "step": 6720
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.387851408009937,
      "learning_rate": 1.8486732067906671e-06,
      "loss": 0.5341,
      "step": 6721
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8806252144508053,
      "learning_rate": 1.8479794780994847e-06,
      "loss": 0.1443,
      "step": 6722
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.937697555267566,
      "learning_rate": 1.8472858032750406e-06,
      "loss": 0.3444,
      "step": 6723
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.578636087283117,
      "learning_rate": 1.8465921823746436e-06,
      "loss": 0.2872,
      "step": 6724
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.91167429984553,
      "learning_rate": 1.845898615455598e-06,
      "loss": 0.5434,
      "step": 6725
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.001549176584138,
      "learning_rate": 1.8452051025752004e-06,
      "loss": 0.1903,
      "step": 6726
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.13516255365456,
      "learning_rate": 1.844511643790748e-06,
      "loss": 0.5828,
      "step": 6727
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.175617348683967,
      "learning_rate": 1.8438182391595291e-06,
      "loss": 0.4277,
      "step": 6728
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.815650814608105,
      "learning_rate": 1.8431248887388303e-06,
      "loss": 0.3109,
      "step": 6729
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.192503876305207,
      "learning_rate": 1.8424315925859337e-06,
      "loss": 0.2226,
      "step": 6730
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.786319938450571,
      "learning_rate": 1.8417383507581138e-06,
      "loss": 0.4267,
      "step": 6731
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.04008232874531,
      "learning_rate": 1.841045163312645e-06,
      "loss": 0.1804,
      "step": 6732
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.0326589714600125,
      "learning_rate": 1.8403520303067935e-06,
      "loss": 0.2159,
      "step": 6733
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.36303456824984,
      "learning_rate": 1.839658951797823e-06,
      "loss": 0.2536,
      "step": 6734
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2758661282507746,
      "learning_rate": 1.8389659278429933e-06,
      "loss": 0.1097,
      "step": 6735
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.930988868744035,
      "learning_rate": 1.8382729584995574e-06,
      "loss": 0.6621,
      "step": 6736
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.99523827958513,
      "learning_rate": 1.8375800438247654e-06,
      "loss": 0.3741,
      "step": 6737
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.6323121548574764,
      "learning_rate": 1.8368871838758616e-06,
      "loss": 0.2359,
      "step": 6738
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.148449876931961,
      "learning_rate": 1.8361943787100883e-06,
      "loss": 0.3497,
      "step": 6739
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.750230704699035,
      "learning_rate": 1.835501628384682e-06,
      "loss": 0.3834,
      "step": 6740
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.770954410313709,
      "learning_rate": 1.834808932956872e-06,
      "loss": 0.3733,
      "step": 6741
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.0573457400143416,
      "learning_rate": 1.8341162924838874e-06,
      "loss": 0.2135,
      "step": 6742
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.03650066558858,
      "learning_rate": 1.8334237070229493e-06,
      "loss": 0.3102,
      "step": 6743
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.9396714779267485,
      "learning_rate": 1.8327311766312766e-06,
      "loss": 0.4273,
      "step": 6744
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8262753094914834,
      "learning_rate": 1.8320387013660835e-06,
      "loss": 0.3574,
      "step": 6745
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.515600679309311,
      "learning_rate": 1.8313462812845772e-06,
      "loss": 0.3119,
      "step": 6746
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.6539820406119916,
      "learning_rate": 1.8306539164439634e-06,
      "loss": 0.2979,
      "step": 6747
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.01064697521323,
      "learning_rate": 1.829961606901441e-06,
      "loss": 0.2147,
      "step": 6748
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.319227756625525,
      "learning_rate": 1.8292693527142053e-06,
      "loss": 0.1215,
      "step": 6749
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.796747305148302,
      "learning_rate": 1.8285771539394487e-06,
      "loss": 0.1784,
      "step": 6750
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.188301482514379,
      "learning_rate": 1.8278850106343546e-06,
      "loss": 0.1687,
      "step": 6751
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7669533884680755,
      "learning_rate": 1.8271929228561062e-06,
      "loss": 0.348,
      "step": 6752
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.180122340738525,
      "learning_rate": 1.8265008906618794e-06,
      "loss": 0.3338,
      "step": 6753
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.016198561935872,
      "learning_rate": 1.825808914108847e-06,
      "loss": 0.3287,
      "step": 6754
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.331422413366424,
      "learning_rate": 1.8251169932541779e-06,
      "loss": 0.3367,
      "step": 6755
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7554881035702,
      "learning_rate": 1.824425128155033e-06,
      "loss": 0.2208,
      "step": 6756
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.4337171192484526,
      "learning_rate": 1.8237333188685719e-06,
      "loss": 0.2701,
      "step": 6757
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.021685119244181,
      "learning_rate": 1.8230415654519487e-06,
      "loss": 0.2388,
      "step": 6758
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.139248856617032,
      "learning_rate": 1.8223498679623119e-06,
      "loss": 0.3595,
      "step": 6759
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.304389564494887,
      "learning_rate": 1.821658226456808e-06,
      "loss": 0.1909,
      "step": 6760
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.101774085601163,
      "learning_rate": 1.8209666409925747e-06,
      "loss": 0.4111,
      "step": 6761
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3059378593960753,
      "learning_rate": 1.820275111626748e-06,
      "loss": 0.0152,
      "step": 6762
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.016321501828005,
      "learning_rate": 1.8195836384164597e-06,
      "loss": 0.2662,
      "step": 6763
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.784253641786622,
      "learning_rate": 1.8188922214188348e-06,
      "loss": 0.2998,
      "step": 6764
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.164835572157805,
      "learning_rate": 1.8182008606909967e-06,
      "loss": 0.2496,
      "step": 6765
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.6286860987863943,
      "learning_rate": 1.8175095562900593e-06,
      "loss": 0.1939,
      "step": 6766
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.68827531009452,
      "learning_rate": 1.8168183082731364e-06,
      "loss": 0.294,
      "step": 6767
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.853132569674757,
      "learning_rate": 1.816127116697336e-06,
      "loss": 0.2176,
      "step": 6768
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0756931358103454,
      "learning_rate": 1.8154359816197598e-06,
      "loss": 0.1004,
      "step": 6769
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.118264334524523,
      "learning_rate": 1.8147449030975078e-06,
      "loss": 0.5563,
      "step": 6770
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.4935176756057915,
      "learning_rate": 1.814053881187671e-06,
      "loss": 0.2542,
      "step": 6771
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.1418830154563517,
      "learning_rate": 1.8133629159473396e-06,
      "loss": 0.2153,
      "step": 6772
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.8677250709687843,
      "learning_rate": 1.8126720074335979e-06,
      "loss": 0.0854,
      "step": 6773
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4651483659283289,
      "learning_rate": 1.811981155703525e-06,
      "loss": 0.0124,
      "step": 6774
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.87378256641817,
      "learning_rate": 1.8112903608141968e-06,
      "loss": 0.3278,
      "step": 6775
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.681690022399206,
      "learning_rate": 1.8105996228226816e-06,
      "loss": 0.3265,
      "step": 6776
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.605339792985247,
      "learning_rate": 1.8099089417860455e-06,
      "loss": 0.2489,
      "step": 6777
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8945885108138563,
      "learning_rate": 1.8092183177613495e-06,
      "loss": 0.2692,
      "step": 6778
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.318895686123558,
      "learning_rate": 1.8085277508056492e-06,
      "loss": 0.323,
      "step": 6779
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.0596154965221847,
      "learning_rate": 1.807837240975997e-06,
      "loss": 0.1875,
      "step": 6780
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.4707186616789856,
      "learning_rate": 1.8071467883294375e-06,
      "loss": 0.187,
      "step": 6781
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8565965403162425,
      "learning_rate": 1.806456392923013e-06,
      "loss": 0.2212,
      "step": 6782
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.675803164522709,
      "learning_rate": 1.805766054813762e-06,
      "loss": 0.2592,
      "step": 6783
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.431950337394819,
      "learning_rate": 1.8050757740587156e-06,
      "loss": 0.4597,
      "step": 6784
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.91596444662437,
      "learning_rate": 1.804385550714902e-06,
      "loss": 0.3648,
      "step": 6785
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.6582948863287035,
      "learning_rate": 1.8036953848393436e-06,
      "loss": 0.3956,
      "step": 6786
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.8881044319663385,
      "learning_rate": 1.8030052764890582e-06,
      "loss": 0.1828,
      "step": 6787
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.324454107177191,
      "learning_rate": 1.8023152257210602e-06,
      "loss": 0.3372,
      "step": 6788
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.843482239293279,
      "learning_rate": 1.8016252325923571e-06,
      "loss": 0.1804,
      "step": 6789
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.590220205952505,
      "learning_rate": 1.8009352971599548e-06,
      "loss": 0.4615,
      "step": 6790
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.180536228573136,
      "learning_rate": 1.8002454194808494e-06,
      "loss": 0.3018,
      "step": 6791
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.295008068070765,
      "learning_rate": 1.799555599612037e-06,
      "loss": 0.4175,
      "step": 6792
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.7894196463513974,
      "learning_rate": 1.7988658376105074e-06,
      "loss": 0.2907,
      "step": 6793
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.78143590452456,
      "learning_rate": 1.7981761335332442e-06,
      "loss": 0.3673,
      "step": 6794
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.378240850289318,
      "learning_rate": 1.7974864874372292e-06,
      "loss": 0.1384,
      "step": 6795
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.0589348537440575,
      "learning_rate": 1.796796899379435e-06,
      "loss": 0.5624,
      "step": 6796
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.103679326695398,
      "learning_rate": 1.7961073694168336e-06,
      "loss": 0.1934,
      "step": 6797
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.0708288402501114,
      "learning_rate": 1.7954178976063907e-06,
      "loss": 0.2259,
      "step": 6798
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.454210642708881,
      "learning_rate": 1.7947284840050667e-06,
      "loss": 0.3318,
      "step": 6799
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.1739618655207122,
      "learning_rate": 1.7940391286698169e-06,
      "loss": 0.1733,
      "step": 6800
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.698715481023651,
      "learning_rate": 1.7933498316575945e-06,
      "loss": 0.34,
      "step": 6801
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.262045419373724,
      "learning_rate": 1.7926605930253435e-06,
      "loss": 0.3894,
      "step": 6802
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.870908115026273,
      "learning_rate": 1.791971412830007e-06,
      "loss": 0.1917,
      "step": 6803
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.230443123601667,
      "learning_rate": 1.7912822911285202e-06,
      "loss": 0.4618,
      "step": 6804
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.694747519076166,
      "learning_rate": 1.790593227977816e-06,
      "loss": 0.4293,
      "step": 6805
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.682429124127558,
      "learning_rate": 1.7899042234348224e-06,
      "loss": 0.3669,
      "step": 6806
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.151202403398376,
      "learning_rate": 1.7892152775564598e-06,
      "loss": 0.278,
      "step": 6807
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.29559514298653,
      "learning_rate": 1.7885263903996462e-06,
      "loss": 0.2657,
      "step": 6808
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.497367612902293,
      "learning_rate": 1.7878375620212935e-06,
      "loss": 0.2731,
      "step": 6809
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.317442659588427,
      "learning_rate": 1.78714879247831e-06,
      "loss": 0.4048,
      "step": 6810
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.891770551768419,
      "learning_rate": 1.7864600818275996e-06,
      "loss": 0.3513,
      "step": 6811
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.0473248200817435,
      "learning_rate": 1.785771430126057e-06,
      "loss": 0.5357,
      "step": 6812
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.637209879160914,
      "learning_rate": 1.785082837430579e-06,
      "loss": 0.4117,
      "step": 6813
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.3717435600632064,
      "learning_rate": 1.7843943037980507e-06,
      "loss": 0.2007,
      "step": 6814
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.646186457088515,
      "learning_rate": 1.7837058292853568e-06,
      "loss": 0.6068,
      "step": 6815
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.07162577252762,
      "learning_rate": 1.7830174139493764e-06,
      "loss": 0.3893,
      "step": 6816
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.235436700697606,
      "learning_rate": 1.7823290578469803e-06,
      "loss": 0.4812,
      "step": 6817
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.781141854803339,
      "learning_rate": 1.781640761035041e-06,
      "loss": 0.454,
      "step": 6818
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.34137691695944,
      "learning_rate": 1.7809525235704187e-06,
      "loss": 0.2322,
      "step": 6819
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.308467124188423,
      "learning_rate": 1.7802643455099738e-06,
      "loss": 0.504,
      "step": 6820
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.8724556619887465,
      "learning_rate": 1.7795762269105607e-06,
      "loss": 0.1371,
      "step": 6821
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.061041994798188,
      "learning_rate": 1.7788881678290265e-06,
      "loss": 0.5361,
      "step": 6822
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.183249309731584,
      "learning_rate": 1.7782001683222178e-06,
      "loss": 0.3001,
      "step": 6823
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.414049501285442,
      "learning_rate": 1.7775122284469715e-06,
      "loss": 0.2256,
      "step": 6824
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9948083897811117,
      "learning_rate": 1.7768243482601224e-06,
      "loss": 0.1771,
      "step": 6825
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.095983889839848,
      "learning_rate": 1.7761365278185007e-06,
      "loss": 0.1693,
      "step": 6826
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.182884341469184,
      "learning_rate": 1.77544876717893e-06,
      "loss": 0.2512,
      "step": 6827
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.283587442115208,
      "learning_rate": 1.7747610663982306e-06,
      "loss": 0.4915,
      "step": 6828
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.329695083023359,
      "learning_rate": 1.7740734255332154e-06,
      "loss": 0.2955,
      "step": 6829
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.417793815259089,
      "learning_rate": 1.7733858446406946e-06,
      "loss": 0.3478,
      "step": 6830
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.460432867702195,
      "learning_rate": 1.7726983237774737e-06,
      "loss": 0.2571,
      "step": 6831
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.968569230563639,
      "learning_rate": 1.7720108630003505e-06,
      "loss": 0.1268,
      "step": 6832
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.377906923043161,
      "learning_rate": 1.7713234623661218e-06,
      "loss": 0.2517,
      "step": 6833
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.286493273601934,
      "learning_rate": 1.7706361219315754e-06,
      "loss": 0.2882,
      "step": 6834
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.308417446310199,
      "learning_rate": 1.7699488417534966e-06,
      "loss": 0.2744,
      "step": 6835
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.643491778552893,
      "learning_rate": 1.769261621888666e-06,
      "loss": 0.5186,
      "step": 6836
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.015927630853283,
      "learning_rate": 1.7685744623938566e-06,
      "loss": 0.2789,
      "step": 6837
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.703555621103088,
      "learning_rate": 1.7678873633258394e-06,
      "loss": 0.3375,
      "step": 6838
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.066384319710058,
      "learning_rate": 1.7672003247413799e-06,
      "loss": 0.1352,
      "step": 6839
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.337742782565338,
      "learning_rate": 1.7665133466972359e-06,
      "loss": 0.4668,
      "step": 6840
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.445131830286539,
      "learning_rate": 1.7658264292501637e-06,
      "loss": 0.1765,
      "step": 6841
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0117991358427627,
      "learning_rate": 1.7651395724569114e-06,
      "loss": 0.2415,
      "step": 6842
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.2565792454994495,
      "learning_rate": 1.7644527763742253e-06,
      "loss": 0.3472,
      "step": 6843
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9924043028099936,
      "learning_rate": 1.7637660410588456e-06,
      "loss": 0.1818,
      "step": 6844
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0484456644224647,
      "learning_rate": 1.763079366567505e-06,
      "loss": 0.1775,
      "step": 6845
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.217195589872964,
      "learning_rate": 1.7623927529569346e-06,
      "loss": 0.2108,
      "step": 6846
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.525648911534629,
      "learning_rate": 1.7617062002838582e-06,
      "loss": 0.301,
      "step": 6847
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9242939894232283,
      "learning_rate": 1.7610197086049957e-06,
      "loss": 0.1513,
      "step": 6848
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.864884525225981,
      "learning_rate": 1.760333277977063e-06,
      "loss": 0.1141,
      "step": 6849
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.016328114559739,
      "learning_rate": 1.7596469084567674e-06,
      "loss": 0.0939,
      "step": 6850
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.202204569278658,
      "learning_rate": 1.7589606001008152e-06,
      "loss": 0.2703,
      "step": 6851
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0749827399269702,
      "learning_rate": 1.7582743529659044e-06,
      "loss": 0.1085,
      "step": 6852
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.015299469033298,
      "learning_rate": 1.7575881671087303e-06,
      "loss": 0.3982,
      "step": 6853
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.8321533764067435,
      "learning_rate": 1.7569020425859825e-06,
      "loss": 0.1863,
      "step": 6854
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.0791419913073135,
      "learning_rate": 1.7562159794543443e-06,
      "loss": 0.1549,
      "step": 6855
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.961169127889867,
      "learning_rate": 1.7555299777704954e-06,
      "loss": 0.3145,
      "step": 6856
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.522899763545951,
      "learning_rate": 1.7548440375911094e-06,
      "loss": 0.1905,
      "step": 6857
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.635839912820425,
      "learning_rate": 1.7541581589728556e-06,
      "loss": 0.4253,
      "step": 6858
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.743659912957205,
      "learning_rate": 1.7534723419723993e-06,
      "loss": 0.2174,
      "step": 6859
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.55593513125131,
      "learning_rate": 1.7527865866463973e-06,
      "loss": 0.0744,
      "step": 6860
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.4288166550817767,
      "learning_rate": 1.7521008930515041e-06,
      "loss": 0.1308,
      "step": 6861
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2518079113447755,
      "learning_rate": 1.751415261244368e-06,
      "loss": 0.0391,
      "step": 6862
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.1806987601067895,
      "learning_rate": 1.750729691281633e-06,
      "loss": 0.1696,
      "step": 6863
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.015618013621148,
      "learning_rate": 1.7500441832199383e-06,
      "loss": 0.1247,
      "step": 6864
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.910244720612196,
      "learning_rate": 1.7493587371159155e-06,
      "loss": 0.1604,
      "step": 6865
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.1540290210088275,
      "learning_rate": 1.748673353026194e-06,
      "loss": 0.1534,
      "step": 6866
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.433114508253748,
      "learning_rate": 1.747988031007396e-06,
      "loss": 0.3245,
      "step": 6867
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.085341260017237,
      "learning_rate": 1.74730277111614e-06,
      "loss": 0.2689,
      "step": 6868
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.775740094885824,
      "learning_rate": 1.74661757340904e-06,
      "loss": 0.2687,
      "step": 6869
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.286952259188915,
      "learning_rate": 1.7459324379427012e-06,
      "loss": 0.2383,
      "step": 6870
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.551799276816805,
      "learning_rate": 1.745247364773728e-06,
      "loss": 0.3955,
      "step": 6871
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0535980851426023,
      "learning_rate": 1.7445623539587167e-06,
      "loss": 0.0982,
      "step": 6872
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.760208233118565,
      "learning_rate": 1.7438774055542599e-06,
      "loss": 0.4034,
      "step": 6873
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.3619331992153905,
      "learning_rate": 1.7431925196169457e-06,
      "loss": 0.4202,
      "step": 6874
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.9306145945904167,
      "learning_rate": 1.7425076962033544e-06,
      "loss": 0.2948,
      "step": 6875
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.359944743673639,
      "learning_rate": 1.741822935370064e-06,
      "loss": 0.1936,
      "step": 6876
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6328466349903568,
      "learning_rate": 1.7411382371736446e-06,
      "loss": 0.0111,
      "step": 6877
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.560493725546474,
      "learning_rate": 1.7404536016706636e-06,
      "loss": 0.0953,
      "step": 6878
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.9567077379648703,
      "learning_rate": 1.7397690289176833e-06,
      "loss": 0.3565,
      "step": 6879
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.386278158574231,
      "learning_rate": 1.739084518971258e-06,
      "loss": 0.244,
      "step": 6880
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.400718083889981,
      "learning_rate": 1.7384000718879384e-06,
      "loss": 0.3043,
      "step": 6881
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.815856508079956,
      "learning_rate": 1.737715687724272e-06,
      "loss": 0.4356,
      "step": 6882
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0491261670312064,
      "learning_rate": 1.7370313665367977e-06,
      "loss": 0.1205,
      "step": 6883
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.3007132498122824,
      "learning_rate": 1.736347108382052e-06,
      "loss": 0.2222,
      "step": 6884
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.275184667796263,
      "learning_rate": 1.7356629133165633e-06,
      "loss": 0.2576,
      "step": 6885
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.955745944844858,
      "learning_rate": 1.7349787813968577e-06,
      "loss": 0.2186,
      "step": 6886
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.4597267916440217,
      "learning_rate": 1.7342947126794546e-06,
      "loss": 0.2328,
      "step": 6887
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.48296467343187,
      "learning_rate": 1.733610707220868e-06,
      "loss": 0.4587,
      "step": 6888
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.365855028788111,
      "learning_rate": 1.7329267650776083e-06,
      "loss": 0.3977,
      "step": 6889
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.2514139961628197,
      "learning_rate": 1.7322428863061775e-06,
      "loss": 0.2509,
      "step": 6890
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.081448185761862,
      "learning_rate": 1.731559070963075e-06,
      "loss": 0.3831,
      "step": 6891
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.979613880824747,
      "learning_rate": 1.7308753191047955e-06,
      "loss": 0.3501,
      "step": 6892
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.227506837967778,
      "learning_rate": 1.7301916307878254e-06,
      "loss": 0.3159,
      "step": 6893
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.9375784630807997,
      "learning_rate": 1.72950800606865e-06,
      "loss": 0.1866,
      "step": 6894
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.8137824490909624,
      "learning_rate": 1.7288244450037444e-06,
      "loss": 0.1703,
      "step": 6895
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.522046280980412,
      "learning_rate": 1.7281409476495826e-06,
      "loss": 0.2439,
      "step": 6896
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.1985799351290805,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 0.5023,
      "step": 6897
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2669738135215,
      "learning_rate": 1.726774144299353e-06,
      "loss": 0.1038,
      "step": 6898
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.643105701711743,
      "learning_rate": 1.7260908384162046e-06,
      "loss": 0.518,
      "step": 6899
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.276309107686845,
      "learning_rate": 1.725407596469636e-06,
      "loss": 0.2663,
      "step": 6900
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.80851551460529,
      "learning_rate": 1.7247244185160938e-06,
      "loss": 0.509,
      "step": 6901
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0372059602792394,
      "learning_rate": 1.7240413046120199e-06,
      "loss": 0.1317,
      "step": 6902
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.4694599129294232,
      "learning_rate": 1.7233582548138484e-06,
      "loss": 0.0466,
      "step": 6903
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.099262242840982,
      "learning_rate": 1.7226752691780116e-06,
      "loss": 0.1929,
      "step": 6904
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.609353979949531,
      "learning_rate": 1.721992347760932e-06,
      "loss": 0.3877,
      "step": 6905
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.06045647777629,
      "learning_rate": 1.7213094906190297e-06,
      "loss": 0.3406,
      "step": 6906
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.470788969361755,
      "learning_rate": 1.7206266978087204e-06,
      "loss": 0.2432,
      "step": 6907
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.580783769456666,
      "learning_rate": 1.7199439693864118e-06,
      "loss": 0.3274,
      "step": 6908
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.782632577185492,
      "learning_rate": 1.719261305408509e-06,
      "loss": 0.3663,
      "step": 6909
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.421291679022141,
      "learning_rate": 1.7185787059314085e-06,
      "loss": 0.1829,
      "step": 6910
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.785048072239248,
      "learning_rate": 1.7178961710115045e-06,
      "loss": 0.409,
      "step": 6911
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.4201225642483095,
      "learning_rate": 1.7172137007051841e-06,
      "loss": 0.3474,
      "step": 6912
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.044016122016041,
      "learning_rate": 1.71653129506883e-06,
      "loss": 0.338,
      "step": 6913
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.680381082614368,
      "learning_rate": 1.7158489541588202e-06,
      "loss": 0.1498,
      "step": 6914
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.579683168498081,
      "learning_rate": 1.715166678031524e-06,
      "loss": 0.2995,
      "step": 6915
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.3853808224497617,
      "learning_rate": 1.7144844667433094e-06,
      "loss": 0.1866,
      "step": 6916
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.268297908257949,
      "learning_rate": 1.7138023203505371e-06,
      "loss": 0.4122,
      "step": 6917
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.2481981153088526,
      "learning_rate": 1.7131202389095625e-06,
      "loss": 0.2181,
      "step": 6918
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.399982256968815,
      "learning_rate": 1.7124382224767356e-06,
      "loss": 0.3216,
      "step": 6919
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.102672892117453,
      "learning_rate": 1.711756271108403e-06,
      "loss": 0.3026,
      "step": 6920
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.274435947498631,
      "learning_rate": 1.7110743848609013e-06,
      "loss": 0.3818,
      "step": 6921
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.852874449192474,
      "learning_rate": 1.7103925637905663e-06,
      "loss": 0.4497,
      "step": 6922
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8813950874048508,
      "learning_rate": 1.7097108079537261e-06,
      "loss": 0.0334,
      "step": 6923
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.8840547499487705,
      "learning_rate": 1.7090291174067047e-06,
      "loss": 0.4478,
      "step": 6924
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.5740437756090953,
      "learning_rate": 1.7083474922058205e-06,
      "loss": 0.2791,
      "step": 6925
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.244207249237595,
      "learning_rate": 1.7076659324073842e-06,
      "loss": 0.3313,
      "step": 6926
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.624960555223291,
      "learning_rate": 1.7069844380677044e-06,
      "loss": 0.4415,
      "step": 6927
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.507857915407244,
      "learning_rate": 1.706303009243082e-06,
      "loss": 0.306,
      "step": 6928
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.162639354980201,
      "learning_rate": 1.7056216459898136e-06,
      "loss": 0.2416,
      "step": 6929
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.3783588134040263,
      "learning_rate": 1.7049403483641914e-06,
      "loss": 0.2651,
      "step": 6930
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.116300599004418,
      "learning_rate": 1.7042591164224986e-06,
      "loss": 0.3823,
      "step": 6931
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.34176343525627,
      "learning_rate": 1.7035779502210165e-06,
      "loss": 0.3216,
      "step": 6932
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.408487246135631,
      "learning_rate": 1.7028968498160192e-06,
      "loss": 0.3271,
      "step": 6933
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.733258683683631,
      "learning_rate": 1.7022158152637763e-06,
      "loss": 0.3415,
      "step": 6934
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.276883564546364,
      "learning_rate": 1.7015348466205521e-06,
      "loss": 0.4025,
      "step": 6935
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.222178124434541,
      "learning_rate": 1.7008539439426037e-06,
      "loss": 0.2026,
      "step": 6936
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1882186787876328,
      "learning_rate": 1.7001731072861843e-06,
      "loss": 0.1108,
      "step": 6937
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.829300406183084,
      "learning_rate": 1.6994923367075414e-06,
      "loss": 0.3305,
      "step": 6938
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.637475126595954,
      "learning_rate": 1.6988116322629167e-06,
      "loss": 0.4593,
      "step": 6939
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.339429280035555,
      "learning_rate": 1.6981309940085483e-06,
      "loss": 0.2481,
      "step": 6940
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.035177007707856,
      "learning_rate": 1.6974504220006643e-06,
      "loss": 0.3566,
      "step": 6941
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.263804562770265,
      "learning_rate": 1.6967699162954927e-06,
      "loss": 0.2338,
      "step": 6942
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.554338575520896,
      "learning_rate": 1.696089476949252e-06,
      "loss": 0.2683,
      "step": 6943
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.529526406167757,
      "learning_rate": 1.6954091040181573e-06,
      "loss": 0.3422,
      "step": 6944
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.12927013977257,
      "learning_rate": 1.6947287975584188e-06,
      "loss": 0.1532,
      "step": 6945
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.925304373252466,
      "learning_rate": 1.6940485576262378e-06,
      "loss": 0.1769,
      "step": 6946
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.155193954358962,
      "learning_rate": 1.6933683842778141e-06,
      "loss": 0.2832,
      "step": 6947
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.369843730731906,
      "learning_rate": 1.6926882775693396e-06,
      "loss": 0.1668,
      "step": 6948
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4172693149021294,
      "learning_rate": 1.6920082375570012e-06,
      "loss": 0.2054,
      "step": 6949
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.840304851062744,
      "learning_rate": 1.6913282642969823e-06,
      "loss": 0.4273,
      "step": 6950
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.137301170583988,
      "learning_rate": 1.6906483578454563e-06,
      "loss": 0.3034,
      "step": 6951
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.483748752002947,
      "learning_rate": 1.6899685182585952e-06,
      "loss": 0.3826,
      "step": 6952
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.627955532876359,
      "learning_rate": 1.6892887455925633e-06,
      "loss": 0.1955,
      "step": 6953
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.7404388054517983,
      "learning_rate": 1.6886090399035206e-06,
      "loss": 0.255,
      "step": 6954
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.579580724430238,
      "learning_rate": 1.6879294012476222e-06,
      "loss": 0.4955,
      "step": 6955
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.091607339341972,
      "learning_rate": 1.6872498296810142e-06,
      "loss": 0.2926,
      "step": 6956
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.188690569992049,
      "learning_rate": 1.6865703252598408e-06,
      "loss": 0.385,
      "step": 6957
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.888294496284609,
      "learning_rate": 1.6858908880402392e-06,
      "loss": 0.3195,
      "step": 6958
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.5611935918314135,
      "learning_rate": 1.6852115180783404e-06,
      "loss": 0.3034,
      "step": 6959
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.327262193790926,
      "learning_rate": 1.6845322154302726e-06,
      "loss": 0.3933,
      "step": 6960
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.191685683832204,
      "learning_rate": 1.6838529801521542e-06,
      "loss": 0.4765,
      "step": 6961
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.623794024271695,
      "learning_rate": 1.6831738123001011e-06,
      "loss": 0.3732,
      "step": 6962
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6001661300670773,
      "learning_rate": 1.6824947119302234e-06,
      "loss": 0.1215,
      "step": 6963
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9576696110005893,
      "learning_rate": 1.6818156790986242e-06,
      "loss": 0.3167,
      "step": 6964
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.3708232893376664,
      "learning_rate": 1.6811367138614033e-06,
      "loss": 0.2423,
      "step": 6965
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9428242998364,
      "learning_rate": 1.6804578162746512e-06,
      "loss": 0.1697,
      "step": 6966
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.154598894235592,
      "learning_rate": 1.6797789863944563e-06,
      "loss": 0.3315,
      "step": 6967
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.2450903688659,
      "learning_rate": 1.6791002242769011e-06,
      "loss": 0.4147,
      "step": 6968
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8391698807674626,
      "learning_rate": 1.6784215299780598e-06,
      "loss": 0.1981,
      "step": 6969
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.142985803668389,
      "learning_rate": 1.6777429035540052e-06,
      "loss": 0.3496,
      "step": 6970
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.395305102942095,
      "learning_rate": 1.6770643450607993e-06,
      "loss": 0.2874,
      "step": 6971
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.543873134568923,
      "learning_rate": 1.6763858545545026e-06,
      "loss": 0.5479,
      "step": 6972
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.324899388707213,
      "learning_rate": 1.675707432091169e-06,
      "loss": 0.3324,
      "step": 6973
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.4088616810536934,
      "learning_rate": 1.6750290777268457e-06,
      "loss": 0.1084,
      "step": 6974
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.430580823611095,
      "learning_rate": 1.6743507915175764e-06,
      "loss": 0.1888,
      "step": 6975
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.175290518842332,
      "learning_rate": 1.6736725735193958e-06,
      "loss": 0.2881,
      "step": 6976
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.339674409962046,
      "learning_rate": 1.6729944237883358e-06,
      "loss": 0.391,
      "step": 6977
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.663916998548889,
      "learning_rate": 1.6723163423804222e-06,
      "loss": 0.4495,
      "step": 6978
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.361732340687766,
      "learning_rate": 1.6716383293516743e-06,
      "loss": 0.4571,
      "step": 6979
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.325693109346079,
      "learning_rate": 1.6709603847581074e-06,
      "loss": 0.4423,
      "step": 6980
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.87964363458759,
      "learning_rate": 1.6702825086557278e-06,
      "loss": 0.4045,
      "step": 6981
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5952114175578167,
      "learning_rate": 1.6696047011005395e-06,
      "loss": 0.1333,
      "step": 6982
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.313965809477165,
      "learning_rate": 1.6689269621485404e-06,
      "loss": 0.289,
      "step": 6983
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9245004077815864,
      "learning_rate": 1.6682492918557203e-06,
      "loss": 0.1498,
      "step": 6984
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.230928721621725,
      "learning_rate": 1.667571690278067e-06,
      "loss": 0.2989,
      "step": 6985
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8788356550596013,
      "learning_rate": 1.6668941574715582e-06,
      "loss": 0.1968,
      "step": 6986
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.0640584776633935,
      "learning_rate": 1.6662166934921697e-06,
      "loss": 0.2516,
      "step": 6987
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.1022391844872343,
      "learning_rate": 1.665539298395871e-06,
      "loss": 0.1912,
      "step": 6988
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.281968140211586,
      "learning_rate": 1.6648619722386236e-06,
      "loss": 0.4853,
      "step": 6989
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.370203103688151,
      "learning_rate": 1.6641847150763867e-06,
      "loss": 0.4206,
      "step": 6990
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.57914049317633,
      "learning_rate": 1.6635075269651096e-06,
      "loss": 0.3113,
      "step": 6991
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4619046160540954,
      "learning_rate": 1.66283040796074e-06,
      "loss": 0.2348,
      "step": 6992
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4473167018331323,
      "learning_rate": 1.6621533581192175e-06,
      "loss": 0.2035,
      "step": 6993
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.3507948897167315,
      "learning_rate": 1.6614763774964768e-06,
      "loss": 0.2454,
      "step": 6994
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.540850994525486,
      "learning_rate": 1.6607994661484474e-06,
      "loss": 0.1526,
      "step": 6995
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.8221062595447326,
      "learning_rate": 1.660122624131051e-06,
      "loss": 0.1507,
      "step": 6996
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.184865689360005,
      "learning_rate": 1.6594458515002057e-06,
      "loss": 0.2693,
      "step": 6997
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.427334113698754,
      "learning_rate": 1.6587691483118235e-06,
      "loss": 0.4623,
      "step": 6998
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.289938587977055,
      "learning_rate": 1.6580925146218096e-06,
      "loss": 0.3731,
      "step": 6999
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.4961895483691072,
      "learning_rate": 1.6574159504860642e-06,
      "loss": 0.1974,
      "step": 7000
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.893480732721543,
      "learning_rate": 1.6567394559604833e-06,
      "loss": 0.4741,
      "step": 7001
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.037068316357253,
      "learning_rate": 1.656063031100953e-06,
      "loss": 0.3067,
      "step": 7002
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.649807700448784,
      "learning_rate": 1.655386675963358e-06,
      "loss": 0.3467,
      "step": 7003
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6433969685233847,
      "learning_rate": 1.6547103906035747e-06,
      "loss": 0.3274,
      "step": 7004
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.41222280160926,
      "learning_rate": 1.6540341750774747e-06,
      "loss": 0.2752,
      "step": 7005
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9385168610824417,
      "learning_rate": 1.6533580294409243e-06,
      "loss": 0.1592,
      "step": 7006
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.6657160051462907,
      "learning_rate": 1.6526819537497819e-06,
      "loss": 0.2284,
      "step": 7007
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.825947552366036,
      "learning_rate": 1.6520059480599026e-06,
      "loss": 0.1471,
      "step": 7008
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.7250588503585544,
      "learning_rate": 1.651330012427134e-06,
      "loss": 0.1466,
      "step": 7009
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.0563677654099695,
      "learning_rate": 1.6506541469073189e-06,
      "loss": 0.2081,
      "step": 7010
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.573911723921668,
      "learning_rate": 1.649978351556295e-06,
      "loss": 0.3404,
      "step": 7011
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.057167741659203,
      "learning_rate": 1.6493026264298908e-06,
      "loss": 0.3535,
      "step": 7012
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.3847864299016437,
      "learning_rate": 1.6486269715839344e-06,
      "loss": 0.2422,
      "step": 7013
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.303413848204608,
      "learning_rate": 1.6479513870742425e-06,
      "loss": 0.39,
      "step": 7014
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.391320290652216,
      "learning_rate": 1.64727587295663e-06,
      "loss": 0.3907,
      "step": 7015
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.607013040006671,
      "learning_rate": 1.6466004292869047e-06,
      "loss": 0.297,
      "step": 7016
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9354098291802002,
      "learning_rate": 1.6459250561208668e-06,
      "loss": 0.0156,
      "step": 7017
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.377425479291724,
      "learning_rate": 1.6452497535143147e-06,
      "loss": 0.2436,
      "step": 7018
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.229051822570936,
      "learning_rate": 1.6445745215230364e-06,
      "loss": 0.3139,
      "step": 7019
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.397642088360675,
      "learning_rate": 1.6438993602028172e-06,
      "loss": 0.3763,
      "step": 7020
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.9199144814199234,
      "learning_rate": 1.6432242696094369e-06,
      "loss": 0.2249,
      "step": 7021
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.205678902678577,
      "learning_rate": 1.642549249798665e-06,
      "loss": 0.3346,
      "step": 7022
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.8394190711526175,
      "learning_rate": 1.641874300826272e-06,
      "loss": 0.3369,
      "step": 7023
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.650141185150985,
      "learning_rate": 1.641199422748016e-06,
      "loss": 0.4024,
      "step": 7024
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.523050580180613,
      "learning_rate": 1.6405246156196531e-06,
      "loss": 0.304,
      "step": 7025
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.1375574001369415,
      "learning_rate": 1.6398498794969334e-06,
      "loss": 0.4091,
      "step": 7026
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.470773192913777,
      "learning_rate": 1.6391752144355988e-06,
      "loss": 0.3074,
      "step": 7027
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.786858420803949,
      "learning_rate": 1.638500620491389e-06,
      "loss": 0.322,
      "step": 7028
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.901754831003708,
      "learning_rate": 1.6378260977200328e-06,
      "loss": 0.5505,
      "step": 7029
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.426991145316721,
      "learning_rate": 1.6371516461772574e-06,
      "loss": 0.1801,
      "step": 7030
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.436503327766157,
      "learning_rate": 1.6364772659187836e-06,
      "loss": 0.0982,
      "step": 7031
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.0231344237221345,
      "learning_rate": 1.6358029570003237e-06,
      "loss": 0.2945,
      "step": 7032
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.848420809537878,
      "learning_rate": 1.6351287194775876e-06,
      "loss": 0.4088,
      "step": 7033
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.226223402076867,
      "learning_rate": 1.6344545534062755e-06,
      "loss": 0.3625,
      "step": 7034
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.690920560735139,
      "learning_rate": 1.6337804588420848e-06,
      "loss": 0.0804,
      "step": 7035
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.042820692252592,
      "learning_rate": 1.6331064358407056e-06,
      "loss": 0.188,
      "step": 7036
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.879800561912365,
      "learning_rate": 1.6324324844578227e-06,
      "loss": 0.3912,
      "step": 7037
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.2881492589954755,
      "learning_rate": 1.6317586047491153e-06,
      "loss": 0.4682,
      "step": 7038
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.747948360867861,
      "learning_rate": 1.6310847967702541e-06,
      "loss": 0.3397,
      "step": 7039
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.1048358383529955,
      "learning_rate": 1.630411060576907e-06,
      "loss": 0.3175,
      "step": 7040
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.3604650180842865,
      "learning_rate": 1.6297373962247352e-06,
      "loss": 0.314,
      "step": 7041
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.5228890514836446,
      "learning_rate": 1.6290638037693923e-06,
      "loss": 0.1406,
      "step": 7042
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.74368762959416,
      "learning_rate": 1.6283902832665282e-06,
      "loss": 0.2822,
      "step": 7043
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.484915397058134,
      "learning_rate": 1.627716834771787e-06,
      "loss": 0.2863,
      "step": 7044
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0168491167653206,
      "learning_rate": 1.627043458340803e-06,
      "loss": 0.075,
      "step": 7045
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.596940842328778,
      "learning_rate": 1.6263701540292088e-06,
      "loss": 0.2953,
      "step": 7046
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2720560471050435,
      "learning_rate": 1.625696921892629e-06,
      "loss": 0.089,
      "step": 7047
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.438920706082827,
      "learning_rate": 1.6250237619866833e-06,
      "loss": 0.3034,
      "step": 7048
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.462430517623583,
      "learning_rate": 1.6243506743669854e-06,
      "loss": 0.3262,
      "step": 7049
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.7713101951946015,
      "learning_rate": 1.6236776590891407e-06,
      "loss": 0.2257,
      "step": 7050
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.049713698479334,
      "learning_rate": 1.623004716208752e-06,
      "loss": 0.337,
      "step": 7051
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.373494460138538,
      "learning_rate": 1.6223318457814136e-06,
      "loss": 0.3469,
      "step": 7052
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.847247623166326,
      "learning_rate": 1.6216590478627148e-06,
      "loss": 0.3918,
      "step": 7053
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.067930232713979,
      "learning_rate": 1.6209863225082403e-06,
      "loss": 0.1549,
      "step": 7054
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.988834331781426,
      "learning_rate": 1.6203136697735652e-06,
      "loss": 0.3305,
      "step": 7055
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.254128327007032,
      "learning_rate": 1.6196410897142624e-06,
      "loss": 0.3284,
      "step": 7056
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.249590122207811,
      "learning_rate": 1.6189685823858964e-06,
      "loss": 0.3563,
      "step": 7057
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.6658741994705504,
      "learning_rate": 1.6182961478440261e-06,
      "loss": 0.0922,
      "step": 7058
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.057074791648175,
      "learning_rate": 1.6176237861442068e-06,
      "loss": 0.3586,
      "step": 7059
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.081906525845877,
      "learning_rate": 1.616951497341983e-06,
      "loss": 0.3116,
      "step": 7060
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.004544764816244,
      "learning_rate": 1.6162792814928976e-06,
      "loss": 0.1996,
      "step": 7061
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.479030909964954,
      "learning_rate": 1.6156071386524848e-06,
      "loss": 0.1276,
      "step": 7062
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.118502011621433,
      "learning_rate": 1.6149350688762744e-06,
      "loss": 0.3057,
      "step": 7063
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.355732662867308,
      "learning_rate": 1.6142630722197905e-06,
      "loss": 0.4108,
      "step": 7064
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.354744189448911,
      "learning_rate": 1.6135911487385477e-06,
      "loss": 0.0541,
      "step": 7065
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.82403255809843,
      "learning_rate": 1.6129192984880593e-06,
      "loss": 0.0896,
      "step": 7066
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.865310269303208,
      "learning_rate": 1.6122475215238284e-06,
      "loss": 0.3156,
      "step": 7067
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8989273155033715,
      "learning_rate": 1.6115758179013549e-06,
      "loss": 0.2751,
      "step": 7068
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.807664106986929,
      "learning_rate": 1.6109041876761328e-06,
      "loss": 0.3449,
      "step": 7069
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.566042268543604,
      "learning_rate": 1.6102326309036464e-06,
      "loss": 0.3264,
      "step": 7070
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8726014062197898,
      "learning_rate": 1.6095611476393786e-06,
      "loss": 0.2042,
      "step": 7071
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.770592108210698,
      "learning_rate": 1.6088897379388027e-06,
      "loss": 0.4496,
      "step": 7072
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.44399890375819,
      "learning_rate": 1.6082184018573876e-06,
      "loss": 0.4101,
      "step": 7073
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.840122715198851,
      "learning_rate": 1.607547139450597e-06,
      "loss": 0.4734,
      "step": 7074
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.1693004362134785,
      "learning_rate": 1.6068759507738851e-06,
      "loss": 0.3988,
      "step": 7075
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.910053552381559,
      "learning_rate": 1.6062048358827043e-06,
      "loss": 0.4509,
      "step": 7076
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.747923645404323,
      "learning_rate": 1.6055337948324973e-06,
      "loss": 0.3707,
      "step": 7077
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.2745349911838364,
      "learning_rate": 1.6048628276787025e-06,
      "loss": 0.2261,
      "step": 7078
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.219265982392538,
      "learning_rate": 1.6041919344767538e-06,
      "loss": 0.3483,
      "step": 7079
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.035863864033851,
      "learning_rate": 1.603521115282074e-06,
      "loss": 0.4184,
      "step": 7080
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.734862146843826,
      "learning_rate": 1.6028503701500848e-06,
      "loss": 0.344,
      "step": 7081
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.4625470071700404,
      "learning_rate": 1.6021796991361998e-06,
      "loss": 0.2793,
      "step": 7082
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.465500331013594,
      "learning_rate": 1.6015091022958263e-06,
      "loss": 0.4316,
      "step": 7083
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.805912656779085,
      "learning_rate": 1.6008385796843663e-06,
      "loss": 0.3171,
      "step": 7084
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.035766339367827,
      "learning_rate": 1.6001681313572133e-06,
      "loss": 0.3732,
      "step": 7085
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.666586134466964,
      "learning_rate": 1.599497757369758e-06,
      "loss": 0.2232,
      "step": 7086
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.044853997745934,
      "learning_rate": 1.5988274577773831e-06,
      "loss": 0.3226,
      "step": 7087
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.836946928622439,
      "learning_rate": 1.5981572326354656e-06,
      "loss": 0.231,
      "step": 7088
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6019040731917475,
      "learning_rate": 1.5974870819993765e-06,
      "loss": 0.2695,
      "step": 7089
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.340206906771675,
      "learning_rate": 1.5968170059244791e-06,
      "loss": 0.3586,
      "step": 7090
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.0541707385291734,
      "learning_rate": 1.596147004466133e-06,
      "loss": 0.1439,
      "step": 7091
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.983998089105042,
      "learning_rate": 1.5954770776796901e-06,
      "loss": 0.5394,
      "step": 7092
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6781328194513923,
      "learning_rate": 1.5948072256204957e-06,
      "loss": 0.316,
      "step": 7093
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.151517411982221,
      "learning_rate": 1.5941374483438918e-06,
      "loss": 0.3201,
      "step": 7094
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.547821594899905,
      "learning_rate": 1.5934677459052096e-06,
      "loss": 0.2666,
      "step": 7095
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6522588132687415,
      "learning_rate": 1.5927981183597776e-06,
      "loss": 0.2416,
      "step": 7096
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.70194700621993,
      "learning_rate": 1.5921285657629183e-06,
      "loss": 0.1829,
      "step": 7097
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.928730704170532,
      "learning_rate": 1.5914590881699446e-06,
      "loss": 0.407,
      "step": 7098
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.94857513608025,
      "learning_rate": 1.5907896856361682e-06,
      "loss": 0.4142,
      "step": 7099
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.7094502331268,
      "learning_rate": 1.590120358216889e-06,
      "loss": 0.2923,
      "step": 7100
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.169660618406799,
      "learning_rate": 1.589451105967405e-06,
      "loss": 0.3173,
      "step": 7101
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.5328340721975175,
      "learning_rate": 1.5887819289430073e-06,
      "loss": 0.3782,
      "step": 7102
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.8419959164404105,
      "learning_rate": 1.5881128271989781e-06,
      "loss": 0.2572,
      "step": 7103
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.708650332733102,
      "learning_rate": 1.587443800790598e-06,
      "loss": 0.2826,
      "step": 7104
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.923351583414813,
      "learning_rate": 1.5867748497731356e-06,
      "loss": 0.3915,
      "step": 7105
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.629583093568575,
      "learning_rate": 1.5861059742018581e-06,
      "loss": 0.2137,
      "step": 7106
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.486044060237522,
      "learning_rate": 1.5854371741320248e-06,
      "loss": 0.2756,
      "step": 7107
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.391357437213046,
      "learning_rate": 1.5847684496188879e-06,
      "loss": 0.1332,
      "step": 7108
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.617184167732644,
      "learning_rate": 1.5840998007176956e-06,
      "loss": 0.2867,
      "step": 7109
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.3848225365677855,
      "learning_rate": 1.5834312274836861e-06,
      "loss": 0.3703,
      "step": 7110
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.8808743252309204,
      "learning_rate": 1.582762729972095e-06,
      "loss": 0.1412,
      "step": 7111
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.512161913301989,
      "learning_rate": 1.5820943082381507e-06,
      "loss": 0.4009,
      "step": 7112
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.234048971618151,
      "learning_rate": 1.5814259623370738e-06,
      "loss": 0.0815,
      "step": 7113
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.101831232090363,
      "learning_rate": 1.5807576923240814e-06,
      "loss": 0.2238,
      "step": 7114
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.464308687210738,
      "learning_rate": 1.5800894982543807e-06,
      "loss": 0.3954,
      "step": 7115
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.194925572677613,
      "learning_rate": 1.5794213801831754e-06,
      "loss": 0.2788,
      "step": 7116
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.338507587740539,
      "learning_rate": 1.5787533381656633e-06,
      "loss": 0.2517,
      "step": 7117
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9571039849194083,
      "learning_rate": 1.5780853722570326e-06,
      "loss": 0.1893,
      "step": 7118
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.7212890292957974,
      "learning_rate": 1.57741748251247e-06,
      "loss": 0.1618,
      "step": 7119
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.735330495286839,
      "learning_rate": 1.5767496689871509e-06,
      "loss": 0.3616,
      "step": 7120
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.249912427068156,
      "learning_rate": 1.5760819317362474e-06,
      "loss": 0.3318,
      "step": 7121
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.464273888179677,
      "learning_rate": 1.5754142708149256e-06,
      "loss": 0.3054,
      "step": 7122
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.6797471863327385,
      "learning_rate": 1.5747466862783434e-06,
      "loss": 0.4443,
      "step": 7123
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.144506012456278,
      "learning_rate": 1.5740791781816534e-06,
      "loss": 0.453,
      "step": 7124
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.450886453469471,
      "learning_rate": 1.5734117465800034e-06,
      "loss": 0.154,
      "step": 7125
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.221964554391813,
      "learning_rate": 1.572744391528531e-06,
      "loss": 0.2331,
      "step": 7126
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.8086712391109696,
      "learning_rate": 1.5720771130823715e-06,
      "loss": 0.288,
      "step": 7127
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.447487152543712,
      "learning_rate": 1.5714099112966508e-06,
      "loss": 0.2539,
      "step": 7128
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.171770379596019,
      "learning_rate": 1.5707427862264907e-06,
      "loss": 0.2057,
      "step": 7129
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.799641338052087,
      "learning_rate": 1.5700757379270066e-06,
      "loss": 0.2405,
      "step": 7130
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.150673015348542,
      "learning_rate": 1.5694087664533051e-06,
      "loss": 0.3753,
      "step": 7131
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.085678323883937,
      "learning_rate": 1.5687418718604892e-06,
      "loss": 0.3203,
      "step": 7132
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.63993938894196,
      "learning_rate": 1.5680750542036538e-06,
      "loss": 0.238,
      "step": 7133
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.7357455174093306,
      "learning_rate": 1.5674083135378882e-06,
      "loss": 0.1332,
      "step": 7134
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.81602059585527,
      "learning_rate": 1.5667416499182764e-06,
      "loss": 0.4883,
      "step": 7135
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.933348194145216,
      "learning_rate": 1.5660750633998933e-06,
      "loss": 0.4611,
      "step": 7136
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.801419549208002,
      "learning_rate": 1.56540855403781e-06,
      "loss": 0.4217,
      "step": 7137
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.960157139683933,
      "learning_rate": 1.5647421218870896e-06,
      "loss": 0.1852,
      "step": 7138
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.506824721757594,
      "learning_rate": 1.5640757670027896e-06,
      "loss": 0.262,
      "step": 7139
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.924013642751717,
      "learning_rate": 1.563409489439962e-06,
      "loss": 0.2249,
      "step": 7140
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.4547269878992486,
      "learning_rate": 1.5627432892536498e-06,
      "loss": 0.1496,
      "step": 7141
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.876665224552324,
      "learning_rate": 1.5620771664988923e-06,
      "loss": 0.4005,
      "step": 7142
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.436978682274988,
      "learning_rate": 1.5614111212307203e-06,
      "loss": 0.1547,
      "step": 7143
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6568186413216734,
      "learning_rate": 1.56074515350416e-06,
      "loss": 0.2609,
      "step": 7144
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.332396494634318,
      "learning_rate": 1.5600792633742312e-06,
      "loss": 0.2669,
      "step": 7145
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.5520446571443856,
      "learning_rate": 1.5594134508959444e-06,
      "loss": 0.1227,
      "step": 7146
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.145725811338305,
      "learning_rate": 1.5587477161243075e-06,
      "loss": 0.2476,
      "step": 7147
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.283919884139985,
      "learning_rate": 1.5580820591143187e-06,
      "loss": 0.4707,
      "step": 7148
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9983124289387573,
      "learning_rate": 1.5574164799209724e-06,
      "loss": 0.1609,
      "step": 7149
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.027443311153944,
      "learning_rate": 1.5567509785992562e-06,
      "loss": 0.205,
      "step": 7150
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.815193019942455,
      "learning_rate": 1.556085555204149e-06,
      "loss": 0.3913,
      "step": 7151
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.4388473014722023,
      "learning_rate": 1.5554202097906257e-06,
      "loss": 0.1808,
      "step": 7152
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.9891780336507563,
      "learning_rate": 1.5547549424136531e-06,
      "loss": 0.1749,
      "step": 7153
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.049746864407419,
      "learning_rate": 1.5540897531281928e-06,
      "loss": 0.2736,
      "step": 7154
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.6418108676398115,
      "learning_rate": 1.553424641989201e-06,
      "loss": 0.2206,
      "step": 7155
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.3153572673660774,
      "learning_rate": 1.5527596090516232e-06,
      "loss": 0.2448,
      "step": 7156
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.300574873193768,
      "learning_rate": 1.5520946543704029e-06,
      "loss": 0.2394,
      "step": 7157
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.525912280721065,
      "learning_rate": 1.5514297780004744e-06,
      "loss": 0.3884,
      "step": 7158
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.971561638877869,
      "learning_rate": 1.5507649799967672e-06,
      "loss": 0.2775,
      "step": 7159
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.656999724451364,
      "learning_rate": 1.550100260414204e-06,
      "loss": 0.3811,
      "step": 7160
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.917506469297454,
      "learning_rate": 1.5494356193076998e-06,
      "loss": 0.3977,
      "step": 7161
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.5622580463815776,
      "learning_rate": 1.548771056732164e-06,
      "loss": 0.205,
      "step": 7162
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.8422850663206907,
      "learning_rate": 1.5481065727425004e-06,
      "loss": 0.1929,
      "step": 7163
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.72324047527044,
      "learning_rate": 1.5474421673936047e-06,
      "loss": 0.3198,
      "step": 7164
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9190907555808003,
      "learning_rate": 1.5467778407403678e-06,
      "loss": 0.2205,
      "step": 7165
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.748283429679403,
      "learning_rate": 1.5461135928376714e-06,
      "loss": 0.1009,
      "step": 7166
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.339438304490097,
      "learning_rate": 1.5454494237403926e-06,
      "loss": 0.2627,
      "step": 7167
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.29679890421365,
      "learning_rate": 1.544785333503404e-06,
      "loss": 0.2435,
      "step": 7168
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.902310872219388,
      "learning_rate": 1.544121322181567e-06,
      "loss": 0.1612,
      "step": 7169
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.3188904011373985,
      "learning_rate": 1.5434573898297412e-06,
      "loss": 0.2973,
      "step": 7170
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.290725852587802,
      "learning_rate": 1.5427935365027745e-06,
      "loss": 0.2185,
      "step": 7171
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.5805093557954404,
      "learning_rate": 1.5421297622555136e-06,
      "loss": 0.2764,
      "step": 7172
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.280101417566462,
      "learning_rate": 1.5414660671427958e-06,
      "loss": 0.4038,
      "step": 7173
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.663465275292041,
      "learning_rate": 1.5408024512194517e-06,
      "loss": 0.3414,
      "step": 7174
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.170776326075762,
      "learning_rate": 1.5401389145403073e-06,
      "loss": 0.3224,
      "step": 7175
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.5091900113824663,
      "learning_rate": 1.5394754571601789e-06,
      "loss": 0.2411,
      "step": 7176
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.109667335887818,
      "learning_rate": 1.538812079133879e-06,
      "loss": 0.373,
      "step": 7177
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.269866538395284,
      "learning_rate": 1.5381487805162129e-06,
      "loss": 0.3822,
      "step": 7178
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.879745874590972,
      "learning_rate": 1.5374855613619786e-06,
      "loss": 0.3045,
      "step": 7179
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.565261847088549,
      "learning_rate": 1.5368224217259692e-06,
      "loss": 0.164,
      "step": 7180
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.763934302619811,
      "learning_rate": 1.5361593616629683e-06,
      "loss": 0.1575,
      "step": 7181
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.232106044563622,
      "learning_rate": 1.5354963812277551e-06,
      "loss": 0.2046,
      "step": 7182
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.756408344131939,
      "learning_rate": 1.5348334804751031e-06,
      "loss": 0.4582,
      "step": 7183
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.603143846715757,
      "learning_rate": 1.534170659459776e-06,
      "loss": 0.4445,
      "step": 7184
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.978680629309807,
      "learning_rate": 1.5335079182365354e-06,
      "loss": 0.3949,
      "step": 7185
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.678790907302208,
      "learning_rate": 1.532845256860131e-06,
      "loss": 0.1217,
      "step": 7186
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.4551770354002365,
      "learning_rate": 1.5321826753853094e-06,
      "loss": 0.5596,
      "step": 7187
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.0625619771596035,
      "learning_rate": 1.5315201738668113e-06,
      "loss": 0.3133,
      "step": 7188
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.040685096645501,
      "learning_rate": 1.5308577523593672e-06,
      "loss": 0.3941,
      "step": 7189
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.754750201585418,
      "learning_rate": 1.5301954109177052e-06,
      "loss": 0.2695,
      "step": 7190
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.177559958920007,
      "learning_rate": 1.5295331495965426e-06,
      "loss": 0.3587,
      "step": 7191
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.6762484316392623,
      "learning_rate": 1.5288709684505936e-06,
      "loss": 0.1329,
      "step": 7192
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.808534158668154,
      "learning_rate": 1.528208867534564e-06,
      "loss": 0.5266,
      "step": 7193
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.67571822107421,
      "learning_rate": 1.5275468469031534e-06,
      "loss": 0.3231,
      "step": 7194
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.457051547387675,
      "learning_rate": 1.5268849066110553e-06,
      "loss": 0.2916,
      "step": 7195
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2730043440217,
      "learning_rate": 1.5262230467129547e-06,
      "loss": 0.1842,
      "step": 7196
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.5816374956651345,
      "learning_rate": 1.525561267263531e-06,
      "loss": 0.4688,
      "step": 7197
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1517341252118087,
      "learning_rate": 1.5248995683174592e-06,
      "loss": 0.2517,
      "step": 7198
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.7684274736190937,
      "learning_rate": 1.5242379499294036e-06,
      "loss": 0.3045,
      "step": 7199
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.0026858807033205,
      "learning_rate": 1.5235764121540258e-06,
      "loss": 0.3528,
      "step": 7200
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.217267088658071,
      "learning_rate": 1.5229149550459765e-06,
      "loss": 0.2226,
      "step": 7201
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.65331520960653,
      "learning_rate": 1.5222535786599034e-06,
      "loss": 0.3251,
      "step": 7202
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.361493597660047,
      "learning_rate": 1.5215922830504466e-06,
      "loss": 0.268,
      "step": 7203
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9571401105658801,
      "learning_rate": 1.5209310682722378e-06,
      "loss": 0.0078,
      "step": 7204
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.1774103859603535,
      "learning_rate": 1.5202699343799039e-06,
      "loss": 0.3174,
      "step": 7205
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0131705415004513,
      "learning_rate": 1.5196088814280656e-06,
      "loss": 0.0273,
      "step": 7206
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.479786278155258,
      "learning_rate": 1.5189479094713345e-06,
      "loss": 0.4489,
      "step": 7207
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1639276360642,
      "learning_rate": 1.5182870185643173e-06,
      "loss": 0.0507,
      "step": 7208
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.669827065005891,
      "learning_rate": 1.5176262087616131e-06,
      "loss": 0.2966,
      "step": 7209
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.8513647029964844,
      "learning_rate": 1.5169654801178153e-06,
      "loss": 0.1359,
      "step": 7210
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.15354729987916,
      "learning_rate": 1.516304832687511e-06,
      "loss": 0.4129,
      "step": 7211
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.4618096443088473,
      "learning_rate": 1.5156442665252773e-06,
      "loss": 0.0855,
      "step": 7212
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.220068374609638,
      "learning_rate": 1.5149837816856893e-06,
      "loss": 0.1942,
      "step": 7213
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.19838607039015,
      "learning_rate": 1.5143233782233113e-06,
      "loss": 0.4163,
      "step": 7214
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2772104577545735,
      "learning_rate": 1.513663056192703e-06,
      "loss": 0.1696,
      "step": 7215
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.591855333350475,
      "learning_rate": 1.5130028156484182e-06,
      "loss": 0.1479,
      "step": 7216
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.042085087240773,
      "learning_rate": 1.5123426566450008e-06,
      "loss": 0.2402,
      "step": 7217
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.454823168485757,
      "learning_rate": 1.511682579236992e-06,
      "loss": 0.4641,
      "step": 7218
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.009401560327983,
      "learning_rate": 1.5110225834789223e-06,
      "loss": 0.371,
      "step": 7219
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.641026426027564,
      "learning_rate": 1.5103626694253179e-06,
      "loss": 0.5821,
      "step": 7220
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.098618314013031,
      "learning_rate": 1.5097028371306988e-06,
      "loss": 0.2038,
      "step": 7221
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.730952661913971,
      "learning_rate": 1.5090430866495748e-06,
      "loss": 0.4683,
      "step": 7222
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.599141802437801,
      "learning_rate": 1.5083834180364538e-06,
      "loss": 0.3357,
      "step": 7223
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.7875415989271706,
      "learning_rate": 1.5077238313458325e-06,
      "loss": 0.2958,
      "step": 7224
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.930121721006549,
      "learning_rate": 1.5070643266322033e-06,
      "loss": 0.2397,
      "step": 7225
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.076960916082744,
      "learning_rate": 1.5064049039500523e-06,
      "loss": 0.3454,
      "step": 7226
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2250032420636465,
      "learning_rate": 1.5057455633538548e-06,
      "loss": 0.2207,
      "step": 7227
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.962419362526952,
      "learning_rate": 1.5050863048980862e-06,
      "loss": 0.273,
      "step": 7228
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.488122649194498,
      "learning_rate": 1.5044271286372086e-06,
      "loss": 0.1249,
      "step": 7229
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.395152066297539,
      "learning_rate": 1.5037680346256804e-06,
      "loss": 0.3102,
      "step": 7230
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.5380843997589775,
      "learning_rate": 1.5031090229179533e-06,
      "loss": 0.5767,
      "step": 7231
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.012988830341083,
      "learning_rate": 1.5024500935684704e-06,
      "loss": 0.289,
      "step": 7232
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.668532960679266,
      "learning_rate": 1.5017912466316716e-06,
      "loss": 0.3121,
      "step": 7233
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.043856906771446,
      "learning_rate": 1.5011324821619849e-06,
      "loss": 0.3142,
      "step": 7234
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.726904314212625,
      "learning_rate": 1.5004738002138353e-06,
      "loss": 0.4244,
      "step": 7235
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.487137968160228,
      "learning_rate": 1.4998152008416406e-06,
      "loss": 0.179,
      "step": 7236
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.084595952804876,
      "learning_rate": 1.4991566840998095e-06,
      "loss": 0.4783,
      "step": 7237
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.435789085937524,
      "learning_rate": 1.4984982500427476e-06,
      "loss": 0.2434,
      "step": 7238
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.4486483320841548,
      "learning_rate": 1.4978398987248488e-06,
      "loss": 0.1958,
      "step": 7239
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.503099016345383,
      "learning_rate": 1.4971816302005047e-06,
      "loss": 0.2864,
      "step": 7240
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.482388418692846,
      "learning_rate": 1.4965234445240977e-06,
      "loss": 0.5401,
      "step": 7241
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.03265585288778,
      "learning_rate": 1.4958653417500042e-06,
      "loss": 0.3427,
      "step": 7242
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.1619389615034517,
      "learning_rate": 1.4952073219325925e-06,
      "loss": 0.2268,
      "step": 7243
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.602008408669286,
      "learning_rate": 1.4945493851262271e-06,
      "loss": 0.3461,
      "step": 7244
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9825082429367904,
      "learning_rate": 1.493891531385261e-06,
      "loss": 0.1801,
      "step": 7245
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.3707218274353306,
      "learning_rate": 1.493233760764044e-06,
      "loss": 0.285,
      "step": 7246
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.833873719248476,
      "learning_rate": 1.4925760733169176e-06,
      "loss": 0.2578,
      "step": 7247
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.057283322784723,
      "learning_rate": 1.4919184690982167e-06,
      "loss": 0.2138,
      "step": 7248
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.638387049745083,
      "learning_rate": 1.4912609481622704e-06,
      "loss": 0.4765,
      "step": 7249
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.00572936109928,
      "learning_rate": 1.4906035105633982e-06,
      "loss": 0.5159,
      "step": 7250
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6260511106017024,
      "learning_rate": 1.4899461563559159e-06,
      "loss": 0.0316,
      "step": 7251
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.4587327335577136,
      "learning_rate": 1.489288885594129e-06,
      "loss": 0.1668,
      "step": 7252
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.481411397959828,
      "learning_rate": 1.4886316983323395e-06,
      "loss": 0.5013,
      "step": 7253
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.220077524705162,
      "learning_rate": 1.487974594624842e-06,
      "loss": 0.2329,
      "step": 7254
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.5502293443052455,
      "learning_rate": 1.48731757452592e-06,
      "loss": 0.2209,
      "step": 7255
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.857726871834593,
      "learning_rate": 1.4866606380898563e-06,
      "loss": 0.2909,
      "step": 7256
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.0970234605910143,
      "learning_rate": 1.4860037853709213e-06,
      "loss": 0.1097,
      "step": 7257
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.576890386915594,
      "learning_rate": 1.485347016423383e-06,
      "loss": 0.2959,
      "step": 7258
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.969718053257797,
      "learning_rate": 1.4846903313015002e-06,
      "loss": 0.2676,
      "step": 7259
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.362451204791467,
      "learning_rate": 1.4840337300595234e-06,
      "loss": 0.4651,
      "step": 7260
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.0543039018746,
      "learning_rate": 1.4833772127516994e-06,
      "loss": 0.1031,
      "step": 7261
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9943199526655233,
      "learning_rate": 1.4827207794322654e-06,
      "loss": 0.1931,
      "step": 7262
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.869257101074904,
      "learning_rate": 1.482064430155453e-06,
      "loss": 0.3073,
      "step": 7263
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.401148781217932,
      "learning_rate": 1.481408164975488e-06,
      "loss": 0.3553,
      "step": 7264
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.9445909118908484,
      "learning_rate": 1.480751983946585e-06,
      "loss": 0.3476,
      "step": 7265
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.727781799141562,
      "learning_rate": 1.480095887122957e-06,
      "loss": 0.1913,
      "step": 7266
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.9983921729911094,
      "learning_rate": 1.4794398745588056e-06,
      "loss": 0.1423,
      "step": 7267
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.687854456529687,
      "learning_rate": 1.4787839463083286e-06,
      "loss": 0.361,
      "step": 7268
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.429040483215981,
      "learning_rate": 1.478128102425716e-06,
      "loss": 0.3749,
      "step": 7269
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.5017076630059867,
      "learning_rate": 1.4774723429651488e-06,
      "loss": 0.2812,
      "step": 7270
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.714757854043945,
      "learning_rate": 1.4768166679808038e-06,
      "loss": 0.0861,
      "step": 7271
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.027849259148009,
      "learning_rate": 1.4761610775268487e-06,
      "loss": 0.3449,
      "step": 7272
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.1604580123325015,
      "learning_rate": 1.4755055716574457e-06,
      "loss": 0.3299,
      "step": 7273
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.435326949195346,
      "learning_rate": 1.4748501504267507e-06,
      "loss": 0.5407,
      "step": 7274
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.560349365771371,
      "learning_rate": 1.4741948138889091e-06,
      "loss": 0.2851,
      "step": 7275
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.851372132087025,
      "learning_rate": 1.4735395620980636e-06,
      "loss": 0.4076,
      "step": 7276
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.325046843675544,
      "learning_rate": 1.472884395108346e-06,
      "loss": 0.2887,
      "step": 7277
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.8694908343633467,
      "learning_rate": 1.4722293129738843e-06,
      "loss": 0.184,
      "step": 7278
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.864003759023804,
      "learning_rate": 1.4715743157487989e-06,
      "loss": 0.2926,
      "step": 7279
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6909856349389207,
      "learning_rate": 1.4709194034872004e-06,
      "loss": 0.0306,
      "step": 7280
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.191373089774898,
      "learning_rate": 1.4702645762431958e-06,
      "loss": 0.382,
      "step": 7281
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.4793923912166305,
      "learning_rate": 1.469609834070883e-06,
      "loss": 0.2152,
      "step": 7282
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6997485101870415,
      "learning_rate": 1.468955177024354e-06,
      "loss": 0.1936,
      "step": 7283
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7006621784646674,
      "learning_rate": 1.4683006051576942e-06,
      "loss": 0.3112,
      "step": 7284
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.251099105116501,
      "learning_rate": 1.4676461185249795e-06,
      "loss": 0.3245,
      "step": 7285
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.68238493339201,
      "learning_rate": 1.466991717180281e-06,
      "loss": 0.2961,
      "step": 7286
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.210112914401284,
      "learning_rate": 1.4663374011776627e-06,
      "loss": 0.2936,
      "step": 7287
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.166796047027963,
      "learning_rate": 1.4656831705711803e-06,
      "loss": 0.4564,
      "step": 7288
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.172034238421294,
      "learning_rate": 1.465029025414884e-06,
      "loss": 0.2578,
      "step": 7289
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.949957260764938,
      "learning_rate": 1.4643749657628148e-06,
      "loss": 0.2062,
      "step": 7290
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.4297372424370396,
      "learning_rate": 1.463720991669008e-06,
      "loss": 0.291,
      "step": 7291
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7949805868164357,
      "learning_rate": 1.4630671031874933e-06,
      "loss": 0.1415,
      "step": 7292
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.910317224927353,
      "learning_rate": 1.4624133003722902e-06,
      "loss": 0.0564,
      "step": 7293
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.517767072656448,
      "learning_rate": 1.461759583277414e-06,
      "loss": 0.3886,
      "step": 7294
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.5235488917880184,
      "learning_rate": 1.4611059519568698e-06,
      "loss": 0.2136,
      "step": 7295
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.366413980620072,
      "learning_rate": 1.4604524064646585e-06,
      "loss": 0.1264,
      "step": 7296
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3061449524306954,
      "learning_rate": 1.4597989468547733e-06,
      "loss": 0.0659,
      "step": 7297
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.54894113531472,
      "learning_rate": 1.4591455731811994e-06,
      "loss": 0.2741,
      "step": 7298
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.6710169101738765,
      "learning_rate": 1.4584922854979147e-06,
      "loss": 0.3838,
      "step": 7299
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.656707558678178,
      "learning_rate": 1.4578390838588904e-06,
      "loss": 0.1529,
      "step": 7300
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.567579651825015,
      "learning_rate": 1.4571859683180928e-06,
      "loss": 0.3242,
      "step": 7301
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.421541444328973,
      "learning_rate": 1.4565329389294774e-06,
      "loss": 0.1973,
      "step": 7302
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.723797196253348,
      "learning_rate": 1.4558799957469953e-06,
      "loss": 0.3081,
      "step": 7303
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.7051153840007958,
      "learning_rate": 1.4552271388245887e-06,
      "loss": 0.0734,
      "step": 7304
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.922166140273065,
      "learning_rate": 1.4545743682161922e-06,
      "loss": 0.1929,
      "step": 7305
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.068061970869841,
      "learning_rate": 1.4539216839757374e-06,
      "loss": 0.5063,
      "step": 7306
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.063319934394519,
      "learning_rate": 1.4532690861571443e-06,
      "loss": 0.2977,
      "step": 7307
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.627171022088653,
      "learning_rate": 1.4526165748143276e-06,
      "loss": 0.3796,
      "step": 7308
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.5876878623163884,
      "learning_rate": 1.4519641500011943e-06,
      "loss": 0.1542,
      "step": 7309
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.121276161104931,
      "learning_rate": 1.4513118117716436e-06,
      "loss": 0.2673,
      "step": 7310
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.567272279240255,
      "learning_rate": 1.450659560179571e-06,
      "loss": 0.1904,
      "step": 7311
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.756887418927082,
      "learning_rate": 1.4500073952788603e-06,
      "loss": 0.1928,
      "step": 7312
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.416837872165663,
      "learning_rate": 1.449355317123391e-06,
      "loss": 0.1416,
      "step": 7313
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.124308565677649,
      "learning_rate": 1.4487033257670346e-06,
      "loss": 0.3004,
      "step": 7314
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.274675947286493,
      "learning_rate": 1.448051421263654e-06,
      "loss": 0.2778,
      "step": 7315
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.5405119317556504,
      "learning_rate": 1.4473996036671084e-06,
      "loss": 0.2973,
      "step": 7316
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.290813382262295,
      "learning_rate": 1.4467478730312472e-06,
      "loss": 0.4075,
      "step": 7317
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.5332312320175916,
      "learning_rate": 1.446096229409913e-06,
      "loss": 0.2813,
      "step": 7318
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.4763982624410414,
      "learning_rate": 1.445444672856941e-06,
      "loss": 0.0897,
      "step": 7319
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.667326795711185,
      "learning_rate": 1.444793203426159e-06,
      "loss": 0.3409,
      "step": 7320
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6386054800576435,
      "learning_rate": 1.44414182117139e-06,
      "loss": 0.3035,
      "step": 7321
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.057877282377474,
      "learning_rate": 1.4434905261464475e-06,
      "loss": 0.3833,
      "step": 7322
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.462849979711901,
      "learning_rate": 1.4428393184051376e-06,
      "loss": 0.244,
      "step": 7323
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.110513769755045,
      "learning_rate": 1.4421881980012594e-06,
      "loss": 0.1979,
      "step": 7324
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.677058962867585,
      "learning_rate": 1.4415371649886073e-06,
      "loss": 0.3793,
      "step": 7325
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.134493054391963,
      "learning_rate": 1.440886219420965e-06,
      "loss": 0.3088,
      "step": 7326
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.314871332534735,
      "learning_rate": 1.4402353613521106e-06,
      "loss": 0.2637,
      "step": 7327
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.336645616272853,
      "learning_rate": 1.4395845908358152e-06,
      "loss": 0.306,
      "step": 7328
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.292610672569607,
      "learning_rate": 1.4389339079258409e-06,
      "loss": 0.2951,
      "step": 7329
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.847559477195154,
      "learning_rate": 1.4382833126759457e-06,
      "loss": 0.3253,
      "step": 7330
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.258407905814321,
      "learning_rate": 1.437632805139878e-06,
      "loss": 0.2105,
      "step": 7331
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.031070199375614,
      "learning_rate": 1.4369823853713796e-06,
      "loss": 0.4207,
      "step": 7332
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.255147250021003,
      "learning_rate": 1.4363320534241848e-06,
      "loss": 0.3629,
      "step": 7333
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.237339772388423,
      "learning_rate": 1.4356818093520197e-06,
      "loss": 0.4598,
      "step": 7334
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6415163058677762,
      "learning_rate": 1.4350316532086066e-06,
      "loss": 0.0569,
      "step": 7335
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.8226182166336145,
      "learning_rate": 1.4343815850476572e-06,
      "loss": 0.4609,
      "step": 7336
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6161189091471817,
      "learning_rate": 1.4337316049228767e-06,
      "loss": 0.2698,
      "step": 7337
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.05654825565563,
      "learning_rate": 1.433081712887963e-06,
      "loss": 0.2137,
      "step": 7338
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.729374858789206,
      "learning_rate": 1.4324319089966069e-06,
      "loss": 0.261,
      "step": 7339
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.6732243276175187,
      "learning_rate": 1.4317821933024933e-06,
      "loss": 0.2809,
      "step": 7340
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.502712066481128,
      "learning_rate": 1.4311325658592976e-06,
      "loss": 0.3036,
      "step": 7341
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7193140621475234,
      "learning_rate": 1.4304830267206888e-06,
      "loss": 0.2296,
      "step": 7342
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.576345599408191,
      "learning_rate": 1.429833575940329e-06,
      "loss": 0.2555,
      "step": 7343
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.275623682488525,
      "learning_rate": 1.4291842135718714e-06,
      "loss": 0.4784,
      "step": 7344
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.3722172799553647,
      "learning_rate": 1.4285349396689652e-06,
      "loss": 0.1449,
      "step": 7345
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.019676047329193,
      "learning_rate": 1.427885754285249e-06,
      "loss": 0.1847,
      "step": 7346
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.2267925986635317,
      "learning_rate": 1.4272366574743558e-06,
      "loss": 0.2332,
      "step": 7347
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1316503813087535,
      "learning_rate": 1.42658764928991e-06,
      "loss": 0.1709,
      "step": 7348
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.13294634394409,
      "learning_rate": 1.4259387297855292e-06,
      "loss": 0.0787,
      "step": 7349
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.099931024858895,
      "learning_rate": 1.4252898990148256e-06,
      "loss": 0.3263,
      "step": 7350
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.396113201638249,
      "learning_rate": 1.4246411570314016e-06,
      "loss": 0.1277,
      "step": 7351
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.07032048680945,
      "learning_rate": 1.423992503888853e-06,
      "loss": 0.0158,
      "step": 7352
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.837978422301225,
      "learning_rate": 1.4233439396407684e-06,
      "loss": 0.2521,
      "step": 7353
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.7096979300470694,
      "learning_rate": 1.4226954643407281e-06,
      "loss": 0.1638,
      "step": 7354
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.2436843817842096,
      "learning_rate": 1.4220470780423074e-06,
      "loss": 0.1927,
      "step": 7355
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.057595659725958,
      "learning_rate": 1.4213987807990725e-06,
      "loss": 0.4201,
      "step": 7356
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.661837790972213,
      "learning_rate": 1.4207505726645821e-06,
      "loss": 0.4909,
      "step": 7357
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.2361053634113395,
      "learning_rate": 1.4201024536923883e-06,
      "loss": 0.5888,
      "step": 7358
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.3941829880599945,
      "learning_rate": 1.4194544239360342e-06,
      "loss": 0.2605,
      "step": 7359
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.9474738678821266,
      "learning_rate": 1.418806483449059e-06,
      "loss": 0.1872,
      "step": 7360
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.571606512542491,
      "learning_rate": 1.4181586322849911e-06,
      "loss": 0.2685,
      "step": 7361
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.613493157251615,
      "learning_rate": 1.4175108704973534e-06,
      "loss": 0.42,
      "step": 7362
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.192221520784077,
      "learning_rate": 1.4168631981396601e-06,
      "loss": 0.3415,
      "step": 7363
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.4985899134095195,
      "learning_rate": 1.4162156152654183e-06,
      "loss": 0.276,
      "step": 7364
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.06514372022691,
      "learning_rate": 1.4155681219281297e-06,
      "loss": 0.2602,
      "step": 7365
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.538165072071313,
      "learning_rate": 1.4149207181812862e-06,
      "loss": 0.3306,
      "step": 7366
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.499079174928037,
      "learning_rate": 1.414273404078372e-06,
      "loss": 0.2439,
      "step": 7367
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.070597914966464,
      "learning_rate": 1.4136261796728685e-06,
      "loss": 0.5721,
      "step": 7368
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.531246378189544,
      "learning_rate": 1.4129790450182412e-06,
      "loss": 0.3062,
      "step": 7369
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.1052856048107578,
      "learning_rate": 1.4123320001679574e-06,
      "loss": 0.1989,
      "step": 7370
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.385814435625212,
      "learning_rate": 1.411685045175471e-06,
      "loss": 0.1209,
      "step": 7371
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.302519783406586,
      "learning_rate": 1.4110381800942294e-06,
      "loss": 0.2286,
      "step": 7372
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.004508781027198,
      "learning_rate": 1.4103914049776766e-06,
      "loss": 0.1559,
      "step": 7373
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.9393633494813445,
      "learning_rate": 1.4097447198792415e-06,
      "loss": 0.3923,
      "step": 7374
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.089595581794872,
      "learning_rate": 1.409098124852354e-06,
      "loss": 0.3701,
      "step": 7375
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.622641376997784,
      "learning_rate": 1.4084516199504306e-06,
      "loss": 0.2699,
      "step": 7376
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.298083873234272,
      "learning_rate": 1.4078052052268818e-06,
      "loss": 0.3197,
      "step": 7377
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.614818512419082,
      "learning_rate": 1.4071588807351145e-06,
      "loss": 0.2754,
      "step": 7378
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.314685580776811,
      "learning_rate": 1.4065126465285201e-06,
      "loss": 0.1814,
      "step": 7379
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.321344512229392,
      "learning_rate": 1.4058665026604909e-06,
      "loss": 0.2799,
      "step": 7380
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.968925070776961,
      "learning_rate": 1.4052204491844074e-06,
      "loss": 0.3318,
      "step": 7381
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.798367306531198,
      "learning_rate": 1.4045744861536415e-06,
      "loss": 0.321,
      "step": 7382
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.046511714239524,
      "learning_rate": 1.4039286136215629e-06,
      "loss": 0.4911,
      "step": 7383
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.785684752338274,
      "learning_rate": 1.403282831641527e-06,
      "loss": 0.3294,
      "step": 7384
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.85949721830636,
      "learning_rate": 1.4026371402668873e-06,
      "loss": 0.2269,
      "step": 7385
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0293199196229224,
      "learning_rate": 1.401991539550987e-06,
      "loss": 0.1386,
      "step": 7386
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.697099764183324,
      "learning_rate": 1.4013460295471615e-06,
      "loss": 0.3173,
      "step": 7387
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.033815067062685,
      "learning_rate": 1.4007006103087427e-06,
      "loss": 0.1753,
      "step": 7388
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.8449616173153247,
      "learning_rate": 1.4000552818890477e-06,
      "loss": 0.2016,
      "step": 7389
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9951575420416767,
      "learning_rate": 1.3994100443413938e-06,
      "loss": 0.3289,
      "step": 7390
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.210541169839444,
      "learning_rate": 1.3987648977190862e-06,
      "loss": 0.3056,
      "step": 7391
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.8560903188717655,
      "learning_rate": 1.3981198420754228e-06,
      "loss": 0.3838,
      "step": 7392
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.887215014388528,
      "learning_rate": 1.3974748774636974e-06,
      "loss": 0.4594,
      "step": 7393
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.395261129615907,
      "learning_rate": 1.39683000393719e-06,
      "loss": 0.257,
      "step": 7394
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.797708273719251,
      "learning_rate": 1.396185221549181e-06,
      "loss": 0.3033,
      "step": 7395
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.269653433652045,
      "learning_rate": 1.3955405303529367e-06,
      "loss": 0.2915,
      "step": 7396
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.199035923513829,
      "learning_rate": 1.394895930401718e-06,
      "loss": 0.361,
      "step": 7397
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.542090603009918,
      "learning_rate": 1.3942514217487816e-06,
      "loss": 0.3304,
      "step": 7398
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.446202796100324,
      "learning_rate": 1.3936070044473692e-06,
      "loss": 0.2702,
      "step": 7399
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.5066746263376585,
      "learning_rate": 1.3929626785507232e-06,
      "loss": 0.2893,
      "step": 7400
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.4725494256287845,
      "learning_rate": 1.392318444112073e-06,
      "loss": 0.1876,
      "step": 7401
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.6931739590078942,
      "learning_rate": 1.3916743011846411e-06,
      "loss": 0.2046,
      "step": 7402
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0520884248988183,
      "learning_rate": 1.3910302498216468e-06,
      "loss": 0.1213,
      "step": 7403
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.249610545931633,
      "learning_rate": 1.3903862900762943e-06,
      "loss": 0.373,
      "step": 7404
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.829084747792312,
      "learning_rate": 1.3897424220017874e-06,
      "loss": 0.349,
      "step": 7405
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.634916291149012,
      "learning_rate": 1.3890986456513184e-06,
      "loss": 0.2437,
      "step": 7406
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.310349971814889,
      "learning_rate": 1.3884549610780716e-06,
      "loss": 0.1582,
      "step": 7407
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.315778003348037,
      "learning_rate": 1.3878113683352281e-06,
      "loss": 0.3231,
      "step": 7408
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.873425756944887,
      "learning_rate": 1.3871678674759548e-06,
      "loss": 0.3847,
      "step": 7409
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.823544393311974,
      "learning_rate": 1.3865244585534166e-06,
      "loss": 0.3843,
      "step": 7410
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.432782165913708,
      "learning_rate": 1.3858811416207701e-06,
      "loss": 0.4485,
      "step": 7411
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.5908998759779776,
      "learning_rate": 1.3852379167311597e-06,
      "loss": 0.1231,
      "step": 7412
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.455824925783109,
      "learning_rate": 1.3845947839377293e-06,
      "loss": 0.2399,
      "step": 7413
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.359383469277152,
      "learning_rate": 1.3839517432936072e-06,
      "loss": 0.3014,
      "step": 7414
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.522161176459299,
      "learning_rate": 1.3833087948519208e-06,
      "loss": 0.2209,
      "step": 7415
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.672287758347547,
      "learning_rate": 1.3826659386657885e-06,
      "loss": 0.248,
      "step": 7416
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.40314650505306,
      "learning_rate": 1.3820231747883163e-06,
      "loss": 0.1436,
      "step": 7417
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.196501629191016,
      "learning_rate": 1.3813805032726103e-06,
      "loss": 0.3629,
      "step": 7418
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.671394777185713,
      "learning_rate": 1.380737924171761e-06,
      "loss": 0.2923,
      "step": 7419
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.594874927343522,
      "learning_rate": 1.380095437538857e-06,
      "loss": 0.2803,
      "step": 7420
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.31841161002258,
      "learning_rate": 1.379453043426979e-06,
      "loss": 0.3746,
      "step": 7421
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.727232285096882,
      "learning_rate": 1.378810741889195e-06,
      "loss": 0.2877,
      "step": 7422
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.570969281286191,
      "learning_rate": 1.3781685329785727e-06,
      "loss": 0.2421,
      "step": 7423
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.9921805864101754,
      "learning_rate": 1.377526416748164e-06,
      "loss": 0.1895,
      "step": 7424
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.6357968884241036,
      "learning_rate": 1.3768843932510196e-06,
      "loss": 0.1497,
      "step": 7425
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.236146438010935,
      "learning_rate": 1.3762424625401825e-06,
      "loss": 0.3414,
      "step": 7426
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9886546818295994,
      "learning_rate": 1.3756006246686814e-06,
      "loss": 0.3633,
      "step": 7427
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.204915879819233,
      "learning_rate": 1.3749588796895464e-06,
      "loss": 0.2188,
      "step": 7428
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.164440243146469,
      "learning_rate": 1.3743172276557903e-06,
      "loss": 0.3391,
      "step": 7429
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.118667923195348,
      "learning_rate": 1.3736756686204268e-06,
      "loss": 0.5891,
      "step": 7430
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9228492619715474,
      "learning_rate": 1.3730342026364587e-06,
      "loss": 0.3487,
      "step": 7431
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.5236022366028683,
      "learning_rate": 1.3723928297568778e-06,
      "loss": 0.1353,
      "step": 7432
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.7563455590663444,
      "learning_rate": 1.3717515500346751e-06,
      "loss": 0.1493,
      "step": 7433
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.329761277159218,
      "learning_rate": 1.3711103635228259e-06,
      "loss": 0.2942,
      "step": 7434
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.175177084302364,
      "learning_rate": 1.370469270274304e-06,
      "loss": 0.0946,
      "step": 7435
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.257784294075334,
      "learning_rate": 1.3698282703420745e-06,
      "loss": 0.1735,
      "step": 7436
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.698661069972636,
      "learning_rate": 1.369187363779091e-06,
      "loss": 0.2954,
      "step": 7437
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.509529034597589,
      "learning_rate": 1.3685465506383041e-06,
      "loss": 0.3027,
      "step": 7438
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.853565756936239,
      "learning_rate": 1.367905830972654e-06,
      "loss": 0.4671,
      "step": 7439
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.077481423749248,
      "learning_rate": 1.3672652048350728e-06,
      "loss": 0.3198,
      "step": 7440
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1129089702318877,
      "learning_rate": 1.366624672278489e-06,
      "loss": 0.2266,
      "step": 7441
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.538040825014489,
      "learning_rate": 1.3659842333558159e-06,
      "loss": 0.1915,
      "step": 7442
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.85980789154958,
      "learning_rate": 1.365343888119967e-06,
      "loss": 0.5373,
      "step": 7443
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.6090276633439804,
      "learning_rate": 1.364703636623843e-06,
      "loss": 0.2752,
      "step": 7444
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.158850983233198,
      "learning_rate": 1.3640634789203374e-06,
      "loss": 0.3049,
      "step": 7445
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.013252558642174,
      "learning_rate": 1.3634234150623398e-06,
      "loss": 0.3129,
      "step": 7446
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.57128444508513,
      "learning_rate": 1.362783445102726e-06,
      "loss": 0.2995,
      "step": 7447
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.012691454495966,
      "learning_rate": 1.362143569094369e-06,
      "loss": 0.39,
      "step": 7448
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.898010820495937,
      "learning_rate": 1.3615037870901316e-06,
      "loss": 0.2395,
      "step": 7449
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.857594594024495,
      "learning_rate": 1.3608640991428687e-06,
      "loss": 0.2076,
      "step": 7450
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.607818671833498,
      "learning_rate": 1.3602245053054312e-06,
      "loss": 0.4371,
      "step": 7451
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.899109082427215,
      "learning_rate": 1.3595850056306548e-06,
      "loss": 0.1708,
      "step": 7452
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.315817347038252,
      "learning_rate": 1.3589456001713751e-06,
      "loss": 0.3725,
      "step": 7453
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.648923333044489,
      "learning_rate": 1.3583062889804158e-06,
      "loss": 0.2523,
      "step": 7454
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.447589112217323,
      "learning_rate": 1.3576670721105921e-06,
      "loss": 0.2678,
      "step": 7455
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.0445041277967144,
      "learning_rate": 1.3570279496147172e-06,
      "loss": 0.2174,
      "step": 7456
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2466496598812067,
      "learning_rate": 1.3563889215455866e-06,
      "loss": 0.0172,
      "step": 7457
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.510233765176803,
      "learning_rate": 1.3557499879559978e-06,
      "loss": 0.359,
      "step": 7458
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.32730580253397,
      "learning_rate": 1.3551111488987351e-06,
      "loss": 0.417,
      "step": 7459
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.2783246201740295,
      "learning_rate": 1.354472404426575e-06,
      "loss": 0.1289,
      "step": 7460
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.174502118203007,
      "learning_rate": 1.3538337545922907e-06,
      "loss": 0.3597,
      "step": 7461
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8768728539638422,
      "learning_rate": 1.3531951994486404e-06,
      "loss": 0.084,
      "step": 7462
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.442125098570098,
      "learning_rate": 1.3525567390483807e-06,
      "loss": 0.3581,
      "step": 7463
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.1310072372057567,
      "learning_rate": 1.3519183734442582e-06,
      "loss": 0.2223,
      "step": 7464
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.8177790730532974,
      "learning_rate": 1.3512801026890095e-06,
      "loss": 0.369,
      "step": 7465
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.5795526737252583,
      "learning_rate": 1.3506419268353694e-06,
      "loss": 0.2617,
      "step": 7466
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3065948388111432,
      "learning_rate": 1.350003845936056e-06,
      "loss": 0.2678,
      "step": 7467
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.2501017576796984,
      "learning_rate": 1.3493658600437876e-06,
      "loss": 0.198,
      "step": 7468
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.937838159683469,
      "learning_rate": 1.3487279692112706e-06,
      "loss": 0.1547,
      "step": 7469
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.316382122017274,
      "learning_rate": 1.3480901734912034e-06,
      "loss": 0.3207,
      "step": 7470
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.368537801409636,
      "learning_rate": 1.347452472936281e-06,
      "loss": 0.1434,
      "step": 7471
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.262802395807379,
      "learning_rate": 1.3468148675991822e-06,
      "loss": 0.3574,
      "step": 7472
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.298286324790434,
      "learning_rate": 1.3461773575325862e-06,
      "loss": 0.3008,
      "step": 7473
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.36723081313859,
      "learning_rate": 1.3455399427891602e-06,
      "loss": 0.2339,
      "step": 7474
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.010424236898206,
      "learning_rate": 1.3449026234215631e-06,
      "loss": 0.3306,
      "step": 7475
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.028316919600596,
      "learning_rate": 1.34426539948245e-06,
      "loss": 0.331,
      "step": 7476
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9102576243286817,
      "learning_rate": 1.3436282710244616e-06,
      "loss": 0.3939,
      "step": 7477
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.4385422838954156,
      "learning_rate": 1.342991238100237e-06,
      "loss": 0.1849,
      "step": 7478
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.055832125062999,
      "learning_rate": 1.3423543007624039e-06,
      "loss": 0.2306,
      "step": 7479
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1758610578098203,
      "learning_rate": 1.3417174590635814e-06,
      "loss": 0.1005,
      "step": 7480
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.726418624935696,
      "learning_rate": 1.3410807130563861e-06,
      "loss": 0.4343,
      "step": 7481
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.852141523685861,
      "learning_rate": 1.3404440627934182e-06,
      "loss": 0.4211,
      "step": 7482
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.8399042911865826,
      "learning_rate": 1.339807508327278e-06,
      "loss": 0.3134,
      "step": 7483
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9646610226551453,
      "learning_rate": 1.3391710497105532e-06,
      "loss": 0.3437,
      "step": 7484
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.999278665393275,
      "learning_rate": 1.338534686995824e-06,
      "loss": 0.6246,
      "step": 7485
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.735681945562453,
      "learning_rate": 1.3378984202356657e-06,
      "loss": 0.3745,
      "step": 7486
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.098661993055096,
      "learning_rate": 1.3372622494826426e-06,
      "loss": 0.2977,
      "step": 7487
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.632782333954424,
      "learning_rate": 1.3366261747893116e-06,
      "loss": 0.3785,
      "step": 7488
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7736213322167091,
      "learning_rate": 1.3359901962082223e-06,
      "loss": 0.0606,
      "step": 7489
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.223634043306002,
      "learning_rate": 1.3353543137919154e-06,
      "loss": 0.3027,
      "step": 7490
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.764565756788592,
      "learning_rate": 1.334718527592926e-06,
      "loss": 0.2571,
      "step": 7491
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.2879704853188745,
      "learning_rate": 1.3340828376637788e-06,
      "loss": 0.3443,
      "step": 7492
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.27608812434371,
      "learning_rate": 1.3334472440569912e-06,
      "loss": 0.3098,
      "step": 7493
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.4183027018272485,
      "learning_rate": 1.3328117468250732e-06,
      "loss": 0.2308,
      "step": 7494
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.187798654335593,
      "learning_rate": 1.332176346020525e-06,
      "loss": 0.2221,
      "step": 7495
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.481836113195422,
      "learning_rate": 1.331541041695843e-06,
      "loss": 0.4517,
      "step": 7496
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.48042169757092,
      "learning_rate": 1.3309058339035113e-06,
      "loss": 0.5078,
      "step": 7497
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.714331963921323,
      "learning_rate": 1.3302707226960077e-06,
      "loss": 0.0807,
      "step": 7498
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.213794187398138,
      "learning_rate": 1.329635708125802e-06,
      "loss": 0.477,
      "step": 7499
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.04981707990332,
      "learning_rate": 1.3290007902453556e-06,
      "loss": 0.2701,
      "step": 7500
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.630038619539009,
      "learning_rate": 1.3283659691071236e-06,
      "loss": 0.1978,
      "step": 7501
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.903535216151279,
      "learning_rate": 1.327731244763551e-06,
      "loss": 0.3269,
      "step": 7502
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.475634222983314,
      "learning_rate": 1.3270966172670755e-06,
      "loss": 0.2825,
      "step": 7503
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.5519654435755608,
      "learning_rate": 1.3264620866701272e-06,
      "loss": 0.2497,
      "step": 7504
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.3007910922102988,
      "learning_rate": 1.3258276530251265e-06,
      "loss": 0.2962,
      "step": 7505
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.257078611593345,
      "learning_rate": 1.32519331638449e-06,
      "loss": 0.2347,
      "step": 7506
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.464345248260924,
      "learning_rate": 1.324559076800621e-06,
      "loss": 0.2798,
      "step": 7507
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.504690689179753,
      "learning_rate": 1.3239249343259192e-06,
      "loss": 0.283,
      "step": 7508
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.4456666982455784,
      "learning_rate": 1.3232908890127726e-06,
      "loss": 0.1942,
      "step": 7509
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.896776624319183,
      "learning_rate": 1.322656940913563e-06,
      "loss": 0.2884,
      "step": 7510
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.867152542349468,
      "learning_rate": 1.3220230900806658e-06,
      "loss": 0.3005,
      "step": 7511
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.085666999190325,
      "learning_rate": 1.3213893365664455e-06,
      "loss": 0.261,
      "step": 7512
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.346387347387228,
      "learning_rate": 1.32075568042326e-06,
      "loss": 0.3626,
      "step": 7513
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.187159718485027,
      "learning_rate": 1.3201221217034583e-06,
      "loss": 0.2831,
      "step": 7514
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.817414361486134,
      "learning_rate": 1.3194886604593815e-06,
      "loss": 0.2874,
      "step": 7515
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.578798181550998,
      "learning_rate": 1.3188552967433648e-06,
      "loss": 0.4448,
      "step": 7516
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.8969927973282443,
      "learning_rate": 1.3182220306077326e-06,
      "loss": 0.1235,
      "step": 7517
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.374774987315696,
      "learning_rate": 1.3175888621048023e-06,
      "loss": 0.4768,
      "step": 7518
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.339937951015301,
      "learning_rate": 1.316955791286883e-06,
      "loss": 0.3024,
      "step": 7519
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.036026543108772,
      "learning_rate": 1.3163228182062754e-06,
      "loss": 0.3322,
      "step": 7520
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.736786069234484,
      "learning_rate": 1.3156899429152743e-06,
      "loss": 0.3032,
      "step": 7521
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.9547815161279,
      "learning_rate": 1.3150571654661632e-06,
      "loss": 0.1359,
      "step": 7522
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.730155862274452,
      "learning_rate": 1.31442448591122e-06,
      "loss": 0.2238,
      "step": 7523
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.6190943885200872,
      "learning_rate": 1.3137919043027134e-06,
      "loss": 0.2159,
      "step": 7524
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.0106678072340864,
      "learning_rate": 1.3131594206929028e-06,
      "loss": 0.1905,
      "step": 7525
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8725831889005686,
      "learning_rate": 1.312527035134043e-06,
      "loss": 0.1557,
      "step": 7526
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.594611732747136,
      "learning_rate": 1.3118947476783777e-06,
      "loss": 0.2046,
      "step": 7527
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6622832873257963,
      "learning_rate": 1.3112625583781436e-06,
      "loss": 0.0054,
      "step": 7528
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.175684239544794,
      "learning_rate": 1.310630467285568e-06,
      "loss": 0.2038,
      "step": 7529
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.694130494234455,
      "learning_rate": 1.3099984744528727e-06,
      "loss": 0.2241,
      "step": 7530
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.337028902648448,
      "learning_rate": 1.3093665799322691e-06,
      "loss": 0.3384,
      "step": 7531
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.823233468869728,
      "learning_rate": 1.3087347837759617e-06,
      "loss": 0.4066,
      "step": 7532
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.612667078012353,
      "learning_rate": 1.3081030860361456e-06,
      "loss": 0.289,
      "step": 7533
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.158961006353022,
      "learning_rate": 1.3074714867650083e-06,
      "loss": 0.4428,
      "step": 7534
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.9643441415748693,
      "learning_rate": 1.3068399860147307e-06,
      "loss": 0.2288,
      "step": 7535
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.185599550823278,
      "learning_rate": 1.3062085838374841e-06,
      "loss": 0.4435,
      "step": 7536
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.270198195743524,
      "learning_rate": 1.3055772802854313e-06,
      "loss": 0.5011,
      "step": 7537
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.460248074290824,
      "learning_rate": 1.304946075410728e-06,
      "loss": 0.3605,
      "step": 7538
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.412230351877195,
      "learning_rate": 1.3043149692655193e-06,
      "loss": 0.2921,
      "step": 7539
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.821677064162605,
      "learning_rate": 1.303683961901947e-06,
      "loss": 0.45,
      "step": 7540
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.068750140826197,
      "learning_rate": 1.3030530533721409e-06,
      "loss": 0.1586,
      "step": 7541
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.4002901097773677,
      "learning_rate": 1.3024222437282225e-06,
      "loss": 0.2613,
      "step": 7542
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.258779816396369,
      "learning_rate": 1.3017915330223075e-06,
      "loss": 0.2338,
      "step": 7543
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.786331804776588,
      "learning_rate": 1.3011609213065e-06,
      "loss": 0.1596,
      "step": 7544
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.197612559384783,
      "learning_rate": 1.3005304086329012e-06,
      "loss": 0.1898,
      "step": 7545
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.924298375569025,
      "learning_rate": 1.2998999950535989e-06,
      "loss": 0.17,
      "step": 7546
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.9755512923551497,
      "learning_rate": 1.2992696806206756e-06,
      "loss": 0.1963,
      "step": 7547
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.244154998454428,
      "learning_rate": 1.298639465386204e-06,
      "loss": 0.2726,
      "step": 7548
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.0026812541513355,
      "learning_rate": 1.2980093494022494e-06,
      "loss": 0.3163,
      "step": 7549
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.007027559781326,
      "learning_rate": 1.29737933272087e-06,
      "loss": 0.3059,
      "step": 7550
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.992999370682765,
      "learning_rate": 1.2967494153941139e-06,
      "loss": 0.2673,
      "step": 7551
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.363948048229719,
      "learning_rate": 1.2961195974740217e-06,
      "loss": 0.667,
      "step": 7552
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.0629834635306175,
      "learning_rate": 1.2954898790126265e-06,
      "loss": 0.4224,
      "step": 7553
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.061077091688082,
      "learning_rate": 1.2948602600619509e-06,
      "loss": 0.3386,
      "step": 7554
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.255859938160769,
      "learning_rate": 1.2942307406740129e-06,
      "loss": 0.4199,
      "step": 7555
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.529575727552058,
      "learning_rate": 1.2936013209008199e-06,
      "loss": 0.2436,
      "step": 7556
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.3687382750659065,
      "learning_rate": 1.2929720007943708e-06,
      "loss": 0.3942,
      "step": 7557
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.988914465390247,
      "learning_rate": 1.2923427804066574e-06,
      "loss": 0.2627,
      "step": 7558
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8398448107075898,
      "learning_rate": 1.2917136597896619e-06,
      "loss": 0.2664,
      "step": 7559
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8793723639467217,
      "learning_rate": 1.2910846389953607e-06,
      "loss": 0.1794,
      "step": 7560
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.295046680973579,
      "learning_rate": 1.2904557180757197e-06,
      "loss": 0.09,
      "step": 7561
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.798803332501388,
      "learning_rate": 1.289826897082697e-06,
      "loss": 0.1886,
      "step": 7562
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.63495428206997,
      "learning_rate": 1.2891981760682431e-06,
      "loss": 0.1637,
      "step": 7563
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.107423759683454,
      "learning_rate": 1.2885695550842987e-06,
      "loss": 0.2615,
      "step": 7564
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.2594490040465214,
      "learning_rate": 1.2879410341827992e-06,
      "loss": 0.0307,
      "step": 7565
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.219528467994612,
      "learning_rate": 1.2873126134156692e-06,
      "loss": 0.2697,
      "step": 7566
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.923882721834003,
      "learning_rate": 1.2866842928348245e-06,
      "loss": 0.2263,
      "step": 7567
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.6663686418247816,
      "learning_rate": 1.2860560724921767e-06,
      "loss": 0.094,
      "step": 7568
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.163263304176852,
      "learning_rate": 1.285427952439623e-06,
      "loss": 0.3124,
      "step": 7569
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.592388384999653,
      "learning_rate": 1.2847999327290577e-06,
      "loss": 0.2823,
      "step": 7570
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.7607442067381696,
      "learning_rate": 1.2841720134123644e-06,
      "loss": 0.1363,
      "step": 7571
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.223686225634922,
      "learning_rate": 1.2835441945414174e-06,
      "loss": 0.3682,
      "step": 7572
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.8316952709997425,
      "learning_rate": 1.2829164761680874e-06,
      "loss": 0.4002,
      "step": 7573
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.266799191294127,
      "learning_rate": 1.2822888583442287e-06,
      "loss": 0.3348,
      "step": 7574
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.5724422552795385,
      "learning_rate": 1.2816613411216956e-06,
      "loss": 0.2408,
      "step": 7575
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.413792587799889,
      "learning_rate": 1.281033924552329e-06,
      "loss": 0.4855,
      "step": 7576
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.6652528561875246,
      "learning_rate": 1.2804066086879626e-06,
      "loss": 0.3225,
      "step": 7577
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.640811355056561,
      "learning_rate": 1.2797793935804248e-06,
      "loss": 0.4481,
      "step": 7578
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.609776230950216,
      "learning_rate": 1.2791522792815292e-06,
      "loss": 0.3005,
      "step": 7579
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.34582548597649,
      "learning_rate": 1.2785252658430874e-06,
      "loss": 0.2837,
      "step": 7580
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.99547356915834,
      "learning_rate": 1.2778983533168997e-06,
      "loss": 0.3005,
      "step": 7581
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.156039523850937,
      "learning_rate": 1.2772715417547576e-06,
      "loss": 0.3912,
      "step": 7582
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.402154859135549,
      "learning_rate": 1.276644831208448e-06,
      "loss": 0.1941,
      "step": 7583
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.0603149042364777,
      "learning_rate": 1.2760182217297427e-06,
      "loss": 0.1552,
      "step": 7584
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.773553371725954,
      "learning_rate": 1.275391713370412e-06,
      "loss": 0.2749,
      "step": 7585
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9036197751899149,
      "learning_rate": 1.2747653061822137e-06,
      "loss": 0.0807,
      "step": 7586
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.057585724026372,
      "learning_rate": 1.2741390002168983e-06,
      "loss": 0.3728,
      "step": 7587
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.572151333064954,
      "learning_rate": 1.2735127955262102e-06,
      "loss": 0.2668,
      "step": 7588
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.9467622533924605,
      "learning_rate": 1.2728866921618798e-06,
      "loss": 0.3607,
      "step": 7589
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.8901772936317713,
      "learning_rate": 1.2722606901756363e-06,
      "loss": 0.3191,
      "step": 7590
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.141989292554304,
      "learning_rate": 1.2716347896191948e-06,
      "loss": 0.3634,
      "step": 7591
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.138400999022347,
      "learning_rate": 1.2710089905442635e-06,
      "loss": 0.0989,
      "step": 7592
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.566433507303103,
      "learning_rate": 1.2703832930025462e-06,
      "loss": 0.2006,
      "step": 7593
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.030386469655825,
      "learning_rate": 1.2697576970457307e-06,
      "loss": 0.2898,
      "step": 7594
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.621783681530027,
      "learning_rate": 1.2691322027255038e-06,
      "loss": 0.2522,
      "step": 7595
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.051611669689462,
      "learning_rate": 1.2685068100935395e-06,
      "loss": 0.4715,
      "step": 7596
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.6271038504020967,
      "learning_rate": 1.2678815192015042e-06,
      "loss": 0.2396,
      "step": 7597
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.669705323981766,
      "learning_rate": 1.2672563301010588e-06,
      "loss": 0.488,
      "step": 7598
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5642202855401381,
      "learning_rate": 1.2666312428438493e-06,
      "loss": 0.0213,
      "step": 7599
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.483865383052613,
      "learning_rate": 1.2660062574815213e-06,
      "loss": 0.0833,
      "step": 7600
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.946531662280304,
      "learning_rate": 1.2653813740657056e-06,
      "loss": 0.1664,
      "step": 7601
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.146027130624078,
      "learning_rate": 1.2647565926480275e-06,
      "loss": 0.322,
      "step": 7602
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.291638556415386,
      "learning_rate": 1.264131913280105e-06,
      "loss": 0.2929,
      "step": 7603
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.090701397157037,
      "learning_rate": 1.263507336013543e-06,
      "loss": 0.4618,
      "step": 7604
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.7054045254153625,
      "learning_rate": 1.2628828608999438e-06,
      "loss": 0.315,
      "step": 7605
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.2917529433476855,
      "learning_rate": 1.2622584879908973e-06,
      "loss": 0.2606,
      "step": 7606
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0632564110759657,
      "learning_rate": 1.2616342173379853e-06,
      "loss": 0.122,
      "step": 7607
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.736102246548025,
      "learning_rate": 1.2610100489927846e-06,
      "loss": 0.0978,
      "step": 7608
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3582256076699895,
      "learning_rate": 1.2603859830068577e-06,
      "loss": 0.2592,
      "step": 7609
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4195341063125966,
      "learning_rate": 1.259762019431764e-06,
      "loss": 0.2081,
      "step": 7610
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.168380016516262,
      "learning_rate": 1.2591381583190516e-06,
      "loss": 0.4678,
      "step": 7611
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6129537770420748,
      "learning_rate": 1.25851439972026e-06,
      "loss": 0.2618,
      "step": 7612
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.736743865076716,
      "learning_rate": 1.2578907436869243e-06,
      "loss": 0.3682,
      "step": 7613
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.456659392758606,
      "learning_rate": 1.2572671902705635e-06,
      "loss": 0.2357,
      "step": 7614
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.0631071308555,
      "learning_rate": 1.2566437395226944e-06,
      "loss": 0.2406,
      "step": 7615
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.563157510734229,
      "learning_rate": 1.256020391494826e-06,
      "loss": 0.4351,
      "step": 7616
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3334566588310373,
      "learning_rate": 1.2553971462384517e-06,
      "loss": 0.2304,
      "step": 7617
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0564988735643892,
      "learning_rate": 1.2547740038050655e-06,
      "loss": 0.007,
      "step": 7618
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.061537576577597,
      "learning_rate": 1.254150964246144e-06,
      "loss": 0.3939,
      "step": 7619
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.748921084906081,
      "learning_rate": 1.2535280276131617e-06,
      "loss": 0.2721,
      "step": 7620
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.352580178800918,
      "learning_rate": 1.2529051939575845e-06,
      "loss": 0.4805,
      "step": 7621
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.208089421604111,
      "learning_rate": 1.252282463330864e-06,
      "loss": 0.0965,
      "step": 7622
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.780149681855687,
      "learning_rate": 1.2516598357844515e-06,
      "loss": 0.3486,
      "step": 7623
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.7677827373565105,
      "learning_rate": 1.2510373113697808e-06,
      "loss": 0.5438,
      "step": 7624
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.008751498810647,
      "learning_rate": 1.2504148901382842e-06,
      "loss": 0.2798,
      "step": 7625
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.931885252302045,
      "learning_rate": 1.249792572141385e-06,
      "loss": 0.293,
      "step": 7626
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7362700495853085,
      "learning_rate": 1.249170357430492e-06,
      "loss": 0.1752,
      "step": 7627
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.520209749977277,
      "learning_rate": 1.2485482460570137e-06,
      "loss": 0.409,
      "step": 7628
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.855291991584864,
      "learning_rate": 1.2479262380723415e-06,
      "loss": 0.3679,
      "step": 7629
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.740312955805126,
      "learning_rate": 1.247304333527865e-06,
      "loss": 0.3048,
      "step": 7630
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3246176890594783,
      "learning_rate": 1.246682532474965e-06,
      "loss": 0.1645,
      "step": 7631
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2811781099946264,
      "learning_rate": 1.2460608349650075e-06,
      "loss": 0.2215,
      "step": 7632
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.1296675527617235,
      "learning_rate": 1.245439241049358e-06,
      "loss": 0.183,
      "step": 7633
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.026273365168599,
      "learning_rate": 1.2448177507793655e-06,
      "loss": 0.4039,
      "step": 7634
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.8519119484277855,
      "learning_rate": 1.2441963642063766e-06,
      "loss": 0.5279,
      "step": 7635
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.254034769270647,
      "learning_rate": 1.2435750813817296e-06,
      "loss": 0.2272,
      "step": 7636
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.5680643904558575,
      "learning_rate": 1.2429539023567473e-06,
      "loss": 0.2943,
      "step": 7637
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.150553018265773,
      "learning_rate": 1.2423328271827524e-06,
      "loss": 0.6414,
      "step": 7638
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.92873678350529,
      "learning_rate": 1.2417118559110513e-06,
      "loss": 0.1106,
      "step": 7639
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.419958481033083,
      "learning_rate": 1.2410909885929478e-06,
      "loss": 0.2654,
      "step": 7640
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.502675425386429,
      "learning_rate": 1.2404702252797366e-06,
      "loss": 0.1838,
      "step": 7641
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.731405267319721,
      "learning_rate": 1.2398495660226984e-06,
      "loss": 0.2791,
      "step": 7642
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.475928000051033,
      "learning_rate": 1.2392290108731117e-06,
      "loss": 0.2836,
      "step": 7643
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.5191672746219114,
      "learning_rate": 1.2386085598822432e-06,
      "loss": 0.1081,
      "step": 7644
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.000541523390229,
      "learning_rate": 1.2379882131013498e-06,
      "loss": 0.2268,
      "step": 7645
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.8176679434164855,
      "learning_rate": 1.2373679705816851e-06,
      "loss": 0.4712,
      "step": 7646
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.711658073153657,
      "learning_rate": 1.2367478323744864e-06,
      "loss": 0.2957,
      "step": 7647
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.538664686808342,
      "learning_rate": 1.2361277985309892e-06,
      "loss": 0.1433,
      "step": 7648
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.213576798668401,
      "learning_rate": 1.235507869102417e-06,
      "loss": 0.2875,
      "step": 7649
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.780684649820035,
      "learning_rate": 1.2348880441399841e-06,
      "loss": 0.2214,
      "step": 7650
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.415703627118019,
      "learning_rate": 1.2342683236949005e-06,
      "loss": 0.2564,
      "step": 7651
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.328244199347334,
      "learning_rate": 1.2336487078183606e-06,
      "loss": 0.1752,
      "step": 7652
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.8617866004323425,
      "learning_rate": 1.2330291965615567e-06,
      "loss": 0.1412,
      "step": 7653
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.1860445986572437,
      "learning_rate": 1.232409789975669e-06,
      "loss": 0.1436,
      "step": 7654
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.566307479112489,
      "learning_rate": 1.231790488111869e-06,
      "loss": 0.2455,
      "step": 7655
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.8942658569961495,
      "learning_rate": 1.2311712910213227e-06,
      "loss": 0.5379,
      "step": 7656
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.1079141275915463,
      "learning_rate": 1.2305521987551821e-06,
      "loss": 0.1708,
      "step": 7657
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.782499510242526,
      "learning_rate": 1.229933211364596e-06,
      "loss": 0.5003,
      "step": 7658
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.304891755389616,
      "learning_rate": 1.2293143289007009e-06,
      "loss": 0.4232,
      "step": 7659
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0731598883800517,
      "learning_rate": 1.228695551414625e-06,
      "loss": 0.1197,
      "step": 7660
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.695776748356557,
      "learning_rate": 1.228076878957492e-06,
      "loss": 0.2727,
      "step": 7661
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.605110137644532,
      "learning_rate": 1.2274583115804096e-06,
      "loss": 0.1479,
      "step": 7662
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.102965984520214,
      "learning_rate": 1.2268398493344832e-06,
      "loss": 0.4307,
      "step": 7663
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.775188046778139,
      "learning_rate": 1.2262214922708069e-06,
      "loss": 0.4915,
      "step": 7664
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4049541483395114,
      "learning_rate": 1.2256032404404644e-06,
      "loss": 0.3033,
      "step": 7665
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.727099283312921,
      "learning_rate": 1.2249850938945366e-06,
      "loss": 0.3631,
      "step": 7666
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.502017730555558,
      "learning_rate": 1.224367052684087e-06,
      "loss": 0.3886,
      "step": 7667
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.0851418167725475,
      "learning_rate": 1.2237491168601784e-06,
      "loss": 0.327,
      "step": 7668
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.248170444468284,
      "learning_rate": 1.2231312864738607e-06,
      "loss": 0.3206,
      "step": 7669
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.061046071063577,
      "learning_rate": 1.222513561576175e-06,
      "loss": 0.2214,
      "step": 7670
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.635449600326452,
      "learning_rate": 1.2218959422181576e-06,
      "loss": 0.3104,
      "step": 7671
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.96638893182436,
      "learning_rate": 1.2212784284508292e-06,
      "loss": 0.2928,
      "step": 7672
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.963810450885632,
      "learning_rate": 1.2206610203252088e-06,
      "loss": 0.4248,
      "step": 7673
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.5532443602021,
      "learning_rate": 1.2200437178923027e-06,
      "loss": 0.3932,
      "step": 7674
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.09657371360348,
      "learning_rate": 1.2194265212031086e-06,
      "loss": 0.3974,
      "step": 7675
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.851856165733098,
      "learning_rate": 1.2188094303086188e-06,
      "loss": 0.3516,
      "step": 7676
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.245880695950445,
      "learning_rate": 1.2181924452598105e-06,
      "loss": 0.436,
      "step": 7677
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.496413158118953,
      "learning_rate": 1.2175755661076588e-06,
      "loss": 0.131,
      "step": 7678
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7069239727014636,
      "learning_rate": 1.2169587929031268e-06,
      "loss": 0.2588,
      "step": 7679
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.596553765523787,
      "learning_rate": 1.2163421256971678e-06,
      "loss": 0.3861,
      "step": 7680
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4686347958160706,
      "learning_rate": 1.2157255645407308e-06,
      "loss": 0.2065,
      "step": 7681
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.461942132549001,
      "learning_rate": 1.2151091094847495e-06,
      "loss": 0.2967,
      "step": 7682
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.668681156777201,
      "learning_rate": 1.2144927605801549e-06,
      "loss": 0.246,
      "step": 7683
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.232198539543104,
      "learning_rate": 1.2138765178778658e-06,
      "loss": 0.4505,
      "step": 7684
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.0050167403953525,
      "learning_rate": 1.2132603814287925e-06,
      "loss": 0.3347,
      "step": 7685
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.192301701060524,
      "learning_rate": 1.2126443512838396e-06,
      "loss": 0.4153,
      "step": 7686
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.873025846237934,
      "learning_rate": 1.212028427493897e-06,
      "loss": 0.2662,
      "step": 7687
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.275398030526327,
      "learning_rate": 1.2114126101098525e-06,
      "loss": 0.3737,
      "step": 7688
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.0784685914436936,
      "learning_rate": 1.21079689918258e-06,
      "loss": 0.2161,
      "step": 7689
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.542109484196927,
      "learning_rate": 1.2101812947629465e-06,
      "loss": 0.0935,
      "step": 7690
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4176617464720898,
      "learning_rate": 1.2095657969018115e-06,
      "loss": 0.2213,
      "step": 7691
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.9859920248472664,
      "learning_rate": 1.208950405650024e-06,
      "loss": 0.2285,
      "step": 7692
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.602270976893665,
      "learning_rate": 1.208335121058424e-06,
      "loss": 0.5088,
      "step": 7693
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.34637849033758,
      "learning_rate": 1.2077199431778436e-06,
      "loss": 0.2468,
      "step": 7694
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.031797690236819,
      "learning_rate": 1.207104872059105e-06,
      "loss": 0.4115,
      "step": 7695
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.3247377119854393,
      "learning_rate": 1.206489907753024e-06,
      "loss": 0.1193,
      "step": 7696
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.4976920006139953,
      "learning_rate": 1.2058750503104053e-06,
      "loss": 0.1257,
      "step": 7697
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.5821526304290474,
      "learning_rate": 1.2052602997820448e-06,
      "loss": 0.2973,
      "step": 7698
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.4792963818011846,
      "learning_rate": 1.204645656218731e-06,
      "loss": 0.3001,
      "step": 7699
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.780091316566511,
      "learning_rate": 1.204031119671241e-06,
      "loss": 0.2815,
      "step": 7700
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.5546969247093045,
      "learning_rate": 1.2034166901903471e-06,
      "loss": 0.3808,
      "step": 7701
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.681967984074751,
      "learning_rate": 1.2028023678268098e-06,
      "loss": 0.1932,
      "step": 7702
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.809568655948546,
      "learning_rate": 1.2021881526313806e-06,
      "loss": 0.2519,
      "step": 7703
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.2991136646999886,
      "learning_rate": 1.2015740446548036e-06,
      "loss": 0.2214,
      "step": 7704
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.761556168633622,
      "learning_rate": 1.2009600439478123e-06,
      "loss": 0.3282,
      "step": 7705
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.9459413711246025,
      "learning_rate": 1.2003461505611342e-06,
      "loss": 0.2695,
      "step": 7706
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.974943187969649,
      "learning_rate": 1.1997323645454852e-06,
      "loss": 0.4739,
      "step": 7707
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.326138815392011,
      "learning_rate": 1.1991186859515735e-06,
      "loss": 0.2878,
      "step": 7708
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.3718909867888156,
      "learning_rate": 1.198505114830098e-06,
      "loss": 0.1754,
      "step": 7709
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.285310039095909,
      "learning_rate": 1.197891651231748e-06,
      "loss": 0.4147,
      "step": 7710
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.689096932622553,
      "learning_rate": 1.197278295207207e-06,
      "loss": 0.1993,
      "step": 7711
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.95254501525409,
      "learning_rate": 1.1966650468071463e-06,
      "loss": 0.3804,
      "step": 7712
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2106644779486344,
      "learning_rate": 1.1960519060822296e-06,
      "loss": 0.0092,
      "step": 7713
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.156375859155833,
      "learning_rate": 1.1954388730831116e-06,
      "loss": 0.3273,
      "step": 7714
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.514236337958735,
      "learning_rate": 1.1948259478604368e-06,
      "loss": 0.2607,
      "step": 7715
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.423974073869869,
      "learning_rate": 1.1942131304648442e-06,
      "loss": 0.2367,
      "step": 7716
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.135448399187395,
      "learning_rate": 1.193600420946961e-06,
      "loss": 0.3632,
      "step": 7717
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7603530843143231,
      "learning_rate": 1.192987819357406e-06,
      "loss": 0.0514,
      "step": 7718
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.376772873446122,
      "learning_rate": 1.1923753257467893e-06,
      "loss": 0.3165,
      "step": 7719
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9527647383214375,
      "learning_rate": 1.1917629401657116e-06,
      "loss": 0.1566,
      "step": 7720
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.421405094079534,
      "learning_rate": 1.1911506626647664e-06,
      "loss": 0.4276,
      "step": 7721
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.313212351943359,
      "learning_rate": 1.190538493294537e-06,
      "loss": 0.1939,
      "step": 7722
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.217703555166767,
      "learning_rate": 1.189926432105597e-06,
      "loss": 0.5268,
      "step": 7723
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.879316563927298,
      "learning_rate": 1.1893144791485123e-06,
      "loss": 0.393,
      "step": 7724
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.618952910127185,
      "learning_rate": 1.1887026344738385e-06,
      "loss": 0.2845,
      "step": 7725
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.948564507031673,
      "learning_rate": 1.1880908981321248e-06,
      "loss": 0.261,
      "step": 7726
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.0072267021544805,
      "learning_rate": 1.1874792701739092e-06,
      "loss": 0.2121,
      "step": 7727
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.703197795362481,
      "learning_rate": 1.1868677506497216e-06,
      "loss": 0.3644,
      "step": 7728
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.9951967217677224,
      "learning_rate": 1.1862563396100815e-06,
      "loss": 0.3216,
      "step": 7729
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.943516460983074,
      "learning_rate": 1.1856450371055025e-06,
      "loss": 0.1956,
      "step": 7730
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.112919732683302,
      "learning_rate": 1.185033843186487e-06,
      "loss": 0.1278,
      "step": 7731
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1724553717521062,
      "learning_rate": 1.184422757903528e-06,
      "loss": 0.2363,
      "step": 7732
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.618009902094684,
      "learning_rate": 1.183811781307111e-06,
      "loss": 0.4113,
      "step": 7733
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.268704268184547,
      "learning_rate": 1.1832009134477107e-06,
      "loss": 0.3721,
      "step": 7734
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.547222731164547,
      "learning_rate": 1.1825901543757959e-06,
      "loss": 0.4913,
      "step": 7735
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.608057575523634,
      "learning_rate": 1.1819795041418237e-06,
      "loss": 0.0942,
      "step": 7736
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.413448282001173,
      "learning_rate": 1.181368962796243e-06,
      "loss": 0.3409,
      "step": 7737
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.3389595719245175,
      "learning_rate": 1.1807585303894936e-06,
      "loss": 0.2801,
      "step": 7738
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.891211179329331,
      "learning_rate": 1.1801482069720056e-06,
      "loss": 0.384,
      "step": 7739
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.166308352667899,
      "learning_rate": 1.1795379925942028e-06,
      "loss": 0.2802,
      "step": 7740
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.833514434830421,
      "learning_rate": 1.178927887306497e-06,
      "loss": 0.1809,
      "step": 7741
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.856037379607251,
      "learning_rate": 1.1783178911592926e-06,
      "loss": 0.0594,
      "step": 7742
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.393572722025054,
      "learning_rate": 1.1777080042029842e-06,
      "loss": 0.1553,
      "step": 7743
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.30456163427894,
      "learning_rate": 1.1770982264879567e-06,
      "loss": 0.377,
      "step": 7744
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.276253839993552,
      "learning_rate": 1.1764885580645885e-06,
      "loss": 0.3557,
      "step": 7745
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.77589051343531,
      "learning_rate": 1.175878998983247e-06,
      "loss": 0.384,
      "step": 7746
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.502217717759653,
      "learning_rate": 1.175269549294291e-06,
      "loss": 0.301,
      "step": 7747
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.63433459111157,
      "learning_rate": 1.1746602090480697e-06,
      "loss": 0.3876,
      "step": 7748
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.43295741426214,
      "learning_rate": 1.1740509782949233e-06,
      "loss": 0.665,
      "step": 7749
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.840456317124515,
      "learning_rate": 1.1734418570851852e-06,
      "loss": 0.1878,
      "step": 7750
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.920584651862398,
      "learning_rate": 1.1728328454691771e-06,
      "loss": 0.2148,
      "step": 7751
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.52796167759496,
      "learning_rate": 1.1722239434972127e-06,
      "loss": 0.4999,
      "step": 7752
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.9610512628909467,
      "learning_rate": 1.1716151512195962e-06,
      "loss": 0.2996,
      "step": 7753
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.6360253568944385,
      "learning_rate": 1.1710064686866223e-06,
      "loss": 0.2096,
      "step": 7754
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.182663880301143,
      "learning_rate": 1.1703978959485792e-06,
      "loss": 0.2136,
      "step": 7755
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.852458896826177,
      "learning_rate": 1.1697894330557433e-06,
      "loss": 0.1753,
      "step": 7756
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.9368178223269212,
      "learning_rate": 1.1691810800583827e-06,
      "loss": 0.2967,
      "step": 7757
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3969001523717095,
      "learning_rate": 1.1685728370067567e-06,
      "loss": 0.0738,
      "step": 7758
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.260469329583855,
      "learning_rate": 1.167964703951114e-06,
      "loss": 0.3463,
      "step": 7759
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.212966986316004,
      "learning_rate": 1.1673566809416984e-06,
      "loss": 0.4121,
      "step": 7760
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.8869340288426866,
      "learning_rate": 1.16674876802874e-06,
      "loss": 0.1911,
      "step": 7761
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.047548870000808,
      "learning_rate": 1.1661409652624614e-06,
      "loss": 0.1973,
      "step": 7762
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.5914288626729425,
      "learning_rate": 1.1655332726930768e-06,
      "loss": 0.2248,
      "step": 7763
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.886389659528634,
      "learning_rate": 1.16492569037079e-06,
      "loss": 0.2926,
      "step": 7764
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1597148072172674,
      "learning_rate": 1.1643182183457978e-06,
      "loss": 0.2092,
      "step": 7765
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.050645843962488,
      "learning_rate": 1.1637108566682865e-06,
      "loss": 0.1987,
      "step": 7766
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1883350690648773,
      "learning_rate": 1.1631036053884321e-06,
      "loss": 0.2321,
      "step": 7767
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.774434719221481,
      "learning_rate": 1.1624964645564038e-06,
      "loss": 0.4216,
      "step": 7768
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.055074826390672,
      "learning_rate": 1.161889434222359e-06,
      "loss": 0.4132,
      "step": 7769
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.2990493089698583,
      "learning_rate": 1.1612825144364495e-06,
      "loss": 0.2088,
      "step": 7770
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.3727315672409497,
      "learning_rate": 1.1606757052488157e-06,
      "loss": 0.0727,
      "step": 7771
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.820436870498074,
      "learning_rate": 1.1600690067095876e-06,
      "loss": 0.1784,
      "step": 7772
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.055263225466427,
      "learning_rate": 1.1594624188688908e-06,
      "loss": 0.2812,
      "step": 7773
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.150401578188887,
      "learning_rate": 1.1588559417768347e-06,
      "loss": 0.3703,
      "step": 7774
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.03098399041668,
      "learning_rate": 1.1582495754835264e-06,
      "loss": 0.182,
      "step": 7775
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.950467262275201,
      "learning_rate": 1.15764332003906e-06,
      "loss": 0.3422,
      "step": 7776
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.181480271914518,
      "learning_rate": 1.1570371754935201e-06,
      "loss": 0.1835,
      "step": 7777
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.943003118047559,
      "learning_rate": 1.1564311418969868e-06,
      "loss": 0.1997,
      "step": 7778
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.1251030884304956,
      "learning_rate": 1.155825219299523e-06,
      "loss": 0.1278,
      "step": 7779
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.294218894584409,
      "learning_rate": 1.1552194077511907e-06,
      "loss": 0.2274,
      "step": 7780
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.102747504182134,
      "learning_rate": 1.154613707302038e-06,
      "loss": 0.2077,
      "step": 7781
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.418557692753734,
      "learning_rate": 1.1540081180021037e-06,
      "loss": 0.6235,
      "step": 7782
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.6517768642763015,
      "learning_rate": 1.1534026399014213e-06,
      "loss": 0.0586,
      "step": 7783
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.3460714578689017,
      "learning_rate": 1.1527972730500092e-06,
      "loss": 0.2278,
      "step": 7784
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.9207409513876526,
      "learning_rate": 1.152192017497882e-06,
      "loss": 0.3378,
      "step": 7785
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.068734862420883,
      "learning_rate": 1.1515868732950428e-06,
      "loss": 0.2027,
      "step": 7786
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.265111394960922,
      "learning_rate": 1.1509818404914837e-06,
      "loss": 0.2447,
      "step": 7787
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.012150807893564,
      "learning_rate": 1.1503769191371936e-06,
      "loss": 0.2837,
      "step": 7788
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.417496347564624,
      "learning_rate": 1.1497721092821434e-06,
      "loss": 0.3484,
      "step": 7789
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.3291838960620135,
      "learning_rate": 1.1491674109763026e-06,
      "loss": 0.3154,
      "step": 7790
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.930360005021829,
      "learning_rate": 1.1485628242696275e-06,
      "loss": 0.2469,
      "step": 7791
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.068820262321864,
      "learning_rate": 1.1479583492120651e-06,
      "loss": 0.38,
      "step": 7792
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.517807092732238,
      "learning_rate": 1.1473539858535577e-06,
      "loss": 0.6468,
      "step": 7793
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.493502104376614,
      "learning_rate": 1.1467497342440297e-06,
      "loss": 0.3852,
      "step": 7794
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.484423351751799,
      "learning_rate": 1.146145594433405e-06,
      "loss": 0.2244,
      "step": 7795
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.840922997103643,
      "learning_rate": 1.1455415664715944e-06,
      "loss": 0.166,
      "step": 7796
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.7677320886841796,
      "learning_rate": 1.1449376504084976e-06,
      "loss": 0.2968,
      "step": 7797
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.9857583594683477,
      "learning_rate": 1.1443338462940104e-06,
      "loss": 0.1933,
      "step": 7798
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.360015178038599,
      "learning_rate": 1.1437301541780127e-06,
      "loss": 0.207,
      "step": 7799
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.077645142784825,
      "learning_rate": 1.1431265741103808e-06,
      "loss": 0.4008,
      "step": 7800
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.208673546643162,
      "learning_rate": 1.1425231061409794e-06,
      "loss": 0.2258,
      "step": 7801
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.404848676735316,
      "learning_rate": 1.1419197503196625e-06,
      "loss": 0.3344,
      "step": 7802
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.576999816902408,
      "learning_rate": 1.1413165066962794e-06,
      "loss": 0.3963,
      "step": 7803
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.65443761067629,
      "learning_rate": 1.1407133753206634e-06,
      "loss": 0.2983,
      "step": 7804
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.8069572986390683,
      "learning_rate": 1.1401103562426448e-06,
      "loss": 0.1135,
      "step": 7805
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.527160202966965,
      "learning_rate": 1.1395074495120417e-06,
      "loss": 0.3521,
      "step": 7806
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.239824927453687,
      "learning_rate": 1.138904655178662e-06,
      "loss": 0.351,
      "step": 7807
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.8605516484544795,
      "learning_rate": 1.1383019732923083e-06,
      "loss": 0.34,
      "step": 7808
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.5643759281352323,
      "learning_rate": 1.1376994039027677e-06,
      "loss": 0.1408,
      "step": 7809
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.805153846712404,
      "learning_rate": 1.1370969470598246e-06,
      "loss": 0.2519,
      "step": 7810
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.43378896255342,
      "learning_rate": 1.1364946028132496e-06,
      "loss": 0.2252,
      "step": 7811
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.132189264198262,
      "learning_rate": 1.1358923712128045e-06,
      "loss": 0.5969,
      "step": 7812
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.280686172700738,
      "learning_rate": 1.1352902523082456e-06,
      "loss": 0.2848,
      "step": 7813
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.681589514613553,
      "learning_rate": 1.1346882461493135e-06,
      "loss": 0.308,
      "step": 7814
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.313028331546428,
      "learning_rate": 1.1340863527857452e-06,
      "loss": 0.3722,
      "step": 7815
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.239185399426156,
      "learning_rate": 1.1334845722672672e-06,
      "loss": 0.4102,
      "step": 7816
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.7519334798433315,
      "learning_rate": 1.1328829046435922e-06,
      "loss": 0.2623,
      "step": 7817
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.067753196567115,
      "learning_rate": 1.1322813499644308e-06,
      "loss": 0.1476,
      "step": 7818
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.443384381236445,
      "learning_rate": 1.1316799082794771e-06,
      "loss": 0.3934,
      "step": 7819
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.265394502701272,
      "learning_rate": 1.1310785796384208e-06,
      "loss": 0.1712,
      "step": 7820
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.634558114269942,
      "learning_rate": 1.1304773640909428e-06,
      "loss": 0.4061,
      "step": 7821
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.397866256397368,
      "learning_rate": 1.1298762616867084e-06,
      "loss": 0.4584,
      "step": 7822
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.602963892068249,
      "learning_rate": 1.129275272475382e-06,
      "loss": 0.3005,
      "step": 7823
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.92989061137846,
      "learning_rate": 1.1286743965066102e-06,
      "loss": 0.3481,
      "step": 7824
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.72448613169474,
      "learning_rate": 1.1280736338300366e-06,
      "loss": 0.3881,
      "step": 7825
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.614090760254896,
      "learning_rate": 1.1274729844952954e-06,
      "loss": 0.2567,
      "step": 7826
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.213280703351922,
      "learning_rate": 1.1268724485520048e-06,
      "loss": 0.3563,
      "step": 7827
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.573995414471811,
      "learning_rate": 1.1262720260497823e-06,
      "loss": 0.42,
      "step": 7828
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.9564175272085476,
      "learning_rate": 1.1256717170382284e-06,
      "loss": 0.3743,
      "step": 7829
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.253271035039273,
      "learning_rate": 1.1250715215669388e-06,
      "loss": 0.2879,
      "step": 7830
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.0433118635057883,
      "learning_rate": 1.1244714396855015e-06,
      "loss": 0.1994,
      "step": 7831
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5248558189561074,
      "learning_rate": 1.123871471443488e-06,
      "loss": 0.2307,
      "step": 7832
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.007851673368254,
      "learning_rate": 1.1232716168904684e-06,
      "loss": 0.4304,
      "step": 7833
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.6475476462412204,
      "learning_rate": 1.1226718760759963e-06,
      "loss": 0.2192,
      "step": 7834
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1224187301555872,
      "learning_rate": 1.1220722490496214e-06,
      "loss": 0.0661,
      "step": 7835
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.431570762705918,
      "learning_rate": 1.1214727358608833e-06,
      "loss": 0.5599,
      "step": 7836
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.17054535103988,
      "learning_rate": 1.1208733365593075e-06,
      "loss": 0.3974,
      "step": 7837
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.355329368528519,
      "learning_rate": 1.1202740511944168e-06,
      "loss": 0.3673,
      "step": 7838
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3608241267957304,
      "learning_rate": 1.1196748798157178e-06,
      "loss": 0.1668,
      "step": 7839
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7827476286060855,
      "learning_rate": 1.1190758224727128e-06,
      "loss": 0.3197,
      "step": 7840
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7074787235420237,
      "learning_rate": 1.1184768792148949e-06,
      "loss": 0.2809,
      "step": 7841
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.2468887876401835,
      "learning_rate": 1.1178780500917423e-06,
      "loss": 0.3864,
      "step": 7842
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.049006624403366,
      "learning_rate": 1.1172793351527298e-06,
      "loss": 0.1562,
      "step": 7843
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.585941056933163,
      "learning_rate": 1.1166807344473194e-06,
      "loss": 0.5568,
      "step": 7844
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.30205822614975,
      "learning_rate": 1.116082248024964e-06,
      "loss": 0.2298,
      "step": 7845
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.44846882847131,
      "learning_rate": 1.11548387593511e-06,
      "loss": 0.2324,
      "step": 7846
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.257626918834146,
      "learning_rate": 1.1148856182271883e-06,
      "loss": 0.0824,
      "step": 7847
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.331147424919384,
      "learning_rate": 1.114287474950627e-06,
      "loss": 0.2856,
      "step": 7848
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.19875802961733,
      "learning_rate": 1.1136894461548409e-06,
      "loss": 0.3802,
      "step": 7849
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.952491045416856,
      "learning_rate": 1.113091531889235e-06,
      "loss": 0.1956,
      "step": 7850
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.741569137082376,
      "learning_rate": 1.1124937322032086e-06,
      "loss": 0.3102,
      "step": 7851
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.269742078366981,
      "learning_rate": 1.111896047146146e-06,
      "loss": 0.4078,
      "step": 7852
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.118951040020756,
      "learning_rate": 1.1112984767674273e-06,
      "loss": 0.2953,
      "step": 7853
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.833990726375626,
      "learning_rate": 1.1107010211164202e-06,
      "loss": 0.1841,
      "step": 7854
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.975813954584036,
      "learning_rate": 1.1101036802424822e-06,
      "loss": 0.3398,
      "step": 7855
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.6757573594186725,
      "learning_rate": 1.1095064541949656e-06,
      "loss": 0.3232,
      "step": 7856
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.176672321126871,
      "learning_rate": 1.108909343023207e-06,
      "loss": 0.2509,
      "step": 7857
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.101641148807683,
      "learning_rate": 1.1083123467765392e-06,
      "loss": 0.2474,
      "step": 7858
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.307691430343617,
      "learning_rate": 1.107715465504282e-06,
      "loss": 0.1963,
      "step": 7859
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.882993008514101,
      "learning_rate": 1.1071186992557465e-06,
      "loss": 0.4092,
      "step": 7860
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3972766374717405,
      "learning_rate": 1.1065220480802365e-06,
      "loss": 0.2725,
      "step": 7861
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.376246979366605,
      "learning_rate": 1.1059255120270416e-06,
      "loss": 0.2629,
      "step": 7862
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.178351633755653,
      "learning_rate": 1.105329091145447e-06,
      "loss": 0.2151,
      "step": 7863
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.572858345202954,
      "learning_rate": 1.104732785484725e-06,
      "loss": 0.2828,
      "step": 7864
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.695599726898285,
      "learning_rate": 1.1041365950941388e-06,
      "loss": 0.2333,
      "step": 7865
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.349670982441027,
      "learning_rate": 1.1035405200229454e-06,
      "loss": 0.4473,
      "step": 7866
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.310775742506098,
      "learning_rate": 1.102944560320386e-06,
      "loss": 0.3121,
      "step": 7867
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.51253035003476,
      "learning_rate": 1.1023487160356983e-06,
      "loss": 0.2364,
      "step": 7868
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.002084992788415,
      "learning_rate": 1.1017529872181076e-06,
      "loss": 0.1169,
      "step": 7869
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.5371161691571467,
      "learning_rate": 1.1011573739168289e-06,
      "loss": 0.2095,
      "step": 7870
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.364924679965691,
      "learning_rate": 1.1005618761810716e-06,
      "loss": 0.2184,
      "step": 7871
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.114321157199646,
      "learning_rate": 1.099966494060029e-06,
      "loss": 0.3211,
      "step": 7872
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.057437291761989,
      "learning_rate": 1.099371227602892e-06,
      "loss": 0.4643,
      "step": 7873
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.289849895040719,
      "learning_rate": 1.0987760768588368e-06,
      "loss": 0.3835,
      "step": 7874
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.921662386254124,
      "learning_rate": 1.0981810418770318e-06,
      "loss": 0.3367,
      "step": 7875
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.552731266670214,
      "learning_rate": 1.0975861227066378e-06,
      "loss": 0.2249,
      "step": 7876
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.3336720800140425,
      "learning_rate": 1.0969913193968013e-06,
      "loss": 0.4862,
      "step": 7877
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.886676317374613,
      "learning_rate": 1.096396631996664e-06,
      "loss": 0.412,
      "step": 7878
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.785038260811812,
      "learning_rate": 1.0958020605553557e-06,
      "loss": 0.2875,
      "step": 7879
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8868805554322696,
      "learning_rate": 1.0952076051219957e-06,
      "loss": 0.0459,
      "step": 7880
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.8020370631840956,
      "learning_rate": 1.094613265745698e-06,
      "loss": 0.1522,
      "step": 7881
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.7691515771643,
      "learning_rate": 1.0940190424755597e-06,
      "loss": 0.2167,
      "step": 7882
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.496728241698276,
      "learning_rate": 1.0934249353606763e-06,
      "loss": 0.2618,
      "step": 7883
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.179195216026421,
      "learning_rate": 1.0928309444501287e-06,
      "loss": 0.3115,
      "step": 7884
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.228204820049673,
      "learning_rate": 1.0922370697929884e-06,
      "loss": 0.3803,
      "step": 7885
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.326108508839045,
      "learning_rate": 1.0916433114383215e-06,
      "loss": 0.2523,
      "step": 7886
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.61032391077536,
      "learning_rate": 1.0910496694351775e-06,
      "loss": 0.1673,
      "step": 7887
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.9405348832678255,
      "learning_rate": 1.0904561438326031e-06,
      "loss": 0.402,
      "step": 7888
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.07537641562312,
      "learning_rate": 1.0898627346796313e-06,
      "loss": 0.3289,
      "step": 7889
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.742774404893435,
      "learning_rate": 1.089269442025286e-06,
      "loss": 0.3264,
      "step": 7890
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.957179679778569,
      "learning_rate": 1.088676265918584e-06,
      "loss": 0.2298,
      "step": 7891
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.366127926610244,
      "learning_rate": 1.0880832064085296e-06,
      "loss": 0.4273,
      "step": 7892
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.1942202730417835,
      "learning_rate": 1.0874902635441182e-06,
      "loss": 0.103,
      "step": 7893
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.3640134728137605,
      "learning_rate": 1.086897437374336e-06,
      "loss": 0.4055,
      "step": 7894
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.29298770673862,
      "learning_rate": 1.0863047279481587e-06,
      "loss": 0.3961,
      "step": 7895
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7701689926193067,
      "learning_rate": 1.0857121353145547e-06,
      "loss": 0.2334,
      "step": 7896
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.9134334419920362,
      "learning_rate": 1.0851196595224803e-06,
      "loss": 0.3247,
      "step": 7897
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.087671500508998,
      "learning_rate": 1.0845273006208829e-06,
      "loss": 0.4367,
      "step": 7898
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.782377752220401,
      "learning_rate": 1.0839350586587e-06,
      "loss": 0.3375,
      "step": 7899
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.5262473303110484,
      "learning_rate": 1.0833429336848592e-06,
      "loss": 0.1147,
      "step": 7900
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.584305551838806,
      "learning_rate": 1.0827509257482806e-06,
      "loss": 0.3251,
      "step": 7901
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.269771308121908,
      "learning_rate": 1.082159034897872e-06,
      "loss": 0.4035,
      "step": 7902
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.7216392304891945,
      "learning_rate": 1.0815672611825325e-06,
      "loss": 0.3434,
      "step": 7903
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.866829854882413,
      "learning_rate": 1.080975604651152e-06,
      "loss": 0.312,
      "step": 7904
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.819816007999829,
      "learning_rate": 1.080384065352609e-06,
      "loss": 0.2865,
      "step": 7905
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.035493783563603,
      "learning_rate": 1.0797926433357752e-06,
      "loss": 0.3052,
      "step": 7906
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.312262097273038,
      "learning_rate": 1.0792013386495108e-06,
      "loss": 0.5436,
      "step": 7907
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.729823713809358,
      "learning_rate": 1.0786101513426655e-06,
      "loss": 0.273,
      "step": 7908
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.60895521163094,
      "learning_rate": 1.0780190814640808e-06,
      "loss": 0.572,
      "step": 7909
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.962263073831718,
      "learning_rate": 1.0774281290625874e-06,
      "loss": 0.4132,
      "step": 7910
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.284580293684214,
      "learning_rate": 1.0768372941870081e-06,
      "loss": 0.2143,
      "step": 7911
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.2946632522715404,
      "learning_rate": 1.0762465768861544e-06,
      "loss": 0.1205,
      "step": 7912
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.719523493423759,
      "learning_rate": 1.0756559772088281e-06,
      "loss": 0.1716,
      "step": 7913
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.436270004827043,
      "learning_rate": 1.075065495203822e-06,
      "loss": 0.431,
      "step": 7914
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.067372489622736,
      "learning_rate": 1.0744751309199174e-06,
      "loss": 0.201,
      "step": 7915
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.69011776564537,
      "learning_rate": 1.0738848844058897e-06,
      "loss": 0.2233,
      "step": 7916
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.3891489240593793,
      "learning_rate": 1.0732947557105006e-06,
      "loss": 0.2705,
      "step": 7917
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.577503864472733,
      "learning_rate": 1.0727047448825043e-06,
      "loss": 0.4066,
      "step": 7918
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.876552453942396,
      "learning_rate": 1.0721148519706443e-06,
      "loss": 0.3495,
      "step": 7919
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.0472298618760245,
      "learning_rate": 1.0715250770236539e-06,
      "loss": 0.3582,
      "step": 7920
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.389565519994612,
      "learning_rate": 1.070935420090259e-06,
      "loss": 0.2352,
      "step": 7921
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.148503624968883,
      "learning_rate": 1.070345881219173e-06,
      "loss": 0.3046,
      "step": 7922
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.7293178687685113,
      "learning_rate": 1.0697564604591015e-06,
      "loss": 0.2879,
      "step": 7923
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.764737806702195,
      "learning_rate": 1.069167157858739e-06,
      "loss": 0.2859,
      "step": 7924
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.504542904932268,
      "learning_rate": 1.06857797346677e-06,
      "loss": 0.2967,
      "step": 7925
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.7380589539759743,
      "learning_rate": 1.067988907331872e-06,
      "loss": 0.0427,
      "step": 7926
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.972915688751554,
      "learning_rate": 1.0673999595027095e-06,
      "loss": 0.3184,
      "step": 7927
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.313193027650622,
      "learning_rate": 1.0668111300279388e-06,
      "loss": 0.3247,
      "step": 7928
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.457555610792065,
      "learning_rate": 1.0662224189562062e-06,
      "loss": 0.2312,
      "step": 7929
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.846413650532037,
      "learning_rate": 1.0656338263361466e-06,
      "loss": 0.2532,
      "step": 7930
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.9380570778999986,
      "learning_rate": 1.0650453522163891e-06,
      "loss": 0.3421,
      "step": 7931
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.2945533618917855,
      "learning_rate": 1.0644569966455492e-06,
      "loss": 0.3135,
      "step": 7932
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.101982487597689,
      "learning_rate": 1.0638687596722343e-06,
      "loss": 0.3793,
      "step": 7933
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.092110304681205,
      "learning_rate": 1.0632806413450408e-06,
      "loss": 0.4182,
      "step": 7934
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.9373683757735023,
      "learning_rate": 1.0626926417125575e-06,
      "loss": 0.0906,
      "step": 7935
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.944931404544964,
      "learning_rate": 1.0621047608233614e-06,
      "loss": 0.3254,
      "step": 7936
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.216300376599261,
      "learning_rate": 1.0615169987260205e-06,
      "loss": 0.2242,
      "step": 7937
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.047093297714543,
      "learning_rate": 1.0609293554690925e-06,
      "loss": 0.341,
      "step": 7938
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.356174571272702,
      "learning_rate": 1.0603418311011248e-06,
      "loss": 0.3576,
      "step": 7939
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.194762489203468,
      "learning_rate": 1.059754425670658e-06,
      "loss": 0.2172,
      "step": 7940
    },
    {
      "epoch": 0.7,
      "grad_norm": 3.4327771027105003,
      "learning_rate": 1.059167139226219e-06,
      "loss": 0.1186,
      "step": 7941
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.519733725326662,
      "learning_rate": 1.0585799718163274e-06,
      "loss": 0.3374,
      "step": 7942
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4871096994050372,
      "learning_rate": 1.0579929234894912e-06,
      "loss": 0.059,
      "step": 7943
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.419762485677104,
      "learning_rate": 1.0574059942942092e-06,
      "loss": 0.2207,
      "step": 7944
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.786038724350996,
      "learning_rate": 1.056819184278972e-06,
      "loss": 0.4922,
      "step": 7945
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.337269659281116,
      "learning_rate": 1.0562324934922582e-06,
      "loss": 0.1588,
      "step": 7946
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.227918068454636,
      "learning_rate": 1.0556459219825376e-06,
      "loss": 0.144,
      "step": 7947
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7366085499345005,
      "learning_rate": 1.0550594697982696e-06,
      "loss": 0.1448,
      "step": 7948
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.706756768430513,
      "learning_rate": 1.0544731369879032e-06,
      "loss": 0.2814,
      "step": 7949
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.0109312511889783,
      "learning_rate": 1.05388692359988e-06,
      "loss": 0.2262,
      "step": 7950
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8905122584672958,
      "learning_rate": 1.0533008296826294e-06,
      "loss": 0.1129,
      "step": 7951
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8905373094342712,
      "learning_rate": 1.0527148552845712e-06,
      "loss": 0.2875,
      "step": 7952
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.494750800848902,
      "learning_rate": 1.0521290004541162e-06,
      "loss": 0.4232,
      "step": 7953
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6138712046328947,
      "learning_rate": 1.0515432652396637e-06,
      "loss": 0.2487,
      "step": 7954
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7530675614702496,
      "learning_rate": 1.0509576496896061e-06,
      "loss": 0.1381,
      "step": 7955
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.527192757959521,
      "learning_rate": 1.0503721538523231e-06,
      "loss": 0.2665,
      "step": 7956
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.252464858771108,
      "learning_rate": 1.0497867777761857e-06,
      "loss": 0.2916,
      "step": 7957
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.431393937510424,
      "learning_rate": 1.0492015215095549e-06,
      "loss": 0.5878,
      "step": 7958
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2749057600178104,
      "learning_rate": 1.0486163851007804e-06,
      "loss": 0.1237,
      "step": 7959
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.012729107619377,
      "learning_rate": 1.0480313685982054e-06,
      "loss": 0.388,
      "step": 7960
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.285509638893047,
      "learning_rate": 1.0474464720501602e-06,
      "loss": 0.1947,
      "step": 7961
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.9913203055550373,
      "learning_rate": 1.046861695504966e-06,
      "loss": 0.167,
      "step": 7962
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.4706765614211257,
      "learning_rate": 1.046277039010934e-06,
      "loss": 0.2014,
      "step": 7963
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.211160516405774,
      "learning_rate": 1.045692502616365e-06,
      "loss": 0.3321,
      "step": 7964
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.982410519045244,
      "learning_rate": 1.045108086369552e-06,
      "loss": 0.3594,
      "step": 7965
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8986920205239497,
      "learning_rate": 1.0445237903187764e-06,
      "loss": 0.3347,
      "step": 7966
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.936614603427612,
      "learning_rate": 1.043939614512309e-06,
      "loss": 0.2365,
      "step": 7967
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.5296128119017993,
      "learning_rate": 1.0433555589984123e-06,
      "loss": 0.1418,
      "step": 7968
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.601968464650039,
      "learning_rate": 1.0427716238253365e-06,
      "loss": 0.2148,
      "step": 7969
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.9280310892387433,
      "learning_rate": 1.042187809041326e-06,
      "loss": 0.3323,
      "step": 7970
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.35763557214826,
      "learning_rate": 1.0416041146946113e-06,
      "loss": 0.2911,
      "step": 7971
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.355476628278919,
      "learning_rate": 1.0410205408334138e-06,
      "loss": 0.4119,
      "step": 7972
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.066643978290284,
      "learning_rate": 1.0404370875059478e-06,
      "loss": 0.3071,
      "step": 7973
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.749790423433223,
      "learning_rate": 1.0398537547604117e-06,
      "loss": 0.5071,
      "step": 7974
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.95952580669225,
      "learning_rate": 1.0392705426450012e-06,
      "loss": 0.2718,
      "step": 7975
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.599888560096297,
      "learning_rate": 1.0386874512078964e-06,
      "loss": 0.2658,
      "step": 7976
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7762489627542193,
      "learning_rate": 1.0381044804972693e-06,
      "loss": 0.2129,
      "step": 7977
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.130758734684824,
      "learning_rate": 1.0375216305612846e-06,
      "loss": 0.3591,
      "step": 7978
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.911511800022975,
      "learning_rate": 1.0369389014480908e-06,
      "loss": 0.1962,
      "step": 7979
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.3258029933660405,
      "learning_rate": 1.0363562932058327e-06,
      "loss": 0.3526,
      "step": 7980
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.803169226613103,
      "learning_rate": 1.0357738058826418e-06,
      "loss": 0.4267,
      "step": 7981
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.482227269710919,
      "learning_rate": 1.0351914395266396e-06,
      "loss": 0.2732,
      "step": 7982
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.088480896113956,
      "learning_rate": 1.0346091941859408e-06,
      "loss": 0.3021,
      "step": 7983
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.0737810151522194,
      "learning_rate": 1.034027069908644e-06,
      "loss": 0.2023,
      "step": 7984
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.189032419325716,
      "learning_rate": 1.0334450667428446e-06,
      "loss": 0.3491,
      "step": 7985
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.454168880312561,
      "learning_rate": 1.0328631847366236e-06,
      "loss": 0.2228,
      "step": 7986
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.5578068864172865,
      "learning_rate": 1.032281423938052e-06,
      "loss": 0.4486,
      "step": 7987
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.676992563687076,
      "learning_rate": 1.0316997843951956e-06,
      "loss": 0.3703,
      "step": 7988
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.49423708733383,
      "learning_rate": 1.0311182661561026e-06,
      "loss": 0.359,
      "step": 7989
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.833110277176276,
      "learning_rate": 1.0305368692688175e-06,
      "loss": 0.5191,
      "step": 7990
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.5678933080532587,
      "learning_rate": 1.0299555937813718e-06,
      "loss": 0.1067,
      "step": 7991
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2242225449080286,
      "learning_rate": 1.0293744397417872e-06,
      "loss": 0.1682,
      "step": 7992
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.7294494962896625,
      "learning_rate": 1.0287934071980777e-06,
      "loss": 0.3884,
      "step": 7993
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.826963317778417,
      "learning_rate": 1.0282124961982423e-06,
      "loss": 0.3322,
      "step": 7994
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.754279631806457,
      "learning_rate": 1.0276317067902755e-06,
      "loss": 0.3877,
      "step": 7995
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.9243025844658828,
      "learning_rate": 1.0270510390221586e-06,
      "loss": 0.2341,
      "step": 7996
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.907554142465052,
      "learning_rate": 1.0264704929418622e-06,
      "loss": 0.2891,
      "step": 7997
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.065076912520467,
      "learning_rate": 1.0258900685973515e-06,
      "loss": 0.4198,
      "step": 7998
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.183510008595119,
      "learning_rate": 1.0253097660365743e-06,
      "loss": 0.2767,
      "step": 7999
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.862269011347802,
      "learning_rate": 1.024729585307475e-06,
      "loss": 0.4935,
      "step": 8000
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.022993282979413,
      "learning_rate": 1.0241495264579845e-06,
      "loss": 0.1997,
      "step": 8001
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.19456564328777,
      "learning_rate": 1.0235695895360238e-06,
      "loss": 0.2672,
      "step": 8002
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.267127213273426,
      "learning_rate": 1.0229897745895065e-06,
      "loss": 0.1106,
      "step": 8003
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.14221093806339,
      "learning_rate": 1.022410081666331e-06,
      "loss": 0.0426,
      "step": 8004
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8432954362482667,
      "learning_rate": 1.021830510814391e-06,
      "loss": 0.1666,
      "step": 8005
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.06776814888677,
      "learning_rate": 1.0212510620815674e-06,
      "loss": 0.3071,
      "step": 8006
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.46111770056584,
      "learning_rate": 1.02067173551573e-06,
      "loss": 0.2517,
      "step": 8007
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.365289379649488,
      "learning_rate": 1.0200925311647431e-06,
      "loss": 0.3038,
      "step": 8008
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.175125288630128,
      "learning_rate": 1.0195134490764537e-06,
      "loss": 0.3835,
      "step": 8009
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.633056141312321,
      "learning_rate": 1.0189344892987057e-06,
      "loss": 0.2482,
      "step": 8010
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.7507170081617485,
      "learning_rate": 1.0183556518793289e-06,
      "loss": 0.4638,
      "step": 8011
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.481735887414481,
      "learning_rate": 1.017776936866143e-06,
      "loss": 0.2317,
      "step": 8012
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7426891254301573,
      "learning_rate": 1.0171983443069617e-06,
      "loss": 0.1911,
      "step": 8013
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.2445983744507485,
      "learning_rate": 1.0166198742495815e-06,
      "loss": 0.2233,
      "step": 8014
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.565522173717786,
      "learning_rate": 1.0160415267417956e-06,
      "loss": 0.2743,
      "step": 8015
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.894499144964879,
      "learning_rate": 1.0154633018313832e-06,
      "loss": 0.158,
      "step": 8016
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.827475864413353,
      "learning_rate": 1.0148851995661136e-06,
      "loss": 0.3777,
      "step": 8017
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.8276162211704343,
      "learning_rate": 1.0143072199937494e-06,
      "loss": 0.2323,
      "step": 8018
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.5100602700071377,
      "learning_rate": 1.013729363162037e-06,
      "loss": 0.2332,
      "step": 8019
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.222000641943792,
      "learning_rate": 1.0131516291187179e-06,
      "loss": 0.4328,
      "step": 8020
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.078102046622693,
      "learning_rate": 1.0125740179115234e-06,
      "loss": 0.3219,
      "step": 8021
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6677008582040656,
      "learning_rate": 1.0119965295881693e-06,
      "loss": 0.1802,
      "step": 8022
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.8465558474121355,
      "learning_rate": 1.0114191641963683e-06,
      "loss": 0.3441,
      "step": 8023
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.221133976894503,
      "learning_rate": 1.0108419217838162e-06,
      "loss": 0.1095,
      "step": 8024
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.606339212539085,
      "learning_rate": 1.0102648023982034e-06,
      "loss": 0.2138,
      "step": 8025
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.203392730527414,
      "learning_rate": 1.0096878060872106e-06,
      "loss": 0.4209,
      "step": 8026
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.9294023949449532,
      "learning_rate": 1.0091109328985032e-06,
      "loss": 0.2077,
      "step": 8027
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.223969450932915,
      "learning_rate": 1.0085341828797427e-06,
      "loss": 0.2503,
      "step": 8028
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6099285218014177,
      "learning_rate": 1.0079575560785736e-06,
      "loss": 0.2638,
      "step": 8029
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.918090478500378,
      "learning_rate": 1.0073810525426366e-06,
      "loss": 0.3409,
      "step": 8030
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.328031569414505,
      "learning_rate": 1.0068046723195606e-06,
      "loss": 0.2071,
      "step": 8031
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.11807179363195,
      "learning_rate": 1.0062284154569604e-06,
      "loss": 0.3468,
      "step": 8032
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.5998195447203836,
      "learning_rate": 1.0056522820024467e-06,
      "loss": 0.1981,
      "step": 8033
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.367694028938498,
      "learning_rate": 1.005076272003613e-06,
      "loss": 0.4032,
      "step": 8034
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.683800479166345,
      "learning_rate": 1.0045003855080486e-06,
      "loss": 0.3087,
      "step": 8035
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.773880756011064,
      "learning_rate": 1.003924622563332e-06,
      "loss": 0.3194,
      "step": 8036
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.5045393405658345,
      "learning_rate": 1.0033489832170265e-06,
      "loss": 0.4679,
      "step": 8037
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.271906464627221,
      "learning_rate": 1.0027734675166922e-06,
      "loss": 0.2174,
      "step": 8038
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.235079978917787,
      "learning_rate": 1.0021980755098717e-06,
      "loss": 0.2327,
      "step": 8039
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.4209452311378974,
      "learning_rate": 1.0016228072441027e-06,
      "loss": 0.2313,
      "step": 8040
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.547695632096783,
      "learning_rate": 1.0010476627669133e-06,
      "loss": 0.3841,
      "step": 8041
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.067954312519738,
      "learning_rate": 1.0004726421258152e-06,
      "loss": 0.2069,
      "step": 8042
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.715356653337662,
      "learning_rate": 9.998977453683174e-07,
      "loss": 0.4024,
      "step": 8043
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.216287534608381,
      "learning_rate": 9.993229725419112e-07,
      "loss": 0.4206,
      "step": 8044
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.4027780044387774,
      "learning_rate": 9.987483236940837e-07,
      "loss": 0.2149,
      "step": 8045
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.848909039931527,
      "learning_rate": 9.981737988723114e-07,
      "loss": 0.4426,
      "step": 8046
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.344373807719834,
      "learning_rate": 9.97599398124055e-07,
      "loss": 0.2942,
      "step": 8047
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.962421928194285,
      "learning_rate": 9.97025121496771e-07,
      "loss": 0.0987,
      "step": 8048
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.3180197707984327,
      "learning_rate": 9.964509690379033e-07,
      "loss": 0.1769,
      "step": 8049
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.7250183413528943,
      "learning_rate": 9.958769407948837e-07,
      "loss": 0.1432,
      "step": 8050
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.031227723168055,
      "learning_rate": 9.95303036815139e-07,
      "loss": 0.1807,
      "step": 8051
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1612695007384515,
      "learning_rate": 9.947292571460784e-07,
      "loss": 0.0976,
      "step": 8052
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.6697263092819914,
      "learning_rate": 9.941556018351075e-07,
      "loss": 0.1115,
      "step": 8053
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.875323202084768,
      "learning_rate": 9.935820709296179e-07,
      "loss": 0.1677,
      "step": 8054
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4668478917643024,
      "learning_rate": 9.930086644769913e-07,
      "loss": 0.1019,
      "step": 8055
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.630110790454691,
      "learning_rate": 9.924353825246023e-07,
      "loss": 0.2815,
      "step": 8056
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.0939109732944585,
      "learning_rate": 9.918622251198086e-07,
      "loss": 0.4311,
      "step": 8057
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.627049180168371,
      "learning_rate": 9.912891923099653e-07,
      "loss": 0.1517,
      "step": 8058
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.996215879569889,
      "learning_rate": 9.907162841424117e-07,
      "loss": 0.3108,
      "step": 8059
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.645549288526271,
      "learning_rate": 9.901435006644786e-07,
      "loss": 0.272,
      "step": 8060
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.599984591444332,
      "learning_rate": 9.895708419234887e-07,
      "loss": 0.2043,
      "step": 8061
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.372643504398348,
      "learning_rate": 9.889983079667486e-07,
      "loss": 0.369,
      "step": 8062
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.627638388887321,
      "learning_rate": 9.884258988415612e-07,
      "loss": 0.4876,
      "step": 8063
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.318679613442378,
      "learning_rate": 9.878536145952153e-07,
      "loss": 0.2888,
      "step": 8064
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.680349451113893,
      "learning_rate": 9.872814552749893e-07,
      "loss": 0.3257,
      "step": 8065
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.113420684818114,
      "learning_rate": 9.867094209281547e-07,
      "loss": 0.2073,
      "step": 8066
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.071440208023291,
      "learning_rate": 9.861375116019664e-07,
      "loss": 0.1866,
      "step": 8067
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.9426659612824877,
      "learning_rate": 9.85565727343676e-07,
      "loss": 0.2969,
      "step": 8068
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.761670926043383,
      "learning_rate": 9.849940682005204e-07,
      "loss": 0.3704,
      "step": 8069
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.545373963931711,
      "learning_rate": 9.844225342197263e-07,
      "loss": 0.187,
      "step": 8070
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.9219606072265645,
      "learning_rate": 9.838511254485135e-07,
      "loss": 0.3777,
      "step": 8071
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.469409067371009,
      "learning_rate": 9.832798419340861e-07,
      "loss": 0.1912,
      "step": 8072
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.174826608953927,
      "learning_rate": 9.827086837236427e-07,
      "loss": 0.3906,
      "step": 8073
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.277543499996601,
      "learning_rate": 9.821376508643688e-07,
      "loss": 0.1869,
      "step": 8074
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.464522769650549,
      "learning_rate": 9.8156674340344e-07,
      "loss": 0.3258,
      "step": 8075
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8168181144932194,
      "learning_rate": 9.809959613880238e-07,
      "loss": 0.3007,
      "step": 8076
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.7413386845600636,
      "learning_rate": 9.804253048652724e-07,
      "loss": 0.307,
      "step": 8077
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.028243528309216,
      "learning_rate": 9.79854773882333e-07,
      "loss": 0.3222,
      "step": 8078
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.4202910052644624,
      "learning_rate": 9.792843684863394e-07,
      "loss": 0.2256,
      "step": 8079
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.517337305347685,
      "learning_rate": 9.787140887244149e-07,
      "loss": 0.3825,
      "step": 8080
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8385730104597577,
      "learning_rate": 9.781439346436753e-07,
      "loss": 0.2097,
      "step": 8081
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.378009830379842,
      "learning_rate": 9.77573906291221e-07,
      "loss": 0.1203,
      "step": 8082
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.5177777940656627,
      "learning_rate": 9.770040037141476e-07,
      "loss": 0.2553,
      "step": 8083
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.994417450046353,
      "learning_rate": 9.764342269595362e-07,
      "loss": 0.3869,
      "step": 8084
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3020226484287467,
      "learning_rate": 9.758645760744587e-07,
      "loss": 0.2092,
      "step": 8085
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.673695278687454,
      "learning_rate": 9.752950511059792e-07,
      "loss": 0.1234,
      "step": 8086
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.610753286804383,
      "learning_rate": 9.747256521011455e-07,
      "loss": 0.3561,
      "step": 8087
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.49628947003665,
      "learning_rate": 9.741563791070014e-07,
      "loss": 0.0372,
      "step": 8088
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.5049171645119355,
      "learning_rate": 9.735872321705763e-07,
      "loss": 0.4547,
      "step": 8089
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.2174938044657635,
      "learning_rate": 9.730182113388898e-07,
      "loss": 0.4326,
      "step": 8090
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.320843257994449,
      "learning_rate": 9.72449316658954e-07,
      "loss": 0.2267,
      "step": 8091
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.061151192422366,
      "learning_rate": 9.718805481777646e-07,
      "loss": 0.3208,
      "step": 8092
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1046494172084658,
      "learning_rate": 9.713119059423132e-07,
      "loss": 0.1232,
      "step": 8093
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.068715235019373,
      "learning_rate": 9.707433899995775e-07,
      "loss": 0.0788,
      "step": 8094
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.520817784704015,
      "learning_rate": 9.701750003965243e-07,
      "loss": 0.2157,
      "step": 8095
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.14369579135392,
      "learning_rate": 9.696067371801135e-07,
      "loss": 0.3171,
      "step": 8096
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.500384349643724,
      "learning_rate": 9.690386003972906e-07,
      "loss": 0.1507,
      "step": 8097
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.3784355815603972,
      "learning_rate": 9.684705900949928e-07,
      "loss": 0.2309,
      "step": 8098
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.371764943237747,
      "learning_rate": 9.679027063201462e-07,
      "loss": 0.4252,
      "step": 8099
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.546216615756755,
      "learning_rate": 9.673349491196657e-07,
      "loss": 0.2391,
      "step": 8100
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.990702348671934,
      "learning_rate": 9.66767318540458e-07,
      "loss": 0.4479,
      "step": 8101
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.848839098006489,
      "learning_rate": 9.661998146294176e-07,
      "loss": 0.4231,
      "step": 8102
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.024041491767088,
      "learning_rate": 9.656324374334288e-07,
      "loss": 0.3762,
      "step": 8103
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.07757346718089,
      "learning_rate": 9.650651869993655e-07,
      "loss": 0.2794,
      "step": 8104
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.7744726882130895,
      "learning_rate": 9.644980633740899e-07,
      "loss": 0.2481,
      "step": 8105
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.413804992994779,
      "learning_rate": 9.639310666044574e-07,
      "loss": 0.3382,
      "step": 8106
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.360208054658699,
      "learning_rate": 9.633641967373088e-07,
      "loss": 0.3181,
      "step": 8107
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.142163592591411,
      "learning_rate": 9.627974538194767e-07,
      "loss": 0.4388,
      "step": 8108
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.367857334489461,
      "learning_rate": 9.622308378977827e-07,
      "loss": 0.1986,
      "step": 8109
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9421507473934145,
      "learning_rate": 9.616643490190364e-07,
      "loss": 0.1471,
      "step": 8110
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.721451557053717,
      "learning_rate": 9.610979872300406e-07,
      "loss": 0.2358,
      "step": 8111
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6374558008369657,
      "learning_rate": 9.605317525775843e-07,
      "loss": 0.1661,
      "step": 8112
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.862997887753607,
      "learning_rate": 9.599656451084468e-07,
      "loss": 0.3311,
      "step": 8113
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.687898528110329,
      "learning_rate": 9.593996648693973e-07,
      "loss": 0.5448,
      "step": 8114
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.382984886848671,
      "learning_rate": 9.588338119071939e-07,
      "loss": 0.2601,
      "step": 8115
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.8469938641219965,
      "learning_rate": 9.582680862685853e-07,
      "loss": 0.4025,
      "step": 8116
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.9888083477485834,
      "learning_rate": 9.57702488000309e-07,
      "loss": 0.2792,
      "step": 8117
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.388497631279313,
      "learning_rate": 9.571370171490918e-07,
      "loss": 0.2077,
      "step": 8118
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.464787539747602,
      "learning_rate": 9.565716737616498e-07,
      "loss": 0.2789,
      "step": 8119
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.740382083649629,
      "learning_rate": 9.560064578846883e-07,
      "loss": 0.4303,
      "step": 8120
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.21611169036712,
      "learning_rate": 9.554413695649043e-07,
      "loss": 0.2307,
      "step": 8121
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.9806720449902087,
      "learning_rate": 9.548764088489817e-07,
      "loss": 0.4403,
      "step": 8122
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.844876750876104,
      "learning_rate": 9.54311575783595e-07,
      "loss": 0.5241,
      "step": 8123
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.6513266372936015,
      "learning_rate": 9.537468704154077e-07,
      "loss": 0.1535,
      "step": 8124
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.296398895397036,
      "learning_rate": 9.531822927910722e-07,
      "loss": 0.1799,
      "step": 8125
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.236464867643115,
      "learning_rate": 9.52617842957233e-07,
      "loss": 0.2449,
      "step": 8126
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.216822303986887,
      "learning_rate": 9.520535209605214e-07,
      "loss": 0.1864,
      "step": 8127
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.343628782257899,
      "learning_rate": 9.514893268475584e-07,
      "loss": 0.2799,
      "step": 8128
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0990777270277055,
      "learning_rate": 9.509252606649555e-07,
      "loss": 0.1732,
      "step": 8129
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.060038474071287,
      "learning_rate": 9.503613224593119e-07,
      "loss": 0.1927,
      "step": 8130
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.43349595234731,
      "learning_rate": 9.49797512277219e-07,
      "loss": 0.298,
      "step": 8131
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.7456579247221358,
      "learning_rate": 9.492338301652557e-07,
      "loss": 0.1954,
      "step": 8132
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.029446078659239,
      "learning_rate": 9.486702761699904e-07,
      "loss": 0.168,
      "step": 8133
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.8104600761784106,
      "learning_rate": 9.481068503379801e-07,
      "loss": 0.2463,
      "step": 8134
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.9823751529875375,
      "learning_rate": 9.475435527157742e-07,
      "loss": 0.305,
      "step": 8135
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7669320617185703,
      "learning_rate": 9.469803833499086e-07,
      "loss": 0.0057,
      "step": 8136
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.224714339550611,
      "learning_rate": 9.464173422869099e-07,
      "loss": 0.3791,
      "step": 8137
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0868703454130637,
      "learning_rate": 9.458544295732936e-07,
      "loss": 0.1415,
      "step": 8138
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.792892166358232,
      "learning_rate": 9.452916452555636e-07,
      "loss": 0.1553,
      "step": 8139
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.382166037363336,
      "learning_rate": 9.447289893802164e-07,
      "loss": 0.2796,
      "step": 8140
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.772049467539659,
      "learning_rate": 9.441664619937352e-07,
      "loss": 0.2981,
      "step": 8141
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.9320488754425353,
      "learning_rate": 9.43604063142593e-07,
      "loss": 0.085,
      "step": 8142
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.637518013320503,
      "learning_rate": 9.430417928732527e-07,
      "loss": 0.3689,
      "step": 8143
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.529953097963664,
      "learning_rate": 9.424796512321651e-07,
      "loss": 0.381,
      "step": 8144
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0928318198747777,
      "learning_rate": 9.419176382657735e-07,
      "loss": 0.1472,
      "step": 8145
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.4526803957437915,
      "learning_rate": 9.413557540205076e-07,
      "loss": 0.4853,
      "step": 8146
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.013875771359988,
      "learning_rate": 9.40793998542788e-07,
      "loss": 0.4646,
      "step": 8147
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.655794018805573,
      "learning_rate": 9.402323718790235e-07,
      "loss": 0.1706,
      "step": 8148
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.804360857351522,
      "learning_rate": 9.396708740756127e-07,
      "loss": 0.28,
      "step": 8149
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.522270398041571,
      "learning_rate": 9.39109505178945e-07,
      "loss": 0.4046,
      "step": 8150
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.414760683866641,
      "learning_rate": 9.385482652353978e-07,
      "loss": 0.4935,
      "step": 8151
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.080708032501242,
      "learning_rate": 9.379871542913371e-07,
      "loss": 0.341,
      "step": 8152
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.542662742135223,
      "learning_rate": 9.374261723931199e-07,
      "loss": 0.2475,
      "step": 8153
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.808342527193517,
      "learning_rate": 9.368653195870902e-07,
      "loss": 0.4152,
      "step": 8154
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.094316274670632,
      "learning_rate": 9.363045959195854e-07,
      "loss": 0.3833,
      "step": 8155
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.354764060898442,
      "learning_rate": 9.357440014369284e-07,
      "loss": 0.2971,
      "step": 8156
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.495346585788684,
      "learning_rate": 9.351835361854328e-07,
      "loss": 0.4874,
      "step": 8157
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.231345228887125,
      "learning_rate": 9.346232002114016e-07,
      "loss": 0.3582,
      "step": 8158
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.0884454895183797,
      "learning_rate": 9.340629935611262e-07,
      "loss": 0.2166,
      "step": 8159
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.6727493916625047,
      "learning_rate": 9.335029162808895e-07,
      "loss": 0.1908,
      "step": 8160
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.2429376752713424,
      "learning_rate": 9.329429684169619e-07,
      "loss": 0.161,
      "step": 8161
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.9810744651656975,
      "learning_rate": 9.323831500156033e-07,
      "loss": 0.4222,
      "step": 8162
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.1985589267819656,
      "learning_rate": 9.318234611230634e-07,
      "loss": 0.2832,
      "step": 8163
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.747303877190647,
      "learning_rate": 9.312639017855799e-07,
      "loss": 0.2063,
      "step": 8164
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.355047106507734,
      "learning_rate": 9.307044720493824e-07,
      "loss": 0.3574,
      "step": 8165
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.933802423702169,
      "learning_rate": 9.301451719606877e-07,
      "loss": 0.5215,
      "step": 8166
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.768061092834999,
      "learning_rate": 9.295860015657024e-07,
      "loss": 0.3955,
      "step": 8167
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.752263080561388,
      "learning_rate": 9.290269609106222e-07,
      "loss": 0.4748,
      "step": 8168
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.212376550447095,
      "learning_rate": 9.284680500416316e-07,
      "loss": 0.1798,
      "step": 8169
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.317597424562535,
      "learning_rate": 9.279092690049069e-07,
      "loss": 0.3391,
      "step": 8170
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.902956867651893,
      "learning_rate": 9.273506178466105e-07,
      "loss": 0.3523,
      "step": 8171
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.443779861103102,
      "learning_rate": 9.26792096612896e-07,
      "loss": 0.4326,
      "step": 8172
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.9792572565310933,
      "learning_rate": 9.262337053499054e-07,
      "loss": 0.1664,
      "step": 8173
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.84510219471958,
      "learning_rate": 9.256754441037696e-07,
      "loss": 0.2989,
      "step": 8174
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.7198818480090132,
      "learning_rate": 9.251173129206106e-07,
      "loss": 0.2787,
      "step": 8175
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.1388213861479564,
      "learning_rate": 9.245593118465385e-07,
      "loss": 0.1832,
      "step": 8176
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.6701992199557796,
      "learning_rate": 9.240014409276507e-07,
      "loss": 0.2216,
      "step": 8177
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.076700456596792,
      "learning_rate": 9.234437002100393e-07,
      "loss": 0.1263,
      "step": 8178
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2064134759958098,
      "learning_rate": 9.228860897397779e-07,
      "loss": 0.0221,
      "step": 8179
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.7729326126729976,
      "learning_rate": 9.223286095629367e-07,
      "loss": 0.1584,
      "step": 8180
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.491865692317227,
      "learning_rate": 9.217712597255707e-07,
      "loss": 0.4952,
      "step": 8181
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.5855567095833853,
      "learning_rate": 9.212140402737249e-07,
      "loss": 0.1016,
      "step": 8182
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.279983229465395,
      "learning_rate": 9.206569512534363e-07,
      "loss": 0.5122,
      "step": 8183
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.273788372856177,
      "learning_rate": 9.200999927107254e-07,
      "loss": 0.3819,
      "step": 8184
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.239441376636026,
      "learning_rate": 9.195431646916084e-07,
      "loss": 0.0097,
      "step": 8185
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8634559005246802,
      "learning_rate": 9.189864672420864e-07,
      "loss": 0.1391,
      "step": 8186
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.445520110663022,
      "learning_rate": 9.184299004081507e-07,
      "loss": 0.394,
      "step": 8187
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.467099872067341,
      "learning_rate": 9.178734642357839e-07,
      "loss": 0.2738,
      "step": 8188
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.082846991474366,
      "learning_rate": 9.173171587709532e-07,
      "loss": 0.2919,
      "step": 8189
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0911113295134984,
      "learning_rate": 9.167609840596201e-07,
      "loss": 0.0755,
      "step": 8190
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.610148930353443,
      "learning_rate": 9.162049401477324e-07,
      "loss": 0.1038,
      "step": 8191
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.7994830356269373,
      "learning_rate": 9.156490270812268e-07,
      "loss": 0.1348,
      "step": 8192
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.570177472842538,
      "learning_rate": 9.150932449060326e-07,
      "loss": 0.4303,
      "step": 8193
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.6086293573757615,
      "learning_rate": 9.145375936680623e-07,
      "loss": 0.2287,
      "step": 8194
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.486655885534055,
      "learning_rate": 9.13982073413224e-07,
      "loss": 0.1119,
      "step": 8195
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.229548253730132,
      "learning_rate": 9.134266841874106e-07,
      "loss": 0.2146,
      "step": 8196
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.661541804641481,
      "learning_rate": 9.128714260365054e-07,
      "loss": 0.2175,
      "step": 8197
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.925349489660926,
      "learning_rate": 9.123162990063835e-07,
      "loss": 0.3031,
      "step": 8198
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.559574555674042,
      "learning_rate": 9.117613031429032e-07,
      "loss": 0.2126,
      "step": 8199
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.003905319692636,
      "learning_rate": 9.112064384919181e-07,
      "loss": 0.1838,
      "step": 8200
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.161457888509066,
      "learning_rate": 9.106517050992678e-07,
      "loss": 0.0802,
      "step": 8201
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.9071724020811294,
      "learning_rate": 9.100971030107802e-07,
      "loss": 0.177,
      "step": 8202
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.801708679218792,
      "learning_rate": 9.095426322722773e-07,
      "loss": 0.6894,
      "step": 8203
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.548021253508987,
      "learning_rate": 9.089882929295623e-07,
      "loss": 0.2574,
      "step": 8204
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.22942175127282,
      "learning_rate": 9.084340850284357e-07,
      "loss": 0.3294,
      "step": 8205
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.374323522939323,
      "learning_rate": 9.078800086146814e-07,
      "loss": 0.425,
      "step": 8206
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.234505770009634,
      "learning_rate": 9.073260637340744e-07,
      "loss": 0.2331,
      "step": 8207
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.533689346818544,
      "learning_rate": 9.067722504323814e-07,
      "loss": 0.3549,
      "step": 8208
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.6076884658040647,
      "learning_rate": 9.062185687553521e-07,
      "loss": 0.1774,
      "step": 8209
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8472608622674835,
      "learning_rate": 9.056650187487315e-07,
      "loss": 0.2448,
      "step": 8210
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.59233398628532,
      "learning_rate": 9.051116004582505e-07,
      "loss": 0.2354,
      "step": 8211
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.687946584205284,
      "learning_rate": 9.045583139296291e-07,
      "loss": 0.43,
      "step": 8212
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.690935998418776,
      "learning_rate": 9.040051592085791e-07,
      "loss": 0.4966,
      "step": 8213
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.1787931186023397,
      "learning_rate": 9.034521363407966e-07,
      "loss": 0.1449,
      "step": 8214
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.3249706035353257,
      "learning_rate": 9.028992453719721e-07,
      "loss": 0.2343,
      "step": 8215
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.297351801608528,
      "learning_rate": 9.023464863477816e-07,
      "loss": 0.4083,
      "step": 8216
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.455769510056125,
      "learning_rate": 9.017938593138909e-07,
      "loss": 0.1949,
      "step": 8217
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.2749747547184067,
      "learning_rate": 9.012413643159573e-07,
      "loss": 0.131,
      "step": 8218
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.773975887530052,
      "learning_rate": 9.006890013996226e-07,
      "loss": 0.3081,
      "step": 8219
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.016777945363405,
      "learning_rate": 9.001367706105213e-07,
      "loss": 0.2414,
      "step": 8220
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3296665423096274,
      "learning_rate": 8.995846719942783e-07,
      "loss": 0.1342,
      "step": 8221
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.157888494489389,
      "learning_rate": 8.990327055965012e-07,
      "loss": 0.1983,
      "step": 8222
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.201556564259509,
      "learning_rate": 8.984808714627949e-07,
      "loss": 0.371,
      "step": 8223
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.2666130580102495,
      "learning_rate": 8.979291696387452e-07,
      "loss": 0.3882,
      "step": 8224
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.176764872702422,
      "learning_rate": 8.973776001699331e-07,
      "loss": 0.2099,
      "step": 8225
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.973228707205303,
      "learning_rate": 8.968261631019282e-07,
      "loss": 0.3708,
      "step": 8226
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.623205263167102,
      "learning_rate": 8.962748584802842e-07,
      "loss": 0.4255,
      "step": 8227
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.772198762643623,
      "learning_rate": 8.9572368635055e-07,
      "loss": 0.4332,
      "step": 8228
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.180528915529125,
      "learning_rate": 8.951726467582581e-07,
      "loss": 0.397,
      "step": 8229
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.543024016645018,
      "learning_rate": 8.94621739748934e-07,
      "loss": 0.314,
      "step": 8230
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.735939626928947,
      "learning_rate": 8.940709653680926e-07,
      "loss": 0.5411,
      "step": 8231
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.5796412806508813,
      "learning_rate": 8.935203236612333e-07,
      "loss": 0.2429,
      "step": 8232
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.13040481018791,
      "learning_rate": 8.9296981467385e-07,
      "loss": 0.2634,
      "step": 8233
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.924415693892976,
      "learning_rate": 8.924194384514204e-07,
      "loss": 0.2453,
      "step": 8234
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.84003756451243,
      "learning_rate": 8.918691950394148e-07,
      "loss": 0.2076,
      "step": 8235
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.252628292745767,
      "learning_rate": 8.913190844832942e-07,
      "loss": 0.1396,
      "step": 8236
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.146736009784042,
      "learning_rate": 8.90769106828502e-07,
      "loss": 0.1724,
      "step": 8237
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.789129834870873,
      "learning_rate": 8.902192621204783e-07,
      "loss": 0.3256,
      "step": 8238
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.4768652106376186,
      "learning_rate": 8.896695504046451e-07,
      "loss": 0.1432,
      "step": 8239
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.6925427031968665,
      "learning_rate": 8.891199717264187e-07,
      "loss": 0.2827,
      "step": 8240
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1286251618262217,
      "learning_rate": 8.885705261312041e-07,
      "loss": 0.0136,
      "step": 8241
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.360690800631623,
      "learning_rate": 8.880212136643906e-07,
      "loss": 0.2336,
      "step": 8242
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.9936822464187625,
      "learning_rate": 8.87472034371363e-07,
      "loss": 0.3042,
      "step": 8243
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.846310264100746,
      "learning_rate": 8.869229882974883e-07,
      "loss": 0.1279,
      "step": 8244
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.925970645049326,
      "learning_rate": 8.863740754881281e-07,
      "loss": 0.42,
      "step": 8245
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.8076456593231605,
      "learning_rate": 8.858252959886321e-07,
      "loss": 0.3779,
      "step": 8246
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.837968273472637,
      "learning_rate": 8.852766498443346e-07,
      "loss": 0.2805,
      "step": 8247
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6203576947267198,
      "learning_rate": 8.847281371005648e-07,
      "loss": 0.0525,
      "step": 8248
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.466886171622072,
      "learning_rate": 8.841797578026368e-07,
      "loss": 0.2899,
      "step": 8249
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.493064414248489,
      "learning_rate": 8.836315119958544e-07,
      "loss": 0.2897,
      "step": 8250
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.1616263918609726,
      "learning_rate": 8.830833997255137e-07,
      "loss": 0.0983,
      "step": 8251
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.416808539674982,
      "learning_rate": 8.825354210368933e-07,
      "loss": 0.2574,
      "step": 8252
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.509069637867112,
      "learning_rate": 8.819875759752675e-07,
      "loss": 0.3122,
      "step": 8253
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.002415570765334,
      "learning_rate": 8.814398645858955e-07,
      "loss": 0.3309,
      "step": 8254
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.906153690029038,
      "learning_rate": 8.808922869140255e-07,
      "loss": 0.1599,
      "step": 8255
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.360381996504404,
      "learning_rate": 8.803448430048983e-07,
      "loss": 0.3964,
      "step": 8256
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.1602279244146283,
      "learning_rate": 8.797975329037379e-07,
      "loss": 0.2382,
      "step": 8257
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.722827994782918,
      "learning_rate": 8.792503566557628e-07,
      "loss": 0.2077,
      "step": 8258
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.873908743178297,
      "learning_rate": 8.787033143061768e-07,
      "loss": 0.1583,
      "step": 8259
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.9716640296022065,
      "learning_rate": 8.781564059001737e-07,
      "loss": 0.4737,
      "step": 8260
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.51958591619148,
      "learning_rate": 8.776096314829386e-07,
      "loss": 0.4027,
      "step": 8261
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.320474020096663,
      "learning_rate": 8.770629910996395e-07,
      "loss": 0.2564,
      "step": 8262
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.4602989742454198,
      "learning_rate": 8.765164847954405e-07,
      "loss": 0.2049,
      "step": 8263
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.8021315134065925,
      "learning_rate": 8.759701126154899e-07,
      "loss": 0.2642,
      "step": 8264
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.63605520303994,
      "learning_rate": 8.754238746049254e-07,
      "loss": 0.2468,
      "step": 8265
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.9868580034824563,
      "learning_rate": 8.748777708088774e-07,
      "loss": 0.3035,
      "step": 8266
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.383455398938201,
      "learning_rate": 8.743318012724586e-07,
      "loss": 0.3324,
      "step": 8267
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.32625426170507,
      "learning_rate": 8.737859660407771e-07,
      "loss": 0.4147,
      "step": 8268
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.552327509203145,
      "learning_rate": 8.732402651589262e-07,
      "loss": 0.2628,
      "step": 8269
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.605391155507953,
      "learning_rate": 8.72694698671988e-07,
      "loss": 0.5077,
      "step": 8270
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.07452338266364,
      "learning_rate": 8.721492666250375e-07,
      "loss": 0.1348,
      "step": 8271
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.041916246353107,
      "learning_rate": 8.716039690631317e-07,
      "loss": 0.1621,
      "step": 8272
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.476705390432785,
      "learning_rate": 8.710588060313236e-07,
      "loss": 0.3061,
      "step": 8273
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.250265088627467,
      "learning_rate": 8.705137775746503e-07,
      "loss": 0.2257,
      "step": 8274
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.143189527295354,
      "learning_rate": 8.699688837381389e-07,
      "loss": 0.4383,
      "step": 8275
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.8880562762808575,
      "learning_rate": 8.694241245668087e-07,
      "loss": 0.1814,
      "step": 8276
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.6363624575007294,
      "learning_rate": 8.688795001056613e-07,
      "loss": 0.3025,
      "step": 8277
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.063708589577131,
      "learning_rate": 8.683350103996935e-07,
      "loss": 0.1504,
      "step": 8278
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.986829003097995,
      "learning_rate": 8.677906554938873e-07,
      "loss": 0.3651,
      "step": 8279
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.6988168707622573,
      "learning_rate": 8.672464354332142e-07,
      "loss": 0.0707,
      "step": 8280
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.725129547335449,
      "learning_rate": 8.66702350262637e-07,
      "loss": 0.3158,
      "step": 8281
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.351792646029808,
      "learning_rate": 8.661584000271029e-07,
      "loss": 0.2909,
      "step": 8282
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.644726322220478,
      "learning_rate": 8.656145847715519e-07,
      "loss": 0.3176,
      "step": 8283
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.1730022312652046,
      "learning_rate": 8.650709045409111e-07,
      "loss": 0.1188,
      "step": 8284
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.5480386823265384,
      "learning_rate": 8.645273593800957e-07,
      "loss": 0.1788,
      "step": 8285
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.534553591188923,
      "learning_rate": 8.63983949334013e-07,
      "loss": 0.101,
      "step": 8286
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.2483077829484905,
      "learning_rate": 8.63440674447554e-07,
      "loss": 0.2712,
      "step": 8287
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7897052972349956,
      "learning_rate": 8.628975347656037e-07,
      "loss": 0.0477,
      "step": 8288
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.072814425857417,
      "learning_rate": 8.623545303330327e-07,
      "loss": 0.3072,
      "step": 8289
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.910646776763889,
      "learning_rate": 8.618116611947008e-07,
      "loss": 0.258,
      "step": 8290
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.435031086147484,
      "learning_rate": 8.612689273954594e-07,
      "loss": 0.3298,
      "step": 8291
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2240141676406595,
      "learning_rate": 8.607263289801435e-07,
      "loss": 0.0653,
      "step": 8292
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.7643526447867086,
      "learning_rate": 8.601838659935819e-07,
      "loss": 0.2457,
      "step": 8293
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.222731807778985,
      "learning_rate": 8.596415384805898e-07,
      "loss": 0.4633,
      "step": 8294
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.492956439588563,
      "learning_rate": 8.590993464859707e-07,
      "loss": 0.5271,
      "step": 8295
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9918639298449348,
      "learning_rate": 8.585572900545197e-07,
      "loss": 0.0779,
      "step": 8296
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.4074454107088696,
      "learning_rate": 8.580153692310175e-07,
      "loss": 0.1129,
      "step": 8297
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.648799342472218,
      "learning_rate": 8.574735840602355e-07,
      "loss": 0.2239,
      "step": 8298
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.046520855104987,
      "learning_rate": 8.569319345869332e-07,
      "loss": 0.468,
      "step": 8299
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.751823757752688,
      "learning_rate": 8.563904208558579e-07,
      "loss": 0.2788,
      "step": 8300
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.825464935319151,
      "learning_rate": 8.558490429117489e-07,
      "loss": 0.2607,
      "step": 8301
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.4177924783456257,
      "learning_rate": 8.553078007993309e-07,
      "loss": 0.1792,
      "step": 8302
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.6234184517057986,
      "learning_rate": 8.54766694563319e-07,
      "loss": 0.2209,
      "step": 8303
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.900600653633024,
      "learning_rate": 8.542257242484167e-07,
      "loss": 0.2425,
      "step": 8304
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.933370796314647,
      "learning_rate": 8.536848898993155e-07,
      "loss": 0.3623,
      "step": 8305
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.177233940150094,
      "learning_rate": 8.531441915606977e-07,
      "loss": 0.2158,
      "step": 8306
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3526486709297854,
      "learning_rate": 8.526036292772333e-07,
      "loss": 0.228,
      "step": 8307
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.28511258124776,
      "learning_rate": 8.520632030935799e-07,
      "loss": 0.1847,
      "step": 8308
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.581938084162937,
      "learning_rate": 8.515229130543854e-07,
      "loss": 0.3296,
      "step": 8309
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.23671653346768,
      "learning_rate": 8.509827592042847e-07,
      "loss": 0.3093,
      "step": 8310
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.1554589827849195,
      "learning_rate": 8.504427415879046e-07,
      "loss": 0.3794,
      "step": 8311
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.35469442171375,
      "learning_rate": 8.499028602498582e-07,
      "loss": 0.3611,
      "step": 8312
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.711581100844185,
      "learning_rate": 8.493631152347472e-07,
      "loss": 0.3608,
      "step": 8313
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.410138196666679,
      "learning_rate": 8.488235065871631e-07,
      "loss": 0.3488,
      "step": 8314
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.807429376380944,
      "learning_rate": 8.482840343516849e-07,
      "loss": 0.4594,
      "step": 8315
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.973613859600531,
      "learning_rate": 8.477446985728824e-07,
      "loss": 0.1963,
      "step": 8316
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.093728982004148,
      "learning_rate": 8.472054992953124e-07,
      "loss": 0.4968,
      "step": 8317
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.511107773838374,
      "learning_rate": 8.466664365635211e-07,
      "loss": 0.5903,
      "step": 8318
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.563455032611807,
      "learning_rate": 8.461275104220426e-07,
      "loss": 0.2426,
      "step": 8319
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.11737825743633,
      "learning_rate": 8.455887209153999e-07,
      "loss": 0.2056,
      "step": 8320
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.13059174284731,
      "learning_rate": 8.450500680881069e-07,
      "loss": 0.3383,
      "step": 8321
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.560823266848485,
      "learning_rate": 8.445115519846636e-07,
      "loss": 0.3273,
      "step": 8322
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.121485864558816,
      "learning_rate": 8.439731726495593e-07,
      "loss": 0.1616,
      "step": 8323
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.8201899809307793,
      "learning_rate": 8.434349301272723e-07,
      "loss": 0.2475,
      "step": 8324
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.4100163889088133,
      "learning_rate": 8.42896824462269e-07,
      "loss": 0.1522,
      "step": 8325
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2460215121107,
      "learning_rate": 8.423588556990064e-07,
      "loss": 0.1012,
      "step": 8326
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.96119226292323,
      "learning_rate": 8.418210238819285e-07,
      "loss": 0.3102,
      "step": 8327
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.725129020441395,
      "learning_rate": 8.412833290554675e-07,
      "loss": 0.3322,
      "step": 8328
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.158701646861289,
      "learning_rate": 8.40745771264046e-07,
      "loss": 0.3547,
      "step": 8329
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.156435657745404,
      "learning_rate": 8.40208350552073e-07,
      "loss": 0.2511,
      "step": 8330
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.974405092262252,
      "learning_rate": 8.396710669639496e-07,
      "loss": 0.153,
      "step": 8331
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.653413109228692,
      "learning_rate": 8.391339205440624e-07,
      "loss": 0.1718,
      "step": 8332
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.4162768184607613,
      "learning_rate": 8.385969113367878e-07,
      "loss": 0.2433,
      "step": 8333
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.689219152699594,
      "learning_rate": 8.38060039386491e-07,
      "loss": 0.1128,
      "step": 8334
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.982098648171531,
      "learning_rate": 8.375233047375248e-07,
      "loss": 0.3692,
      "step": 8335
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.274350542878432,
      "learning_rate": 8.369867074342333e-07,
      "loss": 0.3663,
      "step": 8336
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1671724047610605,
      "learning_rate": 8.364502475209469e-07,
      "loss": 0.0989,
      "step": 8337
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.906202410418922,
      "learning_rate": 8.359139250419848e-07,
      "loss": 0.2789,
      "step": 8338
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.552884315851515,
      "learning_rate": 8.353777400416552e-07,
      "loss": 0.0517,
      "step": 8339
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.092524352124238,
      "learning_rate": 8.348416925642561e-07,
      "loss": 0.3757,
      "step": 8340
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.9066429547581794,
      "learning_rate": 8.343057826540724e-07,
      "loss": 0.258,
      "step": 8341
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.214870040926172,
      "learning_rate": 8.337700103553786e-07,
      "loss": 0.2135,
      "step": 8342
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.4230009571653595,
      "learning_rate": 8.332343757124375e-07,
      "loss": 0.1588,
      "step": 8343
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.096170290431282,
      "learning_rate": 8.326988787694998e-07,
      "loss": 0.427,
      "step": 8344
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.917628209323258,
      "learning_rate": 8.321635195708072e-07,
      "loss": 0.188,
      "step": 8345
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.662992247366623,
      "learning_rate": 8.316282981605875e-07,
      "loss": 0.2679,
      "step": 8346
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.524638282431988,
      "learning_rate": 8.310932145830585e-07,
      "loss": 0.0387,
      "step": 8347
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.912426030390192,
      "learning_rate": 8.305582688824256e-07,
      "loss": 0.2581,
      "step": 8348
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.8015833578585663,
      "learning_rate": 8.300234611028831e-07,
      "loss": 0.1897,
      "step": 8349
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.033765944810466,
      "learning_rate": 8.294887912886154e-07,
      "loss": 0.2931,
      "step": 8350
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.833601336617702,
      "learning_rate": 8.28954259483794e-07,
      "loss": 0.3762,
      "step": 8351
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.343444135993983,
      "learning_rate": 8.284198657325787e-07,
      "loss": 0.5176,
      "step": 8352
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.47045128664034,
      "learning_rate": 8.278856100791189e-07,
      "loss": 0.0338,
      "step": 8353
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.889434218836082,
      "learning_rate": 8.273514925675513e-07,
      "loss": 0.2396,
      "step": 8354
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.036294604416282,
      "learning_rate": 8.268175132420034e-07,
      "loss": 0.1876,
      "step": 8355
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.109725812802697,
      "learning_rate": 8.262836721465897e-07,
      "loss": 0.1008,
      "step": 8356
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.828179668827328,
      "learning_rate": 8.257499693254129e-07,
      "loss": 0.3464,
      "step": 8357
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.9259641343307603,
      "learning_rate": 8.252164048225655e-07,
      "loss": 0.2531,
      "step": 8358
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.969367211382056,
      "learning_rate": 8.246829786821267e-07,
      "loss": 0.5146,
      "step": 8359
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.1196492231952515,
      "learning_rate": 8.241496909481675e-07,
      "loss": 0.2973,
      "step": 8360
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.148576715187873,
      "learning_rate": 8.236165416647446e-07,
      "loss": 0.228,
      "step": 8361
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.951165359286665,
      "learning_rate": 8.230835308759041e-07,
      "loss": 0.182,
      "step": 8362
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.536414555860597,
      "learning_rate": 8.225506586256807e-07,
      "loss": 0.4134,
      "step": 8363
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.42098523994858,
      "learning_rate": 8.220179249580973e-07,
      "loss": 0.4479,
      "step": 8364
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.235533176307316,
      "learning_rate": 8.214853299171669e-07,
      "loss": 0.1442,
      "step": 8365
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.168729167092414,
      "learning_rate": 8.209528735468891e-07,
      "loss": 0.0754,
      "step": 8366
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.188386965658622,
      "learning_rate": 8.204205558912529e-07,
      "loss": 0.2832,
      "step": 8367
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3899555148880136,
      "learning_rate": 8.198883769942359e-07,
      "loss": 0.1712,
      "step": 8368
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.9900164026106957,
      "learning_rate": 8.193563368998034e-07,
      "loss": 0.3535,
      "step": 8369
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.064827560769633,
      "learning_rate": 8.188244356519112e-07,
      "loss": 0.2517,
      "step": 8370
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.566888573367596,
      "learning_rate": 8.182926732945021e-07,
      "loss": 0.1988,
      "step": 8371
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.0737926997355802,
      "learning_rate": 8.177610498715072e-07,
      "loss": 0.1875,
      "step": 8372
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.616663160153066,
      "learning_rate": 8.172295654268467e-07,
      "loss": 0.13,
      "step": 8373
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.630876075664685,
      "learning_rate": 8.166982200044284e-07,
      "loss": 0.2682,
      "step": 8374
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.468742517300733,
      "learning_rate": 8.161670136481514e-07,
      "loss": 0.377,
      "step": 8375
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.7851040086964227,
      "learning_rate": 8.156359464019003e-07,
      "loss": 0.3524,
      "step": 8376
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.117962484157866,
      "learning_rate": 8.151050183095485e-07,
      "loss": 0.2786,
      "step": 8377
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.155069798035007,
      "learning_rate": 8.145742294149611e-07,
      "loss": 0.356,
      "step": 8378
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.3949449426150617,
      "learning_rate": 8.140435797619858e-07,
      "loss": 0.1967,
      "step": 8379
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.18048523099495972,
      "learning_rate": 8.135130693944654e-07,
      "loss": 0.0016,
      "step": 8380
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.7719771201527816,
      "learning_rate": 8.129826983562267e-07,
      "loss": 0.2936,
      "step": 8381
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.564568546172988,
      "learning_rate": 8.124524666910855e-07,
      "loss": 0.113,
      "step": 8382
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.755843043760173,
      "learning_rate": 8.119223744428495e-07,
      "loss": 0.1384,
      "step": 8383
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.5409411369312247,
      "learning_rate": 8.113924216553093e-07,
      "loss": 0.1294,
      "step": 8384
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.045402158541196,
      "learning_rate": 8.10862608372249e-07,
      "loss": 0.2614,
      "step": 8385
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.683306254404071,
      "learning_rate": 8.103329346374386e-07,
      "loss": 0.4829,
      "step": 8386
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.005060426225258,
      "learning_rate": 8.098034004946364e-07,
      "loss": 0.5456,
      "step": 8387
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.6097142736554573,
      "learning_rate": 8.092740059875925e-07,
      "loss": 0.2495,
      "step": 8388
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.309320664444276,
      "learning_rate": 8.087447511600394e-07,
      "loss": 0.1229,
      "step": 8389
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.754690273773429,
      "learning_rate": 8.082156360557037e-07,
      "loss": 0.1054,
      "step": 8390
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.3235847461498285,
      "learning_rate": 8.076866607182979e-07,
      "loss": 0.3949,
      "step": 8391
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.882295716772494,
      "learning_rate": 8.071578251915224e-07,
      "loss": 0.4867,
      "step": 8392
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.302511945551736,
      "learning_rate": 8.066291295190696e-07,
      "loss": 0.0609,
      "step": 8393
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.0001557774775565,
      "learning_rate": 8.061005737446142e-07,
      "loss": 0.21,
      "step": 8394
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.391997598265025,
      "learning_rate": 8.055721579118255e-07,
      "loss": 0.095,
      "step": 8395
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9660242673800665,
      "learning_rate": 8.050438820643577e-07,
      "loss": 0.2055,
      "step": 8396
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.229626553233635,
      "learning_rate": 8.045157462458539e-07,
      "loss": 0.3789,
      "step": 8397
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5946085003606707,
      "learning_rate": 8.039877504999479e-07,
      "loss": 0.1824,
      "step": 8398
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.974083241345115,
      "learning_rate": 8.034598948702574e-07,
      "loss": 0.3398,
      "step": 8399
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.0569228986662536,
      "learning_rate": 8.029321794003936e-07,
      "loss": 0.2325,
      "step": 8400
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.190467626239603,
      "learning_rate": 8.024046041339529e-07,
      "loss": 0.1812,
      "step": 8401
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.093469060434447,
      "learning_rate": 8.0187716911452e-07,
      "loss": 0.334,
      "step": 8402
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.858731502020515,
      "learning_rate": 8.013498743856715e-07,
      "loss": 0.1855,
      "step": 8403
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.19653675123791,
      "learning_rate": 8.008227199909669e-07,
      "loss": 0.3292,
      "step": 8404
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.407807102649598,
      "learning_rate": 8.002957059739591e-07,
      "loss": 0.495,
      "step": 8405
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.523925332753122,
      "learning_rate": 7.997688323781872e-07,
      "loss": 0.4065,
      "step": 8406
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.814737536904159,
      "learning_rate": 7.992420992471775e-07,
      "loss": 0.318,
      "step": 8407
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.820675731372945,
      "learning_rate": 7.98715506624449e-07,
      "loss": 0.2066,
      "step": 8408
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.028569557268173,
      "learning_rate": 7.981890545535026e-07,
      "loss": 0.424,
      "step": 8409
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.7417967546161393,
      "learning_rate": 7.976627430778339e-07,
      "loss": 0.245,
      "step": 8410
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.7542327963469595,
      "learning_rate": 7.971365722409232e-07,
      "loss": 0.2167,
      "step": 8411
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.467355094897801,
      "learning_rate": 7.966105420862394e-07,
      "loss": 0.4668,
      "step": 8412
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36838270792055583,
      "learning_rate": 7.96084652657243e-07,
      "loss": 0.0033,
      "step": 8413
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.316450536029803,
      "learning_rate": 7.955589039973771e-07,
      "loss": 0.2407,
      "step": 8414
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.579439008913187,
      "learning_rate": 7.950332961500792e-07,
      "loss": 0.219,
      "step": 8415
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.362257108320335,
      "learning_rate": 7.94507829158771e-07,
      "loss": 0.1973,
      "step": 8416
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.3147093471861675,
      "learning_rate": 7.939825030668638e-07,
      "loss": 0.4104,
      "step": 8417
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.911183641174631,
      "learning_rate": 7.934573179177596e-07,
      "loss": 0.235,
      "step": 8418
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.465485459033177,
      "learning_rate": 7.929322737548434e-07,
      "loss": 0.2614,
      "step": 8419
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.4416295200226483,
      "learning_rate": 7.924073706214944e-07,
      "loss": 0.2599,
      "step": 8420
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.5077154568262916,
      "learning_rate": 7.918826085610762e-07,
      "loss": 0.1187,
      "step": 8421
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5048476355783373,
      "learning_rate": 7.91357987616942e-07,
      "loss": 0.1961,
      "step": 8422
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.6780668062131006,
      "learning_rate": 7.908335078324353e-07,
      "loss": 0.2491,
      "step": 8423
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.5816378567695337,
      "learning_rate": 7.90309169250883e-07,
      "loss": 0.259,
      "step": 8424
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.496444224666744,
      "learning_rate": 7.897849719156051e-07,
      "loss": 0.433,
      "step": 8425
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.6267041723049687,
      "learning_rate": 7.892609158699094e-07,
      "loss": 0.3239,
      "step": 8426
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.504848535671475,
      "learning_rate": 7.88737001157088e-07,
      "loss": 0.3133,
      "step": 8427
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.231462661451903,
      "learning_rate": 7.882132278204274e-07,
      "loss": 0.5575,
      "step": 8428
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.427201172167387,
      "learning_rate": 7.87689595903196e-07,
      "loss": 0.1656,
      "step": 8429
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.157530390680719,
      "learning_rate": 7.871661054486549e-07,
      "loss": 0.1717,
      "step": 8430
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.3379523824208475,
      "learning_rate": 7.866427565000545e-07,
      "loss": 0.2295,
      "step": 8431
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.881039327190542,
      "learning_rate": 7.861195491006277e-07,
      "loss": 0.2572,
      "step": 8432
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.7069299543815912,
      "learning_rate": 7.855964832936025e-07,
      "loss": 0.1802,
      "step": 8433
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.121390114608878,
      "learning_rate": 7.850735591221892e-07,
      "loss": 0.3504,
      "step": 8434
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.221089658184216,
      "learning_rate": 7.845507766295904e-07,
      "loss": 0.203,
      "step": 8435
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.619376117994914,
      "learning_rate": 7.840281358589976e-07,
      "loss": 0.4206,
      "step": 8436
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.640077259128154,
      "learning_rate": 7.835056368535859e-07,
      "loss": 0.2765,
      "step": 8437
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.842245484609257,
      "learning_rate": 7.829832796565245e-07,
      "loss": 0.2093,
      "step": 8438
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.9657842134319283,
      "learning_rate": 7.824610643109648e-07,
      "loss": 0.2306,
      "step": 8439
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.433185749643536,
      "learning_rate": 7.819389908600514e-07,
      "loss": 0.4859,
      "step": 8440
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.777281855775764,
      "learning_rate": 7.814170593469169e-07,
      "loss": 0.3619,
      "step": 8441
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9430678465162528,
      "learning_rate": 7.808952698146777e-07,
      "loss": 0.3457,
      "step": 8442
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.0370366318902455,
      "learning_rate": 7.803736223064446e-07,
      "loss": 0.1992,
      "step": 8443
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.667437807801282,
      "learning_rate": 7.7985211686531e-07,
      "loss": 0.3594,
      "step": 8444
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.831513161649365,
      "learning_rate": 7.793307535343605e-07,
      "loss": 0.4129,
      "step": 8445
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.156934094333802,
      "learning_rate": 7.788095323566697e-07,
      "loss": 0.3648,
      "step": 8446
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.227989993708689,
      "learning_rate": 7.78288453375295e-07,
      "loss": 0.3867,
      "step": 8447
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.125336432706089,
      "learning_rate": 7.777675166332887e-07,
      "loss": 0.1967,
      "step": 8448
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.672778112166982,
      "learning_rate": 7.772467221736854e-07,
      "loss": 0.3035,
      "step": 8449
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.438096362118105,
      "learning_rate": 7.767260700395113e-07,
      "loss": 0.2045,
      "step": 8450
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.020653650490848,
      "learning_rate": 7.762055602737823e-07,
      "loss": 0.1071,
      "step": 8451
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.3708356180079213,
      "learning_rate": 7.756851929194969e-07,
      "loss": 0.2562,
      "step": 8452
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.91860961327856,
      "learning_rate": 7.75164968019648e-07,
      "loss": 0.1471,
      "step": 8453
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8916208981693776,
      "learning_rate": 7.746448856172128e-07,
      "loss": 0.3549,
      "step": 8454
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.252016683230652,
      "learning_rate": 7.741249457551575e-07,
      "loss": 0.3941,
      "step": 8455
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.199387979744338,
      "learning_rate": 7.736051484764392e-07,
      "loss": 0.2965,
      "step": 8456
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.767919263433525,
      "learning_rate": 7.730854938239978e-07,
      "loss": 0.2525,
      "step": 8457
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9318300461391984,
      "learning_rate": 7.725659818407671e-07,
      "loss": 0.1561,
      "step": 8458
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.898403512993441,
      "learning_rate": 7.720466125696661e-07,
      "loss": 0.481,
      "step": 8459
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.92572305171525,
      "learning_rate": 7.715273860536013e-07,
      "loss": 0.2984,
      "step": 8460
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.728316128714272,
      "learning_rate": 7.710083023354712e-07,
      "loss": 0.4158,
      "step": 8461
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.1561713116841945,
      "learning_rate": 7.704893614581569e-07,
      "loss": 0.5263,
      "step": 8462
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2377012459293266,
      "learning_rate": 7.69970563464533e-07,
      "loss": 0.0597,
      "step": 8463
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.539078285957437,
      "learning_rate": 7.694519083974591e-07,
      "loss": 0.2007,
      "step": 8464
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8303187912889087,
      "learning_rate": 7.689333962997834e-07,
      "loss": 0.3269,
      "step": 8465
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.391951522012367,
      "learning_rate": 7.684150272143451e-07,
      "loss": 0.4429,
      "step": 8466
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.110458587724685,
      "learning_rate": 7.678968011839663e-07,
      "loss": 0.4707,
      "step": 8467
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.64122955925108,
      "learning_rate": 7.673787182514622e-07,
      "loss": 0.4821,
      "step": 8468
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.277846828215742,
      "learning_rate": 7.668607784596341e-07,
      "loss": 0.3902,
      "step": 8469
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.6522116935430318,
      "learning_rate": 7.663429818512705e-07,
      "loss": 0.224,
      "step": 8470
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.176880918547782,
      "learning_rate": 7.658253284691517e-07,
      "loss": 0.3775,
      "step": 8471
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.5829444099885754,
      "learning_rate": 7.653078183560405e-07,
      "loss": 0.1301,
      "step": 8472
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.176061866254142,
      "learning_rate": 7.647904515546933e-07,
      "loss": 0.2314,
      "step": 8473
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.9345950603602398,
      "learning_rate": 7.642732281078519e-07,
      "loss": 0.2745,
      "step": 8474
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.182734897028311,
      "learning_rate": 7.637561480582456e-07,
      "loss": 0.36,
      "step": 8475
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.198174128831156,
      "learning_rate": 7.632392114485956e-07,
      "loss": 0.3467,
      "step": 8476
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.126106554428866,
      "learning_rate": 7.627224183216056e-07,
      "loss": 0.2314,
      "step": 8477
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.620338093491288,
      "learning_rate": 7.622057687199727e-07,
      "loss": 0.3789,
      "step": 8478
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.335596992487283,
      "learning_rate": 7.616892626863792e-07,
      "loss": 0.3459,
      "step": 8479
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.377640749477141,
      "learning_rate": 7.611729002634954e-07,
      "loss": 0.2645,
      "step": 8480
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.758026603792308,
      "learning_rate": 7.606566814939836e-07,
      "loss": 0.4347,
      "step": 8481
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8641435823409163,
      "learning_rate": 7.601406064204872e-07,
      "loss": 0.2442,
      "step": 8482
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.131381274423226,
      "learning_rate": 7.596246750856448e-07,
      "loss": 0.3783,
      "step": 8483
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8009223182775274,
      "learning_rate": 7.591088875320793e-07,
      "loss": 0.2952,
      "step": 8484
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.346482354624301,
      "learning_rate": 7.585932438024013e-07,
      "loss": 0.2391,
      "step": 8485
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.14192479840319,
      "learning_rate": 7.580777439392137e-07,
      "loss": 0.3374,
      "step": 8486
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.816427662083949,
      "learning_rate": 7.57562387985101e-07,
      "loss": 0.2881,
      "step": 8487
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8300826409454336,
      "learning_rate": 7.57047175982642e-07,
      "loss": 0.304,
      "step": 8488
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.178251055256849,
      "learning_rate": 7.565321079743998e-07,
      "loss": 0.4817,
      "step": 8489
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.122420308179506,
      "learning_rate": 7.560171840029265e-07,
      "loss": 0.3647,
      "step": 8490
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.460771507832203,
      "learning_rate": 7.555024041107647e-07,
      "loss": 0.1016,
      "step": 8491
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.8562579909838903,
      "learning_rate": 7.5498776834044e-07,
      "loss": 0.0742,
      "step": 8492
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.3761883710360285,
      "learning_rate": 7.544732767344711e-07,
      "loss": 0.3703,
      "step": 8493
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.463643991727948,
      "learning_rate": 7.539589293353625e-07,
      "loss": 0.4982,
      "step": 8494
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.6389108837557105,
      "learning_rate": 7.534447261856057e-07,
      "loss": 0.3993,
      "step": 8495
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.05160505563199,
      "learning_rate": 7.529306673276845e-07,
      "loss": 0.242,
      "step": 8496
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.008032998164274,
      "learning_rate": 7.524167528040643e-07,
      "loss": 0.3177,
      "step": 8497
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.034470146129082,
      "learning_rate": 7.519029826572052e-07,
      "loss": 0.4257,
      "step": 8498
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.6622795112518975,
      "learning_rate": 7.513893569295511e-07,
      "loss": 0.4817,
      "step": 8499
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.191789709663113,
      "learning_rate": 7.508758756635345e-07,
      "loss": 0.5192,
      "step": 8500
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0425352848274057,
      "learning_rate": 7.503625389015784e-07,
      "loss": 0.0852,
      "step": 8501
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.620065386162318,
      "learning_rate": 7.498493466860913e-07,
      "loss": 0.27,
      "step": 8502
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.414204608864122,
      "learning_rate": 7.493362990594704e-07,
      "loss": 0.3415,
      "step": 8503
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.234178595979735,
      "learning_rate": 7.488233960641017e-07,
      "loss": 0.1514,
      "step": 8504
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.3463731627152606,
      "learning_rate": 7.483106377423577e-07,
      "loss": 0.2549,
      "step": 8505
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.255661411359608,
      "learning_rate": 7.477980241366014e-07,
      "loss": 0.2887,
      "step": 8506
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4370690537889195,
      "learning_rate": 7.47285555289182e-07,
      "loss": 0.0404,
      "step": 8507
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.015547712404416,
      "learning_rate": 7.467732312424367e-07,
      "loss": 0.3723,
      "step": 8508
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.109815879254008,
      "learning_rate": 7.462610520386915e-07,
      "loss": 0.6315,
      "step": 8509
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.215850579115621,
      "learning_rate": 7.457490177202592e-07,
      "loss": 0.2366,
      "step": 8510
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.156055792246166,
      "learning_rate": 7.452371283294432e-07,
      "loss": 0.4067,
      "step": 8511
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.609000550407348,
      "learning_rate": 7.447253839085327e-07,
      "loss": 0.2661,
      "step": 8512
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.354124336391645,
      "learning_rate": 7.442137844998052e-07,
      "loss": 0.2404,
      "step": 8513
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2018557483278856,
      "learning_rate": 7.437023301455268e-07,
      "loss": 0.1489,
      "step": 8514
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.856254860715106,
      "learning_rate": 7.431910208879503e-07,
      "loss": 0.1838,
      "step": 8515
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6644840135794094,
      "learning_rate": 7.426798567693191e-07,
      "loss": 0.121,
      "step": 8516
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.219597854168133,
      "learning_rate": 7.421688378318628e-07,
      "loss": 0.5029,
      "step": 8517
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.40602250385586,
      "learning_rate": 7.416579641177987e-07,
      "loss": 0.2644,
      "step": 8518
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.519623419190251,
      "learning_rate": 7.411472356693331e-07,
      "loss": 0.3702,
      "step": 8519
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.306129649095505,
      "learning_rate": 7.406366525286587e-07,
      "loss": 0.5195,
      "step": 8520
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.721392137547213,
      "learning_rate": 7.401262147379593e-07,
      "loss": 0.3563,
      "step": 8521
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.35481534768445,
      "learning_rate": 7.396159223394039e-07,
      "loss": 0.1896,
      "step": 8522
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.993945922602024,
      "learning_rate": 7.391057753751502e-07,
      "loss": 0.3291,
      "step": 8523
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.8789197374218225,
      "learning_rate": 7.385957738873442e-07,
      "loss": 0.2687,
      "step": 8524
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.211346759379766,
      "learning_rate": 7.380859179181188e-07,
      "loss": 0.2851,
      "step": 8525
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.153024409568016,
      "learning_rate": 7.375762075095974e-07,
      "loss": 0.4108,
      "step": 8526
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.315373283997846,
      "learning_rate": 7.370666427038892e-07,
      "loss": 0.2475,
      "step": 8527
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6516646175337746,
      "learning_rate": 7.365572235430918e-07,
      "loss": 0.1909,
      "step": 8528
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.5298487509025414,
      "learning_rate": 7.360479500692908e-07,
      "loss": 0.2561,
      "step": 8529
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.2229137336824465,
      "learning_rate": 7.355388223245591e-07,
      "loss": 0.4636,
      "step": 8530
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8891144184451525,
      "learning_rate": 7.3502984035096e-07,
      "loss": 0.0813,
      "step": 8531
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.560232854435049,
      "learning_rate": 7.345210041905424e-07,
      "loss": 0.3808,
      "step": 8532
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.203263917043891,
      "learning_rate": 7.340123138853439e-07,
      "loss": 0.2571,
      "step": 8533
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.572632863296835,
      "learning_rate": 7.335037694773897e-07,
      "loss": 0.177,
      "step": 8534
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.045873060024455,
      "learning_rate": 7.329953710086926e-07,
      "loss": 0.3045,
      "step": 8535
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.102400827845679,
      "learning_rate": 7.324871185212556e-07,
      "loss": 0.2672,
      "step": 8536
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.7822355191130033,
      "learning_rate": 7.319790120570672e-07,
      "loss": 0.2107,
      "step": 8537
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.156721894673782,
      "learning_rate": 7.314710516581047e-07,
      "loss": 0.5153,
      "step": 8538
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.042248614397912,
      "learning_rate": 7.309632373663326e-07,
      "loss": 0.1768,
      "step": 8539
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.630277433862052,
      "learning_rate": 7.304555692237053e-07,
      "loss": 0.248,
      "step": 8540
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.935065406437444,
      "learning_rate": 7.299480472721631e-07,
      "loss": 0.2304,
      "step": 8541
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.298427990301134,
      "learning_rate": 7.294406715536357e-07,
      "loss": 0.3312,
      "step": 8542
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.149052893824327,
      "learning_rate": 7.289334421100389e-07,
      "loss": 0.2205,
      "step": 8543
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.846346167311534,
      "learning_rate": 7.284263589832777e-07,
      "loss": 0.5338,
      "step": 8544
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.6497806105530115,
      "learning_rate": 7.279194222152456e-07,
      "loss": 0.3924,
      "step": 8545
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.3977062116863825,
      "learning_rate": 7.274126318478228e-07,
      "loss": 0.2274,
      "step": 8546
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.368750594212543,
      "learning_rate": 7.26905987922878e-07,
      "loss": 0.0933,
      "step": 8547
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.213914739256377,
      "learning_rate": 7.263994904822674e-07,
      "loss": 0.2277,
      "step": 8548
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.944905123564343,
      "learning_rate": 7.258931395678348e-07,
      "loss": 0.2292,
      "step": 8549
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.44336956236419,
      "learning_rate": 7.253869352214135e-07,
      "loss": 0.1163,
      "step": 8550
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.799717302346683,
      "learning_rate": 7.248808774848235e-07,
      "loss": 0.3318,
      "step": 8551
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.196267916470503,
      "learning_rate": 7.24374966399872e-07,
      "loss": 0.2594,
      "step": 8552
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.654000628380239,
      "learning_rate": 7.238692020083557e-07,
      "loss": 0.2901,
      "step": 8553
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.4643955280657415,
      "learning_rate": 7.233635843520572e-07,
      "loss": 0.267,
      "step": 8554
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.8337976154024607,
      "learning_rate": 7.228581134727497e-07,
      "loss": 0.2814,
      "step": 8555
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.367262995852642,
      "learning_rate": 7.223527894121921e-07,
      "loss": 0.2638,
      "step": 8556
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.448710591797555,
      "learning_rate": 7.218476122121315e-07,
      "loss": 0.2633,
      "step": 8557
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.868754919696608,
      "learning_rate": 7.213425819143035e-07,
      "loss": 0.2212,
      "step": 8558
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.159693890312696,
      "learning_rate": 7.208376985604304e-07,
      "loss": 0.2461,
      "step": 8559
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.832867554621343,
      "learning_rate": 7.203329621922245e-07,
      "loss": 0.2865,
      "step": 8560
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.1227538269162003,
      "learning_rate": 7.198283728513841e-07,
      "loss": 0.2405,
      "step": 8561
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.555277439158672,
      "learning_rate": 7.19323930579596e-07,
      "loss": 0.1353,
      "step": 8562
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.207358007926154,
      "learning_rate": 7.188196354185342e-07,
      "loss": 0.2485,
      "step": 8563
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6066076753979206,
      "learning_rate": 7.183154874098611e-07,
      "loss": 0.2998,
      "step": 8564
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.782407313227208,
      "learning_rate": 7.178114865952276e-07,
      "loss": 0.3605,
      "step": 8565
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.692026700307207,
      "learning_rate": 7.173076330162718e-07,
      "loss": 0.1358,
      "step": 8566
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.212769432059856,
      "learning_rate": 7.168039267146193e-07,
      "loss": 0.289,
      "step": 8567
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6889115778124255,
      "learning_rate": 7.163003677318839e-07,
      "loss": 0.2788,
      "step": 8568
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9325282085783495,
      "learning_rate": 7.157969561096662e-07,
      "loss": 0.0743,
      "step": 8569
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.2029257668756514,
      "learning_rate": 7.152936918895572e-07,
      "loss": 0.3469,
      "step": 8570
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.2716915198151773,
      "learning_rate": 7.147905751131335e-07,
      "loss": 0.1657,
      "step": 8571
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.5597751946106335,
      "learning_rate": 7.142876058219599e-07,
      "loss": 0.3653,
      "step": 8572
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.5236463933343045,
      "learning_rate": 7.137847840575898e-07,
      "loss": 0.2386,
      "step": 8573
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.135069697592147,
      "learning_rate": 7.132821098615624e-07,
      "loss": 0.2362,
      "step": 8574
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.420734327278887,
      "learning_rate": 7.127795832754083e-07,
      "loss": 0.219,
      "step": 8575
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.832196633188697,
      "learning_rate": 7.122772043406428e-07,
      "loss": 0.2993,
      "step": 8576
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.177522050862716,
      "learning_rate": 7.117749730987689e-07,
      "loss": 0.1055,
      "step": 8577
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.236901984113754,
      "learning_rate": 7.112728895912813e-07,
      "loss": 0.4521,
      "step": 8578
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.3436608789112965,
      "learning_rate": 7.107709538596563e-07,
      "loss": 0.2153,
      "step": 8579
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.245018940223993,
      "learning_rate": 7.102691659453637e-07,
      "loss": 0.2056,
      "step": 8580
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.776328328482643,
      "learning_rate": 7.097675258898582e-07,
      "loss": 0.318,
      "step": 8581
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.814086948681035,
      "learning_rate": 7.092660337345817e-07,
      "loss": 0.1725,
      "step": 8582
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.408426943914674,
      "learning_rate": 7.087646895209674e-07,
      "loss": 0.332,
      "step": 8583
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.900363867464133,
      "learning_rate": 7.082634932904312e-07,
      "loss": 0.1983,
      "step": 8584
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.078695808441051,
      "learning_rate": 7.077624450843815e-07,
      "loss": 0.2388,
      "step": 8585
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.679958087049836,
      "learning_rate": 7.072615449442116e-07,
      "loss": 0.3075,
      "step": 8586
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.206939151985605,
      "learning_rate": 7.06760792911303e-07,
      "loss": 0.3888,
      "step": 8587
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.334764684587142,
      "learning_rate": 7.062601890270274e-07,
      "loss": 0.1152,
      "step": 8588
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.968823897346262,
      "learning_rate": 7.05759733332739e-07,
      "loss": 0.2857,
      "step": 8589
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7651926394062247,
      "learning_rate": 7.052594258697857e-07,
      "loss": 0.1277,
      "step": 8590
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.5853560916908807,
      "learning_rate": 7.047592666794994e-07,
      "loss": 0.1484,
      "step": 8591
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8821247007113375,
      "learning_rate": 7.042592558032002e-07,
      "loss": 0.1485,
      "step": 8592
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.4148538607958097,
      "learning_rate": 7.037593932821987e-07,
      "loss": 0.2776,
      "step": 8593
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.5293336509413473,
      "learning_rate": 7.032596791577881e-07,
      "loss": 0.3039,
      "step": 8594
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3497768022842513,
      "learning_rate": 7.027601134712547e-07,
      "loss": 0.08,
      "step": 8595
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.097088058072214,
      "learning_rate": 7.022606962638693e-07,
      "loss": 0.2271,
      "step": 8596
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.394688907871998,
      "learning_rate": 7.017614275768906e-07,
      "loss": 0.2372,
      "step": 8597
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.152683619322914,
      "learning_rate": 7.012623074515681e-07,
      "loss": 0.2627,
      "step": 8598
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.248396684664014,
      "learning_rate": 7.007633359291332e-07,
      "loss": 0.3348,
      "step": 8599
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.554852222750421,
      "learning_rate": 7.002645130508112e-07,
      "loss": 0.2629,
      "step": 8600
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.9992952932313885,
      "learning_rate": 6.997658388578116e-07,
      "loss": 0.4989,
      "step": 8601
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.930000689108601,
      "learning_rate": 6.992673133913316e-07,
      "loss": 0.3421,
      "step": 8602
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.615154645813675,
      "learning_rate": 6.98768936692559e-07,
      "loss": 0.3007,
      "step": 8603
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.267601364623423,
      "learning_rate": 6.982707088026644e-07,
      "loss": 0.2954,
      "step": 8604
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.986535480166055,
      "learning_rate": 6.977726297628112e-07,
      "loss": 0.2845,
      "step": 8605
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.986393909003149,
      "learning_rate": 6.972746996141477e-07,
      "loss": 0.27,
      "step": 8606
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.0460730584875675,
      "learning_rate": 6.967769183978093e-07,
      "loss": 0.311,
      "step": 8607
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.6503100141658256,
      "learning_rate": 6.962792861549225e-07,
      "loss": 0.2249,
      "step": 8608
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.3094050684406455,
      "learning_rate": 6.957818029265967e-07,
      "loss": 0.2413,
      "step": 8609
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.293586271628482,
      "learning_rate": 6.952844687539334e-07,
      "loss": 0.2401,
      "step": 8610
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.484252167375673,
      "learning_rate": 6.947872836780192e-07,
      "loss": 0.41,
      "step": 8611
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.7698498326378456,
      "learning_rate": 6.942902477399282e-07,
      "loss": 0.1537,
      "step": 8612
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.804457197253638,
      "learning_rate": 6.93793360980726e-07,
      "loss": 0.3099,
      "step": 8613
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.028697560074589,
      "learning_rate": 6.93296623441459e-07,
      "loss": 0.2819,
      "step": 8614
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.857864631786275,
      "learning_rate": 6.928000351631684e-07,
      "loss": 0.1673,
      "step": 8615
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.499740392260557,
      "learning_rate": 6.923035961868782e-07,
      "loss": 0.263,
      "step": 8616
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.7717314091221974,
      "learning_rate": 6.918073065536016e-07,
      "loss": 0.2454,
      "step": 8617
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.553884615654091,
      "learning_rate": 6.91311166304342e-07,
      "loss": 0.3046,
      "step": 8618
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.525507992679583,
      "learning_rate": 6.908151754800846e-07,
      "loss": 0.3008,
      "step": 8619
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.32685670559447,
      "learning_rate": 6.903193341218081e-07,
      "loss": 0.3636,
      "step": 8620
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.183786470910029,
      "learning_rate": 6.89823642270476e-07,
      "loss": 0.1574,
      "step": 8621
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3340091126960307,
      "learning_rate": 6.893280999670391e-07,
      "loss": 0.2905,
      "step": 8622
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.017070334733244,
      "learning_rate": 6.888327072524384e-07,
      "loss": 0.4144,
      "step": 8623
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.579392273781084,
      "learning_rate": 6.883374641675986e-07,
      "loss": 0.1676,
      "step": 8624
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.5949397668780625,
      "learning_rate": 6.878423707534355e-07,
      "loss": 0.2988,
      "step": 8625
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6563045870395006,
      "learning_rate": 6.873474270508523e-07,
      "loss": 0.071,
      "step": 8626
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.308071427363598,
      "learning_rate": 6.868526331007366e-07,
      "loss": 0.513,
      "step": 8627
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.83422442006342,
      "learning_rate": 6.863579889439684e-07,
      "loss": 0.2862,
      "step": 8628
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.200579788970428,
      "learning_rate": 6.858634946214096e-07,
      "loss": 0.4113,
      "step": 8629
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.37717946732614,
      "learning_rate": 6.853691501739149e-07,
      "loss": 0.3269,
      "step": 8630
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.0111185776047016,
      "learning_rate": 6.848749556423259e-07,
      "loss": 0.158,
      "step": 8631
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.249993713408351,
      "learning_rate": 6.843809110674673e-07,
      "loss": 0.2205,
      "step": 8632
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.587826371946562,
      "learning_rate": 6.838870164901579e-07,
      "loss": 0.0812,
      "step": 8633
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.51760758341763,
      "learning_rate": 6.833932719511979e-07,
      "loss": 0.2778,
      "step": 8634
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.02162093325505,
      "learning_rate": 6.828996774913793e-07,
      "loss": 0.5124,
      "step": 8635
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.078493181662335,
      "learning_rate": 6.824062331514819e-07,
      "loss": 0.1833,
      "step": 8636
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.070428258573241,
      "learning_rate": 6.81912938972269e-07,
      "loss": 0.2918,
      "step": 8637
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.547760317254267,
      "learning_rate": 6.81419794994497e-07,
      "loss": 0.1846,
      "step": 8638
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.173378720300232,
      "learning_rate": 6.80926801258904e-07,
      "loss": 0.3198,
      "step": 8639
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.655505524717246,
      "learning_rate": 6.804339578062203e-07,
      "loss": 0.2986,
      "step": 8640
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.947923552126437,
      "learning_rate": 6.799412646771636e-07,
      "loss": 0.2639,
      "step": 8641
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.8468574252195378,
      "learning_rate": 6.794487219124349e-07,
      "loss": 0.2922,
      "step": 8642
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6811352913289905,
      "learning_rate": 6.789563295527284e-07,
      "loss": 0.1859,
      "step": 8643
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.462785091006466,
      "learning_rate": 6.784640876387208e-07,
      "loss": 0.4973,
      "step": 8644
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.183004980445008,
      "learning_rate": 6.779719962110795e-07,
      "loss": 0.4744,
      "step": 8645
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.15327483140492,
      "learning_rate": 6.774800553104605e-07,
      "loss": 0.3894,
      "step": 8646
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.396925049263398,
      "learning_rate": 6.769882649775022e-07,
      "loss": 0.3051,
      "step": 8647
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.349265733793044,
      "learning_rate": 6.764966252528377e-07,
      "loss": 0.3568,
      "step": 8648
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.932993088819224,
      "learning_rate": 6.7600513617708e-07,
      "loss": 0.3611,
      "step": 8649
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.147823533834106,
      "learning_rate": 6.755137977908355e-07,
      "loss": 0.2595,
      "step": 8650
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.340806153372852,
      "learning_rate": 6.750226101346977e-07,
      "loss": 0.1551,
      "step": 8651
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.233496716999224,
      "learning_rate": 6.745315732492428e-07,
      "loss": 0.2645,
      "step": 8652
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.287352912206515,
      "learning_rate": 6.740406871750402e-07,
      "loss": 0.2181,
      "step": 8653
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.687506931241382,
      "learning_rate": 6.735499519526439e-07,
      "loss": 0.3291,
      "step": 8654
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.736783932675414,
      "learning_rate": 6.730593676225952e-07,
      "loss": 0.4223,
      "step": 8655
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.969558084752858,
      "learning_rate": 6.725689342254261e-07,
      "loss": 0.37,
      "step": 8656
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3867511634597363,
      "learning_rate": 6.720786518016509e-07,
      "loss": 0.2341,
      "step": 8657
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.388736008721275,
      "learning_rate": 6.715885203917761e-07,
      "loss": 0.304,
      "step": 8658
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.003488997648228,
      "learning_rate": 6.710985400362938e-07,
      "loss": 0.0346,
      "step": 8659
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.382356265817594,
      "learning_rate": 6.706087107756825e-07,
      "loss": 0.3333,
      "step": 8660
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.4398963244506304,
      "learning_rate": 6.70119032650412e-07,
      "loss": 0.1541,
      "step": 8661
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.042788191874198,
      "learning_rate": 6.696295057009345e-07,
      "loss": 0.5934,
      "step": 8662
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.9684674573334298,
      "learning_rate": 6.691401299676936e-07,
      "loss": 0.3239,
      "step": 8663
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.043198664919627,
      "learning_rate": 6.686509054911195e-07,
      "loss": 0.5175,
      "step": 8664
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.849364906901437,
      "learning_rate": 6.681618323116279e-07,
      "loss": 0.3074,
      "step": 8665
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.879845876677196,
      "learning_rate": 6.676729104696262e-07,
      "loss": 0.3181,
      "step": 8666
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.771356009885941,
      "learning_rate": 6.671841400055037e-07,
      "loss": 0.1873,
      "step": 8667
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.1021868603660865,
      "learning_rate": 6.666955209596427e-07,
      "loss": 0.2792,
      "step": 8668
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.800752818311662,
      "learning_rate": 6.662070533724094e-07,
      "loss": 0.3489,
      "step": 8669
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.116601960872835,
      "learning_rate": 6.657187372841578e-07,
      "loss": 0.3526,
      "step": 8670
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.087188280378296,
      "learning_rate": 6.652305727352329e-07,
      "loss": 0.1754,
      "step": 8671
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.105168709939368,
      "learning_rate": 6.64742559765961e-07,
      "loss": 0.2546,
      "step": 8672
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.193902014581521,
      "learning_rate": 6.64254698416662e-07,
      "loss": 0.2376,
      "step": 8673
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.06392101158888,
      "learning_rate": 6.637669887276396e-07,
      "loss": 0.2591,
      "step": 8674
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3768046525444,
      "learning_rate": 6.632794307391852e-07,
      "loss": 0.2253,
      "step": 8675
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.6437777097524733,
      "learning_rate": 6.627920244915809e-07,
      "loss": 0.1384,
      "step": 8676
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.622243483114767,
      "learning_rate": 6.623047700250903e-07,
      "loss": 0.1567,
      "step": 8677
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.7203999676445076,
      "learning_rate": 6.618176673799709e-07,
      "loss": 0.2432,
      "step": 8678
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.148151596034677,
      "learning_rate": 6.613307165964638e-07,
      "loss": 0.4537,
      "step": 8679
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.4953899735629395,
      "learning_rate": 6.608439177147977e-07,
      "loss": 0.1286,
      "step": 8680
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.069186382541151,
      "learning_rate": 6.603572707751915e-07,
      "loss": 0.2249,
      "step": 8681
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.796839360050315,
      "learning_rate": 6.598707758178468e-07,
      "loss": 0.139,
      "step": 8682
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.573745917630588,
      "learning_rate": 6.593844328829577e-07,
      "loss": 0.3126,
      "step": 8683
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.230456436502889,
      "learning_rate": 6.588982420107026e-07,
      "loss": 0.2754,
      "step": 8684
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.039738835514301,
      "learning_rate": 6.584122032412479e-07,
      "loss": 0.3306,
      "step": 8685
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.363698846103761,
      "learning_rate": 6.579263166147496e-07,
      "loss": 0.2282,
      "step": 8686
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.35089401211347,
      "learning_rate": 6.574405821713459e-07,
      "loss": 0.2111,
      "step": 8687
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.7361398369018337,
      "learning_rate": 6.569549999511687e-07,
      "loss": 0.2314,
      "step": 8688
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.322870444586881,
      "learning_rate": 6.564695699943335e-07,
      "loss": 0.2377,
      "step": 8689
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.992265301141974,
      "learning_rate": 6.559842923409435e-07,
      "loss": 0.148,
      "step": 8690
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.597259388604972,
      "learning_rate": 6.554991670310918e-07,
      "loss": 0.2746,
      "step": 8691
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.034891184266217,
      "learning_rate": 6.550141941048543e-07,
      "loss": 0.2314,
      "step": 8692
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.140437791212254,
      "learning_rate": 6.545293736022995e-07,
      "loss": 0.4019,
      "step": 8693
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.032003837626759,
      "learning_rate": 6.5404470556348e-07,
      "loss": 0.5578,
      "step": 8694
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.870797336335243,
      "learning_rate": 6.535601900284358e-07,
      "loss": 0.4155,
      "step": 8695
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.622971523206549,
      "learning_rate": 6.530758270371979e-07,
      "loss": 0.3552,
      "step": 8696
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.989729320501146,
      "learning_rate": 6.525916166297783e-07,
      "loss": 0.2826,
      "step": 8697
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.5192983021330666,
      "learning_rate": 6.521075588461828e-07,
      "loss": 0.1488,
      "step": 8698
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.52874395247894,
      "learning_rate": 6.516236537264011e-07,
      "loss": 0.3579,
      "step": 8699
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8367979996914712,
      "learning_rate": 6.511399013104106e-07,
      "loss": 0.0719,
      "step": 8700
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.85786594638445,
      "learning_rate": 6.506563016381772e-07,
      "loss": 0.3782,
      "step": 8701
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.016303932146193,
      "learning_rate": 6.501728547496536e-07,
      "loss": 0.4047,
      "step": 8702
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.35594704030323,
      "learning_rate": 6.496895606847795e-07,
      "loss": 0.2535,
      "step": 8703
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.608716513019182,
      "learning_rate": 6.492064194834824e-07,
      "loss": 0.5547,
      "step": 8704
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.590868569066549,
      "learning_rate": 6.487234311856761e-07,
      "loss": 0.4108,
      "step": 8705
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.346150894753685,
      "learning_rate": 6.482405958312643e-07,
      "loss": 0.3785,
      "step": 8706
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8525574690058895,
      "learning_rate": 6.477579134601361e-07,
      "loss": 0.0611,
      "step": 8707
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.603131368960781,
      "learning_rate": 6.472753841121676e-07,
      "loss": 0.5859,
      "step": 8708
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.3843075803414844,
      "learning_rate": 6.467930078272236e-07,
      "loss": 0.0829,
      "step": 8709
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.7295299278485685,
      "learning_rate": 6.463107846451549e-07,
      "loss": 0.1571,
      "step": 8710
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.493874583149825,
      "learning_rate": 6.458287146058018e-07,
      "loss": 0.2433,
      "step": 8711
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.218488474002577,
      "learning_rate": 6.453467977489896e-07,
      "loss": 0.2316,
      "step": 8712
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.422747998429162,
      "learning_rate": 6.448650341145322e-07,
      "loss": 0.3845,
      "step": 8713
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.087182634126641,
      "learning_rate": 6.443834237422303e-07,
      "loss": 0.2932,
      "step": 8714
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.1242995696741596,
      "learning_rate": 6.439019666718719e-07,
      "loss": 0.1242,
      "step": 8715
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.392405676218042,
      "learning_rate": 6.434206629432335e-07,
      "loss": 0.3138,
      "step": 8716
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.8512057287679817,
      "learning_rate": 6.429395125960777e-07,
      "loss": 0.169,
      "step": 8717
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.589875546126532,
      "learning_rate": 6.424585156701546e-07,
      "loss": 0.425,
      "step": 8718
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.240525315623826,
      "learning_rate": 6.41977672205202e-07,
      "loss": 0.1988,
      "step": 8719
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.117881421479188,
      "learning_rate": 6.414969822409439e-07,
      "loss": 0.2636,
      "step": 8720
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.2605661377637873,
      "learning_rate": 6.410164458170942e-07,
      "loss": 0.0764,
      "step": 8721
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.5326206813969083,
      "learning_rate": 6.405360629733517e-07,
      "loss": 0.2096,
      "step": 8722
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.0707932928798405,
      "learning_rate": 6.400558337494031e-07,
      "loss": 0.5257,
      "step": 8723
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.964090113251455,
      "learning_rate": 6.39575758184923e-07,
      "loss": 0.3232,
      "step": 8724
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.834254907293676,
      "learning_rate": 6.390958363195715e-07,
      "loss": 0.1233,
      "step": 8725
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.422820955838434,
      "learning_rate": 6.386160681929993e-07,
      "loss": 0.2648,
      "step": 8726
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.058489659450271,
      "learning_rate": 6.381364538448417e-07,
      "loss": 0.1963,
      "step": 8727
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.728413630800329,
      "learning_rate": 6.37656993314722e-07,
      "loss": 0.33,
      "step": 8728
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.924111587222456,
      "learning_rate": 6.37177686642251e-07,
      "loss": 0.3365,
      "step": 8729
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.539039278285831,
      "learning_rate": 6.366985338670259e-07,
      "loss": 0.3405,
      "step": 8730
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.317374399757641,
      "learning_rate": 6.362195350286335e-07,
      "loss": 0.2168,
      "step": 8731
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.103447256748323,
      "learning_rate": 6.357406901666452e-07,
      "loss": 0.3804,
      "step": 8732
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.038028505644277,
      "learning_rate": 6.352619993206213e-07,
      "loss": 0.3016,
      "step": 8733
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.406401261774089,
      "learning_rate": 6.347834625301086e-07,
      "loss": 0.4799,
      "step": 8734
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.212324081810354,
      "learning_rate": 6.343050798346409e-07,
      "loss": 0.1868,
      "step": 8735
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.411876164497714,
      "learning_rate": 6.338268512737411e-07,
      "loss": 0.3018,
      "step": 8736
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.902465481913467,
      "learning_rate": 6.333487768869176e-07,
      "loss": 0.4366,
      "step": 8737
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.970867979919887,
      "learning_rate": 6.328708567136665e-07,
      "loss": 0.3426,
      "step": 8738
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9343127960865427,
      "learning_rate": 6.323930907934706e-07,
      "loss": 0.0913,
      "step": 8739
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.214622628416015,
      "learning_rate": 6.319154791658017e-07,
      "loss": 0.3462,
      "step": 8740
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3111000280475245,
      "learning_rate": 6.314380218701172e-07,
      "loss": 0.2719,
      "step": 8741
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.043753398324981,
      "learning_rate": 6.309607189458622e-07,
      "loss": 0.1909,
      "step": 8742
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.048826120599906,
      "learning_rate": 6.304835704324691e-07,
      "loss": 0.2619,
      "step": 8743
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.213023782685722,
      "learning_rate": 6.300065763693572e-07,
      "loss": 0.2336,
      "step": 8744
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.842743664732074,
      "learning_rate": 6.295297367959344e-07,
      "loss": 0.4624,
      "step": 8745
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.793805400065132,
      "learning_rate": 6.290530517515945e-07,
      "loss": 0.3054,
      "step": 8746
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.9628267019280443,
      "learning_rate": 6.285765212757186e-07,
      "loss": 0.1774,
      "step": 8747
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.8640487545090534,
      "learning_rate": 6.281001454076754e-07,
      "loss": 0.1113,
      "step": 8748
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.100215883644822,
      "learning_rate": 6.276239241868198e-07,
      "loss": 0.2406,
      "step": 8749
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3427808045952996,
      "learning_rate": 6.27147857652497e-07,
      "loss": 0.2001,
      "step": 8750
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.865871974962694,
      "learning_rate": 6.26671945844036e-07,
      "loss": 0.3885,
      "step": 8751
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.830649132376615,
      "learning_rate": 6.261961888007542e-07,
      "loss": 0.3397,
      "step": 8752
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.695278522326841,
      "learning_rate": 6.257205865619564e-07,
      "loss": 0.268,
      "step": 8753
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.668369046137469,
      "learning_rate": 6.252451391669342e-07,
      "loss": 0.4341,
      "step": 8754
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.77694234087737,
      "learning_rate": 6.247698466549678e-07,
      "loss": 0.4019,
      "step": 8755
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3332563113948055,
      "learning_rate": 6.242947090653232e-07,
      "loss": 0.1812,
      "step": 8756
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.438510171864616,
      "learning_rate": 6.238197264372534e-07,
      "loss": 0.261,
      "step": 8757
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.660174503145591,
      "learning_rate": 6.233448988099993e-07,
      "loss": 0.2954,
      "step": 8758
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.149750426165508,
      "learning_rate": 6.228702262227882e-07,
      "loss": 0.1555,
      "step": 8759
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.043182060214751,
      "learning_rate": 6.223957087148369e-07,
      "loss": 0.2983,
      "step": 8760
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7688228313804535,
      "learning_rate": 6.219213463253465e-07,
      "loss": 0.1309,
      "step": 8761
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.760263839658084,
      "learning_rate": 6.214471390935067e-07,
      "loss": 0.4469,
      "step": 8762
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.6658841679453182,
      "learning_rate": 6.209730870584943e-07,
      "loss": 0.1084,
      "step": 8763
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.810860379142043,
      "learning_rate": 6.204991902594726e-07,
      "loss": 0.2343,
      "step": 8764
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.178689995350493,
      "learning_rate": 6.200254487355936e-07,
      "loss": 0.201,
      "step": 8765
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.679205509197871,
      "learning_rate": 6.19551862525995e-07,
      "loss": 0.4097,
      "step": 8766
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.4723955772151065,
      "learning_rate": 6.19078431669802e-07,
      "loss": 0.3303,
      "step": 8767
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.282969773479666,
      "learning_rate": 6.186051562061276e-07,
      "loss": 0.2568,
      "step": 8768
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.061248482744649,
      "learning_rate": 6.181320361740706e-07,
      "loss": 0.3195,
      "step": 8769
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.4320643728790725,
      "learning_rate": 6.17659071612719e-07,
      "loss": 0.1585,
      "step": 8770
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.212765883116836,
      "learning_rate": 6.17186262561146e-07,
      "loss": 0.4205,
      "step": 8771
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.944025726588884,
      "learning_rate": 6.167136090584136e-07,
      "loss": 0.4311,
      "step": 8772
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.8351693572205985,
      "learning_rate": 6.162411111435693e-07,
      "loss": 0.3918,
      "step": 8773
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.5735836785699306,
      "learning_rate": 6.15768768855648e-07,
      "loss": 0.0962,
      "step": 8774
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.060787862234207,
      "learning_rate": 6.15296582233674e-07,
      "loss": 0.4217,
      "step": 8775
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.934139485179765,
      "learning_rate": 6.148245513166562e-07,
      "loss": 0.3427,
      "step": 8776
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.9067930145622833,
      "learning_rate": 6.143526761435911e-07,
      "loss": 0.2315,
      "step": 8777
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.7325154002001195,
      "learning_rate": 6.138809567534635e-07,
      "loss": 0.3474,
      "step": 8778
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.672584454967203,
      "learning_rate": 6.134093931852434e-07,
      "loss": 0.2303,
      "step": 8779
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.446513358421451,
      "learning_rate": 6.129379854778902e-07,
      "loss": 0.22,
      "step": 8780
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.106777749471255,
      "learning_rate": 6.124667336703488e-07,
      "loss": 0.1753,
      "step": 8781
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.2124762989334625,
      "learning_rate": 6.119956378015515e-07,
      "loss": 0.4313,
      "step": 8782
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.6026536853769078,
      "learning_rate": 6.115246979104195e-07,
      "loss": 0.2132,
      "step": 8783
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.216130340669621,
      "learning_rate": 6.110539140358565e-07,
      "loss": 0.2718,
      "step": 8784
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.641362884252017,
      "learning_rate": 6.105832862167591e-07,
      "loss": 0.1625,
      "step": 8785
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.879904036785994,
      "learning_rate": 6.101128144920073e-07,
      "loss": 0.3656,
      "step": 8786
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.316517547884165,
      "learning_rate": 6.096424989004684e-07,
      "loss": 0.3461,
      "step": 8787
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.368148966000897,
      "learning_rate": 6.091723394809998e-07,
      "loss": 0.2114,
      "step": 8788
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.494562197336386,
      "learning_rate": 6.087023362724409e-07,
      "loss": 0.251,
      "step": 8789
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.7862445693744595,
      "learning_rate": 6.082324893136229e-07,
      "loss": 0.4368,
      "step": 8790
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.747078321121911,
      "learning_rate": 6.07762798643362e-07,
      "loss": 0.4378,
      "step": 8791
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.549773853370779,
      "learning_rate": 6.072932643004609e-07,
      "loss": 0.4227,
      "step": 8792
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.6162135653629983,
      "learning_rate": 6.068238863237122e-07,
      "loss": 0.1864,
      "step": 8793
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.9010342530231403,
      "learning_rate": 6.063546647518906e-07,
      "loss": 0.2435,
      "step": 8794
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.3503604781034895,
      "learning_rate": 6.058855996237636e-07,
      "loss": 0.3798,
      "step": 8795
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.790514822788293,
      "learning_rate": 6.054166909780818e-07,
      "loss": 0.2665,
      "step": 8796
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3792061501707344,
      "learning_rate": 6.04947938853584e-07,
      "loss": 0.2023,
      "step": 8797
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.352730126531933,
      "learning_rate": 6.044793432889975e-07,
      "loss": 0.2871,
      "step": 8798
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.983598658807315,
      "learning_rate": 6.040109043230333e-07,
      "loss": 0.346,
      "step": 8799
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.487959101515422,
      "learning_rate": 6.035426219943932e-07,
      "loss": 0.2402,
      "step": 8800
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.110002147733225,
      "learning_rate": 6.030744963417639e-07,
      "loss": 0.1799,
      "step": 8801
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.276851243236476,
      "learning_rate": 6.02606527403819e-07,
      "loss": 0.1818,
      "step": 8802
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.361060681560182,
      "learning_rate": 6.021387152192215e-07,
      "loss": 0.1823,
      "step": 8803
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.341579380236455,
      "learning_rate": 6.016710598266176e-07,
      "loss": 0.1259,
      "step": 8804
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.362862850779926,
      "learning_rate": 6.012035612646442e-07,
      "loss": 0.3338,
      "step": 8805
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.164973793396611,
      "learning_rate": 6.007362195719232e-07,
      "loss": 0.1949,
      "step": 8806
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.499512162394424,
      "learning_rate": 6.002690347870638e-07,
      "loss": 0.3236,
      "step": 8807
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.5089576766329587,
      "learning_rate": 5.99802006948664e-07,
      "loss": 0.2125,
      "step": 8808
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.1492076110134315,
      "learning_rate": 5.993351360953048e-07,
      "loss": 0.392,
      "step": 8809
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8235041979821194,
      "learning_rate": 5.988684222655588e-07,
      "loss": 0.0609,
      "step": 8810
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.4809094788032637,
      "learning_rate": 5.984018654979834e-07,
      "loss": 0.1504,
      "step": 8811
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.800154793975819,
      "learning_rate": 5.979354658311218e-07,
      "loss": 0.1953,
      "step": 8812
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.0142184214772785,
      "learning_rate": 5.974692233035082e-07,
      "loss": 0.1685,
      "step": 8813
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.707018659168802,
      "learning_rate": 5.970031379536584e-07,
      "loss": 0.319,
      "step": 8814
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.4946251633661616,
      "learning_rate": 5.9653720982008e-07,
      "loss": 0.2346,
      "step": 8815
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.752361201323292,
      "learning_rate": 5.960714389412653e-07,
      "loss": 0.2605,
      "step": 8816
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.379485440887229,
      "learning_rate": 5.956058253556932e-07,
      "loss": 0.3718,
      "step": 8817
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6861537121263346,
      "learning_rate": 5.951403691018323e-07,
      "loss": 0.0533,
      "step": 8818
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.754407425832473,
      "learning_rate": 5.946750702181336e-07,
      "loss": 0.2224,
      "step": 8819
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1537066546235257,
      "learning_rate": 5.942099287430403e-07,
      "loss": 0.103,
      "step": 8820
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.08976491608408,
      "learning_rate": 5.93744944714979e-07,
      "loss": 0.2627,
      "step": 8821
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.51720068413341,
      "learning_rate": 5.932801181723638e-07,
      "loss": 0.4205,
      "step": 8822
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.26391539784768,
      "learning_rate": 5.928154491535984e-07,
      "loss": 0.1203,
      "step": 8823
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.533249604042976,
      "learning_rate": 5.923509376970688e-07,
      "loss": 0.3344,
      "step": 8824
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.820485008441525,
      "learning_rate": 5.918865838411528e-07,
      "loss": 0.4268,
      "step": 8825
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.940143470410966,
      "learning_rate": 5.914223876242123e-07,
      "loss": 0.3836,
      "step": 8826
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.1349406589170155,
      "learning_rate": 5.909583490845961e-07,
      "loss": 0.2268,
      "step": 8827
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.670775917195618,
      "learning_rate": 5.904944682606428e-07,
      "loss": 0.2527,
      "step": 8828
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.001266165373318,
      "learning_rate": 5.900307451906734e-07,
      "loss": 0.2354,
      "step": 8829
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.930945716753505,
      "learning_rate": 5.895671799129999e-07,
      "loss": 0.2932,
      "step": 8830
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.954670888563065,
      "learning_rate": 5.89103772465921e-07,
      "loss": 0.4388,
      "step": 8831
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.075980476488755,
      "learning_rate": 5.886405228877185e-07,
      "loss": 0.2515,
      "step": 8832
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.458825620495747,
      "learning_rate": 5.881774312166663e-07,
      "loss": 0.3786,
      "step": 8833
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.274440296096603,
      "learning_rate": 5.877144974910204e-07,
      "loss": 0.4314,
      "step": 8834
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.477426563904204,
      "learning_rate": 5.872517217490273e-07,
      "loss": 0.2507,
      "step": 8835
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.262193754810981,
      "learning_rate": 5.867891040289204e-07,
      "loss": 0.5292,
      "step": 8836
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.065559305131551,
      "learning_rate": 5.863266443689164e-07,
      "loss": 0.2041,
      "step": 8837
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.3325713280375044,
      "learning_rate": 5.858643428072244e-07,
      "loss": 0.2136,
      "step": 8838
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.155900004201936,
      "learning_rate": 5.854021993820344e-07,
      "loss": 0.2803,
      "step": 8839
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.1655634288204872,
      "learning_rate": 5.849402141315278e-07,
      "loss": 0.2244,
      "step": 8840
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.864165179559027,
      "learning_rate": 5.844783870938733e-07,
      "loss": 0.3106,
      "step": 8841
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.057303050217594,
      "learning_rate": 5.840167183072215e-07,
      "loss": 0.3252,
      "step": 8842
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.7755332822359255,
      "learning_rate": 5.835552078097164e-07,
      "loss": 0.3203,
      "step": 8843
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.616413330391659,
      "learning_rate": 5.830938556394827e-07,
      "loss": 0.259,
      "step": 8844
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8611873853191854,
      "learning_rate": 5.826326618346368e-07,
      "loss": 0.0424,
      "step": 8845
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.370937585686326,
      "learning_rate": 5.821716264332814e-07,
      "loss": 0.3913,
      "step": 8846
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0643091544278223,
      "learning_rate": 5.817107494735022e-07,
      "loss": 0.1346,
      "step": 8847
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.017608099830611,
      "learning_rate": 5.812500309933772e-07,
      "loss": 0.292,
      "step": 8848
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.3273272085205,
      "learning_rate": 5.807894710309664e-07,
      "loss": 0.4512,
      "step": 8849
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.141396469510742,
      "learning_rate": 5.8032906962432e-07,
      "loss": 0.1755,
      "step": 8850
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.941237955583311,
      "learning_rate": 5.798688268114755e-07,
      "loss": 0.4106,
      "step": 8851
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.754420095716764,
      "learning_rate": 5.794087426304535e-07,
      "loss": 0.2642,
      "step": 8852
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.260173310206639,
      "learning_rate": 5.789488171192662e-07,
      "loss": 0.3377,
      "step": 8853
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.1562258037931668,
      "learning_rate": 5.784890503159076e-07,
      "loss": 0.1185,
      "step": 8854
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.764798880626605,
      "learning_rate": 5.780294422583635e-07,
      "loss": 0.3669,
      "step": 8855
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.903522679038829,
      "learning_rate": 5.775699929846052e-07,
      "loss": 0.1728,
      "step": 8856
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.5895083356921633,
      "learning_rate": 5.771107025325875e-07,
      "loss": 0.2401,
      "step": 8857
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.781080553662627,
      "learning_rate": 5.766515709402568e-07,
      "loss": 0.0544,
      "step": 8858
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.108811139794444,
      "learning_rate": 5.761925982455435e-07,
      "loss": 0.2378,
      "step": 8859
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.36235695933253,
      "learning_rate": 5.757337844863653e-07,
      "loss": 0.2461,
      "step": 8860
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.632798748728685,
      "learning_rate": 5.75275129700629e-07,
      "loss": 0.1559,
      "step": 8861
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.863488778403053,
      "learning_rate": 5.748166339262237e-07,
      "loss": 0.1503,
      "step": 8862
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.304414753515293,
      "learning_rate": 5.743582972010303e-07,
      "loss": 0.3739,
      "step": 8863
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.8891761683558657,
      "learning_rate": 5.739001195629131e-07,
      "loss": 0.3078,
      "step": 8864
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.607266716990231,
      "learning_rate": 5.734421010497246e-07,
      "loss": 0.2139,
      "step": 8865
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.146832297584255,
      "learning_rate": 5.729842416993053e-07,
      "loss": 0.3469,
      "step": 8866
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.9436363794219846,
      "learning_rate": 5.72526541549479e-07,
      "loss": 0.3727,
      "step": 8867
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.084028663229461,
      "learning_rate": 5.720690006380608e-07,
      "loss": 0.3274,
      "step": 8868
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.927534291928609,
      "learning_rate": 5.716116190028495e-07,
      "loss": 0.1217,
      "step": 8869
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.605145360384403,
      "learning_rate": 5.711543966816313e-07,
      "loss": 0.0792,
      "step": 8870
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.881358562750158,
      "learning_rate": 5.706973337121816e-07,
      "loss": 0.4855,
      "step": 8871
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.160209336234598,
      "learning_rate": 5.702404301322578e-07,
      "loss": 0.3677,
      "step": 8872
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.296528803751073,
      "learning_rate": 5.697836859796094e-07,
      "loss": 0.2152,
      "step": 8873
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.6473235563771502,
      "learning_rate": 5.693271012919693e-07,
      "loss": 0.2167,
      "step": 8874
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.614832707139237,
      "learning_rate": 5.688706761070581e-07,
      "loss": 0.3355,
      "step": 8875
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.337711289209221,
      "learning_rate": 5.684144104625849e-07,
      "loss": 0.4144,
      "step": 8876
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.6818381721034266,
      "learning_rate": 5.67958304396242e-07,
      "loss": 0.2147,
      "step": 8877
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.284083762922035,
      "learning_rate": 5.675023579457123e-07,
      "loss": 0.2901,
      "step": 8878
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.526825805616109,
      "learning_rate": 5.670465711486631e-07,
      "loss": 0.4123,
      "step": 8879
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.599118366496398,
      "learning_rate": 5.665909440427489e-07,
      "loss": 0.1867,
      "step": 8880
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.459739130982343,
      "learning_rate": 5.661354766656132e-07,
      "loss": 0.3878,
      "step": 8881
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.255661476127427,
      "learning_rate": 5.656801690548819e-07,
      "loss": 0.3554,
      "step": 8882
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.153889164423724,
      "learning_rate": 5.652250212481722e-07,
      "loss": 0.3119,
      "step": 8883
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.666719837703537,
      "learning_rate": 5.647700332830855e-07,
      "loss": 0.2581,
      "step": 8884
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.174145652769169,
      "learning_rate": 5.643152051972101e-07,
      "loss": 0.2046,
      "step": 8885
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.909476863176918,
      "learning_rate": 5.638605370281236e-07,
      "loss": 0.3416,
      "step": 8886
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.581533074498145,
      "learning_rate": 5.634060288133858e-07,
      "loss": 0.2941,
      "step": 8887
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.06839532137805,
      "learning_rate": 5.62951680590548e-07,
      "loss": 0.1948,
      "step": 8888
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.0318605652519883,
      "learning_rate": 5.624974923971455e-07,
      "loss": 0.123,
      "step": 8889
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3187434895601347,
      "learning_rate": 5.620434642707004e-07,
      "loss": 0.0395,
      "step": 8890
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.289360746627287,
      "learning_rate": 5.615895962487245e-07,
      "loss": 0.1607,
      "step": 8891
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.150057162784638,
      "learning_rate": 5.611358883687113e-07,
      "loss": 0.279,
      "step": 8892
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.7372546718699993,
      "learning_rate": 5.606823406681458e-07,
      "loss": 0.1226,
      "step": 8893
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.735948792020356,
      "learning_rate": 5.602289531844974e-07,
      "loss": 0.3467,
      "step": 8894
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.3815508020079323,
      "learning_rate": 5.59775725955222e-07,
      "loss": 0.1281,
      "step": 8895
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.491874047013076,
      "learning_rate": 5.59322659017765e-07,
      "loss": 0.3019,
      "step": 8896
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.590427120220673,
      "learning_rate": 5.58869752409554e-07,
      "loss": 0.5488,
      "step": 8897
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.9330916571596117,
      "learning_rate": 5.58417006168008e-07,
      "loss": 0.1881,
      "step": 8898
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.943101675085102,
      "learning_rate": 5.579644203305298e-07,
      "loss": 0.2848,
      "step": 8899
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.805782957790294,
      "learning_rate": 5.575119949345093e-07,
      "loss": 0.1947,
      "step": 8900
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.134084270616139,
      "learning_rate": 5.570597300173247e-07,
      "loss": 0.3071,
      "step": 8901
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.672317606937289,
      "learning_rate": 5.566076256163397e-07,
      "loss": 0.2914,
      "step": 8902
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.316705474861862,
      "learning_rate": 5.561556817689046e-07,
      "loss": 0.1342,
      "step": 8903
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.8109209774904307,
      "learning_rate": 5.55703898512357e-07,
      "loss": 0.2244,
      "step": 8904
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.532596125705635,
      "learning_rate": 5.552522758840203e-07,
      "loss": 0.3705,
      "step": 8905
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.621036852694729,
      "learning_rate": 5.548008139212066e-07,
      "loss": 0.3261,
      "step": 8906
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.6767535360166534,
      "learning_rate": 5.543495126612128e-07,
      "loss": 0.1528,
      "step": 8907
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.245969122854178,
      "learning_rate": 5.538983721413236e-07,
      "loss": 0.1962,
      "step": 8908
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.016103224787691,
      "learning_rate": 5.534473923988093e-07,
      "loss": 0.3158,
      "step": 8909
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.872018453573431,
      "learning_rate": 5.529965734709272e-07,
      "loss": 0.2782,
      "step": 8910
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.692085243858401,
      "learning_rate": 5.525459153949236e-07,
      "loss": 0.2698,
      "step": 8911
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.734252270688608,
      "learning_rate": 5.520954182080285e-07,
      "loss": 0.2128,
      "step": 8912
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.171427225044277,
      "learning_rate": 5.516450819474597e-07,
      "loss": 0.2424,
      "step": 8913
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.614140641905951,
      "learning_rate": 5.511949066504222e-07,
      "loss": 0.3115,
      "step": 8914
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.421678298525717,
      "learning_rate": 5.507448923541061e-07,
      "loss": 0.2067,
      "step": 8915
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.176905484732376,
      "learning_rate": 5.502950390956913e-07,
      "loss": 0.3464,
      "step": 8916
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.500805191996123,
      "learning_rate": 5.498453469123413e-07,
      "loss": 0.2924,
      "step": 8917
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.885282065060345,
      "learning_rate": 5.493958158412077e-07,
      "loss": 0.1552,
      "step": 8918
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.919738750249421,
      "learning_rate": 5.489464459194285e-07,
      "loss": 0.3589,
      "step": 8919
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.201615999402383,
      "learning_rate": 5.484972371841277e-07,
      "loss": 0.2498,
      "step": 8920
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.514941464802766,
      "learning_rate": 5.480481896724183e-07,
      "loss": 0.421,
      "step": 8921
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.7925742383989696,
      "learning_rate": 5.475993034213975e-07,
      "loss": 0.1039,
      "step": 8922
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.043909160864891,
      "learning_rate": 5.471505784681499e-07,
      "loss": 0.2916,
      "step": 8923
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.5430546463815182,
      "learning_rate": 5.467020148497476e-07,
      "loss": 0.252,
      "step": 8924
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.233644802789416,
      "learning_rate": 5.462536126032475e-07,
      "loss": 0.4294,
      "step": 8925
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.716987844348675,
      "learning_rate": 5.458053717656958e-07,
      "loss": 0.2303,
      "step": 8926
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.708008314576804,
      "learning_rate": 5.453572923741232e-07,
      "loss": 0.3498,
      "step": 8927
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.340553373037235,
      "learning_rate": 5.449093744655485e-07,
      "loss": 0.5663,
      "step": 8928
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7814938546182986,
      "learning_rate": 5.444616180769757e-07,
      "loss": 0.0615,
      "step": 8929
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.6185589186354585,
      "learning_rate": 5.440140232453959e-07,
      "loss": 0.2977,
      "step": 8930
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.00650096591709,
      "learning_rate": 5.435665900077883e-07,
      "loss": 0.3049,
      "step": 8931
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.701280838063547,
      "learning_rate": 5.431193184011175e-07,
      "loss": 0.2623,
      "step": 8932
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.640992124919013,
      "learning_rate": 5.426722084623343e-07,
      "loss": 0.3527,
      "step": 8933
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.1768636926074905,
      "learning_rate": 5.42225260228377e-07,
      "loss": 0.357,
      "step": 8934
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.4146307073472,
      "learning_rate": 5.417784737361698e-07,
      "loss": 0.4059,
      "step": 8935
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.719954846298756,
      "learning_rate": 5.413318490226249e-07,
      "loss": 0.181,
      "step": 8936
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.6236085708004686,
      "learning_rate": 5.408853861246396e-07,
      "loss": 0.2102,
      "step": 8937
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.865815803959275,
      "learning_rate": 5.404390850790992e-07,
      "loss": 0.3055,
      "step": 8938
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.754842612762156,
      "learning_rate": 5.399929459228739e-07,
      "loss": 0.1617,
      "step": 8939
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.889087402760445,
      "learning_rate": 5.395469686928217e-07,
      "loss": 0.3349,
      "step": 8940
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.067584914816354,
      "learning_rate": 5.391011534257878e-07,
      "loss": 0.3539,
      "step": 8941
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.167292295736868,
      "learning_rate": 5.386555001586031e-07,
      "loss": 0.2702,
      "step": 8942
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.153818863203408,
      "learning_rate": 5.382100089280848e-07,
      "loss": 0.2042,
      "step": 8943
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.6408217035178776,
      "learning_rate": 5.377646797710368e-07,
      "loss": 0.2108,
      "step": 8944
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.692834233080037,
      "learning_rate": 5.373195127242512e-07,
      "loss": 0.1021,
      "step": 8945
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.704099395403822,
      "learning_rate": 5.36874507824505e-07,
      "loss": 0.2945,
      "step": 8946
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.778574564247967,
      "learning_rate": 5.364296651085624e-07,
      "loss": 0.1586,
      "step": 8947
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.540881914227837,
      "learning_rate": 5.35984984613174e-07,
      "loss": 0.4057,
      "step": 8948
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.880692565171862,
      "learning_rate": 5.355404663750763e-07,
      "loss": 0.295,
      "step": 8949
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.403202553585321,
      "learning_rate": 5.350961104309949e-07,
      "loss": 0.2685,
      "step": 8950
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.750694934302755,
      "learning_rate": 5.346519168176392e-07,
      "loss": 0.2252,
      "step": 8951
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.782682055316991,
      "learning_rate": 5.342078855717067e-07,
      "loss": 0.1746,
      "step": 8952
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.1225331911114855,
      "learning_rate": 5.337640167298811e-07,
      "loss": 0.2743,
      "step": 8953
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.285847512860967,
      "learning_rate": 5.333203103288315e-07,
      "loss": 0.3033,
      "step": 8954
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.692844818789066,
      "learning_rate": 5.328767664052168e-07,
      "loss": 0.2561,
      "step": 8955
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.545229663666343,
      "learning_rate": 5.324333849956789e-07,
      "loss": 0.2532,
      "step": 8956
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.3005643182820106,
      "learning_rate": 5.319901661368487e-07,
      "loss": 0.1943,
      "step": 8957
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.2782522898477637,
      "learning_rate": 5.315471098653421e-07,
      "loss": 0.1231,
      "step": 8958
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.280057792451141,
      "learning_rate": 5.311042162177618e-07,
      "loss": 0.2578,
      "step": 8959
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9418435022667095,
      "learning_rate": 5.306614852306991e-07,
      "loss": 0.1627,
      "step": 8960
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.955511345134078,
      "learning_rate": 5.30218916940729e-07,
      "loss": 0.153,
      "step": 8961
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9867113890087165,
      "learning_rate": 5.29776511384415e-07,
      "loss": 0.1679,
      "step": 8962
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9204323685106046,
      "learning_rate": 5.29334268598306e-07,
      "loss": 0.2316,
      "step": 8963
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.4541301785792617,
      "learning_rate": 5.288921886189374e-07,
      "loss": 0.2797,
      "step": 8964
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.8512007460338715,
      "learning_rate": 5.28450271482833e-07,
      "loss": 0.3368,
      "step": 8965
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8188366340944593,
      "learning_rate": 5.280085172265012e-07,
      "loss": 0.3367,
      "step": 8966
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.014514220406024,
      "learning_rate": 5.275669258864375e-07,
      "loss": 0.5087,
      "step": 8967
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8535541959789716,
      "learning_rate": 5.271254974991241e-07,
      "loss": 0.3162,
      "step": 8968
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.542252007783291,
      "learning_rate": 5.266842321010291e-07,
      "loss": 0.3971,
      "step": 8969
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2935326757311927,
      "learning_rate": 5.262431297286085e-07,
      "loss": 0.0405,
      "step": 8970
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.419132696671542,
      "learning_rate": 5.258021904183042e-07,
      "loss": 0.342,
      "step": 8971
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.066772010291447,
      "learning_rate": 5.253614142065436e-07,
      "loss": 0.3001,
      "step": 8972
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.70868570104377,
      "learning_rate": 5.249208011297418e-07,
      "loss": 0.3296,
      "step": 8973
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.6899382544675507,
      "learning_rate": 5.244803512242996e-07,
      "loss": 0.2597,
      "step": 8974
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.153464578016254,
      "learning_rate": 5.240400645266061e-07,
      "loss": 0.408,
      "step": 8975
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.455963864332385,
      "learning_rate": 5.235999410730347e-07,
      "loss": 0.381,
      "step": 8976
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.488206900493692,
      "learning_rate": 5.231599808999466e-07,
      "loss": 0.3033,
      "step": 8977
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.927022825869537,
      "learning_rate": 5.227201840436885e-07,
      "loss": 0.3579,
      "step": 8978
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.6604738523793507,
      "learning_rate": 5.222805505405946e-07,
      "loss": 0.124,
      "step": 8979
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.8467391905504287,
      "learning_rate": 5.218410804269858e-07,
      "loss": 0.2159,
      "step": 8980
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9109934982046113,
      "learning_rate": 5.214017737391682e-07,
      "loss": 0.1865,
      "step": 8981
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.819667091840925,
      "learning_rate": 5.209626305134352e-07,
      "loss": 0.1738,
      "step": 8982
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.505873864111549,
      "learning_rate": 5.205236507860683e-07,
      "loss": 0.3192,
      "step": 8983
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.594173734324266,
      "learning_rate": 5.200848345933313e-07,
      "loss": 0.3681,
      "step": 8984
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0125062808391507,
      "learning_rate": 5.196461819714788e-07,
      "loss": 0.0858,
      "step": 8985
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9148882798734164,
      "learning_rate": 5.192076929567496e-07,
      "loss": 0.2953,
      "step": 8986
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.855816244501095,
      "learning_rate": 5.187693675853689e-07,
      "loss": 0.0606,
      "step": 8987
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.525321976819882,
      "learning_rate": 5.183312058935508e-07,
      "loss": 0.5272,
      "step": 8988
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.542591770291816,
      "learning_rate": 5.178932079174915e-07,
      "loss": 0.592,
      "step": 8989
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.772781566195327,
      "learning_rate": 5.174553736933786e-07,
      "loss": 0.3061,
      "step": 8990
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.91367013633958,
      "learning_rate": 5.170177032573822e-07,
      "loss": 0.5355,
      "step": 8991
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.073655871134276,
      "learning_rate": 5.16580196645661e-07,
      "loss": 0.3291,
      "step": 8992
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.5635317786613085,
      "learning_rate": 5.161428538943611e-07,
      "loss": 0.1227,
      "step": 8993
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.437475287001465,
      "learning_rate": 5.157056750396108e-07,
      "loss": 0.3049,
      "step": 8994
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.16466670218206,
      "learning_rate": 5.152686601175297e-07,
      "loss": 0.2421,
      "step": 8995
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9226706667019884,
      "learning_rate": 5.148318091642216e-07,
      "loss": 0.2077,
      "step": 8996
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.991198846932611,
      "learning_rate": 5.143951222157756e-07,
      "loss": 0.3922,
      "step": 8997
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.069173905857143,
      "learning_rate": 5.139585993082713e-07,
      "loss": 0.3685,
      "step": 8998
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.144313288662258,
      "learning_rate": 5.135222404777693e-07,
      "loss": 0.2469,
      "step": 8999
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.757120889110641,
      "learning_rate": 5.130860457603209e-07,
      "loss": 0.4215,
      "step": 9000
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.539379672737755,
      "learning_rate": 5.126500151919625e-07,
      "loss": 0.2317,
      "step": 9001
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.8025711008118983,
      "learning_rate": 5.122141488087154e-07,
      "loss": 0.1512,
      "step": 9002
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.08685293149885,
      "learning_rate": 5.117784466465911e-07,
      "loss": 0.5062,
      "step": 9003
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1439934363823303,
      "learning_rate": 5.113429087415827e-07,
      "loss": 0.0878,
      "step": 9004
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.342725133476312,
      "learning_rate": 5.109075351296738e-07,
      "loss": 0.4287,
      "step": 9005
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.11196950682526,
      "learning_rate": 5.104723258468325e-07,
      "loss": 0.2149,
      "step": 9006
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.067925354992909,
      "learning_rate": 5.100372809290127e-07,
      "loss": 0.0747,
      "step": 9007
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.903560996823707,
      "learning_rate": 5.096024004121577e-07,
      "loss": 0.3324,
      "step": 9008
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.7706428085225823,
      "learning_rate": 5.09167684332193e-07,
      "loss": 0.1378,
      "step": 9009
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.860307996863526,
      "learning_rate": 5.087331327250342e-07,
      "loss": 0.3584,
      "step": 9010
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.537997152729579,
      "learning_rate": 5.082987456265812e-07,
      "loss": 0.5007,
      "step": 9011
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.906837136721504,
      "learning_rate": 5.078645230727205e-07,
      "loss": 0.277,
      "step": 9012
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.508233321537644,
      "learning_rate": 5.074304650993273e-07,
      "loss": 0.2661,
      "step": 9013
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.378477457397081,
      "learning_rate": 5.069965717422587e-07,
      "loss": 0.2183,
      "step": 9014
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.654768080532316,
      "learning_rate": 5.065628430373629e-07,
      "loss": 0.1768,
      "step": 9015
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.965606431016747,
      "learning_rate": 5.061292790204716e-07,
      "loss": 0.0935,
      "step": 9016
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.502296955826223,
      "learning_rate": 5.056958797274034e-07,
      "loss": 0.4036,
      "step": 9017
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.309126467938634,
      "learning_rate": 5.052626451939652e-07,
      "loss": 0.1995,
      "step": 9018
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.731621658632704,
      "learning_rate": 5.048295754559462e-07,
      "loss": 0.1444,
      "step": 9019
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.2626084048140065,
      "learning_rate": 5.043966705491266e-07,
      "loss": 0.4064,
      "step": 9020
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.755947715548907,
      "learning_rate": 5.039639305092703e-07,
      "loss": 0.1523,
      "step": 9021
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.625389120804192,
      "learning_rate": 5.035313553721271e-07,
      "loss": 0.27,
      "step": 9022
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.0264879872928505,
      "learning_rate": 5.030989451734366e-07,
      "loss": 0.2816,
      "step": 9023
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.217642055611893,
      "learning_rate": 5.026666999489199e-07,
      "loss": 0.272,
      "step": 9024
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.1540529235570602,
      "learning_rate": 5.022346197342884e-07,
      "loss": 0.1394,
      "step": 9025
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.383248515668089,
      "learning_rate": 5.018027045652382e-07,
      "loss": 0.1776,
      "step": 9026
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.499174802206997,
      "learning_rate": 5.013709544774515e-07,
      "loss": 0.3153,
      "step": 9027
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.451338571159618,
      "learning_rate": 5.009393695065987e-07,
      "loss": 0.3204,
      "step": 9028
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9390123019659353,
      "learning_rate": 5.00507949688333e-07,
      "loss": 0.2418,
      "step": 9029
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.9332302854311365,
      "learning_rate": 5.000766950582977e-07,
      "loss": 0.5471,
      "step": 9030
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.619500349172152,
      "learning_rate": 4.996456056521223e-07,
      "loss": 0.4705,
      "step": 9031
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.808440381143129,
      "learning_rate": 4.99214681505418e-07,
      "loss": 0.3752,
      "step": 9032
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.344220665996377,
      "learning_rate": 4.987839226537891e-07,
      "loss": 0.1321,
      "step": 9033
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.941284101098663,
      "learning_rate": 4.983533291328196e-07,
      "loss": 0.1502,
      "step": 9034
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.290148005452687,
      "learning_rate": 4.979229009780845e-07,
      "loss": 0.3766,
      "step": 9035
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.349474056879919,
      "learning_rate": 4.974926382251449e-07,
      "loss": 0.3521,
      "step": 9036
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.6507476898450064,
      "learning_rate": 4.970625409095444e-07,
      "loss": 0.0818,
      "step": 9037
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.0793522079090785,
      "learning_rate": 4.966326090668184e-07,
      "loss": 0.1755,
      "step": 9038
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.1986479038698175,
      "learning_rate": 4.962028427324828e-07,
      "loss": 0.2727,
      "step": 9039
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.466183830614315,
      "learning_rate": 4.957732419420444e-07,
      "loss": 0.2314,
      "step": 9040
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.338157357142586,
      "learning_rate": 4.953438067309957e-07,
      "loss": 0.3017,
      "step": 9041
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.108061607316047,
      "learning_rate": 4.949145371348121e-07,
      "loss": 0.2415,
      "step": 9042
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.7636074471958207,
      "learning_rate": 4.944854331889606e-07,
      "loss": 0.1607,
      "step": 9043
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.6436152704575555,
      "learning_rate": 4.940564949288884e-07,
      "loss": 0.3247,
      "step": 9044
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.635343530102377,
      "learning_rate": 4.936277223900343e-07,
      "loss": 0.4678,
      "step": 9045
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.8826237122544836,
      "learning_rate": 4.931991156078222e-07,
      "loss": 0.6614,
      "step": 9046
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.588290260713593,
      "learning_rate": 4.927706746176592e-07,
      "loss": 0.4287,
      "step": 9047
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.215376974978421,
      "learning_rate": 4.923423994549434e-07,
      "loss": 0.2216,
      "step": 9048
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.601869530332023,
      "learning_rate": 4.919142901550541e-07,
      "loss": 0.1005,
      "step": 9049
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.785738513883449,
      "learning_rate": 4.914863467533609e-07,
      "loss": 0.4232,
      "step": 9050
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.707324033908159,
      "learning_rate": 4.910585692852201e-07,
      "loss": 0.1989,
      "step": 9051
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.211858763008401,
      "learning_rate": 4.906309577859697e-07,
      "loss": 0.1424,
      "step": 9052
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.115761550334378,
      "learning_rate": 4.902035122909393e-07,
      "loss": 0.1672,
      "step": 9053
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9337479944633604,
      "learning_rate": 4.897762328354403e-07,
      "loss": 0.3718,
      "step": 9054
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.775655428451525,
      "learning_rate": 4.893491194547728e-07,
      "loss": 0.4964,
      "step": 9055
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.778225554335005,
      "learning_rate": 4.88922172184225e-07,
      "loss": 0.4056,
      "step": 9056
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.656638398890276,
      "learning_rate": 4.884953910590662e-07,
      "loss": 0.25,
      "step": 9057
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.26399976466887,
      "learning_rate": 4.880687761145569e-07,
      "loss": 0.3444,
      "step": 9058
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.230798916975702,
      "learning_rate": 4.876423273859413e-07,
      "loss": 0.125,
      "step": 9059
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.0696169673083356,
      "learning_rate": 4.872160449084501e-07,
      "loss": 0.1664,
      "step": 9060
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.9110711795117465,
      "learning_rate": 4.867899287173023e-07,
      "loss": 0.2407,
      "step": 9061
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.861836405316314,
      "learning_rate": 4.86363978847699e-07,
      "loss": 0.2572,
      "step": 9062
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.7105808433336,
      "learning_rate": 4.859381953348322e-07,
      "loss": 0.2379,
      "step": 9063
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.367538104023535,
      "learning_rate": 4.855125782138772e-07,
      "loss": 0.3372,
      "step": 9064
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.206224837032614,
      "learning_rate": 4.850871275199959e-07,
      "loss": 0.2002,
      "step": 9065
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.054538276801671,
      "learning_rate": 4.846618432883388e-07,
      "loss": 0.3528,
      "step": 9066
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.18687918875796,
      "learning_rate": 4.842367255540381e-07,
      "loss": 0.3972,
      "step": 9067
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.6304149664303957,
      "learning_rate": 4.838117743522172e-07,
      "loss": 0.2137,
      "step": 9068
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.167596385666117,
      "learning_rate": 4.833869897179827e-07,
      "loss": 0.4458,
      "step": 9069
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.089004064810101,
      "learning_rate": 4.829623716864273e-07,
      "loss": 0.3164,
      "step": 9070
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.054509469948028,
      "learning_rate": 4.825379202926333e-07,
      "loss": 0.3151,
      "step": 9071
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7355056147198207,
      "learning_rate": 4.82113635571664e-07,
      "loss": 0.2483,
      "step": 9072
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.182831723833043,
      "learning_rate": 4.816895175585734e-07,
      "loss": 0.2727,
      "step": 9073
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.555543504500051,
      "learning_rate": 4.812655662883997e-07,
      "loss": 0.225,
      "step": 9074
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.763429581527369,
      "learning_rate": 4.808417817961669e-07,
      "loss": 0.3235,
      "step": 9075
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.5191015489637545,
      "learning_rate": 4.804181641168881e-07,
      "loss": 0.2173,
      "step": 9076
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.15481779496356,
      "learning_rate": 4.799947132855578e-07,
      "loss": 0.3637,
      "step": 9077
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.935891749085168,
      "learning_rate": 4.795714293371612e-07,
      "loss": 0.2918,
      "step": 9078
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.5070815256950505,
      "learning_rate": 4.791483123066676e-07,
      "loss": 0.2847,
      "step": 9079
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.743826750419954,
      "learning_rate": 4.787253622290322e-07,
      "loss": 0.3184,
      "step": 9080
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.6460103333249383,
      "learning_rate": 4.783025791391987e-07,
      "loss": 0.1468,
      "step": 9081
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4612879412596902,
      "learning_rate": 4.778799630720932e-07,
      "loss": 0.0148,
      "step": 9082
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.159018084418265,
      "learning_rate": 4.774575140626317e-07,
      "loss": 0.4373,
      "step": 9083
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.80310903818018,
      "learning_rate": 4.770352321457145e-07,
      "loss": 0.432,
      "step": 9084
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.3460972656250245,
      "learning_rate": 4.7661311735622744e-07,
      "loss": 0.3337,
      "step": 9085
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.184404407158843,
      "learning_rate": 4.761911697290461e-07,
      "loss": 0.4457,
      "step": 9086
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.8028779352326203,
      "learning_rate": 4.757693892990267e-07,
      "loss": 0.2645,
      "step": 9087
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.046603989108502,
      "learning_rate": 4.753477761010167e-07,
      "loss": 0.3426,
      "step": 9088
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.4536047802425927,
      "learning_rate": 4.74926330169847e-07,
      "loss": 0.2295,
      "step": 9089
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.178851644411129,
      "learning_rate": 4.7450505154033515e-07,
      "loss": 0.3066,
      "step": 9090
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.403036317431598,
      "learning_rate": 4.740839402472866e-07,
      "loss": 0.2249,
      "step": 9091
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.973468088073221,
      "learning_rate": 4.7366299632548907e-07,
      "loss": 0.2476,
      "step": 9092
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.547529502359525,
      "learning_rate": 4.7324221980972085e-07,
      "loss": 0.3174,
      "step": 9093
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7152706748247066,
      "learning_rate": 4.7282161073474384e-07,
      "loss": 0.1831,
      "step": 9094
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.072918921937422,
      "learning_rate": 4.7240116913530586e-07,
      "loss": 0.4538,
      "step": 9095
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.3520284341189335,
      "learning_rate": 4.719808950461438e-07,
      "loss": 0.2843,
      "step": 9096
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.47121277726958,
      "learning_rate": 4.715607885019763e-07,
      "loss": 0.1613,
      "step": 9097
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.363005912814623,
      "learning_rate": 4.7114084953751195e-07,
      "loss": 0.3491,
      "step": 9098
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.1323920992904135,
      "learning_rate": 4.7072107818744354e-07,
      "loss": 0.3298,
      "step": 9099
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.3178083387816555,
      "learning_rate": 4.7030147448645036e-07,
      "loss": 0.1981,
      "step": 9100
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.26485024328431,
      "learning_rate": 4.6988203846919935e-07,
      "loss": 0.3418,
      "step": 9101
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.742463326638424,
      "learning_rate": 4.6946277017034e-07,
      "loss": 0.2773,
      "step": 9102
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5646910598085633,
      "learning_rate": 4.690436696245118e-07,
      "loss": 0.1543,
      "step": 9103
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.464387348658252,
      "learning_rate": 4.686247368663385e-07,
      "loss": 0.3349,
      "step": 9104
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5956891678247396,
      "learning_rate": 4.6820597193042967e-07,
      "loss": 0.2031,
      "step": 9105
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.140725666356199,
      "learning_rate": 4.6778737485138255e-07,
      "loss": 0.2256,
      "step": 9106
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7860718922123464,
      "learning_rate": 4.6736894566377924e-07,
      "loss": 0.3119,
      "step": 9107
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.8307457592476344,
      "learning_rate": 4.669506844021879e-07,
      "loss": 0.1858,
      "step": 9108
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.3747666829984064,
      "learning_rate": 4.6653259110116396e-07,
      "loss": 0.2035,
      "step": 9109
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.973341961658494,
      "learning_rate": 4.6611466579524675e-07,
      "loss": 0.3571,
      "step": 9110
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.09565942503534,
      "learning_rate": 4.6569690851896497e-07,
      "loss": 0.4023,
      "step": 9111
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.275811387285268,
      "learning_rate": 4.6527931930683113e-07,
      "loss": 0.4373,
      "step": 9112
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.309274829142306,
      "learning_rate": 4.648618981933442e-07,
      "loss": 0.2391,
      "step": 9113
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8028723271804066,
      "learning_rate": 4.644446452129894e-07,
      "loss": 0.0196,
      "step": 9114
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.84327505664472,
      "learning_rate": 4.6402756040023757e-07,
      "loss": 0.3369,
      "step": 9115
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5387909315425556,
      "learning_rate": 4.636106437895477e-07,
      "loss": 0.2558,
      "step": 9116
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.8799491908201964,
      "learning_rate": 4.6319389541536225e-07,
      "loss": 0.1984,
      "step": 9117
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.4088680541472844,
      "learning_rate": 4.627773153121115e-07,
      "loss": 0.2371,
      "step": 9118
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.885431385127828,
      "learning_rate": 4.623609035142107e-07,
      "loss": 0.5851,
      "step": 9119
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9517433460587494,
      "learning_rate": 4.6194466005606176e-07,
      "loss": 0.014,
      "step": 9120
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9524047836899547,
      "learning_rate": 4.615285849720533e-07,
      "loss": 0.1515,
      "step": 9121
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.875739891890848,
      "learning_rate": 4.6111267829655916e-07,
      "loss": 0.4745,
      "step": 9122
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.3891926303750037,
      "learning_rate": 4.6069694006393967e-07,
      "loss": 0.2434,
      "step": 9123
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.4888776792953493,
      "learning_rate": 4.602813703085404e-07,
      "loss": 0.0831,
      "step": 9124
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.767046494301327,
      "learning_rate": 4.5986596906469367e-07,
      "loss": 0.23,
      "step": 9125
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.616928563194551,
      "learning_rate": 4.594507363667192e-07,
      "loss": 0.3866,
      "step": 9126
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.029467096247181,
      "learning_rate": 4.590356722489203e-07,
      "loss": 0.3474,
      "step": 9127
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.6983376415622162,
      "learning_rate": 4.58620776745588e-07,
      "loss": 0.2448,
      "step": 9128
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.383517853396763,
      "learning_rate": 4.5820604989099876e-07,
      "loss": 0.3565,
      "step": 9129
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.9753819251907525,
      "learning_rate": 4.5779149171941455e-07,
      "loss": 0.2122,
      "step": 9130
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.979832340233512,
      "learning_rate": 4.5737710226508553e-07,
      "loss": 0.329,
      "step": 9131
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.930960219078533,
      "learning_rate": 4.5696288156224624e-07,
      "loss": 0.3206,
      "step": 9132
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.594140979136295,
      "learning_rate": 4.5654882964511674e-07,
      "loss": 0.2418,
      "step": 9133
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.845255368006273,
      "learning_rate": 4.561349465479048e-07,
      "loss": 0.2935,
      "step": 9134
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.482269552084935,
      "learning_rate": 4.5572123230480214e-07,
      "loss": 0.2655,
      "step": 9135
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.300389582746559,
      "learning_rate": 4.553076869499895e-07,
      "loss": 0.2416,
      "step": 9136
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3494664325317465,
      "learning_rate": 4.548943105176309e-07,
      "loss": 0.0155,
      "step": 9137
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.379746784145853,
      "learning_rate": 4.5448110304187806e-07,
      "loss": 0.4014,
      "step": 9138
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.7369253929051367,
      "learning_rate": 4.5406806455686763e-07,
      "loss": 0.2349,
      "step": 9139
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.5039034331584156,
      "learning_rate": 4.5365519509672273e-07,
      "loss": 0.2654,
      "step": 9140
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.605301463616669,
      "learning_rate": 4.532424946955533e-07,
      "loss": 0.2266,
      "step": 9141
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.386394427838382,
      "learning_rate": 4.528299633874542e-07,
      "loss": 0.3943,
      "step": 9142
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.585315426815794,
      "learning_rate": 4.5241760120650706e-07,
      "loss": 0.1958,
      "step": 9143
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.99782034439052,
      "learning_rate": 4.5200540818677807e-07,
      "loss": 0.3035,
      "step": 9144
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9710874355400274,
      "learning_rate": 4.5159338436232213e-07,
      "loss": 0.1445,
      "step": 9145
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9025572900208685,
      "learning_rate": 4.5118152976717816e-07,
      "loss": 0.1116,
      "step": 9146
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.4943286276691805,
      "learning_rate": 4.5076984443537115e-07,
      "loss": 0.2904,
      "step": 9147
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.94826465435638,
      "learning_rate": 4.50358328400913e-07,
      "loss": 0.1447,
      "step": 9148
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.4374103493763455,
      "learning_rate": 4.4994698169779997e-07,
      "loss": 0.2438,
      "step": 9149
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.897976094902197,
      "learning_rate": 4.4953580436001734e-07,
      "loss": 0.4414,
      "step": 9150
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.947918444757439,
      "learning_rate": 4.491247964215337e-07,
      "loss": 0.2808,
      "step": 9151
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.283673004815233,
      "learning_rate": 4.487139579163044e-07,
      "loss": 0.2514,
      "step": 9152
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.552143083533885,
      "learning_rate": 4.4830328887827097e-07,
      "loss": 0.4506,
      "step": 9153
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.066256586908642,
      "learning_rate": 4.478927893413601e-07,
      "loss": 0.3301,
      "step": 9154
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.256664387687101,
      "learning_rate": 4.474824593394869e-07,
      "loss": 0.0909,
      "step": 9155
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.321320134115385,
      "learning_rate": 4.4707229890654984e-07,
      "loss": 0.2813,
      "step": 9156
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.1545699866815498,
      "learning_rate": 4.466623080764343e-07,
      "loss": 0.1823,
      "step": 9157
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.598803380768821,
      "learning_rate": 4.462524868830123e-07,
      "loss": 0.2685,
      "step": 9158
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.618006737624096,
      "learning_rate": 4.4584283536013987e-07,
      "loss": 0.195,
      "step": 9159
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.6629557183292873,
      "learning_rate": 4.454333535416622e-07,
      "loss": 0.2893,
      "step": 9160
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.194667962408967,
      "learning_rate": 4.4502404146140795e-07,
      "loss": 0.2114,
      "step": 9161
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.357532319124216,
      "learning_rate": 4.4461489915319204e-07,
      "loss": 0.1123,
      "step": 9162
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.829585012796214,
      "learning_rate": 4.442059266508164e-07,
      "loss": 0.1862,
      "step": 9163
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.026196508036165,
      "learning_rate": 4.4379712398806757e-07,
      "loss": 0.3695,
      "step": 9164
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.927698598662732,
      "learning_rate": 4.433884911987199e-07,
      "loss": 0.3502,
      "step": 9165
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.2884522089289305,
      "learning_rate": 4.42980028316532e-07,
      "loss": 0.2474,
      "step": 9166
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.8791587161467094,
      "learning_rate": 4.42571735375249e-07,
      "loss": 0.2602,
      "step": 9167
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.592301125321855,
      "learning_rate": 4.4216361240860235e-07,
      "loss": 0.4544,
      "step": 9168
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.127662839900107,
      "learning_rate": 4.4175565945030837e-07,
      "loss": 0.2366,
      "step": 9169
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.104764710347647,
      "learning_rate": 4.4134787653407147e-07,
      "loss": 0.2704,
      "step": 9170
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.3363001471459217,
      "learning_rate": 4.409402636935797e-07,
      "loss": 0.2658,
      "step": 9171
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.623685773092339,
      "learning_rate": 4.4053282096250874e-07,
      "loss": 0.0349,
      "step": 9172
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.9543236285302465,
      "learning_rate": 4.401255483745187e-07,
      "loss": 0.2336,
      "step": 9173
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.52313443362204,
      "learning_rate": 4.397184459632564e-07,
      "loss": 0.2062,
      "step": 9174
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.969710707000514,
      "learning_rate": 4.3931151376235577e-07,
      "loss": 0.2959,
      "step": 9175
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.161510234827653,
      "learning_rate": 4.3890475180543486e-07,
      "loss": 0.3384,
      "step": 9176
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.642275024723356,
      "learning_rate": 4.384981601260982e-07,
      "loss": 0.1144,
      "step": 9177
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.520607872292256,
      "learning_rate": 4.380917387579367e-07,
      "loss": 0.162,
      "step": 9178
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.7887315133336745,
      "learning_rate": 4.37685487734526e-07,
      "loss": 0.4226,
      "step": 9179
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.67521861554301,
      "learning_rate": 4.372794070894298e-07,
      "loss": 0.2326,
      "step": 9180
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.294586414565315,
      "learning_rate": 4.3687349685619636e-07,
      "loss": 0.3025,
      "step": 9181
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.359975091646389,
      "learning_rate": 4.3646775706835954e-07,
      "loss": 0.2314,
      "step": 9182
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.822056389499341,
      "learning_rate": 4.360621877594395e-07,
      "loss": 0.3465,
      "step": 9183
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.3307186068579515,
      "learning_rate": 4.3565678896294207e-07,
      "loss": 0.1082,
      "step": 9184
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.63539240280415,
      "learning_rate": 4.3525156071236056e-07,
      "loss": 0.2078,
      "step": 9185
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.778362435769869,
      "learning_rate": 4.348465030411719e-07,
      "loss": 0.3971,
      "step": 9186
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.3258169432457345,
      "learning_rate": 4.344416159828399e-07,
      "loss": 0.2929,
      "step": 9187
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.663813638358913,
      "learning_rate": 4.340368995708158e-07,
      "loss": 0.1761,
      "step": 9188
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8699067778622345,
      "learning_rate": 4.336323538385329e-07,
      "loss": 0.3344,
      "step": 9189
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8659639959438707,
      "learning_rate": 4.332279788194149e-07,
      "loss": 0.1207,
      "step": 9190
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.7291502878623906,
      "learning_rate": 4.328237745468683e-07,
      "loss": 0.1861,
      "step": 9191
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.447434467712385,
      "learning_rate": 4.3241974105428604e-07,
      "loss": 0.1857,
      "step": 9192
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.68469782846401,
      "learning_rate": 4.3201587837504915e-07,
      "loss": 0.4023,
      "step": 9193
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.152784221905315,
      "learning_rate": 4.3161218654252045e-07,
      "loss": 0.344,
      "step": 9194
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.442408285063668,
      "learning_rate": 4.312086655900527e-07,
      "loss": 0.2287,
      "step": 9195
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6143891918710445,
      "learning_rate": 4.3080531555098226e-07,
      "loss": 0.3332,
      "step": 9196
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9952472134885113,
      "learning_rate": 4.304021364586314e-07,
      "loss": 0.1691,
      "step": 9197
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.432751674016792,
      "learning_rate": 4.299991283463106e-07,
      "loss": 0.2447,
      "step": 9198
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.5619326381845653,
      "learning_rate": 4.2959629124731173e-07,
      "loss": 0.1798,
      "step": 9199
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.034375879736207,
      "learning_rate": 4.291936251949175e-07,
      "loss": 0.4152,
      "step": 9200
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.367123526852972,
      "learning_rate": 4.2879113022239293e-07,
      "loss": 0.2142,
      "step": 9201
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.540464272713725,
      "learning_rate": 4.283888063629904e-07,
      "loss": 0.3651,
      "step": 9202
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.8771279800688037,
      "learning_rate": 4.279866536499491e-07,
      "loss": 0.1743,
      "step": 9203
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.461969424516099,
      "learning_rate": 4.2758467211649083e-07,
      "loss": 0.224,
      "step": 9204
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9991807929578145,
      "learning_rate": 4.271828617958271e-07,
      "loss": 0.313,
      "step": 9205
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.436946954904972,
      "learning_rate": 4.2678122272115263e-07,
      "loss": 0.2631,
      "step": 9206
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.378162625278876,
      "learning_rate": 4.2637975492564877e-07,
      "loss": 0.2226,
      "step": 9207
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.1659908662267595,
      "learning_rate": 4.259784584424842e-07,
      "loss": 0.2147,
      "step": 9208
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.0188441197832647,
      "learning_rate": 4.2557733330480993e-07,
      "loss": 0.2272,
      "step": 9209
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.016189436661961,
      "learning_rate": 4.251763795457664e-07,
      "loss": 0.3013,
      "step": 9210
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.791618770679475,
      "learning_rate": 4.2477559719847823e-07,
      "loss": 0.4,
      "step": 9211
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.077231863064907,
      "learning_rate": 4.243749862960553e-07,
      "loss": 0.3131,
      "step": 9212
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.86386290698661,
      "learning_rate": 4.2397454687159614e-07,
      "loss": 0.1698,
      "step": 9213
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.594777853570692,
      "learning_rate": 4.235742789581804e-07,
      "loss": 0.4969,
      "step": 9214
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.471468601787573,
      "learning_rate": 4.23174182588878e-07,
      "loss": 0.2802,
      "step": 9215
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.146624786158862,
      "learning_rate": 4.227742577967428e-07,
      "loss": 0.4602,
      "step": 9216
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.7635528643723575,
      "learning_rate": 4.223745046148134e-07,
      "loss": 0.3184,
      "step": 9217
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.544086138791781,
      "learning_rate": 4.2197492307611775e-07,
      "loss": 0.1834,
      "step": 9218
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9401641429901986,
      "learning_rate": 4.215755132136648e-07,
      "loss": 0.349,
      "step": 9219
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.4488266687716114,
      "learning_rate": 4.211762750604534e-07,
      "loss": 0.1204,
      "step": 9220
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.63887008346664,
      "learning_rate": 4.2077720864946656e-07,
      "loss": 0.3881,
      "step": 9221
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.353048931915217,
      "learning_rate": 4.203783140136719e-07,
      "loss": 0.1678,
      "step": 9222
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.448803163068678,
      "learning_rate": 4.199795911860266e-07,
      "loss": 0.4302,
      "step": 9223
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.014512385183641,
      "learning_rate": 4.1958104019946823e-07,
      "loss": 0.2552,
      "step": 9224
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.329232145124466,
      "learning_rate": 4.191826610869254e-07,
      "loss": 0.2464,
      "step": 9225
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.418085326897973,
      "learning_rate": 4.1878445388130935e-07,
      "loss": 0.4626,
      "step": 9226
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.863871442897281,
      "learning_rate": 4.183864186155173e-07,
      "loss": 0.2102,
      "step": 9227
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.32883746298936,
      "learning_rate": 4.17988555322435e-07,
      "loss": 0.4347,
      "step": 9228
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.0319592261519617,
      "learning_rate": 4.1759086403492966e-07,
      "loss": 0.2045,
      "step": 9229
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.205281934396549,
      "learning_rate": 4.171933447858581e-07,
      "loss": 0.2465,
      "step": 9230
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.237492324623599,
      "learning_rate": 4.16795997608061e-07,
      "loss": 0.26,
      "step": 9231
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.380304475628687,
      "learning_rate": 4.1639882253436456e-07,
      "loss": 0.2552,
      "step": 9232
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.7300054518440584,
      "learning_rate": 4.160018195975832e-07,
      "loss": 0.2938,
      "step": 9233
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.172553634949707,
      "learning_rate": 4.1560498883051303e-07,
      "loss": 0.0144,
      "step": 9234
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.429653618492404,
      "learning_rate": 4.152083302659396e-07,
      "loss": 0.3312,
      "step": 9235
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.744899349921562,
      "learning_rate": 4.1481184393663385e-07,
      "loss": 0.2072,
      "step": 9236
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6980018141256217,
      "learning_rate": 4.144155298753491e-07,
      "loss": 0.1278,
      "step": 9237
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.840088831727061,
      "learning_rate": 4.1401938811482966e-07,
      "loss": 0.2712,
      "step": 9238
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.201991235428539,
      "learning_rate": 4.1362341868779994e-07,
      "loss": 0.3561,
      "step": 9239
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.6281048086733825,
      "learning_rate": 4.1322762162697453e-07,
      "loss": 0.2797,
      "step": 9240
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.112608244261901,
      "learning_rate": 4.1283199696505324e-07,
      "loss": 0.2984,
      "step": 9241
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.4577071380583844,
      "learning_rate": 4.124365447347184e-07,
      "loss": 0.2149,
      "step": 9242
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.300731286731129,
      "learning_rate": 4.1204126496864253e-07,
      "loss": 0.201,
      "step": 9243
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.243091379797953,
      "learning_rate": 4.116461576994796e-07,
      "loss": 0.3548,
      "step": 9244
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.635148895017867,
      "learning_rate": 4.112512229598725e-07,
      "loss": 0.361,
      "step": 9245
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.993714620192397,
      "learning_rate": 4.1085646078244974e-07,
      "loss": 0.2867,
      "step": 9246
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35279612101467245,
      "learning_rate": 4.1046187119982264e-07,
      "loss": 0.005,
      "step": 9247
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.238174531639016,
      "learning_rate": 4.100674542445926e-07,
      "loss": 0.3333,
      "step": 9248
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.383503614504453,
      "learning_rate": 4.096732099493417e-07,
      "loss": 0.376,
      "step": 9249
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.148043383717671,
      "learning_rate": 4.0927913834664185e-07,
      "loss": 0.3592,
      "step": 9250
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9800269884774406,
      "learning_rate": 4.088852394690507e-07,
      "loss": 0.0811,
      "step": 9251
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6148048538552464,
      "learning_rate": 4.084915133491077e-07,
      "loss": 0.0533,
      "step": 9252
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.303224527801266,
      "learning_rate": 4.080979600193427e-07,
      "loss": 0.371,
      "step": 9253
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.7952277166138013,
      "learning_rate": 4.0770457951226723e-07,
      "loss": 0.1584,
      "step": 9254
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.679263187834224,
      "learning_rate": 4.0731137186038135e-07,
      "loss": 0.1742,
      "step": 9255
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.5931501773911005,
      "learning_rate": 4.0691833709617127e-07,
      "loss": 0.3851,
      "step": 9256
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.746941942160687,
      "learning_rate": 4.06525475252105e-07,
      "loss": 0.3703,
      "step": 9257
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.5107582196526876,
      "learning_rate": 4.0613278636064153e-07,
      "loss": 0.2248,
      "step": 9258
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.233757884903066,
      "learning_rate": 4.057402704542204e-07,
      "loss": 0.3668,
      "step": 9259
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.7655578862612167,
      "learning_rate": 4.053479275652705e-07,
      "loss": 0.2311,
      "step": 9260
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.4682194031125433,
      "learning_rate": 4.049557577262067e-07,
      "loss": 0.2137,
      "step": 9261
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.399741534066923,
      "learning_rate": 4.045637609694253e-07,
      "loss": 0.3968,
      "step": 9262
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8659768827554966,
      "learning_rate": 4.0417193732731316e-07,
      "loss": 0.2628,
      "step": 9263
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.547924696794093,
      "learning_rate": 4.037802868322405e-07,
      "loss": 0.1595,
      "step": 9264
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.9597832796444226,
      "learning_rate": 4.033888095165625e-07,
      "loss": 0.199,
      "step": 9265
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.951203365135937,
      "learning_rate": 4.029975054126231e-07,
      "loss": 0.3384,
      "step": 9266
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.7981046759783523,
      "learning_rate": 4.026063745527478e-07,
      "loss": 0.335,
      "step": 9267
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.3101969396582227,
      "learning_rate": 4.0221541696925134e-07,
      "loss": 0.1389,
      "step": 9268
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.267832006611836,
      "learning_rate": 4.0182463269443223e-07,
      "loss": 0.3287,
      "step": 9269
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.328493111042738,
      "learning_rate": 4.0143402176057443e-07,
      "loss": 0.0984,
      "step": 9270
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.665748154697875,
      "learning_rate": 4.0104358419995047e-07,
      "loss": 0.4582,
      "step": 9271
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.072728143393557,
      "learning_rate": 4.006533200448137e-07,
      "loss": 0.47,
      "step": 9272
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.182137479553119,
      "learning_rate": 4.0026322932740754e-07,
      "loss": 0.2559,
      "step": 9273
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.663123348588128,
      "learning_rate": 3.9987331207995916e-07,
      "loss": 0.3053,
      "step": 9274
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.402193847645892,
      "learning_rate": 3.994835683346804e-07,
      "loss": 0.2449,
      "step": 9275
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.998968526930851,
      "learning_rate": 3.9909399812377264e-07,
      "loss": 0.3044,
      "step": 9276
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.33789853751419,
      "learning_rate": 3.9870460147941684e-07,
      "loss": 0.3259,
      "step": 9277
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.611652085790328,
      "learning_rate": 3.9831537843378563e-07,
      "loss": 0.3046,
      "step": 9278
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.714253281447115,
      "learning_rate": 3.979263290190341e-07,
      "loss": 0.4265,
      "step": 9279
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.6831739133041324,
      "learning_rate": 3.9753745326730234e-07,
      "loss": 0.1374,
      "step": 9280
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.1280058299918965,
      "learning_rate": 3.971487512107197e-07,
      "loss": 0.1753,
      "step": 9281
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.879073664564951,
      "learning_rate": 3.9676022288139657e-07,
      "loss": 0.1754,
      "step": 9282
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.643630851979538,
      "learning_rate": 3.9637186831143257e-07,
      "loss": 0.3146,
      "step": 9283
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.043793740146546,
      "learning_rate": 3.959836875329112e-07,
      "loss": 0.385,
      "step": 9284
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.198888243936436,
      "learning_rate": 3.9559568057790154e-07,
      "loss": 0.3107,
      "step": 9285
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.030325338750401,
      "learning_rate": 3.9520784747846067e-07,
      "loss": 0.069,
      "step": 9286
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.505332815360947,
      "learning_rate": 3.948201882666272e-07,
      "loss": 0.4431,
      "step": 9287
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.2540562132530715,
      "learning_rate": 3.944327029744291e-07,
      "loss": 0.3125,
      "step": 9288
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.736582668374588,
      "learning_rate": 3.94045391633878e-07,
      "loss": 0.2952,
      "step": 9289
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.77434692462079,
      "learning_rate": 3.9365825427697094e-07,
      "loss": 0.1874,
      "step": 9290
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.775616946819947,
      "learning_rate": 3.9327129093569336e-07,
      "loss": 0.2381,
      "step": 9291
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.239044340156684,
      "learning_rate": 3.928845016420119e-07,
      "loss": 0.276,
      "step": 9292
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.058186089202461,
      "learning_rate": 3.924978864278828e-07,
      "loss": 0.2309,
      "step": 9293
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.2504440599038364,
      "learning_rate": 3.9211144532524586e-07,
      "loss": 0.1986,
      "step": 9294
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.8306081186514165,
      "learning_rate": 3.917251783660264e-07,
      "loss": 0.1723,
      "step": 9295
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.271231752638115,
      "learning_rate": 3.913390855821375e-07,
      "loss": 0.3381,
      "step": 9296
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.697604157160422,
      "learning_rate": 3.9095316700547394e-07,
      "loss": 0.3058,
      "step": 9297
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.048051695264145,
      "learning_rate": 3.9056742266792016e-07,
      "loss": 0.3814,
      "step": 9298
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7606881355994886,
      "learning_rate": 3.9018185260134375e-07,
      "loss": 0.1702,
      "step": 9299
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.363164603913762,
      "learning_rate": 3.897964568375984e-07,
      "loss": 0.3731,
      "step": 9300
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.714320145615825,
      "learning_rate": 3.894112354085253e-07,
      "loss": 0.3281,
      "step": 9301
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.5919518322934,
      "learning_rate": 3.89026188345947e-07,
      "loss": 0.3587,
      "step": 9302
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.304510856417021,
      "learning_rate": 3.8864131568167594e-07,
      "loss": 0.287,
      "step": 9303
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.7724795846321975,
      "learning_rate": 3.88256617447508e-07,
      "loss": 0.1817,
      "step": 9304
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.367806852566984,
      "learning_rate": 3.8787209367522477e-07,
      "loss": 0.0747,
      "step": 9305
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2375608124194755,
      "learning_rate": 3.8748774439659407e-07,
      "loss": 0.2166,
      "step": 9306
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.6066906808735015,
      "learning_rate": 3.871035696433692e-07,
      "loss": 0.2451,
      "step": 9307
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.9951622857497755,
      "learning_rate": 3.8671956944728807e-07,
      "loss": 0.3319,
      "step": 9308
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.234021076795367,
      "learning_rate": 3.8633574384007554e-07,
      "loss": 0.456,
      "step": 9309
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4984824379344763,
      "learning_rate": 3.859520928534405e-07,
      "loss": 0.2468,
      "step": 9310
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.630816796861273,
      "learning_rate": 3.855686165190792e-07,
      "loss": 0.3192,
      "step": 9311
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3864815450450707,
      "learning_rate": 3.8518531486867265e-07,
      "loss": 0.0391,
      "step": 9312
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.33525837046343,
      "learning_rate": 3.84802187933887e-07,
      "loss": 0.2969,
      "step": 9313
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.625432542237824,
      "learning_rate": 3.844192357463741e-07,
      "loss": 0.1246,
      "step": 9314
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.370558087785415,
      "learning_rate": 3.840364583377715e-07,
      "loss": 0.5755,
      "step": 9315
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.8160853963534422,
      "learning_rate": 3.8365385573970295e-07,
      "loss": 0.2507,
      "step": 9316
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.361810295739783,
      "learning_rate": 3.832714279837771e-07,
      "loss": 0.2478,
      "step": 9317
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.539860851388952,
      "learning_rate": 3.828891751015884e-07,
      "loss": 0.327,
      "step": 9318
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2146019935182593,
      "learning_rate": 3.825070971247161e-07,
      "loss": 0.2726,
      "step": 9319
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.9773004678702217,
      "learning_rate": 3.8212519408472535e-07,
      "loss": 0.1757,
      "step": 9320
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.38295152153268,
      "learning_rate": 3.817434660131683e-07,
      "loss": 0.2514,
      "step": 9321
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.9615331830176945,
      "learning_rate": 3.8136191294158075e-07,
      "loss": 0.3244,
      "step": 9322
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.653643998356154,
      "learning_rate": 3.8098053490148504e-07,
      "loss": 0.4634,
      "step": 9323
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.674120469792287,
      "learning_rate": 3.805993319243886e-07,
      "loss": 0.3808,
      "step": 9324
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.268308077467454,
      "learning_rate": 3.8021830404178365e-07,
      "loss": 0.4907,
      "step": 9325
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.146218444869314,
      "learning_rate": 3.798374512851505e-07,
      "loss": 0.3547,
      "step": 9326
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.326075178405727,
      "learning_rate": 3.7945677368595225e-07,
      "loss": 0.4331,
      "step": 9327
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.3883411729093,
      "learning_rate": 3.790762712756391e-07,
      "loss": 0.3216,
      "step": 9328
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.662092426750122,
      "learning_rate": 3.7869594408564607e-07,
      "loss": 0.2674,
      "step": 9329
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.6966681582881025,
      "learning_rate": 3.7831579214739347e-07,
      "loss": 0.1297,
      "step": 9330
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.8953409818600826,
      "learning_rate": 3.7793581549228826e-07,
      "loss": 0.2612,
      "step": 9331
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.898594784762063,
      "learning_rate": 3.7755601415172234e-07,
      "loss": 0.4713,
      "step": 9332
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3274150186175384,
      "learning_rate": 3.771763881570728e-07,
      "loss": 0.1491,
      "step": 9333
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.6253006421527143,
      "learning_rate": 3.767969375397024e-07,
      "loss": 0.288,
      "step": 9334
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.297689495679991,
      "learning_rate": 3.764176623309587e-07,
      "loss": 0.5313,
      "step": 9335
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.303818314248304,
      "learning_rate": 3.7603856256217735e-07,
      "loss": 0.2319,
      "step": 9336
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.094022891425856,
      "learning_rate": 3.7565963826467654e-07,
      "loss": 0.1849,
      "step": 9337
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.519177992634037,
      "learning_rate": 3.7528088946976125e-07,
      "loss": 0.313,
      "step": 9338
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.6229600195055203,
      "learning_rate": 3.7490231620872203e-07,
      "loss": 0.2463,
      "step": 9339
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.943899312073724,
      "learning_rate": 3.745239185128341e-07,
      "loss": 0.3145,
      "step": 9340
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.4211112812499564,
      "learning_rate": 3.741456964133597e-07,
      "loss": 0.4589,
      "step": 9341
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.112587552551278,
      "learning_rate": 3.737676499415457e-07,
      "loss": 0.3751,
      "step": 9342
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.8143510811846815,
      "learning_rate": 3.7338977912862394e-07,
      "loss": 0.1195,
      "step": 9343
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.948612384893383,
      "learning_rate": 3.730120840058124e-07,
      "loss": 0.4474,
      "step": 9344
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.1297375980944695,
      "learning_rate": 3.726345646043139e-07,
      "loss": 0.3251,
      "step": 9345
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.971071143962724,
      "learning_rate": 3.722572209553182e-07,
      "loss": 0.5586,
      "step": 9346
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.6115886212748345,
      "learning_rate": 3.7188005308999937e-07,
      "loss": 0.2525,
      "step": 9347
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.6056418697272394,
      "learning_rate": 3.715030610395168e-07,
      "loss": 0.2555,
      "step": 9348
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.127282670665531,
      "learning_rate": 3.711262448350153e-07,
      "loss": 0.395,
      "step": 9349
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.54374094087997,
      "learning_rate": 3.70749604507627e-07,
      "loss": 0.3502,
      "step": 9350
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.536937264506309,
      "learning_rate": 3.7037314008846694e-07,
      "loss": 0.3345,
      "step": 9351
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.9580019422847035,
      "learning_rate": 3.699968516086369e-07,
      "loss": 0.2268,
      "step": 9352
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.661606374894025,
      "learning_rate": 3.696207390992243e-07,
      "loss": 0.3533,
      "step": 9353
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.288672722623781,
      "learning_rate": 3.692448025913012e-07,
      "loss": 0.2794,
      "step": 9354
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.778661089135148,
      "learning_rate": 3.688690421159263e-07,
      "loss": 0.2954,
      "step": 9355
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.943861090553794,
      "learning_rate": 3.684934577041427e-07,
      "loss": 0.3099,
      "step": 9356
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.044428833443906,
      "learning_rate": 3.681180493869796e-07,
      "loss": 0.2012,
      "step": 9357
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.093562810586067,
      "learning_rate": 3.6774281719545114e-07,
      "loss": 0.2755,
      "step": 9358
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4261456580548844,
      "learning_rate": 3.6736776116055665e-07,
      "loss": 0.1463,
      "step": 9359
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.615217114297448,
      "learning_rate": 3.6699288131328255e-07,
      "loss": 0.2686,
      "step": 9360
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4897920210871223,
      "learning_rate": 3.6661817768459923e-07,
      "loss": 0.2578,
      "step": 9361
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.126563303222722,
      "learning_rate": 3.662436503054623e-07,
      "loss": 0.4157,
      "step": 9362
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.74771499166004,
      "learning_rate": 3.6586929920681414e-07,
      "loss": 0.2956,
      "step": 9363
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.7180296079928326,
      "learning_rate": 3.6549512441958034e-07,
      "loss": 0.4921,
      "step": 9364
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.23098735725296,
      "learning_rate": 3.6512112597467534e-07,
      "loss": 0.3418,
      "step": 9365
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.21768046572261,
      "learning_rate": 3.6474730390299614e-07,
      "loss": 0.3161,
      "step": 9366
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.81630518720885,
      "learning_rate": 3.6437365823542625e-07,
      "loss": 0.3989,
      "step": 9367
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.086228028262036,
      "learning_rate": 3.6400018900283415e-07,
      "loss": 0.4062,
      "step": 9368
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.774140656669381,
      "learning_rate": 3.636268962360734e-07,
      "loss": 0.2038,
      "step": 9369
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.16406684414436,
      "learning_rate": 3.632537799659852e-07,
      "loss": 0.4087,
      "step": 9370
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2046675388507655,
      "learning_rate": 3.6288084022339383e-07,
      "loss": 0.1542,
      "step": 9371
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.464922513234247,
      "learning_rate": 3.6250807703910983e-07,
      "loss": 0.2614,
      "step": 9372
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.252202124191459,
      "learning_rate": 3.6213549044392863e-07,
      "loss": 0.1442,
      "step": 9373
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.611133288076002,
      "learning_rate": 3.6176308046863144e-07,
      "loss": 0.2788,
      "step": 9374
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.7439055642334043,
      "learning_rate": 3.6139084714398584e-07,
      "loss": 0.1268,
      "step": 9375
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.282953678801409,
      "learning_rate": 3.610187905007434e-07,
      "loss": 0.5228,
      "step": 9376
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.183727320972399,
      "learning_rate": 3.606469105696417e-07,
      "loss": 0.6179,
      "step": 9377
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.3737031268288193,
      "learning_rate": 3.6027520738140366e-07,
      "loss": 0.1721,
      "step": 9378
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.4460838823649,
      "learning_rate": 3.5990368096673664e-07,
      "loss": 0.4192,
      "step": 9379
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.099862146356922,
      "learning_rate": 3.5953233135633594e-07,
      "loss": 0.3706,
      "step": 9380
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.852536631781468,
      "learning_rate": 3.591611585808799e-07,
      "loss": 0.3293,
      "step": 9381
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.9897453670007703,
      "learning_rate": 3.5879016267103274e-07,
      "loss": 0.2785,
      "step": 9382
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.3482146665065455,
      "learning_rate": 3.5841934365744485e-07,
      "loss": 0.38,
      "step": 9383
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.92095352938456,
      "learning_rate": 3.580487015707504e-07,
      "loss": 0.3247,
      "step": 9384
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.682351157987579,
      "learning_rate": 3.5767823644157153e-07,
      "loss": 0.0728,
      "step": 9385
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.270368581155445,
      "learning_rate": 3.5730794830051345e-07,
      "loss": 0.2026,
      "step": 9386
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.4460725082389,
      "learning_rate": 3.5693783717816704e-07,
      "loss": 0.3519,
      "step": 9387
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.249318069100588,
      "learning_rate": 3.5656790310511085e-07,
      "loss": 0.2927,
      "step": 9388
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.316703943043044,
      "learning_rate": 3.561981461119046e-07,
      "loss": 0.2817,
      "step": 9389
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.388328601414643,
      "learning_rate": 3.558285662290975e-07,
      "loss": 0.1276,
      "step": 9390
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.525596832773664,
      "learning_rate": 3.554591634872223e-07,
      "loss": 0.4122,
      "step": 9391
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.746216204046719,
      "learning_rate": 3.5508993791679603e-07,
      "loss": 0.376,
      "step": 9392
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.80201685560572,
      "learning_rate": 3.547208895483242e-07,
      "loss": 0.3517,
      "step": 9393
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.650965055959909,
      "learning_rate": 3.5435201841229393e-07,
      "loss": 0.4585,
      "step": 9394
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2692114118916837,
      "learning_rate": 3.5398332453918077e-07,
      "loss": 0.2086,
      "step": 9395
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.571276055675763,
      "learning_rate": 3.53614807959444e-07,
      "loss": 0.4214,
      "step": 9396
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.537080442673252,
      "learning_rate": 3.5324646870352806e-07,
      "loss": 0.335,
      "step": 9397
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.25209360532118913,
      "learning_rate": 3.5287830680186514e-07,
      "loss": 0.0025,
      "step": 9398
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.210248908751285,
      "learning_rate": 3.525103222848683e-07,
      "loss": 0.2027,
      "step": 9399
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.969721543865682,
      "learning_rate": 3.521425151829408e-07,
      "loss": 0.2028,
      "step": 9400
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.651351993141943,
      "learning_rate": 3.5177488552646853e-07,
      "loss": 0.3999,
      "step": 9401
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.452173782351135,
      "learning_rate": 3.5140743334582227e-07,
      "loss": 0.4293,
      "step": 9402
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.281598715163998,
      "learning_rate": 3.5104015867136106e-07,
      "loss": 0.2928,
      "step": 9403
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.393878265093833,
      "learning_rate": 3.50673061533425e-07,
      "loss": 0.2422,
      "step": 9404
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.469290972037537,
      "learning_rate": 3.503061419623435e-07,
      "loss": 0.2651,
      "step": 9405
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.751564652832937,
      "learning_rate": 3.499393999884293e-07,
      "loss": 0.236,
      "step": 9406
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.2003792359297303,
      "learning_rate": 3.4957283564198e-07,
      "loss": 0.2422,
      "step": 9407
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.510950190455714,
      "learning_rate": 3.492064489532812e-07,
      "loss": 0.3991,
      "step": 9408
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.187728132898207,
      "learning_rate": 3.4884023995259973e-07,
      "loss": 0.2698,
      "step": 9409
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.994922892246824,
      "learning_rate": 3.484742086701917e-07,
      "loss": 0.1695,
      "step": 9410
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.771897367889627,
      "learning_rate": 3.4810835513629626e-07,
      "loss": 0.2626,
      "step": 9411
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.962417300952501,
      "learning_rate": 3.4774267938113756e-07,
      "loss": 0.2889,
      "step": 9412
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9920914552472184,
      "learning_rate": 3.4737718143492804e-07,
      "loss": 0.0791,
      "step": 9413
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.505510046449945,
      "learning_rate": 3.4701186132786084e-07,
      "loss": 0.4229,
      "step": 9414
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.0599008901078495,
      "learning_rate": 3.4664671909011875e-07,
      "loss": 0.237,
      "step": 9415
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.684817859616017,
      "learning_rate": 3.462817547518674e-07,
      "loss": 0.3312,
      "step": 9416
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5643249573353324,
      "learning_rate": 3.4591696834325754e-07,
      "loss": 0.1712,
      "step": 9417
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.598495544068035,
      "learning_rate": 3.4555235989442825e-07,
      "loss": 0.2102,
      "step": 9418
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.469753107399522,
      "learning_rate": 3.451879294354993e-07,
      "loss": 0.3003,
      "step": 9419
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.495745519181699,
      "learning_rate": 3.448236769965793e-07,
      "loss": 0.3614,
      "step": 9420
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.101699249104842,
      "learning_rate": 3.444596026077612e-07,
      "loss": 0.1379,
      "step": 9421
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.162347833576515,
      "learning_rate": 3.4409570629912207e-07,
      "loss": 0.4079,
      "step": 9422
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9791097588181628,
      "learning_rate": 3.437319881007267e-07,
      "loss": 0.1487,
      "step": 9423
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.536170586272887,
      "learning_rate": 3.433684480426222e-07,
      "loss": 0.172,
      "step": 9424
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.750528422114262,
      "learning_rate": 3.430050861548434e-07,
      "loss": 0.4324,
      "step": 9425
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.072207850882751,
      "learning_rate": 3.426419024674091e-07,
      "loss": 0.2004,
      "step": 9426
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.683532392824596,
      "learning_rate": 3.4227889701032357e-07,
      "loss": 0.3781,
      "step": 9427
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.244655496705043,
      "learning_rate": 3.419160698135779e-07,
      "loss": 0.4254,
      "step": 9428
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.16311191813995,
      "learning_rate": 3.415534209071447e-07,
      "loss": 0.1941,
      "step": 9429
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.939418632001551,
      "learning_rate": 3.411909503209862e-07,
      "loss": 0.2801,
      "step": 9430
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.668657407970279,
      "learning_rate": 3.408286580850476e-07,
      "loss": 0.4436,
      "step": 9431
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.246327504186439,
      "learning_rate": 3.4046654422925845e-07,
      "loss": 0.3669,
      "step": 9432
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.599556377216871,
      "learning_rate": 3.401046087835372e-07,
      "loss": 0.2306,
      "step": 9433
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.475999860626261,
      "learning_rate": 3.397428517777826e-07,
      "loss": 0.0085,
      "step": 9434
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.645049192503659,
      "learning_rate": 3.393812732418822e-07,
      "loss": 0.1548,
      "step": 9435
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.3315334687966893,
      "learning_rate": 3.390198732057093e-07,
      "loss": 0.1987,
      "step": 9436
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.063886507082756,
      "learning_rate": 3.386586516991189e-07,
      "loss": 0.249,
      "step": 9437
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5482159902687216,
      "learning_rate": 3.3829760875195515e-07,
      "loss": 0.2647,
      "step": 9438
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.07843642276344,
      "learning_rate": 3.3793674439404345e-07,
      "loss": 0.4411,
      "step": 9439
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.871968245601699,
      "learning_rate": 3.3757605865519796e-07,
      "loss": 0.2445,
      "step": 9440
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.670548197553006,
      "learning_rate": 3.3721555156521784e-07,
      "loss": 0.1503,
      "step": 9441
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.563350889826039,
      "learning_rate": 3.3685522315388434e-07,
      "loss": 0.4405,
      "step": 9442
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.419410282556681,
      "learning_rate": 3.364950734509678e-07,
      "loss": 0.3103,
      "step": 9443
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.359437341839115,
      "learning_rate": 3.3613510248622027e-07,
      "loss": 0.5739,
      "step": 9444
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.3128960355103505,
      "learning_rate": 3.357753102893818e-07,
      "loss": 0.4716,
      "step": 9445
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5922763075351245,
      "learning_rate": 3.3541569689017755e-07,
      "loss": 0.1474,
      "step": 9446
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.567893310660708,
      "learning_rate": 3.350562623183151e-07,
      "loss": 0.3633,
      "step": 9447
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.808649198498416,
      "learning_rate": 3.34697006603491e-07,
      "loss": 0.326,
      "step": 9448
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.965791284014009,
      "learning_rate": 3.343379297753832e-07,
      "loss": 0.2994,
      "step": 9449
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.594175050889286,
      "learning_rate": 3.339790318636582e-07,
      "loss": 0.2322,
      "step": 9450
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.454089131126899,
      "learning_rate": 3.33620312897967e-07,
      "loss": 0.1061,
      "step": 9451
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.293342359734503,
      "learning_rate": 3.332617729079435e-07,
      "loss": 0.3532,
      "step": 9452
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.494744722138192,
      "learning_rate": 3.3290341192321023e-07,
      "loss": 0.2894,
      "step": 9453
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.003667685056496,
      "learning_rate": 3.325452299733714e-07,
      "loss": 0.3322,
      "step": 9454
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.8395279134621383,
      "learning_rate": 3.321872270880191e-07,
      "loss": 0.2113,
      "step": 9455
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.729089727489507,
      "learning_rate": 3.318294032967312e-07,
      "loss": 0.2436,
      "step": 9456
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.103801780807034,
      "learning_rate": 3.3147175862906666e-07,
      "loss": 0.2328,
      "step": 9457
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.0883340242502815,
      "learning_rate": 3.31114293114575e-07,
      "loss": 0.3535,
      "step": 9458
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.447489132108724,
      "learning_rate": 3.307570067827859e-07,
      "loss": 0.2557,
      "step": 9459
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.463833809112086,
      "learning_rate": 3.303998996632177e-07,
      "loss": 0.3885,
      "step": 9460
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.945855625835781,
      "learning_rate": 3.300429717853737e-07,
      "loss": 0.2928,
      "step": 9461
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.357105307795199,
      "learning_rate": 3.2968622317873956e-07,
      "loss": 0.2706,
      "step": 9462
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.742860139567832,
      "learning_rate": 3.2932965387278996e-07,
      "loss": 0.38,
      "step": 9463
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.546406727343277,
      "learning_rate": 3.28973263896982e-07,
      "loss": 0.2598,
      "step": 9464
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5421609681765778,
      "learning_rate": 3.286170532807584e-07,
      "loss": 0.0525,
      "step": 9465
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.638958049466665,
      "learning_rate": 3.2826102205354927e-07,
      "loss": 0.4388,
      "step": 9466
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.908845304085821,
      "learning_rate": 3.279051702447661e-07,
      "loss": 0.3447,
      "step": 9467
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.959046997545225,
      "learning_rate": 3.2754949788380897e-07,
      "loss": 0.2344,
      "step": 9468
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.694329457048012,
      "learning_rate": 3.2719400500006133e-07,
      "loss": 0.3045,
      "step": 9469
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.7491823489351646,
      "learning_rate": 3.268386916228916e-07,
      "loss": 0.4245,
      "step": 9470
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.23861659914413,
      "learning_rate": 3.2648355778165607e-07,
      "loss": 0.2019,
      "step": 9471
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.4109517047680775,
      "learning_rate": 3.2612860350569184e-07,
      "loss": 0.3073,
      "step": 9472
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.812785212008319,
      "learning_rate": 3.257738288243248e-07,
      "loss": 0.1746,
      "step": 9473
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.2844972920757844,
      "learning_rate": 3.2541923376686433e-07,
      "loss": 0.1836,
      "step": 9474
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.823080757325993,
      "learning_rate": 3.2506481836260483e-07,
      "loss": 0.3264,
      "step": 9475
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.8447611863528905,
      "learning_rate": 3.247105826408281e-07,
      "loss": 0.4213,
      "step": 9476
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.685161809966681,
      "learning_rate": 3.2435652663079716e-07,
      "loss": 0.2761,
      "step": 9477
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9310354041659767,
      "learning_rate": 3.240026503617638e-07,
      "loss": 0.1907,
      "step": 9478
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.600088468422065,
      "learning_rate": 3.2364895386296307e-07,
      "loss": 0.2349,
      "step": 9479
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9013706590199217,
      "learning_rate": 3.2329543716361543e-07,
      "loss": 0.2127,
      "step": 9480
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.69519171369699,
      "learning_rate": 3.229421002929278e-07,
      "loss": 0.2618,
      "step": 9481
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0052607365984296,
      "learning_rate": 3.225889432800891e-07,
      "loss": 0.1032,
      "step": 9482
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.2897714393250697,
      "learning_rate": 3.222359661542776e-07,
      "loss": 0.2447,
      "step": 9483
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.6080220713704736,
      "learning_rate": 3.218831689446533e-07,
      "loss": 0.183,
      "step": 9484
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.464969879665358,
      "learning_rate": 3.2153055168036215e-07,
      "loss": 0.2942,
      "step": 9485
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.6834334488827025,
      "learning_rate": 3.211781143905376e-07,
      "loss": 0.4593,
      "step": 9486
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.973113280498575,
      "learning_rate": 3.208258571042941e-07,
      "loss": 0.3618,
      "step": 9487
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.992047527412119,
      "learning_rate": 3.2047377985073486e-07,
      "loss": 0.2449,
      "step": 9488
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.8932279492303796,
      "learning_rate": 3.2012188265894614e-07,
      "loss": 0.0913,
      "step": 9489
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.748778365555115,
      "learning_rate": 3.197701655579996e-07,
      "loss": 0.0176,
      "step": 9490
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9500197154192547,
      "learning_rate": 3.1941862857695404e-07,
      "loss": 0.1237,
      "step": 9491
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.002643530306656,
      "learning_rate": 3.1906727174484955e-07,
      "loss": 0.0908,
      "step": 9492
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.590626038041699,
      "learning_rate": 3.1871609509071507e-07,
      "loss": 0.2701,
      "step": 9493
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.9132666854172746,
      "learning_rate": 3.183650986435624e-07,
      "loss": 0.2143,
      "step": 9494
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.323213459298969,
      "learning_rate": 3.18014282432389e-07,
      "loss": 0.2254,
      "step": 9495
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.471087708287034,
      "learning_rate": 3.1766364648617874e-07,
      "loss": 0.2005,
      "step": 9496
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.915355202435628,
      "learning_rate": 3.173131908338975e-07,
      "loss": 0.2424,
      "step": 9497
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.155063004763289,
      "learning_rate": 3.169629155045001e-07,
      "loss": 0.3078,
      "step": 9498
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.8232246059763906,
      "learning_rate": 3.166128205269234e-07,
      "loss": 0.2732,
      "step": 9499
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.12467522729291,
      "learning_rate": 3.162629059300906e-07,
      "loss": 0.3432,
      "step": 9500
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.328860554870308,
      "learning_rate": 3.159131717429112e-07,
      "loss": 0.2966,
      "step": 9501
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.305928422045405,
      "learning_rate": 3.1556361799427644e-07,
      "loss": 0.1954,
      "step": 9502
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.9918294345655605,
      "learning_rate": 3.1521424471306655e-07,
      "loss": 0.1792,
      "step": 9503
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.5127492617576888,
      "learning_rate": 3.148650519281443e-07,
      "loss": 0.2282,
      "step": 9504
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6249378690996172,
      "learning_rate": 3.1451603966835777e-07,
      "loss": 0.1196,
      "step": 9505
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.4217846909000067,
      "learning_rate": 3.1416720796254214e-07,
      "loss": 0.1062,
      "step": 9506
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.0655227042060096,
      "learning_rate": 3.138185568395144e-07,
      "loss": 0.2045,
      "step": 9507
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.964237283956179,
      "learning_rate": 3.134700863280796e-07,
      "loss": 0.234,
      "step": 9508
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.014890298948918,
      "learning_rate": 3.131217964570263e-07,
      "loss": 0.5724,
      "step": 9509
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.899429850829437,
      "learning_rate": 3.127736872551279e-07,
      "loss": 0.2665,
      "step": 9510
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.621373241450464,
      "learning_rate": 3.124257587511445e-07,
      "loss": 0.1385,
      "step": 9511
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.439733196194649,
      "learning_rate": 3.1207801097382e-07,
      "loss": 0.2783,
      "step": 9512
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.741494743731999,
      "learning_rate": 3.117304439518834e-07,
      "loss": 0.3758,
      "step": 9513
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9199386912477778,
      "learning_rate": 3.1138305771404893e-07,
      "loss": 0.081,
      "step": 9514
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.204250574119713,
      "learning_rate": 3.110358522890156e-07,
      "loss": 0.4264,
      "step": 9515
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.186090643735146,
      "learning_rate": 3.106888277054687e-07,
      "loss": 0.2349,
      "step": 9516
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.908973689390395,
      "learning_rate": 3.103419839920774e-07,
      "loss": 0.3065,
      "step": 9517
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.851099018103423,
      "learning_rate": 3.099953211774959e-07,
      "loss": 0.356,
      "step": 9518
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.0236832770065756,
      "learning_rate": 3.0964883929036405e-07,
      "loss": 0.206,
      "step": 9519
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.131444156774441,
      "learning_rate": 3.093025383593057e-07,
      "loss": 0.1826,
      "step": 9520
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.018257612203782,
      "learning_rate": 3.089564184129318e-07,
      "loss": 0.2716,
      "step": 9521
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.603045572169107,
      "learning_rate": 3.086104794798367e-07,
      "loss": 0.2995,
      "step": 9522
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.012008692775299,
      "learning_rate": 3.0826472158859976e-07,
      "loss": 0.3068,
      "step": 9523
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3600754092872616,
      "learning_rate": 3.0791914476778596e-07,
      "loss": 0.193,
      "step": 9524
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.460672639120016,
      "learning_rate": 3.075737490459449e-07,
      "loss": 0.532,
      "step": 9525
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.251956260827073,
      "learning_rate": 3.072285344516124e-07,
      "loss": 0.4096,
      "step": 9526
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.670567827168527,
      "learning_rate": 3.068835010133076e-07,
      "loss": 0.1513,
      "step": 9527
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.8881289611291745,
      "learning_rate": 3.0653864875953567e-07,
      "loss": 0.5284,
      "step": 9528
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.800085345824791,
      "learning_rate": 3.061939777187867e-07,
      "loss": 0.1418,
      "step": 9529
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3887176783018234,
      "learning_rate": 3.0584948791953504e-07,
      "loss": 0.258,
      "step": 9530
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.691490684822772,
      "learning_rate": 3.055051793902419e-07,
      "loss": 0.454,
      "step": 9531
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5316766501170025,
      "learning_rate": 3.0516105215935206e-07,
      "loss": 0.2533,
      "step": 9532
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8835392822139987,
      "learning_rate": 3.0481710625529513e-07,
      "loss": 0.0785,
      "step": 9533
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.268914782546442,
      "learning_rate": 3.044733417064863e-07,
      "loss": 0.2646,
      "step": 9534
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8283161521373685,
      "learning_rate": 3.041297585413258e-07,
      "loss": 0.1692,
      "step": 9535
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.041342995500878,
      "learning_rate": 3.0378635678819906e-07,
      "loss": 0.3101,
      "step": 9536
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8481660829122064,
      "learning_rate": 3.0344313647547614e-07,
      "loss": 0.2835,
      "step": 9537
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.347120388417301,
      "learning_rate": 3.031000976315121e-07,
      "loss": 0.4264,
      "step": 9538
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.1106803683072375,
      "learning_rate": 3.0275724028464737e-07,
      "loss": 0.2683,
      "step": 9539
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.9552344096244156,
      "learning_rate": 3.024145644632062e-07,
      "loss": 0.1788,
      "step": 9540
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.46035101444164,
      "learning_rate": 3.0207207019550044e-07,
      "loss": 0.3678,
      "step": 9541
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.047548737274054,
      "learning_rate": 3.017297575098241e-07,
      "loss": 0.1883,
      "step": 9542
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.265175755033977,
      "learning_rate": 3.013876264344576e-07,
      "loss": 0.0877,
      "step": 9543
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.331067973631491,
      "learning_rate": 3.0104567699766637e-07,
      "loss": 0.2963,
      "step": 9544
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.583038413812417,
      "learning_rate": 3.007039092276995e-07,
      "loss": 0.3412,
      "step": 9545
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.7941652248402775,
      "learning_rate": 3.0036232315279414e-07,
      "loss": 0.2631,
      "step": 9546
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.275421595149816,
      "learning_rate": 3.000209188011691e-07,
      "loss": 0.3607,
      "step": 9547
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.7044441797557575,
      "learning_rate": 2.9967969620102983e-07,
      "loss": 0.4405,
      "step": 9548
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.479545734810358,
      "learning_rate": 2.9933865538056605e-07,
      "loss": 0.2549,
      "step": 9549
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.613725319284093,
      "learning_rate": 2.9899779636795356e-07,
      "loss": 0.3286,
      "step": 9550
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.127532140783209,
      "learning_rate": 2.986571191913526e-07,
      "loss": 0.271,
      "step": 9551
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.6530265538866855,
      "learning_rate": 2.983166238789076e-07,
      "loss": 0.2229,
      "step": 9552
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.741577132235813,
      "learning_rate": 2.979763104587488e-07,
      "loss": 0.2369,
      "step": 9553
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.783181746423654,
      "learning_rate": 2.9763617895899097e-07,
      "loss": 0.3024,
      "step": 9554
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.689844576720516,
      "learning_rate": 2.9729622940773474e-07,
      "loss": 0.3057,
      "step": 9555
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.4175497087681554,
      "learning_rate": 2.96956461833065e-07,
      "loss": 0.3453,
      "step": 9556
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.005661570673823,
      "learning_rate": 2.966168762630514e-07,
      "loss": 0.4697,
      "step": 9557
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.453900962914778,
      "learning_rate": 2.962774727257486e-07,
      "loss": 0.2542,
      "step": 9558
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.182101051597392,
      "learning_rate": 2.959382512491965e-07,
      "loss": 0.4642,
      "step": 9559
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.3257134686435394,
      "learning_rate": 2.9559921186142067e-07,
      "loss": 0.3857,
      "step": 9560
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.330900883296725,
      "learning_rate": 2.952603545904303e-07,
      "loss": 0.1972,
      "step": 9561
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.844382879635099,
      "learning_rate": 2.949216794642201e-07,
      "loss": 0.2966,
      "step": 9562
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6908005564129693,
      "learning_rate": 2.945831865107698e-07,
      "loss": 0.1356,
      "step": 9563
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.842966270046879,
      "learning_rate": 2.942448757580438e-07,
      "loss": 0.4141,
      "step": 9564
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0689134231864745,
      "learning_rate": 2.939067472339921e-07,
      "loss": 0.1831,
      "step": 9565
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.105737443210973,
      "learning_rate": 2.935688009665494e-07,
      "loss": 0.3875,
      "step": 9566
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.984996731886963,
      "learning_rate": 2.932310369836344e-07,
      "loss": 0.2171,
      "step": 9567
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.476282892846762,
      "learning_rate": 2.928934553131524e-07,
      "loss": 0.2735,
      "step": 9568
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.463095966907978,
      "learning_rate": 2.9255605598299127e-07,
      "loss": 0.2082,
      "step": 9569
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.06682703438428,
      "learning_rate": 2.9221883902102704e-07,
      "loss": 0.2133,
      "step": 9570
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.7881715742781275,
      "learning_rate": 2.9188180445511834e-07,
      "loss": 0.3551,
      "step": 9571
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.138228523558797,
      "learning_rate": 2.915449523131089e-07,
      "loss": 0.1214,
      "step": 9572
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6830241131411556,
      "learning_rate": 2.9120828262282797e-07,
      "loss": 0.1623,
      "step": 9573
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.6215105000397574,
      "learning_rate": 2.908717954120893e-07,
      "loss": 0.2606,
      "step": 9574
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.0899956940401765,
      "learning_rate": 2.905354907086924e-07,
      "loss": 0.3432,
      "step": 9575
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.4577316432158423,
      "learning_rate": 2.9019936854042087e-07,
      "loss": 0.1519,
      "step": 9576
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.411645315650886,
      "learning_rate": 2.8986342893504356e-07,
      "loss": 0.2805,
      "step": 9577
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.7450453822865155,
      "learning_rate": 2.895276719203141e-07,
      "loss": 0.184,
      "step": 9578
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.4547086172134978,
      "learning_rate": 2.891920975239704e-07,
      "loss": 0.1024,
      "step": 9579
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.2701921608633935,
      "learning_rate": 2.888567057737371e-07,
      "loss": 0.3547,
      "step": 9580
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.378280098216411,
      "learning_rate": 2.885214966973221e-07,
      "loss": 0.3412,
      "step": 9581
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.3135608356419946,
      "learning_rate": 2.881864703224188e-07,
      "loss": 0.5071,
      "step": 9582
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.768127207962402,
      "learning_rate": 2.8785162667670536e-07,
      "loss": 0.4044,
      "step": 9583
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.403215275857888,
      "learning_rate": 2.875169657878446e-07,
      "loss": 0.3224,
      "step": 9584
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.6851620460003724,
      "learning_rate": 2.8718248768348526e-07,
      "loss": 0.4453,
      "step": 9585
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.551965496790294,
      "learning_rate": 2.8684819239126e-07,
      "loss": 0.03,
      "step": 9586
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8166128055159714,
      "learning_rate": 2.865140799387864e-07,
      "loss": 0.1368,
      "step": 9587
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3900189287059552,
      "learning_rate": 2.861801503536674e-07,
      "loss": 0.1181,
      "step": 9588
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.4222812254239035,
      "learning_rate": 2.858464036634903e-07,
      "loss": 0.2807,
      "step": 9589
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.363614364041202,
      "learning_rate": 2.855128398958282e-07,
      "loss": 0.3934,
      "step": 9590
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.784005201019246,
      "learning_rate": 2.8517945907823833e-07,
      "loss": 0.1263,
      "step": 9591
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.135029758558329,
      "learning_rate": 2.8484626123826243e-07,
      "loss": 0.3265,
      "step": 9592
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.3033397306099594,
      "learning_rate": 2.845132464034292e-07,
      "loss": 0.1329,
      "step": 9593
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.360384712374914,
      "learning_rate": 2.8418041460124835e-07,
      "loss": 0.16,
      "step": 9594
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.907999746772092,
      "learning_rate": 2.838477658592187e-07,
      "loss": 0.3069,
      "step": 9595
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.126842090497215,
      "learning_rate": 2.8351530020482134e-07,
      "loss": 0.123,
      "step": 9596
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.539233398592226,
      "learning_rate": 2.831830176655226e-07,
      "loss": 0.1824,
      "step": 9597
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.770592183603225,
      "learning_rate": 2.8285091826877553e-07,
      "loss": 0.3235,
      "step": 9598
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.827180023631432,
      "learning_rate": 2.825190020420146e-07,
      "loss": 0.2829,
      "step": 9599
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.397676670308149,
      "learning_rate": 2.821872690126623e-07,
      "loss": 0.2572,
      "step": 9600
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.7584547287757477,
      "learning_rate": 2.818557192081245e-07,
      "loss": 0.2045,
      "step": 9601
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.532673902306661,
      "learning_rate": 2.8152435265579176e-07,
      "loss": 0.1958,
      "step": 9602
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.4834823146013307,
      "learning_rate": 2.811931693830414e-07,
      "loss": 0.1851,
      "step": 9603
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.7161691868471602,
      "learning_rate": 2.80862169417232e-07,
      "loss": 0.2704,
      "step": 9604
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.268102802438197,
      "learning_rate": 2.805313527857109e-07,
      "loss": 0.1233,
      "step": 9605
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.8885065661826745,
      "learning_rate": 2.802007195158082e-07,
      "loss": 0.2139,
      "step": 9606
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.206299266041931,
      "learning_rate": 2.798702696348382e-07,
      "loss": 0.27,
      "step": 9607
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.786854081231408,
      "learning_rate": 2.795400031701029e-07,
      "loss": 0.3112,
      "step": 9608
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.2966391198038334,
      "learning_rate": 2.7920992014888515e-07,
      "loss": 0.1139,
      "step": 9609
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.472799381953279,
      "learning_rate": 2.7888002059845656e-07,
      "loss": 0.2712,
      "step": 9610
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.525388592156662,
      "learning_rate": 2.785503045460708e-07,
      "loss": 0.3107,
      "step": 9611
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.7132034815158326,
      "learning_rate": 2.782207720189675e-07,
      "loss": 0.2314,
      "step": 9612
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.06835697421199,
      "learning_rate": 2.778914230443719e-07,
      "loss": 0.2666,
      "step": 9613
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.139548325392302,
      "learning_rate": 2.7756225764949183e-07,
      "loss": 0.2881,
      "step": 9614
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.111337561791751,
      "learning_rate": 2.772332758615223e-07,
      "loss": 0.1219,
      "step": 9615
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.498054498967812,
      "learning_rate": 2.7690447770764186e-07,
      "loss": 0.383,
      "step": 9616
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.983544456889647,
      "learning_rate": 2.7657586321501395e-07,
      "loss": 0.3542,
      "step": 9617
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.678927662914105,
      "learning_rate": 2.7624743241078806e-07,
      "loss": 0.3533,
      "step": 9618
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.462592100056966,
      "learning_rate": 2.7591918532209604e-07,
      "loss": 0.5514,
      "step": 9619
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.576014653092155,
      "learning_rate": 2.755911219760574e-07,
      "loss": 0.5796,
      "step": 9620
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.4583400777363558,
      "learning_rate": 2.7526324239977446e-07,
      "loss": 0.1795,
      "step": 9621
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.9809960148867654,
      "learning_rate": 2.749355466203346e-07,
      "loss": 0.3091,
      "step": 9622
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.347825898243688,
      "learning_rate": 2.74608034664812e-07,
      "loss": 0.3259,
      "step": 9623
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.655460350056957,
      "learning_rate": 2.7428070656026207e-07,
      "loss": 0.1518,
      "step": 9624
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.5146489069843256,
      "learning_rate": 2.739535623337286e-07,
      "loss": 0.3038,
      "step": 9625
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.0404079219187645,
      "learning_rate": 2.73626602012238e-07,
      "loss": 0.2862,
      "step": 9626
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.9500480419092976,
      "learning_rate": 2.732998256228017e-07,
      "loss": 0.2221,
      "step": 9627
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.8918547396018353,
      "learning_rate": 2.7297323319241816e-07,
      "loss": 0.1398,
      "step": 9628
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.7558357259468065,
      "learning_rate": 2.7264682474806627e-07,
      "loss": 0.1269,
      "step": 9629
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.6339405874585404,
      "learning_rate": 2.723206003167142e-07,
      "loss": 0.1232,
      "step": 9630
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.884879141609599,
      "learning_rate": 2.7199455992531235e-07,
      "loss": 0.2814,
      "step": 9631
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.900575042790948,
      "learning_rate": 2.716687036007962e-07,
      "loss": 0.4688,
      "step": 9632
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.433407816032432,
      "learning_rate": 2.7134303137008794e-07,
      "loss": 0.2836,
      "step": 9633
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.171848698902372,
      "learning_rate": 2.710175432600906e-07,
      "loss": 0.3839,
      "step": 9634
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9907177838956542,
      "learning_rate": 2.706922392976966e-07,
      "loss": 0.1433,
      "step": 9635
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.148985870723691,
      "learning_rate": 2.7036711950977983e-07,
      "loss": 0.402,
      "step": 9636
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.988574746534376,
      "learning_rate": 2.700421839231998e-07,
      "loss": 0.2644,
      "step": 9637
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.632472689495881,
      "learning_rate": 2.6971743256480274e-07,
      "loss": 0.08,
      "step": 9638
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.057220307804211,
      "learning_rate": 2.6939286546141614e-07,
      "loss": 0.3295,
      "step": 9639
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.7809124122447773,
      "learning_rate": 2.690684826398546e-07,
      "loss": 0.242,
      "step": 9640
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8195251424477236,
      "learning_rate": 2.6874428412691834e-07,
      "loss": 0.0055,
      "step": 9641
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.019199589086108,
      "learning_rate": 2.684202699493893e-07,
      "loss": 0.3336,
      "step": 9642
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.638706220889156,
      "learning_rate": 2.6809644013403775e-07,
      "loss": 0.3603,
      "step": 9643
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.50629848390324,
      "learning_rate": 2.677727947076145e-07,
      "loss": 0.1902,
      "step": 9644
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9176229249560626,
      "learning_rate": 2.6744933369685924e-07,
      "loss": 0.1939,
      "step": 9645
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.7331348715544213,
      "learning_rate": 2.671260571284953e-07,
      "loss": 0.1146,
      "step": 9646
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.544168636659747,
      "learning_rate": 2.6680296502922834e-07,
      "loss": 0.1893,
      "step": 9647
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.4292676826517496,
      "learning_rate": 2.6648005742575284e-07,
      "loss": 0.0722,
      "step": 9648
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.023123645334491,
      "learning_rate": 2.661573343447435e-07,
      "loss": 0.1868,
      "step": 9649
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.3753298848371305,
      "learning_rate": 2.658347958128632e-07,
      "loss": 0.3722,
      "step": 9650
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.573074866981362,
      "learning_rate": 2.655124418567595e-07,
      "loss": 0.4454,
      "step": 9651
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.842390171493308,
      "learning_rate": 2.6519027250306196e-07,
      "loss": 0.1467,
      "step": 9652
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.0674410962810255,
      "learning_rate": 2.648682877783881e-07,
      "loss": 0.2119,
      "step": 9653
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.25023991369945,
      "learning_rate": 2.6454648770933735e-07,
      "loss": 0.2261,
      "step": 9654
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.750966714446043,
      "learning_rate": 2.642248723224958e-07,
      "loss": 0.3945,
      "step": 9655
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.251457037591061,
      "learning_rate": 2.6390344164443496e-07,
      "loss": 0.2225,
      "step": 9656
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1559244191238447,
      "learning_rate": 2.635821957017076e-07,
      "loss": 0.2246,
      "step": 9657
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.032770460303428,
      "learning_rate": 2.632611345208558e-07,
      "loss": 0.3091,
      "step": 9658
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.470852777740987,
      "learning_rate": 2.6294025812840213e-07,
      "loss": 0.0714,
      "step": 9659
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.767087490360899,
      "learning_rate": 2.626195665508563e-07,
      "loss": 0.1164,
      "step": 9660
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8177822474210776,
      "learning_rate": 2.6229905981471376e-07,
      "loss": 0.021,
      "step": 9661
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.257453815572063,
      "learning_rate": 2.6197873794645084e-07,
      "loss": 0.315,
      "step": 9662
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.4427632099298675,
      "learning_rate": 2.616586009725333e-07,
      "loss": 0.2514,
      "step": 9663
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6391769467577553,
      "learning_rate": 2.6133864891940726e-07,
      "loss": 0.0772,
      "step": 9664
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.970095228606527,
      "learning_rate": 2.610188818135065e-07,
      "loss": 0.1807,
      "step": 9665
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.1328448968788694,
      "learning_rate": 2.606992996812493e-07,
      "loss": 0.1468,
      "step": 9666
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.7806388549911487,
      "learning_rate": 2.603799025490364e-07,
      "loss": 0.1417,
      "step": 9667
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.521783160072767,
      "learning_rate": 2.6006069044325623e-07,
      "loss": 0.3746,
      "step": 9668
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.17339912106986,
      "learning_rate": 2.5974166339028e-07,
      "loss": 0.2152,
      "step": 9669
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.72553842559851,
      "learning_rate": 2.5942282141646375e-07,
      "loss": 0.4525,
      "step": 9670
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.6366768463566457,
      "learning_rate": 2.5910416454814986e-07,
      "loss": 0.2394,
      "step": 9671
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.016889886313531,
      "learning_rate": 2.587856928116625e-07,
      "loss": 0.1814,
      "step": 9672
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.874585017007024,
      "learning_rate": 2.584674062333134e-07,
      "loss": 0.259,
      "step": 9673
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.27331861116998,
      "learning_rate": 2.5814930483939767e-07,
      "loss": 0.3398,
      "step": 9674
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.932348140972879,
      "learning_rate": 2.5783138865619476e-07,
      "loss": 0.3823,
      "step": 9675
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.16446977754414,
      "learning_rate": 2.5751365770997075e-07,
      "loss": 0.1873,
      "step": 9676
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.8672818462359935,
      "learning_rate": 2.571961120269731e-07,
      "loss": 0.3675,
      "step": 9677
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.623513600258101,
      "learning_rate": 2.56878751633437e-07,
      "loss": 0.5616,
      "step": 9678
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.9005796038086795,
      "learning_rate": 2.565615765555812e-07,
      "loss": 0.1531,
      "step": 9679
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.916246693853986,
      "learning_rate": 2.562445868196084e-07,
      "loss": 0.3285,
      "step": 9680
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.2217943500959856,
      "learning_rate": 2.559277824517084e-07,
      "loss": 0.1787,
      "step": 9681
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.403496629587988,
      "learning_rate": 2.55611163478052e-07,
      "loss": 0.2517,
      "step": 9682
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.0239956297472133,
      "learning_rate": 2.5529472992479794e-07,
      "loss": 0.1726,
      "step": 9683
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.842255856352642,
      "learning_rate": 2.549784818180881e-07,
      "loss": 0.1004,
      "step": 9684
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.208869012469839,
      "learning_rate": 2.5466241918404884e-07,
      "loss": 0.1784,
      "step": 9685
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.6655309834452448,
      "learning_rate": 2.543465420487931e-07,
      "loss": 0.2455,
      "step": 9686
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.655118205163774,
      "learning_rate": 2.540308504384154e-07,
      "loss": 0.1884,
      "step": 9687
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.298091287095551,
      "learning_rate": 2.5371534437899803e-07,
      "loss": 0.3536,
      "step": 9688
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.862937721021787,
      "learning_rate": 2.534000238966058e-07,
      "loss": 0.4583,
      "step": 9689
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.8751150131076244,
      "learning_rate": 2.530848890172888e-07,
      "loss": 0.2689,
      "step": 9690
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.514793940385832,
      "learning_rate": 2.52769939767083e-07,
      "loss": 0.1969,
      "step": 9691
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.0141892993309005,
      "learning_rate": 2.5245517617200624e-07,
      "loss": 0.2684,
      "step": 9692
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.5867869146060753,
      "learning_rate": 2.5214059825806434e-07,
      "loss": 0.0627,
      "step": 9693
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.7141848755031233,
      "learning_rate": 2.5182620605124523e-07,
      "loss": 0.3114,
      "step": 9694
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.405045136799402,
      "learning_rate": 2.5151199957752253e-07,
      "loss": 0.2064,
      "step": 9695
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.735159381172989,
      "learning_rate": 2.5119797886285573e-07,
      "loss": 0.3432,
      "step": 9696
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.786849462073597,
      "learning_rate": 2.508841439331855e-07,
      "loss": 0.3592,
      "step": 9697
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.485519408231103,
      "learning_rate": 2.5057049481444107e-07,
      "loss": 0.2522,
      "step": 9698
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.053483079628835,
      "learning_rate": 2.5025703153253395e-07,
      "loss": 0.4052,
      "step": 9699
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.256009573885011,
      "learning_rate": 2.4994375411336063e-07,
      "loss": 0.3416,
      "step": 9700
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.7328567176040846,
      "learning_rate": 2.496306625828038e-07,
      "loss": 0.2075,
      "step": 9701
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.406600727952778,
      "learning_rate": 2.493177569667277e-07,
      "loss": 0.1936,
      "step": 9702
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.9812961572596235,
      "learning_rate": 2.4900503729098443e-07,
      "loss": 0.5396,
      "step": 9703
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.800176663069178,
      "learning_rate": 2.486925035814089e-07,
      "loss": 0.28,
      "step": 9704
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.785968731198424,
      "learning_rate": 2.4838015586382075e-07,
      "loss": 0.3144,
      "step": 9705
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.485235605858948,
      "learning_rate": 2.4806799416402597e-07,
      "loss": 0.1568,
      "step": 9706
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.809811284025651,
      "learning_rate": 2.477560185078121e-07,
      "loss": 0.288,
      "step": 9707
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.2873088169317715,
      "learning_rate": 2.4744422892095413e-07,
      "loss": 0.4814,
      "step": 9708
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.782340488471335,
      "learning_rate": 2.471326254292103e-07,
      "loss": 0.2046,
      "step": 9709
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.903043690574586,
      "learning_rate": 2.468212080583232e-07,
      "loss": 0.1783,
      "step": 9710
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.664597821577764,
      "learning_rate": 2.465099768340218e-07,
      "loss": 0.3149,
      "step": 9711
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.456476498600517,
      "learning_rate": 2.4619893178201767e-07,
      "loss": 0.2949,
      "step": 9712
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.6635380099702615,
      "learning_rate": 2.4588807292800806e-07,
      "loss": 0.1699,
      "step": 9713
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.7646711015981698,
      "learning_rate": 2.4557740029767464e-07,
      "loss": 0.3107,
      "step": 9714
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.275595909845997,
      "learning_rate": 2.45266913916683e-07,
      "loss": 0.2272,
      "step": 9715
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9291623908911566,
      "learning_rate": 2.4495661381068516e-07,
      "loss": 0.1322,
      "step": 9716
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.4937639301550347,
      "learning_rate": 2.446465000053158e-07,
      "loss": 0.2193,
      "step": 9717
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.464745256856856,
      "learning_rate": 2.443365725261954e-07,
      "loss": 0.1534,
      "step": 9718
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.09012275981771,
      "learning_rate": 2.440268313989283e-07,
      "loss": 0.3526,
      "step": 9719
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.7210285186224805,
      "learning_rate": 2.437172766491036e-07,
      "loss": 0.2725,
      "step": 9720
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.4913628588031957,
      "learning_rate": 2.434079083022961e-07,
      "loss": 0.2488,
      "step": 9721
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.845953919548028,
      "learning_rate": 2.430987263840634e-07,
      "loss": 0.476,
      "step": 9722
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.356772798442472,
      "learning_rate": 2.4278973091994924e-07,
      "loss": 0.3453,
      "step": 9723
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.284246713875962,
      "learning_rate": 2.4248092193548097e-07,
      "loss": 0.2913,
      "step": 9724
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.476397972053339,
      "learning_rate": 2.4217229945617037e-07,
      "loss": 0.4011,
      "step": 9725
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.3783469620703497,
      "learning_rate": 2.4186386350751515e-07,
      "loss": 0.1877,
      "step": 9726
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2377270624576,
      "learning_rate": 2.4155561411499687e-07,
      "loss": 0.0921,
      "step": 9727
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.04735164723174,
      "learning_rate": 2.41247551304081e-07,
      "loss": 0.2718,
      "step": 9728
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.328299252497513,
      "learning_rate": 2.4093967510021826e-07,
      "loss": 0.1997,
      "step": 9729
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.8304995664235952,
      "learning_rate": 2.406319855288436e-07,
      "loss": 0.1371,
      "step": 9730
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.640005556290945,
      "learning_rate": 2.403244826153778e-07,
      "loss": 0.0979,
      "step": 9731
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.732755269566134,
      "learning_rate": 2.4001716638522465e-07,
      "loss": 0.3355,
      "step": 9732
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.1066611068513525,
      "learning_rate": 2.397100368637728e-07,
      "loss": 0.3575,
      "step": 9733
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.797489840655113,
      "learning_rate": 2.394030940763964e-07,
      "loss": 0.3874,
      "step": 9734
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.89989688162179,
      "learning_rate": 2.390963380484529e-07,
      "loss": 0.4703,
      "step": 9735
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.153986464166129,
      "learning_rate": 2.387897688052854e-07,
      "loss": 0.3038,
      "step": 9736
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.358226546875491,
      "learning_rate": 2.384833863722213e-07,
      "loss": 0.3073,
      "step": 9737
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.310176921366194,
      "learning_rate": 2.3817719077457225e-07,
      "loss": 0.1096,
      "step": 9738
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.291588882413349,
      "learning_rate": 2.378711820376345e-07,
      "loss": 0.2821,
      "step": 9739
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.782092521247755,
      "learning_rate": 2.375653601866887e-07,
      "loss": 0.4085,
      "step": 9740
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.4828799246190094,
      "learning_rate": 2.3725972524700103e-07,
      "loss": 0.2855,
      "step": 9741
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.778715778813208,
      "learning_rate": 2.369542772438213e-07,
      "loss": 0.2918,
      "step": 9742
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.8541283765995877,
      "learning_rate": 2.3664901620238433e-07,
      "loss": 0.1502,
      "step": 9743
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.1125471134933616,
      "learning_rate": 2.3634394214790886e-07,
      "loss": 0.3378,
      "step": 9744
    },
    {
      "epoch": 0.86,
      "grad_norm": 3.5616465974638407,
      "learning_rate": 2.3603905510559798e-07,
      "loss": 0.279,
      "step": 9745
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.562785487872893,
      "learning_rate": 2.3573435510064162e-07,
      "loss": 0.0902,
      "step": 9746
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.14816673400672,
      "learning_rate": 2.354298421582116e-07,
      "loss": 0.1087,
      "step": 9747
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.508648006334808,
      "learning_rate": 2.3512551630346546e-07,
      "loss": 0.1206,
      "step": 9748
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.701983680732476,
      "learning_rate": 2.3482137756154515e-07,
      "loss": 0.2072,
      "step": 9749
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.803315871061326,
      "learning_rate": 2.3451742595757637e-07,
      "loss": 0.1274,
      "step": 9750
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.469793662937259,
      "learning_rate": 2.342136615166715e-07,
      "loss": 0.1228,
      "step": 9751
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.383789004906043,
      "learning_rate": 2.3391008426392524e-07,
      "loss": 0.2027,
      "step": 9752
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.690547814550125,
      "learning_rate": 2.336066942244178e-07,
      "loss": 0.2257,
      "step": 9753
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.865447071581157,
      "learning_rate": 2.3330349142321352e-07,
      "loss": 0.3812,
      "step": 9754
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.341580531764783,
      "learning_rate": 2.3300047588536183e-07,
      "loss": 0.301,
      "step": 9755
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.249700709387213,
      "learning_rate": 2.3269764763589664e-07,
      "loss": 0.3426,
      "step": 9756
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.862407043203184,
      "learning_rate": 2.3239500669983567e-07,
      "loss": 0.1854,
      "step": 9757
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4393425181767387,
      "learning_rate": 2.320925531021817e-07,
      "loss": 0.1092,
      "step": 9758
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.361314633427667,
      "learning_rate": 2.317902868679217e-07,
      "loss": 0.3418,
      "step": 9759
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.059157916338315,
      "learning_rate": 2.314882080220282e-07,
      "loss": 0.3844,
      "step": 9760
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.974519169391528,
      "learning_rate": 2.3118631658945706e-07,
      "loss": 0.3174,
      "step": 9761
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.193596069155372,
      "learning_rate": 2.308846125951489e-07,
      "loss": 0.0655,
      "step": 9762
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.538010502570891,
      "learning_rate": 2.3058309606402928e-07,
      "loss": 0.4392,
      "step": 9763
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.137477999519585,
      "learning_rate": 2.3028176702100753e-07,
      "loss": 0.3975,
      "step": 9764
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.952115719728736,
      "learning_rate": 2.2998062549097893e-07,
      "loss": 0.0648,
      "step": 9765
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.7322844840121605,
      "learning_rate": 2.296796714988214e-07,
      "loss": 0.1492,
      "step": 9766
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.3124999442614564,
      "learning_rate": 2.2937890506939898e-07,
      "loss": 0.256,
      "step": 9767
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.036663900722824,
      "learning_rate": 2.2907832622755892e-07,
      "loss": 0.1712,
      "step": 9768
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.263214067910406,
      "learning_rate": 2.287779349981334e-07,
      "loss": 0.2852,
      "step": 9769
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.119312745489119,
      "learning_rate": 2.2847773140594027e-07,
      "loss": 0.1902,
      "step": 9770
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9366080815281101,
      "learning_rate": 2.2817771547578028e-07,
      "loss": 0.0707,
      "step": 9771
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.224715896958122,
      "learning_rate": 2.2787788723243914e-07,
      "loss": 0.364,
      "step": 9772
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.127272391692241,
      "learning_rate": 2.2757824670068735e-07,
      "loss": 0.335,
      "step": 9773
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.326606388961616,
      "learning_rate": 2.2727879390527924e-07,
      "loss": 0.1502,
      "step": 9774
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.4114456385890057,
      "learning_rate": 2.2697952887095532e-07,
      "loss": 0.086,
      "step": 9775
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.7504710492313182,
      "learning_rate": 2.2668045162243857e-07,
      "loss": 0.2458,
      "step": 9776
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.510940679813774,
      "learning_rate": 2.263815621844373e-07,
      "loss": 0.2783,
      "step": 9777
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.165969147324264,
      "learning_rate": 2.2608286058164481e-07,
      "loss": 0.3418,
      "step": 9778
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.82954341775784,
      "learning_rate": 2.2578434683873718e-07,
      "loss": 0.3176,
      "step": 9779
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.200000011279569,
      "learning_rate": 2.2548602098037748e-07,
      "loss": 0.3453,
      "step": 9780
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.111918024080998,
      "learning_rate": 2.2518788303121125e-07,
      "loss": 0.1107,
      "step": 9781
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.643070830824475,
      "learning_rate": 2.2488993301586937e-07,
      "loss": 0.1498,
      "step": 9782
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.228107861991068,
      "learning_rate": 2.2459217095896712e-07,
      "loss": 0.2356,
      "step": 9783
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.533991548792398,
      "learning_rate": 2.2429459688510345e-07,
      "loss": 0.4212,
      "step": 9784
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.521376218517377,
      "learning_rate": 2.2399721081886345e-07,
      "loss": 0.2616,
      "step": 9785
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.4721142698823884,
      "learning_rate": 2.237000127848152e-07,
      "loss": 0.1934,
      "step": 9786
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.319391636094901,
      "learning_rate": 2.234030028075118e-07,
      "loss": 0.2088,
      "step": 9787
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.190552295648015,
      "learning_rate": 2.2310618091149068e-07,
      "loss": 0.2671,
      "step": 9788
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.478024341761899,
      "learning_rate": 2.228095471212735e-07,
      "loss": 0.3359,
      "step": 9789
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.9385021209136877,
      "learning_rate": 2.2251310146136767e-07,
      "loss": 0.1684,
      "step": 9790
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.056802454152368,
      "learning_rate": 2.222168439562633e-07,
      "loss": 0.1076,
      "step": 9791
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.817828932846354,
      "learning_rate": 2.2192077463043555e-07,
      "loss": 0.212,
      "step": 9792
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9027686823663355,
      "learning_rate": 2.2162489350834538e-07,
      "loss": 0.0745,
      "step": 9793
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.7163242068207865,
      "learning_rate": 2.2132920061443547e-07,
      "loss": 0.5474,
      "step": 9794
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.965193111260821,
      "learning_rate": 2.2103369597313574e-07,
      "loss": 0.3633,
      "step": 9795
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.637208521745984,
      "learning_rate": 2.2073837960885885e-07,
      "loss": 0.1832,
      "step": 9796
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.954825227325027,
      "learning_rate": 2.2044325154600199e-07,
      "loss": 0.1688,
      "step": 9797
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.0460262879690876,
      "learning_rate": 2.2014831180894842e-07,
      "loss": 0.1164,
      "step": 9798
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.500183126416593,
      "learning_rate": 2.198535604220628e-07,
      "loss": 0.4571,
      "step": 9799
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.736300116546982,
      "learning_rate": 2.1955899740969788e-07,
      "loss": 0.1976,
      "step": 9800
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.340350162585657,
      "learning_rate": 2.192646227961878e-07,
      "loss": 0.2312,
      "step": 9801
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.249902480186162,
      "learning_rate": 2.189704366058526e-07,
      "loss": 0.2701,
      "step": 9802
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.346465899813066,
      "learning_rate": 2.186764388629975e-07,
      "loss": 0.3007,
      "step": 9803
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.9243485686279094,
      "learning_rate": 2.183826295919092e-07,
      "loss": 0.1398,
      "step": 9804
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.9849264555802377,
      "learning_rate": 2.1808900881686245e-07,
      "loss": 0.2966,
      "step": 9805
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.450705930165649,
      "learning_rate": 2.1779557656211425e-07,
      "loss": 0.362,
      "step": 9806
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.364640104208416,
      "learning_rate": 2.1750233285190603e-07,
      "loss": 0.2463,
      "step": 9807
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.96486181553971,
      "learning_rate": 2.1720927771046534e-07,
      "loss": 0.3004,
      "step": 9808
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.111649875936366,
      "learning_rate": 2.1691641116200118e-07,
      "loss": 0.2851,
      "step": 9809
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.7128035493788123,
      "learning_rate": 2.1662373323071055e-07,
      "loss": 0.142,
      "step": 9810
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.885313783929336,
      "learning_rate": 2.1633124394077222e-07,
      "loss": 0.5176,
      "step": 9811
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.878230627316359,
      "learning_rate": 2.1603894331634984e-07,
      "loss": 0.4475,
      "step": 9812
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5880440426057496,
      "learning_rate": 2.1574683138159303e-07,
      "loss": 0.2628,
      "step": 9813
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.751994639654134,
      "learning_rate": 2.1545490816063303e-07,
      "loss": 0.2693,
      "step": 9814
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.395484710342793,
      "learning_rate": 2.151631736775886e-07,
      "loss": 0.136,
      "step": 9815
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.452916023098141,
      "learning_rate": 2.1487162795656076e-07,
      "loss": 0.304,
      "step": 9816
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.778717634131051,
      "learning_rate": 2.145802710216352e-07,
      "loss": 0.1349,
      "step": 9817
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.602533656513242,
      "learning_rate": 2.1428910289688382e-07,
      "loss": 0.0489,
      "step": 9818
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.329434783763345,
      "learning_rate": 2.1399812360635958e-07,
      "loss": 0.5638,
      "step": 9819
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.9658839933320436,
      "learning_rate": 2.1370733317410298e-07,
      "loss": 0.2696,
      "step": 9820
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.590212174789549,
      "learning_rate": 2.1341673162413757e-07,
      "loss": 0.4683,
      "step": 9821
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8146854936530334,
      "learning_rate": 2.1312631898047083e-07,
      "loss": 0.1602,
      "step": 9822
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.099280445689195,
      "learning_rate": 2.1283609526709658e-07,
      "loss": 0.3767,
      "step": 9823
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.167705014611048,
      "learning_rate": 2.1254606050798986e-07,
      "loss": 0.4306,
      "step": 9824
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.07178453383251,
      "learning_rate": 2.1225621472711317e-07,
      "loss": 0.2635,
      "step": 9825
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.208944167780403,
      "learning_rate": 2.1196655794841203e-07,
      "loss": 0.2338,
      "step": 9826
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.312113944448981,
      "learning_rate": 2.1167709019581568e-07,
      "loss": 0.3177,
      "step": 9827
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.039263837254163,
      "learning_rate": 2.1138781149323995e-07,
      "loss": 0.2957,
      "step": 9828
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.860291737574215,
      "learning_rate": 2.1109872186458185e-07,
      "loss": 0.2057,
      "step": 9829
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.508126185847549,
      "learning_rate": 2.108098213337259e-07,
      "loss": 0.2035,
      "step": 9830
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.3530249067985665,
      "learning_rate": 2.1052110992453938e-07,
      "loss": 0.2646,
      "step": 9831
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.2518252486426755,
      "learning_rate": 2.102325876608735e-07,
      "loss": 0.1575,
      "step": 9832
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.5418463285957746,
      "learning_rate": 2.0994425456656614e-07,
      "loss": 0.2179,
      "step": 9833
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.803740418590396,
      "learning_rate": 2.09656110665436e-07,
      "loss": 0.239,
      "step": 9834
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.689599083985559,
      "learning_rate": 2.0936815598128934e-07,
      "loss": 0.0701,
      "step": 9835
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.599967613789277,
      "learning_rate": 2.0908039053791545e-07,
      "loss": 0.2945,
      "step": 9836
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.8063909017473923,
      "learning_rate": 2.0879281435908728e-07,
      "loss": 0.1559,
      "step": 9837
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.269354453578364,
      "learning_rate": 2.0850542746856466e-07,
      "loss": 0.3738,
      "step": 9838
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.5728588381618174,
      "learning_rate": 2.082182298900881e-07,
      "loss": 0.1417,
      "step": 9839
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.338444032362636,
      "learning_rate": 2.0793122164738526e-07,
      "loss": 0.2289,
      "step": 9840
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.513409884800344,
      "learning_rate": 2.076444027641686e-07,
      "loss": 0.3681,
      "step": 9841
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.108583614397566,
      "learning_rate": 2.073577732641313e-07,
      "loss": 0.1595,
      "step": 9842
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.676199768891952,
      "learning_rate": 2.070713331709559e-07,
      "loss": 0.3845,
      "step": 9843
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.209246951887903,
      "learning_rate": 2.0678508250830425e-07,
      "loss": 0.3426,
      "step": 9844
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.838282307439133,
      "learning_rate": 2.0649902129982579e-07,
      "loss": 0.1384,
      "step": 9845
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.272855925921765,
      "learning_rate": 2.0621314956915495e-07,
      "loss": 0.3204,
      "step": 9846
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.900901945783882,
      "learning_rate": 2.0592746733990698e-07,
      "loss": 0.4376,
      "step": 9847
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.774590772368878,
      "learning_rate": 2.0564197463568525e-07,
      "loss": 0.2839,
      "step": 9848
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.278806458928446,
      "learning_rate": 2.0535667148007392e-07,
      "loss": 0.1255,
      "step": 9849
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.629734109427151,
      "learning_rate": 2.0507155789664473e-07,
      "loss": 0.362,
      "step": 9850
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.703971736434895,
      "learning_rate": 2.0478663390895242e-07,
      "loss": 0.0978,
      "step": 9851
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.182304969390213,
      "learning_rate": 2.0450189954053513e-07,
      "loss": 0.2546,
      "step": 9852
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.1273550623425015,
      "learning_rate": 2.0421735481491734e-07,
      "loss": 0.2416,
      "step": 9853
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.375381669155394,
      "learning_rate": 2.0393299975560527e-07,
      "loss": 0.3757,
      "step": 9854
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.118827825694272,
      "learning_rate": 2.0364883438609152e-07,
      "loss": 0.3105,
      "step": 9855
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.199233842264893,
      "learning_rate": 2.0336485872985394e-07,
      "loss": 0.2935,
      "step": 9856
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.889823188058972,
      "learning_rate": 2.0308107281035077e-07,
      "loss": 0.3017,
      "step": 9857
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.548664726727995,
      "learning_rate": 2.0279747665102907e-07,
      "loss": 0.2761,
      "step": 9858
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.3997059079055205,
      "learning_rate": 2.0251407027531644e-07,
      "loss": 0.3574,
      "step": 9859
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.793797421626995,
      "learning_rate": 2.0223085370662755e-07,
      "loss": 0.3495,
      "step": 9860
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.1624989662063765,
      "learning_rate": 2.0194782696836084e-07,
      "loss": 0.3208,
      "step": 9861
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.140808852454061,
      "learning_rate": 2.0166499008389712e-07,
      "loss": 0.2194,
      "step": 9862
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.768577239589653,
      "learning_rate": 2.013823430766046e-07,
      "loss": 0.5101,
      "step": 9863
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.069551619991949,
      "learning_rate": 2.0109988596983237e-07,
      "loss": 0.2725,
      "step": 9864
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3186395401209694,
      "learning_rate": 2.0081761878691707e-07,
      "loss": 0.1883,
      "step": 9865
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.346312800841999,
      "learning_rate": 2.0053554155117839e-07,
      "loss": 0.2216,
      "step": 9866
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.018705759379353,
      "learning_rate": 2.0025365428591875e-07,
      "loss": 0.2351,
      "step": 9867
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.5322832048513257,
      "learning_rate": 1.9997195701442763e-07,
      "loss": 0.3049,
      "step": 9868
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.392582613003293,
      "learning_rate": 1.996904497599769e-07,
      "loss": 0.2355,
      "step": 9869
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.932648631067604,
      "learning_rate": 1.9940913254582307e-07,
      "loss": 0.3893,
      "step": 9870
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3755493536120658,
      "learning_rate": 1.9912800539520855e-07,
      "loss": 0.2327,
      "step": 9871
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.177535263680919,
      "learning_rate": 1.9884706833135647e-07,
      "loss": 0.3024,
      "step": 9872
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.4180133389496286,
      "learning_rate": 1.9856632137747828e-07,
      "loss": 0.1358,
      "step": 9873
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5785408188446626,
      "learning_rate": 1.9828576455676706e-07,
      "loss": 0.0216,
      "step": 9874
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.152909584804003,
      "learning_rate": 1.9800539789240124e-07,
      "loss": 0.3622,
      "step": 9875
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.01129116664836,
      "learning_rate": 1.9772522140754396e-07,
      "loss": 0.2267,
      "step": 9876
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.515509713761657,
      "learning_rate": 1.9744523512534058e-07,
      "loss": 0.1715,
      "step": 9877
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.596712076876909,
      "learning_rate": 1.9716543906892372e-07,
      "loss": 0.4392,
      "step": 9878
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.239030925503328,
      "learning_rate": 1.9688583326140793e-07,
      "loss": 0.4478,
      "step": 9879
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.620158186826326,
      "learning_rate": 1.966064177258925e-07,
      "loss": 0.342,
      "step": 9880
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.053361549646234,
      "learning_rate": 1.963271924854629e-07,
      "loss": 0.2772,
      "step": 9881
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9076727340164283,
      "learning_rate": 1.9604815756318535e-07,
      "loss": 0.2679,
      "step": 9882
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.039147156109239,
      "learning_rate": 1.9576931298211393e-07,
      "loss": 0.531,
      "step": 9883
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6092306063946586,
      "learning_rate": 1.9549065876528494e-07,
      "loss": 0.1574,
      "step": 9884
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.203801662987757,
      "learning_rate": 1.9521219493571857e-07,
      "loss": 0.5335,
      "step": 9885
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.253294935676727,
      "learning_rate": 1.9493392151642198e-07,
      "loss": 0.2759,
      "step": 9886
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.7090665288188247,
      "learning_rate": 1.946558385303829e-07,
      "loss": 0.1667,
      "step": 9887
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.251137817267574,
      "learning_rate": 1.9437794600057626e-07,
      "loss": 0.2454,
      "step": 9888
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.975496088197841,
      "learning_rate": 1.941002439499598e-07,
      "loss": 0.2825,
      "step": 9889
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.879219169282509,
      "learning_rate": 1.9382273240147547e-07,
      "loss": 0.1379,
      "step": 9890
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.9343230100206545,
      "learning_rate": 1.9354541137805133e-07,
      "loss": 0.4606,
      "step": 9891
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.675543987862493,
      "learning_rate": 1.9326828090259653e-07,
      "loss": 0.4471,
      "step": 9892
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.632064237578835,
      "learning_rate": 1.929913409980075e-07,
      "loss": 0.3578,
      "step": 9893
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9353943347876954,
      "learning_rate": 1.9271459168716338e-07,
      "loss": 0.2348,
      "step": 9894
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.215191663302837,
      "learning_rate": 1.9243803299292707e-07,
      "loss": 0.3308,
      "step": 9895
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5897485476304554,
      "learning_rate": 1.9216166493814804e-07,
      "loss": 0.1097,
      "step": 9896
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.543373259243275,
      "learning_rate": 1.918854875456566e-07,
      "loss": 0.3975,
      "step": 9897
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.1737581136184665,
      "learning_rate": 1.9160950083827068e-07,
      "loss": 0.1768,
      "step": 9898
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9519244928188,
      "learning_rate": 1.9133370483879033e-07,
      "loss": 0.2576,
      "step": 9899
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.10039713449081,
      "learning_rate": 1.9105809956999981e-07,
      "loss": 0.1536,
      "step": 9900
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6059044633209627,
      "learning_rate": 1.9078268505467012e-07,
      "loss": 0.2197,
      "step": 9901
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.074849453016816,
      "learning_rate": 1.9050746131555247e-07,
      "loss": 0.2309,
      "step": 9902
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6773173664786825,
      "learning_rate": 1.9023242837538618e-07,
      "loss": 0.2447,
      "step": 9903
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.963953619974164,
      "learning_rate": 1.8995758625689225e-07,
      "loss": 0.3954,
      "step": 9904
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.2551669541970076,
      "learning_rate": 1.896829349827767e-07,
      "loss": 0.1906,
      "step": 9905
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.740532313860607,
      "learning_rate": 1.8940847457573102e-07,
      "loss": 0.3288,
      "step": 9906
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.071367761396012,
      "learning_rate": 1.891342050584283e-07,
      "loss": 0.346,
      "step": 9907
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.3430356871192832,
      "learning_rate": 1.888601264535281e-07,
      "loss": 0.1169,
      "step": 9908
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.987875490599252,
      "learning_rate": 1.8858623878367344e-07,
      "loss": 0.1765,
      "step": 9909
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.738740807121709,
      "learning_rate": 1.8831254207149125e-07,
      "loss": 0.242,
      "step": 9910
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.6794168174063016,
      "learning_rate": 1.880390363395937e-07,
      "loss": 0.4264,
      "step": 9911
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.56983724404181,
      "learning_rate": 1.877657216105755e-07,
      "loss": 0.3589,
      "step": 9912
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6693886291808613,
      "learning_rate": 1.8749259790701746e-07,
      "loss": 0.111,
      "step": 9913
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.6157959183817177,
      "learning_rate": 1.8721966525148322e-07,
      "loss": 0.2158,
      "step": 9914
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.998189464522005,
      "learning_rate": 1.8694692366652112e-07,
      "loss": 0.2252,
      "step": 9915
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.497574127716588,
      "learning_rate": 1.8667437317466425e-07,
      "loss": 0.3711,
      "step": 9916
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.503787469625297,
      "learning_rate": 1.8640201379842903e-07,
      "loss": 0.1329,
      "step": 9917
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.942783172643543,
      "learning_rate": 1.8612984556031638e-07,
      "loss": 0.142,
      "step": 9918
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.2016423838969015,
      "learning_rate": 1.8585786848281163e-07,
      "loss": 0.1903,
      "step": 9919
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3187825126408743,
      "learning_rate": 1.8558608258838372e-07,
      "loss": 0.2071,
      "step": 9920
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.27830613934005,
      "learning_rate": 1.8531448789948724e-07,
      "loss": 0.3637,
      "step": 9921
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.229415657533484,
      "learning_rate": 1.8504308443855923e-07,
      "loss": 0.1943,
      "step": 9922
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.094974208191738,
      "learning_rate": 1.847718722280223e-07,
      "loss": 0.3697,
      "step": 9923
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.744886692801761,
      "learning_rate": 1.8450085129028212e-07,
      "loss": 0.3539,
      "step": 9924
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.607473209703301,
      "learning_rate": 1.8423002164772885e-07,
      "loss": 0.5038,
      "step": 9925
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.421115009787333,
      "learning_rate": 1.8395938332273821e-07,
      "loss": 0.1874,
      "step": 9926
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.434667576824248,
      "learning_rate": 1.8368893633766815e-07,
      "loss": 0.0522,
      "step": 9927
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.1560192997523377,
      "learning_rate": 1.8341868071486218e-07,
      "loss": 0.1596,
      "step": 9928
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.596337806398352,
      "learning_rate": 1.8314861647664688e-07,
      "loss": 0.4167,
      "step": 9929
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.6475744931772556,
      "learning_rate": 1.8287874364533386e-07,
      "loss": 0.1558,
      "step": 9930
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.940892421030762,
      "learning_rate": 1.8260906224321916e-07,
      "loss": 0.1951,
      "step": 9931
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.098197207223663,
      "learning_rate": 1.823395722925822e-07,
      "loss": 0.32,
      "step": 9932
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.021319712305369,
      "learning_rate": 1.8207027381568682e-07,
      "loss": 0.4093,
      "step": 9933
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.8945007957802766,
      "learning_rate": 1.8180116683478106e-07,
      "loss": 0.1524,
      "step": 9934
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9833250310956294,
      "learning_rate": 1.8153225137209712e-07,
      "loss": 0.243,
      "step": 9935
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.0828233173535304,
      "learning_rate": 1.8126352744985227e-07,
      "loss": 0.1544,
      "step": 9936
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.746748971235443,
      "learning_rate": 1.8099499509024649e-07,
      "loss": 0.2894,
      "step": 9937
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.588293240153364,
      "learning_rate": 1.807266543154648e-07,
      "loss": 0.3696,
      "step": 9938
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.179859941475444,
      "learning_rate": 1.8045850514767622e-07,
      "loss": 0.2298,
      "step": 9939
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.564241629784693,
      "learning_rate": 1.801905476090332e-07,
      "loss": 0.2633,
      "step": 9940
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.462453954720817,
      "learning_rate": 1.799227817216745e-07,
      "loss": 0.2688,
      "step": 9941
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.221722739716454,
      "learning_rate": 1.7965520750772043e-07,
      "loss": 0.1523,
      "step": 9942
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.064203283641534,
      "learning_rate": 1.7938782498927748e-07,
      "loss": 0.2915,
      "step": 9943
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.249993548160784,
      "learning_rate": 1.7912063418843468e-07,
      "loss": 0.2018,
      "step": 9944
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.4936437840448145,
      "learning_rate": 1.7885363512726627e-07,
      "loss": 0.2381,
      "step": 9945
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.8378014751249303,
      "learning_rate": 1.7858682782783105e-07,
      "loss": 0.1502,
      "step": 9946
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.247334334250937,
      "learning_rate": 1.783202123121705e-07,
      "loss": 0.1631,
      "step": 9947
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.741671371957797,
      "learning_rate": 1.7805378860231177e-07,
      "loss": 0.2599,
      "step": 9948
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.9239784893611906,
      "learning_rate": 1.77787556720265e-07,
      "loss": 0.2304,
      "step": 9949
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.93980549047672,
      "learning_rate": 1.7752151668802452e-07,
      "loss": 0.152,
      "step": 9950
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.272812088849393,
      "learning_rate": 1.772556685275706e-07,
      "loss": 0.3841,
      "step": 9951
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.29405292438301,
      "learning_rate": 1.769900122608653e-07,
      "loss": 0.303,
      "step": 9952
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.228370470542147,
      "learning_rate": 1.7672454790985615e-07,
      "loss": 0.1515,
      "step": 9953
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.822054992534564,
      "learning_rate": 1.764592754964739e-07,
      "loss": 0.2374,
      "step": 9954
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.285130573432829,
      "learning_rate": 1.7619419504263518e-07,
      "loss": 0.0961,
      "step": 9955
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.660469619978311,
      "learning_rate": 1.7592930657023916e-07,
      "loss": 0.1833,
      "step": 9956
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.51951496141914,
      "learning_rate": 1.7566461010116913e-07,
      "loss": 0.2805,
      "step": 9957
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.1619817857301245,
      "learning_rate": 1.7540010565729376e-07,
      "loss": 0.0653,
      "step": 9958
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.8794476981184256,
      "learning_rate": 1.7513579326046414e-07,
      "loss": 0.2968,
      "step": 9959
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.160485252172202,
      "learning_rate": 1.7487167293251782e-07,
      "loss": 0.3,
      "step": 9960
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.1864203206708215,
      "learning_rate": 1.7460774469527402e-07,
      "loss": 0.3776,
      "step": 9961
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.999523492338231,
      "learning_rate": 1.743440085705378e-07,
      "loss": 0.2128,
      "step": 9962
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.3838642590202532,
      "learning_rate": 1.7408046458009753e-07,
      "loss": 0.2179,
      "step": 9963
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.357153033855,
      "learning_rate": 1.7381711274572528e-07,
      "loss": 0.199,
      "step": 9964
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.0738864713317504,
      "learning_rate": 1.7355395308917916e-07,
      "loss": 0.1167,
      "step": 9965
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.4477555136529436,
      "learning_rate": 1.732909856321993e-07,
      "loss": 0.2447,
      "step": 9966
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.2780161933797825,
      "learning_rate": 1.7302821039651113e-07,
      "loss": 0.2601,
      "step": 9967
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.320528224078119,
      "learning_rate": 1.7276562740382364e-07,
      "loss": 0.2229,
      "step": 9968
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.023706347178893,
      "learning_rate": 1.7250323667582975e-07,
      "loss": 0.5359,
      "step": 9969
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.280788363250021,
      "learning_rate": 1.7224103823420767e-07,
      "loss": 0.2558,
      "step": 9970
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.0186172703234733,
      "learning_rate": 1.7197903210061872e-07,
      "loss": 0.2147,
      "step": 9971
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.031595820876689,
      "learning_rate": 1.7171721829670833e-07,
      "loss": 0.2246,
      "step": 9972
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.5834675569719603,
      "learning_rate": 1.7145559684410673e-07,
      "loss": 0.1664,
      "step": 9973
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.796247845786646,
      "learning_rate": 1.7119416776442687e-07,
      "loss": 0.2585,
      "step": 9974
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.180038562265914,
      "learning_rate": 1.7093293107926762e-07,
      "loss": 0.2067,
      "step": 9975
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.169981822175069,
      "learning_rate": 1.7067188681021085e-07,
      "loss": 0.1594,
      "step": 9976
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.069106719840202,
      "learning_rate": 1.7041103497882293e-07,
      "loss": 0.4152,
      "step": 9977
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7884980143549845,
      "learning_rate": 1.7015037560665387e-07,
      "loss": 0.2227,
      "step": 9978
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.25349912341557,
      "learning_rate": 1.6988990871523754e-07,
      "loss": 0.2022,
      "step": 9979
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.082198583317164,
      "learning_rate": 1.6962963432609364e-07,
      "loss": 0.3899,
      "step": 9980
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.8272273380369235,
      "learning_rate": 1.6936955246072417e-07,
      "loss": 0.2947,
      "step": 9981
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.223768198889303,
      "learning_rate": 1.6910966314061584e-07,
      "loss": 0.1838,
      "step": 9982
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.014671511185173,
      "learning_rate": 1.688499663872392e-07,
      "loss": 0.4807,
      "step": 9983
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.107440127756817,
      "learning_rate": 1.6859046222204912e-07,
      "loss": 0.337,
      "step": 9984
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.660417294485303,
      "learning_rate": 1.6833115066648503e-07,
      "loss": 0.1645,
      "step": 9985
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2888118148791374,
      "learning_rate": 1.6807203174196958e-07,
      "loss": 0.1947,
      "step": 9986
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.216997133937149,
      "learning_rate": 1.678131054699103e-07,
      "loss": 0.2411,
      "step": 9987
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.557610298235775,
      "learning_rate": 1.6755437187169822e-07,
      "loss": 0.4223,
      "step": 9988
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.317266200075214,
      "learning_rate": 1.6729583096870784e-07,
      "loss": 0.1725,
      "step": 9989
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.359825116064199,
      "learning_rate": 1.6703748278229987e-07,
      "loss": 0.3321,
      "step": 9990
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.202165490593508,
      "learning_rate": 1.6677932733381726e-07,
      "loss": 0.173,
      "step": 9991
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1135885443914564,
      "learning_rate": 1.6652136464458734e-07,
      "loss": 0.2135,
      "step": 9992
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.304555132644627,
      "learning_rate": 1.6626359473592173e-07,
      "loss": 0.2097,
      "step": 9993
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0652626874932727,
      "learning_rate": 1.6600601762911617e-07,
      "loss": 0.2109,
      "step": 9994
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.183373805010607,
      "learning_rate": 1.6574863334545056e-07,
      "loss": 0.3583,
      "step": 9995
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.450606116624129,
      "learning_rate": 1.6549144190618872e-07,
      "loss": 0.4332,
      "step": 9996
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.548350999834788,
      "learning_rate": 1.652344433325781e-07,
      "loss": 0.2542,
      "step": 9997
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.981017338634201,
      "learning_rate": 1.6497763764585173e-07,
      "loss": 0.3434,
      "step": 9998
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.947638483347514,
      "learning_rate": 1.647210248672243e-07,
      "loss": 0.3395,
      "step": 9999
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.181523072551666,
      "learning_rate": 1.6446460501789691e-07,
      "loss": 0.294,
      "step": 10000
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0833546391854565,
      "learning_rate": 1.6420837811905342e-07,
      "loss": 0.0811,
      "step": 10001
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.6654430858794607,
      "learning_rate": 1.639523441918614e-07,
      "loss": 0.2815,
      "step": 10002
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7908155355201774,
      "learning_rate": 1.6369650325747444e-07,
      "loss": 0.1167,
      "step": 10003
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.537233311235478,
      "learning_rate": 1.6344085533702757e-07,
      "loss": 0.278,
      "step": 10004
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.157250704932134,
      "learning_rate": 1.631854004516417e-07,
      "loss": 0.2694,
      "step": 10005
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3935266655803553,
      "learning_rate": 1.6293013862242164e-07,
      "loss": 0.167,
      "step": 10006
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7543671765981048,
      "learning_rate": 1.626750698704549e-07,
      "loss": 0.0757,
      "step": 10007
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.137079997900556,
      "learning_rate": 1.6242019421681548e-07,
      "loss": 0.285,
      "step": 10008
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.8445534638369563,
      "learning_rate": 1.6216551168255823e-07,
      "loss": 0.3177,
      "step": 10009
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.996461236356794,
      "learning_rate": 1.6191102228872492e-07,
      "loss": 0.1902,
      "step": 10010
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.282538028231858,
      "learning_rate": 1.616567260563401e-07,
      "loss": 0.3519,
      "step": 10011
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.334259638575612,
      "learning_rate": 1.6140262300641197e-07,
      "loss": 0.3224,
      "step": 10012
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2121118373396977,
      "learning_rate": 1.6114871315993404e-07,
      "loss": 0.0956,
      "step": 10013
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.687565479883795,
      "learning_rate": 1.6089499653788225e-07,
      "loss": 0.3566,
      "step": 10014
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.9201748390704445,
      "learning_rate": 1.60641473161218e-07,
      "loss": 0.1893,
      "step": 10015
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.516409012798413,
      "learning_rate": 1.6038814305088607e-07,
      "loss": 0.2051,
      "step": 10016
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4519424700107337,
      "learning_rate": 1.601350062278148e-07,
      "loss": 0.0531,
      "step": 10017
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.601559305670913,
      "learning_rate": 1.598820627129183e-07,
      "loss": 0.3484,
      "step": 10018
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1469966328445573,
      "learning_rate": 1.59629312527092e-07,
      "loss": 0.1923,
      "step": 10019
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.809836359357729,
      "learning_rate": 1.593767556912182e-07,
      "loss": 0.5984,
      "step": 10020
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.201787134018627,
      "learning_rate": 1.5912439222616148e-07,
      "loss": 0.4302,
      "step": 10021
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.009909909200929,
      "learning_rate": 1.5887222215277026e-07,
      "loss": 0.3355,
      "step": 10022
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.516732972682045,
      "learning_rate": 1.5862024549187893e-07,
      "loss": 0.1848,
      "step": 10023
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.393221019054918,
      "learning_rate": 1.583684622643028e-07,
      "loss": 0.323,
      "step": 10024
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.094378173490401,
      "learning_rate": 1.581168724908444e-07,
      "loss": 0.1818,
      "step": 10025
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.9219331917207665,
      "learning_rate": 1.578654761922885e-07,
      "loss": 0.4596,
      "step": 10026
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.713534760103401,
      "learning_rate": 1.576142733894037e-07,
      "loss": 0.2498,
      "step": 10027
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.295987312085036,
      "learning_rate": 1.5736326410294405e-07,
      "loss": 0.2314,
      "step": 10028
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.796755696395785,
      "learning_rate": 1.571124483536457e-07,
      "loss": 0.3202,
      "step": 10029
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7537435432563875,
      "learning_rate": 1.568618261622304e-07,
      "loss": 0.2118,
      "step": 10030
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.5449288122560345,
      "learning_rate": 1.5661139754940353e-07,
      "loss": 0.2581,
      "step": 10031
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3150520831942645,
      "learning_rate": 1.5636116253585354e-07,
      "loss": 0.1173,
      "step": 10032
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.418409838877001,
      "learning_rate": 1.5611112114225474e-07,
      "loss": 0.1269,
      "step": 10033
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.2039051267965153,
      "learning_rate": 1.558612733892628e-07,
      "loss": 0.1883,
      "step": 10034
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.5574215371713365,
      "learning_rate": 1.5561161929752045e-07,
      "loss": 0.2444,
      "step": 10035
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.181036205604112,
      "learning_rate": 1.5536215888765194e-07,
      "loss": 0.248,
      "step": 10036
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4816391108449658,
      "learning_rate": 1.5511289218026637e-07,
      "loss": 0.0047,
      "step": 10037
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.8077424541541305,
      "learning_rate": 1.5486381919595812e-07,
      "loss": 0.3323,
      "step": 10038
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7409288455308767,
      "learning_rate": 1.5461493995530263e-07,
      "loss": 0.251,
      "step": 10039
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.901697520124555,
      "learning_rate": 1.543662544788624e-07,
      "loss": 0.3515,
      "step": 10040
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.584809327539787,
      "learning_rate": 1.5411776278718233e-07,
      "loss": 0.2971,
      "step": 10041
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.115615096526308,
      "learning_rate": 1.5386946490079103e-07,
      "loss": 0.1115,
      "step": 10042
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.764633804123802,
      "learning_rate": 1.536213608402029e-07,
      "loss": 0.1462,
      "step": 10043
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.474020210802252,
      "learning_rate": 1.5337345062591325e-07,
      "loss": 0.4655,
      "step": 10044
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.332754330246797,
      "learning_rate": 1.5312573427840454e-07,
      "loss": 0.4151,
      "step": 10045
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.092786708866617,
      "learning_rate": 1.528782118181421e-07,
      "loss": 0.3036,
      "step": 10046
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.31554917986725,
      "learning_rate": 1.52630883265574e-07,
      "loss": 0.3017,
      "step": 10047
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.6677574486352675,
      "learning_rate": 1.523837486411342e-07,
      "loss": 0.0759,
      "step": 10048
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.3934989356348995,
      "learning_rate": 1.5213680796523883e-07,
      "loss": 0.2862,
      "step": 10049
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.666932659924919,
      "learning_rate": 1.5189006125828936e-07,
      "loss": 0.2602,
      "step": 10050
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.206904829876253,
      "learning_rate": 1.516435085406717e-07,
      "loss": 0.3715,
      "step": 10051
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0532147902812428,
      "learning_rate": 1.5139714983275295e-07,
      "loss": 0.1808,
      "step": 10052
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.018352676504925,
      "learning_rate": 1.5115098515488813e-07,
      "loss": 0.3799,
      "step": 10053
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.2858127279826075,
      "learning_rate": 1.509050145274124e-07,
      "loss": 0.4513,
      "step": 10054
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.86371938223359,
      "learning_rate": 1.506592379706473e-07,
      "loss": 0.2236,
      "step": 10055
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4358263648544485,
      "learning_rate": 1.504136555048985e-07,
      "loss": 0.2116,
      "step": 10056
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.193191864793739,
      "learning_rate": 1.5016826715045313e-07,
      "loss": 0.0899,
      "step": 10057
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.674922007347308,
      "learning_rate": 1.4992307292758605e-07,
      "loss": 0.3278,
      "step": 10058
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.168716054922217,
      "learning_rate": 1.4967807285655195e-07,
      "loss": 0.4007,
      "step": 10059
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.217936720615384,
      "learning_rate": 1.4943326695759237e-07,
      "loss": 0.2269,
      "step": 10060
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.5194373673918733,
      "learning_rate": 1.491886552509328e-07,
      "loss": 0.1505,
      "step": 10061
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.8959872806479803,
      "learning_rate": 1.4894423775678069e-07,
      "loss": 0.2044,
      "step": 10062
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.350179163502309,
      "learning_rate": 1.4870001449532962e-07,
      "loss": 0.1209,
      "step": 10063
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.3457571986687733,
      "learning_rate": 1.4845598548675454e-07,
      "loss": 0.1942,
      "step": 10064
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.339615470274331,
      "learning_rate": 1.4821215075121742e-07,
      "loss": 0.309,
      "step": 10065
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.388007868220577,
      "learning_rate": 1.4796851030886267e-07,
      "loss": 0.2142,
      "step": 10066
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.09198547360148,
      "learning_rate": 1.4772506417981753e-07,
      "loss": 0.3408,
      "step": 10067
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.4189708403335235,
      "learning_rate": 1.4748181238419535e-07,
      "loss": 0.2203,
      "step": 10068
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.173041425056966,
      "learning_rate": 1.47238754942092e-07,
      "loss": 0.429,
      "step": 10069
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.0582056296931306,
      "learning_rate": 1.4699589187358725e-07,
      "loss": 0.1818,
      "step": 10070
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.89978875232229,
      "learning_rate": 1.4675322319874673e-07,
      "loss": 0.2591,
      "step": 10071
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.358091568177816,
      "learning_rate": 1.4651074893761692e-07,
      "loss": 0.3631,
      "step": 10072
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.89149215222344,
      "learning_rate": 1.4626846911023062e-07,
      "loss": 0.3739,
      "step": 10073
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4885354561957933,
      "learning_rate": 1.4602638373660383e-07,
      "loss": 0.1359,
      "step": 10074
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.2109741291893465,
      "learning_rate": 1.457844928367358e-07,
      "loss": 0.1933,
      "step": 10075
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.7209011951854816,
      "learning_rate": 1.4554279643061164e-07,
      "loss": 0.0789,
      "step": 10076
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.7529801603575925,
      "learning_rate": 1.4530129453819792e-07,
      "loss": 0.1516,
      "step": 10077
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.8458076839545225,
      "learning_rate": 1.450599871794467e-07,
      "loss": 0.1848,
      "step": 10078
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.722065195572098,
      "learning_rate": 1.4481887437429397e-07,
      "loss": 0.461,
      "step": 10079
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9794547807579939,
      "learning_rate": 1.4457795614265857e-07,
      "loss": 0.0713,
      "step": 10080
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.19932425655059,
      "learning_rate": 1.443372325044451e-07,
      "loss": 0.4855,
      "step": 10081
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.996987447116089,
      "learning_rate": 1.4409670347953958e-07,
      "loss": 0.3002,
      "step": 10082
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0864471943867744,
      "learning_rate": 1.4385636908781448e-07,
      "loss": 0.0677,
      "step": 10083
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.041626635783796,
      "learning_rate": 1.4361622934912445e-07,
      "loss": 0.1419,
      "step": 10084
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.381436545108713,
      "learning_rate": 1.4337628428330864e-07,
      "loss": 0.4448,
      "step": 10085
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.661818191502862,
      "learning_rate": 1.4313653391019116e-07,
      "loss": 0.2156,
      "step": 10086
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.297585460295797,
      "learning_rate": 1.4289697824957705e-07,
      "loss": 0.4345,
      "step": 10087
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9034930535128383,
      "learning_rate": 1.4265761732125905e-07,
      "loss": 0.1807,
      "step": 10088
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.602680713485937,
      "learning_rate": 1.4241845114501108e-07,
      "loss": 0.1945,
      "step": 10089
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.792594336351961,
      "learning_rate": 1.4217947974059175e-07,
      "loss": 0.2885,
      "step": 10090
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6150988852076136,
      "learning_rate": 1.4194070312774477e-07,
      "loss": 0.0427,
      "step": 10091
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.6751372843128935,
      "learning_rate": 1.4170212132619543e-07,
      "loss": 0.4305,
      "step": 10092
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7632239797260585,
      "learning_rate": 1.414637343556549e-07,
      "loss": 0.1237,
      "step": 10093
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7186244992177944,
      "learning_rate": 1.412255422358172e-07,
      "loss": 0.2319,
      "step": 10094
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.477532874303072,
      "learning_rate": 1.409875449863607e-07,
      "loss": 0.2756,
      "step": 10095
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.442510957913314,
      "learning_rate": 1.407497426269483e-07,
      "loss": 0.2901,
      "step": 10096
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.707880498331033,
      "learning_rate": 1.4051213517722427e-07,
      "loss": 0.3089,
      "step": 10097
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.026032210543919,
      "learning_rate": 1.4027472265682045e-07,
      "loss": 0.3342,
      "step": 10098
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6896283257320124,
      "learning_rate": 1.4003750508534974e-07,
      "loss": 0.2148,
      "step": 10099
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.71190431284692,
      "learning_rate": 1.398004824824098e-07,
      "loss": 0.1976,
      "step": 10100
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.746314408793106,
      "learning_rate": 1.3956365486758327e-07,
      "loss": 0.3214,
      "step": 10101
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.164466056738327,
      "learning_rate": 1.3932702226043427e-07,
      "loss": 0.3152,
      "step": 10102
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.4878972618960855,
      "learning_rate": 1.390905846805135e-07,
      "loss": 0.4385,
      "step": 10103
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.827301067942611,
      "learning_rate": 1.3885434214735343e-07,
      "loss": 0.4528,
      "step": 10104
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.6179502910681345,
      "learning_rate": 1.386182946804712e-07,
      "loss": 0.3252,
      "step": 10105
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3906966607935813,
      "learning_rate": 1.3838244229936904e-07,
      "loss": 0.1044,
      "step": 10106
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.0142257534122825,
      "learning_rate": 1.3814678502353046e-07,
      "loss": 0.2243,
      "step": 10107
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.648600448116173,
      "learning_rate": 1.379113228724252e-07,
      "loss": 0.3574,
      "step": 10108
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9864720896754224,
      "learning_rate": 1.3767605586550577e-07,
      "loss": 0.0806,
      "step": 10109
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.1875969732016225,
      "learning_rate": 1.3744098402220824e-07,
      "loss": 0.1416,
      "step": 10110
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.2108063567323004,
      "learning_rate": 1.372061073619546e-07,
      "loss": 0.2178,
      "step": 10111
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.118876194776084,
      "learning_rate": 1.3697142590414713e-07,
      "loss": 0.3851,
      "step": 10112
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.919158078173349,
      "learning_rate": 1.3673693966817586e-07,
      "loss": 0.2726,
      "step": 10113
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.838063106689988,
      "learning_rate": 1.3650264867341222e-07,
      "loss": 0.3948,
      "step": 10114
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.004965557888275,
      "learning_rate": 1.362685529392116e-07,
      "loss": 0.2395,
      "step": 10115
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.47833928483785,
      "learning_rate": 1.360346524849146e-07,
      "loss": 0.3498,
      "step": 10116
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.351029271411513,
      "learning_rate": 1.3580094732984493e-07,
      "loss": 0.1591,
      "step": 10117
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.691038175074694,
      "learning_rate": 1.3556743749330996e-07,
      "loss": 0.3201,
      "step": 10118
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.020322691191302,
      "learning_rate": 1.3533412299460091e-07,
      "loss": 0.3406,
      "step": 10119
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.529603253297196,
      "learning_rate": 1.3510100385299317e-07,
      "loss": 0.2787,
      "step": 10120
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.7414890840545,
      "learning_rate": 1.3486808008774638e-07,
      "loss": 0.3054,
      "step": 10121
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.8937035143277274,
      "learning_rate": 1.3463535171810317e-07,
      "loss": 0.3195,
      "step": 10122
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.218049122528655,
      "learning_rate": 1.344028187632901e-07,
      "loss": 0.3847,
      "step": 10123
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.962009094722815,
      "learning_rate": 1.3417048124251876e-07,
      "loss": 0.4973,
      "step": 10124
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.0334408464552505,
      "learning_rate": 1.339383391749824e-07,
      "loss": 0.3871,
      "step": 10125
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5763170765220185,
      "learning_rate": 1.3370639257986095e-07,
      "loss": 0.2335,
      "step": 10126
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.831179254987879,
      "learning_rate": 1.3347464147631605e-07,
      "loss": 0.2113,
      "step": 10127
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.970515903046348,
      "learning_rate": 1.332430858834935e-07,
      "loss": 0.4039,
      "step": 10128
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.782876703658037,
      "learning_rate": 1.3301172582052408e-07,
      "loss": 0.3068,
      "step": 10129
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.4699375028579897,
      "learning_rate": 1.3278056130652028e-07,
      "loss": 0.139,
      "step": 10130
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.489748964289207,
      "learning_rate": 1.325495923605813e-07,
      "loss": 0.1593,
      "step": 10131
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.126026871077764,
      "learning_rate": 1.3231881900178794e-07,
      "loss": 0.3626,
      "step": 10132
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.493600729276178,
      "learning_rate": 1.3208824124920584e-07,
      "loss": 0.1262,
      "step": 10133
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.701043164074623,
      "learning_rate": 1.3185785912188387e-07,
      "loss": 0.1172,
      "step": 10134
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.14890777152305,
      "learning_rate": 1.316276726388549e-07,
      "loss": 0.0717,
      "step": 10135
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.952968362062301,
      "learning_rate": 1.313976818191362e-07,
      "loss": 0.2856,
      "step": 10136
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.680559004712572,
      "learning_rate": 1.311678866817287e-07,
      "loss": 0.2448,
      "step": 10137
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.230436872166253,
      "learning_rate": 1.3093828724561664e-07,
      "loss": 0.1962,
      "step": 10138
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.778346547282839,
      "learning_rate": 1.307088835297682e-07,
      "loss": 0.3808,
      "step": 10139
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.4189542166552784,
      "learning_rate": 1.304796755531354e-07,
      "loss": 0.1758,
      "step": 10140
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.2540096806718406,
      "learning_rate": 1.3025066333465535e-07,
      "loss": 0.2095,
      "step": 10141
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5310344873351074,
      "learning_rate": 1.3002184689324709e-07,
      "loss": 0.172,
      "step": 10142
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.2837003765369395,
      "learning_rate": 1.2979322624781437e-07,
      "loss": 0.1417,
      "step": 10143
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7876819347491693,
      "learning_rate": 1.2956480141724486e-07,
      "loss": 0.1671,
      "step": 10144
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.689122691335596,
      "learning_rate": 1.2933657242040932e-07,
      "loss": 0.1472,
      "step": 10145
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.550499430629142,
      "learning_rate": 1.2910853927616402e-07,
      "loss": 0.4285,
      "step": 10146
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7295219906703636,
      "learning_rate": 1.2888070200334724e-07,
      "loss": 0.1418,
      "step": 10147
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.540890311177231,
      "learning_rate": 1.28653060620782e-07,
      "loss": 0.2973,
      "step": 10148
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.894958217148485,
      "learning_rate": 1.284256151472746e-07,
      "loss": 0.2733,
      "step": 10149
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.864586011894568,
      "learning_rate": 1.2819836560161557e-07,
      "loss": 0.2774,
      "step": 10150
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.002674244915102,
      "learning_rate": 1.279713120025794e-07,
      "loss": 0.0295,
      "step": 10151
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5021864439198507,
      "learning_rate": 1.2774445436892408e-07,
      "loss": 0.1166,
      "step": 10152
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.768385193431882,
      "learning_rate": 1.2751779271939157e-07,
      "loss": 0.1201,
      "step": 10153
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6301010539748364,
      "learning_rate": 1.2729132707270718e-07,
      "loss": 0.2231,
      "step": 10154
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.978108578295238,
      "learning_rate": 1.2706505744758013e-07,
      "loss": 0.2825,
      "step": 10155
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.480497598768282,
      "learning_rate": 1.268389838627046e-07,
      "loss": 0.248,
      "step": 10156
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.289077366282432,
      "learning_rate": 1.2661310633675733e-07,
      "loss": 0.4151,
      "step": 10157
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.544723973725101,
      "learning_rate": 1.263874248883992e-07,
      "loss": 0.277,
      "step": 10158
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.161234638875655,
      "learning_rate": 1.2616193953627425e-07,
      "loss": 0.2046,
      "step": 10159
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.370946187005405,
      "learning_rate": 1.2593665029901197e-07,
      "loss": 0.1808,
      "step": 10160
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.269503532917139,
      "learning_rate": 1.2571155719522443e-07,
      "loss": 0.4138,
      "step": 10161
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.764047109077235,
      "learning_rate": 1.254866602435073e-07,
      "loss": 0.2568,
      "step": 10162
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.226589818221532,
      "learning_rate": 1.2526195946244074e-07,
      "loss": 0.2264,
      "step": 10163
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.679621688918868,
      "learning_rate": 1.250374548705882e-07,
      "loss": 0.26,
      "step": 10164
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.493827800892967,
      "learning_rate": 1.248131464864974e-07,
      "loss": 0.3006,
      "step": 10165
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.473001108998162,
      "learning_rate": 1.2458903432869957e-07,
      "loss": 0.2348,
      "step": 10166
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.840146035458551,
      "learning_rate": 1.2436511841570992e-07,
      "loss": 0.4168,
      "step": 10167
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.0097927571852634,
      "learning_rate": 1.24141398766027e-07,
      "loss": 0.092,
      "step": 10168
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.75764503367319,
      "learning_rate": 1.2391787539813294e-07,
      "loss": 0.4684,
      "step": 10169
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5267885110505857,
      "learning_rate": 1.236945483304952e-07,
      "loss": 0.2214,
      "step": 10170
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7022621359760026,
      "learning_rate": 1.234714175815635e-07,
      "loss": 0.2485,
      "step": 10171
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9395292362863845,
      "learning_rate": 1.2324848316977195e-07,
      "loss": 0.2717,
      "step": 10172
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.153831394796417,
      "learning_rate": 1.2302574511353776e-07,
      "loss": 0.1016,
      "step": 10173
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.212790981926701,
      "learning_rate": 1.228032034312629e-07,
      "loss": 0.3232,
      "step": 10174
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.8574102310155993,
      "learning_rate": 1.2258085814133263e-07,
      "loss": 0.1909,
      "step": 10175
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.402645011226367,
      "learning_rate": 1.223587092621162e-07,
      "loss": 0.1299,
      "step": 10176
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.328492683982525,
      "learning_rate": 1.2213675681196607e-07,
      "loss": 0.2101,
      "step": 10177
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.299147591792965,
      "learning_rate": 1.219150008092193e-07,
      "loss": 0.2597,
      "step": 10178
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.640586052456623,
      "learning_rate": 1.216934412721954e-07,
      "loss": 0.146,
      "step": 10179
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.913327774289551,
      "learning_rate": 1.214720782191997e-07,
      "loss": 0.273,
      "step": 10180
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.5132453493623244,
      "learning_rate": 1.2125091166851983e-07,
      "loss": 0.1959,
      "step": 10181
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.180751459470585,
      "learning_rate": 1.2102994163842696e-07,
      "loss": 0.1735,
      "step": 10182
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6416771037947213,
      "learning_rate": 1.208091681471768e-07,
      "loss": 0.2455,
      "step": 10183
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.3897559953762553,
      "learning_rate": 1.2058859121300837e-07,
      "loss": 0.2398,
      "step": 10184
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.4191903963473465,
      "learning_rate": 1.203682108541454e-07,
      "loss": 0.2242,
      "step": 10185
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.6404830518176943,
      "learning_rate": 1.2014802708879392e-07,
      "loss": 0.1679,
      "step": 10186
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.193351912496063,
      "learning_rate": 1.1992803993514462e-07,
      "loss": 0.5049,
      "step": 10187
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.111680968798034,
      "learning_rate": 1.1970824941137214e-07,
      "loss": 0.2837,
      "step": 10188
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9918396044833675,
      "learning_rate": 1.194886555356334e-07,
      "loss": 0.3634,
      "step": 10189
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.571873169615521,
      "learning_rate": 1.1926925832607157e-07,
      "loss": 0.2916,
      "step": 10190
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.2341699125932486,
      "learning_rate": 1.1905005780081136e-07,
      "loss": 0.2087,
      "step": 10191
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.0256613458716135,
      "learning_rate": 1.1883105397796219e-07,
      "loss": 0.177,
      "step": 10192
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.7357274920762245,
      "learning_rate": 1.1861224687561734e-07,
      "loss": 0.2154,
      "step": 10193
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1996115099862905,
      "learning_rate": 1.1839363651185265e-07,
      "loss": 0.0785,
      "step": 10194
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.493972284028822,
      "learning_rate": 1.1817522290472977e-07,
      "loss": 0.4239,
      "step": 10195
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.444841162035516,
      "learning_rate": 1.179570060722926e-07,
      "loss": 0.2123,
      "step": 10196
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.9802922877177638,
      "learning_rate": 1.1773898603256894e-07,
      "loss": 0.3361,
      "step": 10197
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.6921786935457193,
      "learning_rate": 1.1752116280357107e-07,
      "loss": 0.2712,
      "step": 10198
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.9815531776736157,
      "learning_rate": 1.1730353640329344e-07,
      "loss": 0.3297,
      "step": 10199
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.5794991305902375,
      "learning_rate": 1.1708610684971617e-07,
      "loss": 0.1471,
      "step": 10200
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3473645857549226,
      "learning_rate": 1.1686887416080179e-07,
      "loss": 0.0127,
      "step": 10201
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.401185816181251,
      "learning_rate": 1.1665183835449707e-07,
      "loss": 0.3067,
      "step": 10202
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.469476819995939,
      "learning_rate": 1.1643499944873321e-07,
      "loss": 0.2825,
      "step": 10203
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.374942518943572,
      "learning_rate": 1.1621835746142313e-07,
      "loss": 0.3138,
      "step": 10204
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3126332949575144,
      "learning_rate": 1.1600191241046554e-07,
      "loss": 0.1326,
      "step": 10205
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.8007818206306956,
      "learning_rate": 1.1578566431374172e-07,
      "loss": 0.272,
      "step": 10206
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.859591118043511,
      "learning_rate": 1.155696131891168e-07,
      "loss": 0.1664,
      "step": 10207
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.6578775957738925,
      "learning_rate": 1.1535375905444097e-07,
      "loss": 0.2239,
      "step": 10208
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.367810804589893,
      "learning_rate": 1.1513810192754576e-07,
      "loss": 0.2103,
      "step": 10209
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.082066082180814,
      "learning_rate": 1.1492264182624834e-07,
      "loss": 0.1073,
      "step": 10210
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.057932478523878,
      "learning_rate": 1.1470737876834892e-07,
      "loss": 0.1478,
      "step": 10211
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.005041387511929,
      "learning_rate": 1.1449231277163104e-07,
      "loss": 0.3092,
      "step": 10212
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.674425354381003,
      "learning_rate": 1.1427744385386352e-07,
      "loss": 0.2813,
      "step": 10213
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1155055581773903,
      "learning_rate": 1.1406277203279608e-07,
      "loss": 0.1992,
      "step": 10214
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.228958888600255,
      "learning_rate": 1.1384829732616537e-07,
      "loss": 0.1808,
      "step": 10215
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.040334574191254,
      "learning_rate": 1.136340197516897e-07,
      "loss": 0.2518,
      "step": 10216
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.5038146386109137,
      "learning_rate": 1.1341993932707102e-07,
      "loss": 0.1303,
      "step": 10217
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.639556463545122,
      "learning_rate": 1.1320605606999713e-07,
      "loss": 0.2236,
      "step": 10218
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.800734151961072,
      "learning_rate": 1.129923699981364e-07,
      "loss": 0.0176,
      "step": 10219
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.322085178717749,
      "learning_rate": 1.1277888112914331e-07,
      "loss": 0.2014,
      "step": 10220
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.182036785344244,
      "learning_rate": 1.1256558948065515e-07,
      "loss": 0.405,
      "step": 10221
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1411227523443659,
      "learning_rate": 1.1235249507029284e-07,
      "loss": 0.0284,
      "step": 10222
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.237115648959342,
      "learning_rate": 1.1213959791566198e-07,
      "loss": 0.2926,
      "step": 10223
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.367648832072851,
      "learning_rate": 1.1192689803434992e-07,
      "loss": 0.39,
      "step": 10224
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.1976086207732277,
      "learning_rate": 1.1171439544392954e-07,
      "loss": 0.1032,
      "step": 10225
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.926616988914841,
      "learning_rate": 1.1150209016195706e-07,
      "loss": 0.3545,
      "step": 10226
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.269362932791369,
      "learning_rate": 1.1128998220597098e-07,
      "loss": 0.1531,
      "step": 10227
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.656876621964192,
      "learning_rate": 1.1107807159349643e-07,
      "loss": 0.3388,
      "step": 10228
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.040847262879763,
      "learning_rate": 1.1086635834203829e-07,
      "loss": 0.3403,
      "step": 10229
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.830470306593721,
      "learning_rate": 1.1065484246908898e-07,
      "loss": 0.341,
      "step": 10230
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.8659538860738785,
      "learning_rate": 1.10443523992122e-07,
      "loss": 0.2201,
      "step": 10231
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.8632017935593543,
      "learning_rate": 1.1023240292859561e-07,
      "loss": 0.1832,
      "step": 10232
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.482414843852079,
      "learning_rate": 1.1002147929595225e-07,
      "loss": 0.3002,
      "step": 10233
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.936368593262057,
      "learning_rate": 1.0981075311161604e-07,
      "loss": 0.5158,
      "step": 10234
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.669271614413388,
      "learning_rate": 1.096002243929975e-07,
      "loss": 0.2894,
      "step": 10235
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.655591732754236,
      "learning_rate": 1.0938989315748882e-07,
      "loss": 0.2305,
      "step": 10236
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.314893637791801,
      "learning_rate": 1.0917975942246611e-07,
      "loss": 0.2915,
      "step": 10237
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.247212725833831,
      "learning_rate": 1.0896982320529103e-07,
      "loss": 0.3406,
      "step": 10238
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.25414321796785,
      "learning_rate": 1.0876008452330582e-07,
      "loss": 0.2307,
      "step": 10239
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.9864817260464926,
      "learning_rate": 1.0855054339383914e-07,
      "loss": 0.0949,
      "step": 10240
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.352672167038536,
      "learning_rate": 1.0834119983420183e-07,
      "loss": 0.2513,
      "step": 10241
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.713426164657278,
      "learning_rate": 1.0813205386168896e-07,
      "loss": 0.3216,
      "step": 10242
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.981041390402119,
      "learning_rate": 1.0792310549357948e-07,
      "loss": 0.3416,
      "step": 10243
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.3854831030461416,
      "learning_rate": 1.077143547471346e-07,
      "loss": 0.1741,
      "step": 10244
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.863397729819696,
      "learning_rate": 1.0750580163960105e-07,
      "loss": 0.3704,
      "step": 10245
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.4185696311689995,
      "learning_rate": 1.0729744618820897e-07,
      "loss": 0.0166,
      "step": 10246
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.22017432527918,
      "learning_rate": 1.0708928841017069e-07,
      "loss": 0.3988,
      "step": 10247
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.770731063102791,
      "learning_rate": 1.0688132832268383e-07,
      "loss": 0.1209,
      "step": 10248
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.665653705329667,
      "learning_rate": 1.0667356594292854e-07,
      "loss": 0.2452,
      "step": 10249
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.407446883146841,
      "learning_rate": 1.0646600128806917e-07,
      "loss": 0.3898,
      "step": 10250
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.934840212931296,
      "learning_rate": 1.0625863437525474e-07,
      "loss": 0.2466,
      "step": 10251
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.974859372014786,
      "learning_rate": 1.0605146522161519e-07,
      "loss": 0.3002,
      "step": 10252
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.9918220026850832,
      "learning_rate": 1.0584449384426736e-07,
      "loss": 0.4015,
      "step": 10253
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.930051085420111,
      "learning_rate": 1.05637720260309e-07,
      "loss": 0.3248,
      "step": 10254
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.340081449768206,
      "learning_rate": 1.0543114448682335e-07,
      "loss": 0.4577,
      "step": 10255
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.71690754701432,
      "learning_rate": 1.0522476654087705e-07,
      "loss": 0.4121,
      "step": 10256
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.147028201333759,
      "learning_rate": 1.0501858643951923e-07,
      "loss": 0.4844,
      "step": 10257
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.6783582035948115,
      "learning_rate": 1.0481260419978406e-07,
      "loss": 0.4613,
      "step": 10258
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.512593155957478,
      "learning_rate": 1.046068198386882e-07,
      "loss": 0.3359,
      "step": 10259
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.8398969407640875,
      "learning_rate": 1.0440123337323277e-07,
      "loss": 0.3522,
      "step": 10260
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.951804889656352,
      "learning_rate": 1.0419584482040335e-07,
      "loss": 0.4938,
      "step": 10261
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.1778393098347784,
      "learning_rate": 1.0399065419716636e-07,
      "loss": 0.1226,
      "step": 10262
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.697590814843145,
      "learning_rate": 1.037856615204752e-07,
      "loss": 0.2465,
      "step": 10263
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.747388440861455,
      "learning_rate": 1.0358086680726409e-07,
      "loss": 0.4351,
      "step": 10264
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.6603823031683964,
      "learning_rate": 1.0337627007445255e-07,
      "loss": 0.164,
      "step": 10265
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.250591938166395,
      "learning_rate": 1.031718713389443e-07,
      "loss": 0.3765,
      "step": 10266
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.720190864899054,
      "learning_rate": 1.029676706176247e-07,
      "loss": 0.3637,
      "step": 10267
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.140220987912869,
      "learning_rate": 1.0276366792736442e-07,
      "loss": 0.1852,
      "step": 10268
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.665050936515874,
      "learning_rate": 1.0255986328501638e-07,
      "loss": 0.2958,
      "step": 10269
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.6993712104026866,
      "learning_rate": 1.0235625670741822e-07,
      "loss": 0.1529,
      "step": 10270
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.378666634806296,
      "learning_rate": 1.0215284821139205e-07,
      "loss": 0.1525,
      "step": 10271
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.831056361477971,
      "learning_rate": 1.0194963781374078e-07,
      "loss": 0.1765,
      "step": 10272
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.049362856334923,
      "learning_rate": 1.017466255312538e-07,
      "loss": 0.0677,
      "step": 10273
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.505260294219125,
      "learning_rate": 1.015438113807024e-07,
      "loss": 0.2444,
      "step": 10274
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.293127803726427,
      "learning_rate": 1.0134119537884235e-07,
      "loss": 0.1847,
      "step": 10275
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.920702347689428,
      "learning_rate": 1.0113877754241302e-07,
      "loss": 0.5579,
      "step": 10276
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.246228257644653,
      "learning_rate": 1.009365578881366e-07,
      "loss": 0.053,
      "step": 10277
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.8366058442743807,
      "learning_rate": 1.0073453643271974e-07,
      "loss": 0.0336,
      "step": 10278
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.161813859675761,
      "learning_rate": 1.0053271319285268e-07,
      "loss": 0.2407,
      "step": 10279
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.135720110701824,
      "learning_rate": 1.0033108818520876e-07,
      "loss": 0.3222,
      "step": 10280
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.785855834467171,
      "learning_rate": 1.0012966142644575e-07,
      "loss": 0.2636,
      "step": 10281
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.476722082332576,
      "learning_rate": 9.99284329332037e-08,
      "loss": 0.2492,
      "step": 10282
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.001300699495517,
      "learning_rate": 9.97274027221079e-08,
      "loss": 0.2807,
      "step": 10283
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.5224260802012495,
      "learning_rate": 9.952657080976619e-08,
      "loss": 0.4481,
      "step": 10284
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.157783663344168,
      "learning_rate": 9.932593721277001e-08,
      "loss": 0.1975,
      "step": 10285
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.862612456313278,
      "learning_rate": 9.912550194769583e-08,
      "loss": 0.288,
      "step": 10286
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.766107989531802,
      "learning_rate": 9.892526503110095e-08,
      "loss": 0.4407,
      "step": 10287
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.005842425585413,
      "learning_rate": 9.872522647952909e-08,
      "loss": 0.288,
      "step": 10288
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.848529158605377,
      "learning_rate": 9.852538630950648e-08,
      "loss": 0.085,
      "step": 10289
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.6929858595058676,
      "learning_rate": 9.832574453754212e-08,
      "loss": 0.2614,
      "step": 10290
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.502926138597503,
      "learning_rate": 9.812630118013062e-08,
      "loss": 0.3867,
      "step": 10291
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.4259300989912544,
      "learning_rate": 9.792705625374793e-08,
      "loss": 0.1513,
      "step": 10292
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.75138901986234,
      "learning_rate": 9.772800977485508e-08,
      "loss": 0.3196,
      "step": 10293
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.913128847432528,
      "learning_rate": 9.752916175989668e-08,
      "loss": 0.3991,
      "step": 10294
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.4604648900881703,
      "learning_rate": 9.733051222529987e-08,
      "loss": 0.1629,
      "step": 10295
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.094515186077972,
      "learning_rate": 9.713206118747709e-08,
      "loss": 0.0764,
      "step": 10296
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.237977073998667,
      "learning_rate": 9.693380866282215e-08,
      "loss": 0.3194,
      "step": 10297
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.928765694888776,
      "learning_rate": 9.673575466771445e-08,
      "loss": 0.2471,
      "step": 10298
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.78366561156596,
      "learning_rate": 9.653789921851648e-08,
      "loss": 0.2819,
      "step": 10299
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.394609845552299,
      "learning_rate": 9.634024233157318e-08,
      "loss": 0.2487,
      "step": 10300
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.9174833062672563,
      "learning_rate": 9.614278402321541e-08,
      "loss": 0.1253,
      "step": 10301
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.6282810770150227,
      "learning_rate": 9.594552430975457e-08,
      "loss": 0.1407,
      "step": 10302
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.226303592250457,
      "learning_rate": 9.574846320748843e-08,
      "loss": 0.0909,
      "step": 10303
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.403620122732656,
      "learning_rate": 9.555160073269676e-08,
      "loss": 0.3223,
      "step": 10304
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.5424257160700052,
      "learning_rate": 9.535493690164321e-08,
      "loss": 0.2392,
      "step": 10305
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.49395738073694,
      "learning_rate": 9.515847173057591e-08,
      "loss": 0.2579,
      "step": 10306
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.465587932984746,
      "learning_rate": 9.496220523572492e-08,
      "loss": 0.3844,
      "step": 10307
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.220679267509998,
      "learning_rate": 9.476613743330559e-08,
      "loss": 0.0812,
      "step": 10308
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.9257776999729375,
      "learning_rate": 9.457026833951555e-08,
      "loss": 0.2851,
      "step": 10309
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.907219466546589,
      "learning_rate": 9.437459797053683e-08,
      "loss": 0.2333,
      "step": 10310
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.527735735001754,
      "learning_rate": 9.417912634253485e-08,
      "loss": 0.3356,
      "step": 10311
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.916901685315782,
      "learning_rate": 9.398385347165805e-08,
      "loss": 0.228,
      "step": 10312
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.68923600987385,
      "learning_rate": 9.378877937403941e-08,
      "loss": 0.2451,
      "step": 10313
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.4519754227805475,
      "learning_rate": 9.359390406579488e-08,
      "loss": 0.2622,
      "step": 10314
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3903430810567334,
      "learning_rate": 9.339922756302356e-08,
      "loss": 0.2574,
      "step": 10315
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.814637893042017,
      "learning_rate": 9.320474988180977e-08,
      "loss": 0.2498,
      "step": 10316
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3994654993978957,
      "learning_rate": 9.30104710382193e-08,
      "loss": 0.3066,
      "step": 10317
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.469392235470414,
      "learning_rate": 9.281639104830292e-08,
      "loss": 0.3585,
      "step": 10318
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.839243284195748,
      "learning_rate": 9.262250992809446e-08,
      "loss": 0.4597,
      "step": 10319
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.346534878736967,
      "learning_rate": 9.242882769361139e-08,
      "loss": 0.2045,
      "step": 10320
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9401751680177304,
      "learning_rate": 9.223534436085508e-08,
      "loss": 0.2485,
      "step": 10321
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.400486411851755,
      "learning_rate": 9.204205994581022e-08,
      "loss": 0.2301,
      "step": 10322
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8256637568354086,
      "learning_rate": 9.184897446444463e-08,
      "loss": 0.0769,
      "step": 10323
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.2423471743475,
      "learning_rate": 9.165608793270997e-08,
      "loss": 0.2716,
      "step": 10324
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9441414503018297,
      "learning_rate": 9.146340036654183e-08,
      "loss": 0.1863,
      "step": 10325
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.836173486450012,
      "learning_rate": 9.127091178185943e-08,
      "loss": 0.277,
      "step": 10326
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.106455126356078,
      "learning_rate": 9.107862219456504e-08,
      "loss": 0.4801,
      "step": 10327
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.4508013413986407,
      "learning_rate": 9.088653162054428e-08,
      "loss": 0.2293,
      "step": 10328
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1038514635569663,
      "learning_rate": 9.069464007566725e-08,
      "loss": 0.0357,
      "step": 10329
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.820309847646241,
      "learning_rate": 9.050294757578653e-08,
      "loss": 0.1593,
      "step": 10330
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5789292610796206,
      "learning_rate": 9.031145413673948e-08,
      "loss": 0.2081,
      "step": 10331
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.943723889786638,
      "learning_rate": 9.012015977434591e-08,
      "loss": 0.1333,
      "step": 10332
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.3825057201619018,
      "learning_rate": 8.992906450440986e-08,
      "loss": 0.1009,
      "step": 10333
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.9966097232924405,
      "learning_rate": 8.973816834271842e-08,
      "loss": 0.4588,
      "step": 10334
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.348501714965561,
      "learning_rate": 8.95474713050426e-08,
      "loss": 0.2681,
      "step": 10335
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.35463524876942,
      "learning_rate": 8.9356973407137e-08,
      "loss": 0.2039,
      "step": 10336
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.572130487606889,
      "learning_rate": 8.916667466473932e-08,
      "loss": 0.2668,
      "step": 10337
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.4821851841571565,
      "learning_rate": 8.897657509357171e-08,
      "loss": 0.4302,
      "step": 10338
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.6714998944953203,
      "learning_rate": 8.878667470933855e-08,
      "loss": 0.3716,
      "step": 10339
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.50039932946925,
      "learning_rate": 8.859697352772867e-08,
      "loss": 0.25,
      "step": 10340
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6723870096498037,
      "learning_rate": 8.840747156441453e-08,
      "loss": 0.1068,
      "step": 10341
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.8782824242087948,
      "learning_rate": 8.821816883505168e-08,
      "loss": 0.353,
      "step": 10342
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.678733218653662,
      "learning_rate": 8.802906535527927e-08,
      "loss": 0.382,
      "step": 10343
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0509516206194807,
      "learning_rate": 8.784016114072064e-08,
      "loss": 0.1938,
      "step": 10344
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.432605095735349,
      "learning_rate": 8.765145620698107e-08,
      "loss": 0.4188,
      "step": 10345
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.099510479871563,
      "learning_rate": 8.746295056965143e-08,
      "loss": 0.3146,
      "step": 10346
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.6638412490023216,
      "learning_rate": 8.727464424430481e-08,
      "loss": 0.519,
      "step": 10347
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.090306723328519,
      "learning_rate": 8.708653724649824e-08,
      "loss": 0.1035,
      "step": 10348
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.4095788946920678,
      "learning_rate": 8.689862959177203e-08,
      "loss": 0.1288,
      "step": 10349
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.6272134580819624,
      "learning_rate": 8.67109212956499e-08,
      "loss": 0.2059,
      "step": 10350
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.5490025181717995,
      "learning_rate": 8.652341237363998e-08,
      "loss": 0.3643,
      "step": 10351
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.6941380275430538,
      "learning_rate": 8.633610284123323e-08,
      "loss": 0.3139,
      "step": 10352
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.463145850573943,
      "learning_rate": 8.614899271390393e-08,
      "loss": 0.4349,
      "step": 10353
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9947975607908797,
      "learning_rate": 8.596208200711054e-08,
      "loss": 0.3183,
      "step": 10354
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.384727782241831,
      "learning_rate": 8.577537073629405e-08,
      "loss": 0.1358,
      "step": 10355
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.708194044367993,
      "learning_rate": 8.558885891688074e-08,
      "loss": 0.3165,
      "step": 10356
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.337922955986124,
      "learning_rate": 8.540254656427827e-08,
      "loss": 0.3312,
      "step": 10357
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.699733384266253,
      "learning_rate": 8.52164336938796e-08,
      "loss": 0.3405,
      "step": 10358
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.442474475264452,
      "learning_rate": 8.503052032105968e-08,
      "loss": 0.1063,
      "step": 10359
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9664244496693075,
      "learning_rate": 8.484480646117871e-08,
      "loss": 0.3528,
      "step": 10360
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.00414370070018,
      "learning_rate": 8.465929212957885e-08,
      "loss": 0.3579,
      "step": 10361
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.226071428926727,
      "learning_rate": 8.447397734158647e-08,
      "loss": 0.2964,
      "step": 10362
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.7301924669768574,
      "learning_rate": 8.428886211251125e-08,
      "loss": 0.1502,
      "step": 10363
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.078955394409984,
      "learning_rate": 8.410394645764653e-08,
      "loss": 0.3773,
      "step": 10364
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.2789473985829622,
      "learning_rate": 8.391923039226951e-08,
      "loss": 0.2072,
      "step": 10365
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0003727013505204,
      "learning_rate": 8.373471393164023e-08,
      "loss": 0.1491,
      "step": 10366
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9720396485969176,
      "learning_rate": 8.35503970910026e-08,
      "loss": 0.2131,
      "step": 10367
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.916127324674014,
      "learning_rate": 8.336627988558388e-08,
      "loss": 0.2322,
      "step": 10368
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9790451216156804,
      "learning_rate": 8.318236233059496e-08,
      "loss": 0.3719,
      "step": 10369
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.587326086714646,
      "learning_rate": 8.299864444123035e-08,
      "loss": 0.3585,
      "step": 10370
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.88163267043713,
      "learning_rate": 8.281512623266763e-08,
      "loss": 0.2764,
      "step": 10371
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.090451255985619,
      "learning_rate": 8.263180772006857e-08,
      "loss": 0.4896,
      "step": 10372
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.238349107002602,
      "learning_rate": 8.244868891857799e-08,
      "loss": 0.3062,
      "step": 10373
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.949338663738667,
      "learning_rate": 8.226576984332351e-08,
      "loss": 0.2764,
      "step": 10374
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.373856930304224,
      "learning_rate": 8.208305050941806e-08,
      "loss": 0.3691,
      "step": 10375
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.369427117213654,
      "learning_rate": 8.190053093195676e-08,
      "loss": 0.308,
      "step": 10376
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.869493336538428,
      "learning_rate": 8.171821112601814e-08,
      "loss": 0.3224,
      "step": 10377
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.586595289497804,
      "learning_rate": 8.153609110666455e-08,
      "loss": 0.3108,
      "step": 10378
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3264643583265188,
      "learning_rate": 8.135417088894176e-08,
      "loss": 0.1133,
      "step": 10379
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.6577928940359192,
      "learning_rate": 8.117245048787998e-08,
      "loss": 0.2529,
      "step": 10380
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.6658970738386807,
      "learning_rate": 8.099092991849106e-08,
      "loss": 0.2795,
      "step": 10381
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.072044907746597,
      "learning_rate": 8.080960919577219e-08,
      "loss": 0.3988,
      "step": 10382
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.841887702122753,
      "learning_rate": 8.06284883347025e-08,
      "loss": 0.2507,
      "step": 10383
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5230104336523915,
      "learning_rate": 8.044756735024528e-08,
      "loss": 0.2241,
      "step": 10384
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.0015785884608657,
      "learning_rate": 8.026684625734776e-08,
      "loss": 0.1544,
      "step": 10385
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.081463835154924,
      "learning_rate": 8.008632507094022e-08,
      "loss": 0.3042,
      "step": 10386
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.205336019770011,
      "learning_rate": 7.990600380593655e-08,
      "loss": 0.154,
      "step": 10387
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.474580747778477,
      "learning_rate": 7.972588247723345e-08,
      "loss": 0.2222,
      "step": 10388
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.135955714628068,
      "learning_rate": 7.95459610997118e-08,
      "loss": 0.1847,
      "step": 10389
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.467107001722118,
      "learning_rate": 7.936623968823636e-08,
      "loss": 0.1998,
      "step": 10390
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.012168229239756,
      "learning_rate": 7.918671825765445e-08,
      "loss": 0.2493,
      "step": 10391
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.353575849776886,
      "learning_rate": 7.900739682279696e-08,
      "loss": 0.3806,
      "step": 10392
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.6506566981796102,
      "learning_rate": 7.882827539847925e-08,
      "loss": 0.0611,
      "step": 10393
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9286841568092714,
      "learning_rate": 7.86493539994987e-08,
      "loss": 0.2335,
      "step": 10394
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.034253814603348,
      "learning_rate": 7.84706326406373e-08,
      "loss": 0.257,
      "step": 10395
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.762843020559855,
      "learning_rate": 7.829211133666026e-08,
      "loss": 0.416,
      "step": 10396
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.912254915297415,
      "learning_rate": 7.811379010231601e-08,
      "loss": 0.3333,
      "step": 10397
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.569709316814724,
      "learning_rate": 7.793566895233667e-08,
      "loss": 0.3731,
      "step": 10398
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.402600535104165,
      "learning_rate": 7.775774790143714e-08,
      "loss": 0.0948,
      "step": 10399
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.500806038150744,
      "learning_rate": 7.758002696431732e-08,
      "loss": 0.2817,
      "step": 10400
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.633926696362923,
      "learning_rate": 7.740250615565909e-08,
      "loss": 0.3513,
      "step": 10401
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.786444732999832,
      "learning_rate": 7.722518549012819e-08,
      "loss": 0.2776,
      "step": 10402
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.562305956681777,
      "learning_rate": 7.704806498237489e-08,
      "loss": 0.3506,
      "step": 10403
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.4909357960564837,
      "learning_rate": 7.687114464703077e-08,
      "loss": 0.1437,
      "step": 10404
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.151853026280685,
      "learning_rate": 7.669442449871278e-08,
      "loss": 0.2,
      "step": 10405
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.139809587957418,
      "learning_rate": 7.651790455202091e-08,
      "loss": 0.3382,
      "step": 10406
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.5009304941692334,
      "learning_rate": 7.634158482153764e-08,
      "loss": 0.1395,
      "step": 10407
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.497002233329385,
      "learning_rate": 7.61654653218305e-08,
      "loss": 0.3456,
      "step": 10408
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.721129036007975,
      "learning_rate": 7.598954606744896e-08,
      "loss": 0.3137,
      "step": 10409
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.313607478195961,
      "learning_rate": 7.581382707292694e-08,
      "loss": 0.3115,
      "step": 10410
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.3452050982045325,
      "learning_rate": 7.563830835278146e-08,
      "loss": 0.2632,
      "step": 10411
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.514439964659453,
      "learning_rate": 7.546298992151257e-08,
      "loss": 0.3267,
      "step": 10412
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.654107416705928,
      "learning_rate": 7.528787179360509e-08,
      "loss": 0.1339,
      "step": 10413
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.9746129725566184,
      "learning_rate": 7.511295398352547e-08,
      "loss": 0.2402,
      "step": 10414
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.305616792374337,
      "learning_rate": 7.493823650572523e-08,
      "loss": 0.2397,
      "step": 10415
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.221313411306523,
      "learning_rate": 7.476371937463839e-08,
      "loss": 0.3255,
      "step": 10416
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.3728783341952076,
      "learning_rate": 7.458940260468256e-08,
      "loss": 0.2823,
      "step": 10417
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.595395347947323,
      "learning_rate": 7.441528621025984e-08,
      "loss": 0.2394,
      "step": 10418
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.684532235095166,
      "learning_rate": 7.424137020575345e-08,
      "loss": 0.2311,
      "step": 10419
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.637175827185012,
      "learning_rate": 7.406765460553245e-08,
      "loss": 0.4976,
      "step": 10420
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.985898483134038,
      "learning_rate": 7.389413942394813e-08,
      "loss": 0.2197,
      "step": 10421
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.235587855961309,
      "learning_rate": 7.372082467533542e-08,
      "loss": 0.2906,
      "step": 10422
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.3568877371902435,
      "learning_rate": 7.354771037401315e-08,
      "loss": 0.4087,
      "step": 10423
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.498864420583875,
      "learning_rate": 7.33747965342821e-08,
      "loss": 0.2936,
      "step": 10424
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.658675718382019,
      "learning_rate": 7.32020831704286e-08,
      "loss": 0.2922,
      "step": 10425
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.131952507153375,
      "learning_rate": 7.302957029672125e-08,
      "loss": 0.3701,
      "step": 10426
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.8970733113599865,
      "learning_rate": 7.285725792741144e-08,
      "loss": 0.3321,
      "step": 10427
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.174673460253052,
      "learning_rate": 7.26851460767361e-08,
      "loss": 0.3428,
      "step": 10428
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.651416363343906,
      "learning_rate": 7.251323475891276e-08,
      "loss": 0.3793,
      "step": 10429
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.908802207510499,
      "learning_rate": 7.234152398814504e-08,
      "loss": 0.1544,
      "step": 10430
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.312163870028915,
      "learning_rate": 7.217001377861854e-08,
      "loss": 0.1906,
      "step": 10431
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.4989103148751246,
      "learning_rate": 7.199870414450221e-08,
      "loss": 0.237,
      "step": 10432
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7112527409025513,
      "learning_rate": 7.182759509994946e-08,
      "loss": 0.1167,
      "step": 10433
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.687844929949333,
      "learning_rate": 7.165668665909564e-08,
      "loss": 0.1877,
      "step": 10434
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.443724818261422,
      "learning_rate": 7.148597883606113e-08,
      "loss": 0.3626,
      "step": 10435
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.846303839403572,
      "learning_rate": 7.131547164494878e-08,
      "loss": 0.2043,
      "step": 10436
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.720083980212511,
      "learning_rate": 7.114516509984459e-08,
      "loss": 0.3133,
      "step": 10437
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.037599836516663,
      "learning_rate": 7.097505921481923e-08,
      "loss": 0.2787,
      "step": 10438
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.978091574104936,
      "learning_rate": 7.080515400392534e-08,
      "loss": 0.2472,
      "step": 10439
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.33396932596189,
      "learning_rate": 7.063544948120005e-08,
      "loss": 0.1973,
      "step": 10440
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.472994759206455,
      "learning_rate": 7.046594566066323e-08,
      "loss": 0.3357,
      "step": 10441
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.275601092542961,
      "learning_rate": 7.029664255631869e-08,
      "loss": 0.1333,
      "step": 10442
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.011265444353192,
      "learning_rate": 7.012754018215357e-08,
      "loss": 0.3694,
      "step": 10443
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.510221113538512,
      "learning_rate": 6.99586385521378e-08,
      "loss": 0.4042,
      "step": 10444
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0084969207585577,
      "learning_rate": 6.978993768022552e-08,
      "loss": 0.0353,
      "step": 10445
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.67512894952402,
      "learning_rate": 6.962143758035416e-08,
      "loss": 0.2721,
      "step": 10446
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.9475026610577633,
      "learning_rate": 6.945313826644346e-08,
      "loss": 0.2695,
      "step": 10447
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.624212180644679,
      "learning_rate": 6.928503975239892e-08,
      "loss": 0.1677,
      "step": 10448
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.480512495463591,
      "learning_rate": 6.911714205210668e-08,
      "loss": 0.1193,
      "step": 10449
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.001145549345608,
      "learning_rate": 6.894944517943841e-08,
      "loss": 0.1357,
      "step": 10450
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.4758371511698565,
      "learning_rate": 6.878194914824832e-08,
      "loss": 0.2665,
      "step": 10451
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.349629707579302,
      "learning_rate": 6.861465397237366e-08,
      "loss": 0.1475,
      "step": 10452
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.2998718086473335,
      "learning_rate": 6.844755966563643e-08,
      "loss": 0.4958,
      "step": 10453
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.236250235000793,
      "learning_rate": 6.828066624184005e-08,
      "loss": 0.4225,
      "step": 10454
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.71536557519446,
      "learning_rate": 6.811397371477264e-08,
      "loss": 0.2351,
      "step": 10455
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.235240336463258,
      "learning_rate": 6.794748209820678e-08,
      "loss": 0.3771,
      "step": 10456
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.897831659588125,
      "learning_rate": 6.778119140589539e-08,
      "loss": 0.2959,
      "step": 10457
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.4421289841312848,
      "learning_rate": 6.761510165157803e-08,
      "loss": 0.2478,
      "step": 10458
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.173485916800353,
      "learning_rate": 6.74492128489751e-08,
      "loss": 0.2609,
      "step": 10459
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.161651887052988,
      "learning_rate": 6.728352501179203e-08,
      "loss": 0.3215,
      "step": 10460
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.8407644323357792,
      "learning_rate": 6.711803815371787e-08,
      "loss": 0.3051,
      "step": 10461
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.933009381055094,
      "learning_rate": 6.695275228842307e-08,
      "loss": 0.1391,
      "step": 10462
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.807561112226115,
      "learning_rate": 6.678766742956367e-08,
      "loss": 0.3167,
      "step": 10463
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.80117142334302,
      "learning_rate": 6.662278359077734e-08,
      "loss": 0.1949,
      "step": 10464
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6338091359229474,
      "learning_rate": 6.645810078568626e-08,
      "loss": 0.2195,
      "step": 10465
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.8517652237990436,
      "learning_rate": 6.629361902789649e-08,
      "loss": 0.1388,
      "step": 10466
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.344664613146834,
      "learning_rate": 6.612933833099577e-08,
      "loss": 0.3539,
      "step": 10467
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6700249021709204,
      "learning_rate": 6.596525870855686e-08,
      "loss": 0.1294,
      "step": 10468
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.894112995327701,
      "learning_rate": 6.580138017413418e-08,
      "loss": 0.3115,
      "step": 10469
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.2526653214232395,
      "learning_rate": 6.563770274126746e-08,
      "loss": 0.3059,
      "step": 10470
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.033354158129085,
      "learning_rate": 6.547422642347922e-08,
      "loss": 0.2254,
      "step": 10471
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.950677515232079,
      "learning_rate": 6.531095123427395e-08,
      "loss": 0.3478,
      "step": 10472
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.05078813616934,
      "learning_rate": 6.514787718714138e-08,
      "loss": 0.2662,
      "step": 10473
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.953104492421977,
      "learning_rate": 6.49850042955541e-08,
      "loss": 0.1768,
      "step": 10474
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.9750503876559073,
      "learning_rate": 6.482233257296688e-08,
      "loss": 0.3083,
      "step": 10475
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.7439103952670103,
      "learning_rate": 6.465986203282037e-08,
      "loss": 0.2517,
      "step": 10476
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.783433146805806,
      "learning_rate": 6.449759268853551e-08,
      "loss": 0.1594,
      "step": 10477
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.38994545556534,
      "learning_rate": 6.433552455351932e-08,
      "loss": 0.214,
      "step": 10478
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.114828003247239,
      "learning_rate": 6.417365764116057e-08,
      "loss": 0.3469,
      "step": 10479
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.629739143517505,
      "learning_rate": 6.40119919648316e-08,
      "loss": 0.2399,
      "step": 10480
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.237552979561107,
      "learning_rate": 6.38505275378895e-08,
      "loss": 0.3787,
      "step": 10481
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.844448503784464,
      "learning_rate": 6.368926437367252e-08,
      "loss": 0.3762,
      "step": 10482
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.899194024439024,
      "learning_rate": 6.35282024855044e-08,
      "loss": 0.3348,
      "step": 10483
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.85901164718608,
      "learning_rate": 6.336734188669036e-08,
      "loss": 0.2944,
      "step": 10484
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.832604784092317,
      "learning_rate": 6.32066825905206e-08,
      "loss": 0.2632,
      "step": 10485
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.461573199673201,
      "learning_rate": 6.304622461026782e-08,
      "loss": 0.3759,
      "step": 10486
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.4592708256149445,
      "learning_rate": 6.288596795918811e-08,
      "loss": 0.241,
      "step": 10487
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.6804245149135584,
      "learning_rate": 6.272591265052113e-08,
      "loss": 0.1451,
      "step": 10488
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.8734153158132685,
      "learning_rate": 6.256605869749021e-08,
      "loss": 0.2514,
      "step": 10489
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.61105771857965,
      "learning_rate": 6.24064061133009e-08,
      "loss": 0.4706,
      "step": 10490
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.323051898028678,
      "learning_rate": 6.22469549111443e-08,
      "loss": 0.21,
      "step": 10491
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.382790434503297,
      "learning_rate": 6.208770510419182e-08,
      "loss": 0.2948,
      "step": 10492
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.508933680848752,
      "learning_rate": 6.192865670560099e-08,
      "loss": 0.309,
      "step": 10493
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.109434026029743,
      "learning_rate": 6.176980972851132e-08,
      "loss": 0.0817,
      "step": 10494
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.918484673761477,
      "learning_rate": 6.16111641860459e-08,
      "loss": 0.1931,
      "step": 10495
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.3575475808066,
      "learning_rate": 6.145272009131176e-08,
      "loss": 0.2894,
      "step": 10496
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6328946234887605,
      "learning_rate": 6.129447745739758e-08,
      "loss": 0.2458,
      "step": 10497
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.722955749190235,
      "learning_rate": 6.113643629737792e-08,
      "loss": 0.3297,
      "step": 10498
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.878192183227869,
      "learning_rate": 6.097859662430871e-08,
      "loss": 0.3082,
      "step": 10499
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.974381859647219,
      "learning_rate": 6.082095845122954e-08,
      "loss": 0.3499,
      "step": 10500
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.687871818406166,
      "learning_rate": 6.06635217911647e-08,
      "loss": 0.356,
      "step": 10501
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.014461377873275,
      "learning_rate": 6.050628665711988e-08,
      "loss": 0.3201,
      "step": 10502
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.181941621690425,
      "learning_rate": 6.034925306208583e-08,
      "loss": 0.2048,
      "step": 10503
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.790493430426852,
      "learning_rate": 6.019242101903522e-08,
      "loss": 0.2676,
      "step": 10504
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.947987793763536,
      "learning_rate": 6.00357905409249e-08,
      "loss": 0.3139,
      "step": 10505
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.9906927156327487,
      "learning_rate": 5.987936164069563e-08,
      "loss": 0.2316,
      "step": 10506
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.972931714245264,
      "learning_rate": 5.972313433126986e-08,
      "loss": 0.2568,
      "step": 10507
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.696855058819173,
      "learning_rate": 5.956710862555476e-08,
      "loss": 0.4192,
      "step": 10508
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.772058391536253,
      "learning_rate": 5.941128453644029e-08,
      "loss": 0.4604,
      "step": 10509
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.74896409274193,
      "learning_rate": 5.925566207679978e-08,
      "loss": 0.2449,
      "step": 10510
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.296392014148789,
      "learning_rate": 5.910024125949071e-08,
      "loss": 0.2223,
      "step": 10511
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.668049103398622,
      "learning_rate": 5.894502209735198e-08,
      "loss": 0.1173,
      "step": 10512
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6088373429790086,
      "learning_rate": 5.879000460320805e-08,
      "loss": 0.2252,
      "step": 10513
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.605284281050659,
      "learning_rate": 5.8635188789865075e-08,
      "loss": 0.4422,
      "step": 10514
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.503326179565996,
      "learning_rate": 5.848057467011364e-08,
      "loss": 0.5655,
      "step": 10515
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.113191436389464,
      "learning_rate": 5.832616225672716e-08,
      "loss": 0.3102,
      "step": 10516
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.3267525914338556,
      "learning_rate": 5.817195156246181e-08,
      "loss": 0.2148,
      "step": 10517
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.265030962598339,
      "learning_rate": 5.8017942600058494e-08,
      "loss": 0.2433,
      "step": 10518
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6247386012479637,
      "learning_rate": 5.7864135382240394e-08,
      "loss": 0.2704,
      "step": 10519
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.893618473385234,
      "learning_rate": 5.7710529921714e-08,
      "loss": 0.1615,
      "step": 10520
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6291920364537846,
      "learning_rate": 5.755712623116999e-08,
      "loss": 0.2268,
      "step": 10521
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.2407402014165343,
      "learning_rate": 5.740392432328157e-08,
      "loss": 0.2257,
      "step": 10522
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.825406141932143,
      "learning_rate": 5.725092421070527e-08,
      "loss": 0.4082,
      "step": 10523
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.1162379172392916,
      "learning_rate": 5.709812590608183e-08,
      "loss": 0.2692,
      "step": 10524
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.725393952725771,
      "learning_rate": 5.6945529422033906e-08,
      "loss": 0.3469,
      "step": 10525
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.820115731831641,
      "learning_rate": 5.679313477116893e-08,
      "loss": 0.0924,
      "step": 10526
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.423598021228167,
      "learning_rate": 5.664094196607683e-08,
      "loss": 0.2338,
      "step": 10527
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.802015876244494,
      "learning_rate": 5.6488951019330885e-08,
      "loss": 0.1749,
      "step": 10528
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.417405061568057,
      "learning_rate": 5.633716194348798e-08,
      "loss": 0.2599,
      "step": 10529
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.585003791293194,
      "learning_rate": 5.6185574751087825e-08,
      "loss": 0.3031,
      "step": 10530
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6599391492600004,
      "learning_rate": 5.603418945465427e-08,
      "loss": 0.1371,
      "step": 10531
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.264008516856804,
      "learning_rate": 5.5883006066694e-08,
      "loss": 0.2986,
      "step": 10532
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.067304311389626,
      "learning_rate": 5.573202459969673e-08,
      "loss": 0.2273,
      "step": 10533
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.296616784821721,
      "learning_rate": 5.558124506613638e-08,
      "loss": 0.4924,
      "step": 10534
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.564581630566145,
      "learning_rate": 5.5430667478468535e-08,
      "loss": 0.1311,
      "step": 10535
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3664392900434272,
      "learning_rate": 5.528029184913436e-08,
      "loss": 0.0856,
      "step": 10536
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.652577420863712,
      "learning_rate": 5.5130118190556694e-08,
      "loss": 0.2731,
      "step": 10537
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.4759691059984528,
      "learning_rate": 5.4980146515142e-08,
      "loss": 0.2791,
      "step": 10538
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.196834636137251,
      "learning_rate": 5.483037683528036e-08,
      "loss": 0.2704,
      "step": 10539
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.404315424733515,
      "learning_rate": 5.468080916334467e-08,
      "loss": 0.3181,
      "step": 10540
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3958234471018063,
      "learning_rate": 5.453144351169198e-08,
      "loss": 0.1318,
      "step": 10541
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.6722681620957625,
      "learning_rate": 5.4382279892661874e-08,
      "loss": 0.4085,
      "step": 10542
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.861947797355405,
      "learning_rate": 5.423331831857781e-08,
      "loss": 0.224,
      "step": 10543
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.361017882973993,
      "learning_rate": 5.408455880174579e-08,
      "loss": 0.2497,
      "step": 10544
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.214505848706607,
      "learning_rate": 5.393600135445598e-08,
      "loss": 0.2341,
      "step": 10545
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.958215730023193,
      "learning_rate": 5.3787645988981343e-08,
      "loss": 0.2934,
      "step": 10546
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.797136456973325,
      "learning_rate": 5.363949271757818e-08,
      "loss": 0.2652,
      "step": 10547
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.293393095281925,
      "learning_rate": 5.349154155248643e-08,
      "loss": 0.3414,
      "step": 10548
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.0979227756753485,
      "learning_rate": 5.3343792505928807e-08,
      "loss": 0.3549,
      "step": 10549
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.479959572603356,
      "learning_rate": 5.319624559011138e-08,
      "loss": 0.3526,
      "step": 10550
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.11720396936476,
      "learning_rate": 5.304890081722469e-08,
      "loss": 0.2735,
      "step": 10551
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.01339374342516,
      "learning_rate": 5.290175819944066e-08,
      "loss": 0.3138,
      "step": 10552
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.524290909887453,
      "learning_rate": 5.275481774891622e-08,
      "loss": 0.2116,
      "step": 10553
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.2508291969810195,
      "learning_rate": 5.2608079477790565e-08,
      "loss": 0.2673,
      "step": 10554
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.9902904414180482,
      "learning_rate": 5.2461543398185934e-08,
      "loss": 0.3393,
      "step": 10555
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.409850431150604,
      "learning_rate": 5.231520952220931e-08,
      "loss": 0.3896,
      "step": 10556
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.12967452281364,
      "learning_rate": 5.2169077861949916e-08,
      "loss": 0.2736,
      "step": 10557
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.547124956510072,
      "learning_rate": 5.202314842948003e-08,
      "loss": 0.3533,
      "step": 10558
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.717078109437378,
      "learning_rate": 5.187742123685585e-08,
      "loss": 0.1742,
      "step": 10559
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.082622863580739,
      "learning_rate": 5.1731896296116355e-08,
      "loss": 0.3076,
      "step": 10560
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.627955712332072,
      "learning_rate": 5.158657361928471e-08,
      "loss": 0.2914,
      "step": 10561
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.277252768466115,
      "learning_rate": 5.1441453218366575e-08,
      "loss": 0.2546,
      "step": 10562
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.95213990761212,
      "learning_rate": 5.129653510535071e-08,
      "loss": 0.2907,
      "step": 10563
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.34426049546176,
      "learning_rate": 5.115181929220975e-08,
      "loss": 0.2955,
      "step": 10564
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.7484223817740574,
      "learning_rate": 5.100730579089941e-08,
      "loss": 0.2823,
      "step": 10565
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.229328240329951,
      "learning_rate": 5.0862994613358754e-08,
      "loss": 0.2731,
      "step": 10566
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.415040996360127,
      "learning_rate": 5.0718885771510185e-08,
      "loss": 0.232,
      "step": 10567
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.7488946306249495,
      "learning_rate": 5.057497927725891e-08,
      "loss": 0.3495,
      "step": 10568
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.9511965242436515,
      "learning_rate": 5.043127514249374e-08,
      "loss": 0.3267,
      "step": 10569
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.067546634054275,
      "learning_rate": 5.0287773379087416e-08,
      "loss": 0.2488,
      "step": 10570
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.753742182843639,
      "learning_rate": 5.014447399889488e-08,
      "loss": 0.3659,
      "step": 10571
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.827258544722015,
      "learning_rate": 5.000137701375474e-08,
      "loss": 0.0993,
      "step": 10572
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.004160234130625,
      "learning_rate": 4.985848243548919e-08,
      "loss": 0.0539,
      "step": 10573
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.8969948706933915,
      "learning_rate": 4.9715790275903505e-08,
      "loss": 0.319,
      "step": 10574
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.983862246930282,
      "learning_rate": 4.957330054678605e-08,
      "loss": 0.3384,
      "step": 10575
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.27686279627614,
      "learning_rate": 4.9431013259908786e-08,
      "loss": 0.2126,
      "step": 10576
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.228996660875394,
      "learning_rate": 4.928892842702676e-08,
      "loss": 0.377,
      "step": 10577
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.325037380111799,
      "learning_rate": 4.914704605987808e-08,
      "loss": 0.2479,
      "step": 10578
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.130890245537766,
      "learning_rate": 4.900536617018475e-08,
      "loss": 0.2417,
      "step": 10579
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.331853029664971,
      "learning_rate": 4.886388876965131e-08,
      "loss": 0.3534,
      "step": 10580
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4953421211128406,
      "learning_rate": 4.8722613869966184e-08,
      "loss": 0.1825,
      "step": 10581
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.481523948107679,
      "learning_rate": 4.8581541482800866e-08,
      "loss": 0.4955,
      "step": 10582
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.205196814690083,
      "learning_rate": 4.844067161980992e-08,
      "loss": 0.546,
      "step": 10583
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.891839587961164,
      "learning_rate": 4.830000429263099e-08,
      "loss": 0.295,
      "step": 10584
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.963875315458366,
      "learning_rate": 4.8159539512886157e-08,
      "loss": 0.3129,
      "step": 10585
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.9023903946788314,
      "learning_rate": 4.80192772921792e-08,
      "loss": 0.2499,
      "step": 10586
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.919403588665067,
      "learning_rate": 4.787921764209807e-08,
      "loss": 0.2662,
      "step": 10587
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.746024210833015,
      "learning_rate": 4.7739360574214074e-08,
      "loss": 0.1505,
      "step": 10588
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.8688609264145053,
      "learning_rate": 4.759970610008102e-08,
      "loss": 0.1466,
      "step": 10589
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.635479192368004,
      "learning_rate": 4.746025423123718e-08,
      "loss": 0.2291,
      "step": 10590
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.233708302419185,
      "learning_rate": 4.7321004979202776e-08,
      "loss": 0.1558,
      "step": 10591
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.487144051585976,
      "learning_rate": 4.718195835548223e-08,
      "loss": 0.0693,
      "step": 10592
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.461490233564692,
      "learning_rate": 4.704311437156245e-08,
      "loss": 0.4304,
      "step": 10593
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.5985333390273695,
      "learning_rate": 4.690447303891454e-08,
      "loss": 0.2992,
      "step": 10594
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.676206069088071,
      "learning_rate": 4.676603436899213e-08,
      "loss": 0.5404,
      "step": 10595
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.8489815195872366,
      "learning_rate": 4.662779837323245e-08,
      "loss": 0.2818,
      "step": 10596
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.335220409557764,
      "learning_rate": 4.648976506305608e-08,
      "loss": 0.2012,
      "step": 10597
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.2755138526129204,
      "learning_rate": 4.635193444986613e-08,
      "loss": 0.1994,
      "step": 10598
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.747874108144957,
      "learning_rate": 4.6214306545049606e-08,
      "loss": 0.5056,
      "step": 10599
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.89210595595748,
      "learning_rate": 4.607688135997712e-08,
      "loss": 0.1426,
      "step": 10600
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.523823227488751,
      "learning_rate": 4.593965890600155e-08,
      "loss": 0.2453,
      "step": 10601
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.094081325229073,
      "learning_rate": 4.580263919445993e-08,
      "loss": 0.1783,
      "step": 10602
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.195124270463726,
      "learning_rate": 4.56658222366721e-08,
      "loss": 0.3473,
      "step": 10603
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.857986739334216,
      "learning_rate": 4.552920804394068e-08,
      "loss": 0.3049,
      "step": 10604
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.0110388883180255,
      "learning_rate": 4.539279662755275e-08,
      "loss": 0.3268,
      "step": 10605
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.7067303726983165,
      "learning_rate": 4.525658799877763e-08,
      "loss": 0.2669,
      "step": 10606
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0398842849520666,
      "learning_rate": 4.5120582168868e-08,
      "loss": 0.0189,
      "step": 10607
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1629067578627224,
      "learning_rate": 4.498477914906069e-08,
      "loss": 0.0669,
      "step": 10608
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.127450188402349,
      "learning_rate": 4.4849178950574226e-08,
      "loss": 0.3931,
      "step": 10609
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.169922382424996,
      "learning_rate": 4.471378158461187e-08,
      "loss": 0.3936,
      "step": 10610
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.793439127693295,
      "learning_rate": 4.45785870623594e-08,
      "loss": 0.4808,
      "step": 10611
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.793700217840455,
      "learning_rate": 4.4443595394985384e-08,
      "loss": 0.4064,
      "step": 10612
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.036464089761573,
      "learning_rate": 4.4308806593643115e-08,
      "loss": 0.2868,
      "step": 10613
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.876857956203779,
      "learning_rate": 4.417422066946703e-08,
      "loss": 0.2924,
      "step": 10614
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.085124987951906,
      "learning_rate": 4.403983763357683e-08,
      "loss": 0.2821,
      "step": 10615
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.218895747114248,
      "learning_rate": 4.3905657497074485e-08,
      "loss": 0.1993,
      "step": 10616
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.929853192139104,
      "learning_rate": 4.377168027104473e-08,
      "loss": 0.1535,
      "step": 10617
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7691443691066688,
      "learning_rate": 4.3637905966557035e-08,
      "loss": 0.0457,
      "step": 10618
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.185973821702417,
      "learning_rate": 4.3504334594662304e-08,
      "loss": 0.298,
      "step": 10619
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.3207586259214725,
      "learning_rate": 4.3370966166396136e-08,
      "loss": 0.2755,
      "step": 10620
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.734929801900172,
      "learning_rate": 4.323780069277639e-08,
      "loss": 0.4265,
      "step": 10621
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.7479686122112175,
      "learning_rate": 4.3104838184804546e-08,
      "loss": 0.3024,
      "step": 10622
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.8155128980678317,
      "learning_rate": 4.29720786534657e-08,
      "loss": 0.3214,
      "step": 10623
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.7047117164553143,
      "learning_rate": 4.283952210972747e-08,
      "loss": 0.2235,
      "step": 10624
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.23992606801632,
      "learning_rate": 4.27071685645411e-08,
      "loss": 0.241,
      "step": 10625
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.467982562453929,
      "learning_rate": 4.2575018028841176e-08,
      "loss": 0.1795,
      "step": 10626
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.3639373610488548,
      "learning_rate": 4.2443070513544794e-08,
      "loss": 0.2491,
      "step": 10627
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.096923332217472,
      "learning_rate": 4.2311326029553514e-08,
      "loss": 0.3501,
      "step": 10628
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1771825686339357,
      "learning_rate": 4.217978458775085e-08,
      "loss": 0.1207,
      "step": 10629
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.0660073047013676,
      "learning_rate": 4.20484461990045e-08,
      "loss": 0.1215,
      "step": 10630
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.968292618685185,
      "learning_rate": 4.1917310874164673e-08,
      "loss": 0.2751,
      "step": 10631
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.2031737103721944,
      "learning_rate": 4.1786378624065205e-08,
      "loss": 0.4576,
      "step": 10632
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.976226281154215,
      "learning_rate": 4.165564945952383e-08,
      "loss": 0.3793,
      "step": 10633
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.131617461443884,
      "learning_rate": 4.1525123391339405e-08,
      "loss": 0.3718,
      "step": 10634
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.062236883774021,
      "learning_rate": 4.139480043029609e-08,
      "loss": 0.3816,
      "step": 10635
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.110062650245295,
      "learning_rate": 4.126468058716082e-08,
      "loss": 0.2418,
      "step": 10636
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.73741237597236,
      "learning_rate": 4.1134763872682784e-08,
      "loss": 0.2741,
      "step": 10637
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.8323259700593493,
      "learning_rate": 4.100505029759561e-08,
      "loss": 0.2789,
      "step": 10638
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.84124635735788,
      "learning_rate": 4.0875539872615166e-08,
      "loss": 0.2983,
      "step": 10639
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1025030259771946,
      "learning_rate": 4.074623260844124e-08,
      "loss": 0.0913,
      "step": 10640
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.467941180139237,
      "learning_rate": 4.061712851575667e-08,
      "loss": 0.3277,
      "step": 10641
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.419289096028134,
      "learning_rate": 4.048822760522681e-08,
      "loss": 0.4071,
      "step": 10642
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.6300343643750597,
      "learning_rate": 4.035952988750202e-08,
      "loss": 0.3026,
      "step": 10643
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.074475779077144,
      "learning_rate": 4.023103537321327e-08,
      "loss": 0.1548,
      "step": 10644
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.900519368234698,
      "learning_rate": 4.010274407297704e-08,
      "loss": 0.2168,
      "step": 10645
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.536820686708727,
      "learning_rate": 3.997465599739181e-08,
      "loss": 0.4181,
      "step": 10646
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.547711207193165,
      "learning_rate": 3.984677115703939e-08,
      "loss": 0.3796,
      "step": 10647
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4497042161882043,
      "learning_rate": 3.9719089562485766e-08,
      "loss": 0.099,
      "step": 10648
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.874608791772189,
      "learning_rate": 3.959161122427835e-08,
      "loss": 0.2684,
      "step": 10649
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.047514532837928,
      "learning_rate": 3.946433615294926e-08,
      "loss": 0.2208,
      "step": 10650
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.812406675096705,
      "learning_rate": 3.933726435901397e-08,
      "loss": 0.1473,
      "step": 10651
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.60511633464117,
      "learning_rate": 3.921039585296937e-08,
      "loss": 0.4268,
      "step": 10652
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.3542423498177687,
      "learning_rate": 3.908373064529763e-08,
      "loss": 0.216,
      "step": 10653
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.9713325202289655,
      "learning_rate": 3.8957268746462603e-08,
      "loss": 0.1409,
      "step": 10654
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.384461763682886,
      "learning_rate": 3.883101016691204e-08,
      "loss": 0.1974,
      "step": 10655
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.009288925638122,
      "learning_rate": 3.870495491707732e-08,
      "loss": 0.4984,
      "step": 10656
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.865123133352467,
      "learning_rate": 3.8579103007371775e-08,
      "loss": 0.3283,
      "step": 10657
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.2547432218950645,
      "learning_rate": 3.845345444819348e-08,
      "loss": 0.2326,
      "step": 10658
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8234691106524763,
      "learning_rate": 3.8328009249922197e-08,
      "loss": 0.1833,
      "step": 10659
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.232506098173512,
      "learning_rate": 3.820276742292184e-08,
      "loss": 0.1987,
      "step": 10660
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.044952634930463,
      "learning_rate": 3.807772897753942e-08,
      "loss": 0.2478,
      "step": 10661
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.773014371963446,
      "learning_rate": 3.7952893924104725e-08,
      "loss": 0.3013,
      "step": 10662
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.294318179897413,
      "learning_rate": 3.782826227293146e-08,
      "loss": 0.3409,
      "step": 10663
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.018295228713566,
      "learning_rate": 3.7703834034315546e-08,
      "loss": 0.2041,
      "step": 10664
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.514854974882,
      "learning_rate": 3.757960921853654e-08,
      "loss": 0.3613,
      "step": 10665
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.523583059422809,
      "learning_rate": 3.745558783585845e-08,
      "loss": 0.4096,
      "step": 10666
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.30357083061787,
      "learning_rate": 3.733176989652587e-08,
      "loss": 0.2107,
      "step": 10667
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.4828310289211086,
      "learning_rate": 3.720815541076894e-08,
      "loss": 0.1269,
      "step": 10668
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.264406903087973,
      "learning_rate": 3.70847443887995e-08,
      "loss": 0.3202,
      "step": 10669
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.787048859895033,
      "learning_rate": 3.696153684081355e-08,
      "loss": 0.3232,
      "step": 10670
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.441260661041055,
      "learning_rate": 3.683853277699018e-08,
      "loss": 0.3657,
      "step": 10671
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.85231868896328,
      "learning_rate": 3.671573220749042e-08,
      "loss": 0.4193,
      "step": 10672
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.931970868111424,
      "learning_rate": 3.65931351424606e-08,
      "loss": 0.3158,
      "step": 10673
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.305512317458081,
      "learning_rate": 3.6470741592027904e-08,
      "loss": 0.4119,
      "step": 10674
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.86894987017192,
      "learning_rate": 3.634855156630479e-08,
      "loss": 0.5047,
      "step": 10675
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.539698755739962,
      "learning_rate": 3.6226565075385966e-08,
      "loss": 0.2751,
      "step": 10676
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.5676278063616635,
      "learning_rate": 3.610478212934865e-08,
      "loss": 0.4136,
      "step": 10677
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.106227761723862,
      "learning_rate": 3.598320273825478e-08,
      "loss": 0.3071,
      "step": 10678
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.976578161488868,
      "learning_rate": 3.586182691214829e-08,
      "loss": 0.3343,
      "step": 10679
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.004288777311646,
      "learning_rate": 3.574065466105614e-08,
      "loss": 0.299,
      "step": 10680
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7655690819095398,
      "learning_rate": 3.561968599499005e-08,
      "loss": 0.2779,
      "step": 10681
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.256127364543058,
      "learning_rate": 3.5498920923943135e-08,
      "loss": 0.3655,
      "step": 10682
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.426295305920997,
      "learning_rate": 3.537835945789242e-08,
      "loss": 0.2849,
      "step": 10683
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.650278918911423,
      "learning_rate": 3.525800160679854e-08,
      "loss": 0.1353,
      "step": 10684
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7050998510512927,
      "learning_rate": 3.5137847380604105e-08,
      "loss": 0.2052,
      "step": 10685
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.028710346131098,
      "learning_rate": 3.501789678923673e-08,
      "loss": 0.162,
      "step": 10686
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.481082098584476,
      "learning_rate": 3.4898149842605146e-08,
      "loss": 0.2159,
      "step": 10687
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.0290052602111124,
      "learning_rate": 3.477860655060283e-08,
      "loss": 0.1588,
      "step": 10688
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.988610705892544,
      "learning_rate": 3.465926692310578e-08,
      "loss": 0.2607,
      "step": 10689
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.813091340412305,
      "learning_rate": 3.454013096997305e-08,
      "loss": 0.3274,
      "step": 10690
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.9860642411174654,
      "learning_rate": 3.44211987010476e-08,
      "loss": 0.187,
      "step": 10691
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.005505272887017,
      "learning_rate": 3.4302470126154066e-08,
      "loss": 0.243,
      "step": 10692
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.291063528559399,
      "learning_rate": 3.41839452551021e-08,
      "loss": 0.0702,
      "step": 10693
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.292918172976505,
      "learning_rate": 3.406562409768333e-08,
      "loss": 0.2321,
      "step": 10694
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.644066253139931,
      "learning_rate": 3.394750666367241e-08,
      "loss": 0.1755,
      "step": 10695
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.6728745019269855,
      "learning_rate": 3.3829592962828785e-08,
      "loss": 0.3503,
      "step": 10696
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.85482718028515,
      "learning_rate": 3.37118830048927e-08,
      "loss": 0.2505,
      "step": 10697
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.383956512372733,
      "learning_rate": 3.359437679958943e-08,
      "loss": 0.272,
      "step": 10698
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.159737215982804,
      "learning_rate": 3.3477074356626784e-08,
      "loss": 0.4133,
      "step": 10699
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8795295055276453,
      "learning_rate": 3.3359975685695324e-08,
      "loss": 0.2578,
      "step": 10700
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.9446107629152753,
      "learning_rate": 3.324308079646982e-08,
      "loss": 0.1847,
      "step": 10701
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.334888517932094,
      "learning_rate": 3.3126389698606434e-08,
      "loss": 0.3338,
      "step": 10702
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.1533083717139725,
      "learning_rate": 3.30099024017469e-08,
      "loss": 0.4164,
      "step": 10703
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.086506401119566,
      "learning_rate": 3.2893618915513795e-08,
      "loss": 0.2836,
      "step": 10704
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.8342329886006232,
      "learning_rate": 3.277753924951443e-08,
      "loss": 0.2056,
      "step": 10705
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.128571181306619,
      "learning_rate": 3.266166341333893e-08,
      "loss": 0.4261,
      "step": 10706
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4633485853835237,
      "learning_rate": 3.254599141655962e-08,
      "loss": 0.0147,
      "step": 10707
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.5860147778150173,
      "learning_rate": 3.243052326873358e-08,
      "loss": 0.2613,
      "step": 10708
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.443794183124427,
      "learning_rate": 3.2315258979399856e-08,
      "loss": 0.0416,
      "step": 10709
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.6015584570748223,
      "learning_rate": 3.220019855808082e-08,
      "loss": 0.1822,
      "step": 10710
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.044183224821336,
      "learning_rate": 3.208534201428276e-08,
      "loss": 0.2175,
      "step": 10711
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7320874594344047,
      "learning_rate": 3.197068935749364e-08,
      "loss": 0.2521,
      "step": 10712
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.572112828013661,
      "learning_rate": 3.185624059718645e-08,
      "loss": 0.3392,
      "step": 10713
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.878148243551709,
      "learning_rate": 3.1741995742816113e-08,
      "loss": 0.3307,
      "step": 10714
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.789067934769656,
      "learning_rate": 3.162795480382064e-08,
      "loss": 0.3489,
      "step": 10715
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.036009966669306,
      "learning_rate": 3.1514117789621935e-08,
      "loss": 0.3345,
      "step": 10716
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.060530379504351,
      "learning_rate": 3.1400484709624144e-08,
      "loss": 0.1755,
      "step": 10717
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.069603714985338,
      "learning_rate": 3.128705557321588e-08,
      "loss": 0.3001,
      "step": 10718
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7196380074750404,
      "learning_rate": 3.11738303897674e-08,
      "loss": 0.1278,
      "step": 10719
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.7987746624832806,
      "learning_rate": 3.106080916863291e-08,
      "loss": 0.0781,
      "step": 10720
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.200121109557572,
      "learning_rate": 3.094799191915021e-08,
      "loss": 0.3094,
      "step": 10721
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.549042543708025,
      "learning_rate": 3.083537865063879e-08,
      "loss": 0.1728,
      "step": 10722
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.8138397612481145,
      "learning_rate": 3.0722969372403154e-08,
      "loss": 0.3162,
      "step": 10723
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7068238910484794,
      "learning_rate": 3.061076409372948e-08,
      "loss": 0.266,
      "step": 10724
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.285692983206174,
      "learning_rate": 3.0498762823887306e-08,
      "loss": 0.5017,
      "step": 10725
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.109032380688434,
      "learning_rate": 3.038696557213061e-08,
      "loss": 0.3856,
      "step": 10726
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.628714437328145,
      "learning_rate": 3.027537234769479e-08,
      "loss": 0.1233,
      "step": 10727
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.5212239667903513,
      "learning_rate": 3.016398315979913e-08,
      "loss": 0.2676,
      "step": 10728
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.001980637771731,
      "learning_rate": 3.005279801764627e-08,
      "loss": 0.1965,
      "step": 10729
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.5947302750776435,
      "learning_rate": 2.994181693042164e-08,
      "loss": 0.288,
      "step": 10730
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.964860146837418,
      "learning_rate": 2.98310399072943e-08,
      "loss": 0.2525,
      "step": 10731
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.340247823710128,
      "learning_rate": 2.9720466957415816e-08,
      "loss": 0.2954,
      "step": 10732
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.91989586294005,
      "learning_rate": 2.9610098089921378e-08,
      "loss": 0.2609,
      "step": 10733
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.828722314514788,
      "learning_rate": 2.9499933313928696e-08,
      "loss": 0.3639,
      "step": 10734
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.986339565107202,
      "learning_rate": 2.93899726385391e-08,
      "loss": 0.2765,
      "step": 10735
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.574421590581603,
      "learning_rate": 2.9280216072837542e-08,
      "loss": 0.1703,
      "step": 10736
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.0429041275596065,
      "learning_rate": 2.9170663625891493e-08,
      "loss": 0.2235,
      "step": 10737
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.742271680780672,
      "learning_rate": 2.9061315306750936e-08,
      "loss": 0.2683,
      "step": 10738
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.950586369819294,
      "learning_rate": 2.895217112445031e-08,
      "loss": 0.2596,
      "step": 10739
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.050993385026516,
      "learning_rate": 2.8843231088006296e-08,
      "loss": 0.2224,
      "step": 10740
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.2397940213271488,
      "learning_rate": 2.8734495206419468e-08,
      "loss": 0.2022,
      "step": 10741
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.554358369256011,
      "learning_rate": 2.8625963488672358e-08,
      "loss": 0.3576,
      "step": 10742
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.016411900825227,
      "learning_rate": 2.851763594373169e-08,
      "loss": 0.2525,
      "step": 10743
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.310748681553,
      "learning_rate": 2.8409512580546973e-08,
      "loss": 0.1601,
      "step": 10744
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.5637108006854317,
      "learning_rate": 2.8301593408050777e-08,
      "loss": 0.1749,
      "step": 10745
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.818948458190796,
      "learning_rate": 2.819387843515875e-08,
      "loss": 0.4276,
      "step": 10746
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29794322776807497,
      "learning_rate": 2.8086367670769886e-08,
      "loss": 0.0024,
      "step": 10747
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.885188079065414,
      "learning_rate": 2.797906112376625e-08,
      "loss": 0.1777,
      "step": 10748
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.873193735228387,
      "learning_rate": 2.7871958803012965e-08,
      "loss": 0.1673,
      "step": 10749
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.174313450249031,
      "learning_rate": 2.7765060717358238e-08,
      "loss": 0.3543,
      "step": 10750
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.921719323521148,
      "learning_rate": 2.7658366875633335e-08,
      "loss": 0.1562,
      "step": 10751
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0471425966303993,
      "learning_rate": 2.755187728665315e-08,
      "loss": 0.0311,
      "step": 10752
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.043353656300405,
      "learning_rate": 2.744559195921509e-08,
      "loss": 0.1635,
      "step": 10753
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.86155862729888,
      "learning_rate": 2.733951090209963e-08,
      "loss": 0.5643,
      "step": 10754
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.842697398713115,
      "learning_rate": 2.723363412407115e-08,
      "loss": 0.3886,
      "step": 10755
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.418985595016847,
      "learning_rate": 2.7127961633876544e-08,
      "loss": 0.4194,
      "step": 10756
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.7374288361324353,
      "learning_rate": 2.7022493440246046e-08,
      "loss": 0.2164,
      "step": 10757
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.874994733217052,
      "learning_rate": 2.6917229551892687e-08,
      "loss": 0.2934,
      "step": 10758
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.676054317539993,
      "learning_rate": 2.6812169977512847e-08,
      "loss": 0.2678,
      "step": 10759
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.580989633034274,
      "learning_rate": 2.670731472578597e-08,
      "loss": 0.1613,
      "step": 10760
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.854188222505255,
      "learning_rate": 2.6602663805375128e-08,
      "loss": 0.4343,
      "step": 10761
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.485676488419444,
      "learning_rate": 2.6498217224925903e-08,
      "loss": 0.5007,
      "step": 10762
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.009341847148158,
      "learning_rate": 2.6393974993066675e-08,
      "loss": 0.3401,
      "step": 10763
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.4223251150349383,
      "learning_rate": 2.628993711840999e-08,
      "loss": 0.2516,
      "step": 10764
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.018187294138535,
      "learning_rate": 2.618610360955065e-08,
      "loss": 0.1585,
      "step": 10765
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.502839092983888,
      "learning_rate": 2.6082474475067343e-08,
      "loss": 0.193,
      "step": 10766
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.381824492238123,
      "learning_rate": 2.597904972352072e-08,
      "loss": 0.1098,
      "step": 10767
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.973595288110562,
      "learning_rate": 2.587582936345562e-08,
      "loss": 0.6181,
      "step": 10768
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.3690430685359827,
      "learning_rate": 2.5772813403399377e-08,
      "loss": 0.1298,
      "step": 10769
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.7879571629460473,
      "learning_rate": 2.567000185186297e-08,
      "loss": 0.3136,
      "step": 10770
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.7676666680571196,
      "learning_rate": 2.5567394717339878e-08,
      "loss": 0.1218,
      "step": 10771
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.417364264715908,
      "learning_rate": 2.546499200830721e-08,
      "loss": 0.3024,
      "step": 10772
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.875531065869993,
      "learning_rate": 2.5362793733225144e-08,
      "loss": 0.3233,
      "step": 10773
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.3871794013585035,
      "learning_rate": 2.5260799900536094e-08,
      "loss": 0.2366,
      "step": 10774
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.647517281884049,
      "learning_rate": 2.5159010518667205e-08,
      "loss": 0.086,
      "step": 10775
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.3897937349904605,
      "learning_rate": 2.505742559602703e-08,
      "loss": 0.1975,
      "step": 10776
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.0365059342393113,
      "learning_rate": 2.495604514100858e-08,
      "loss": 0.1083,
      "step": 10777
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6926893282377193,
      "learning_rate": 2.4854869161987095e-08,
      "loss": 0.1297,
      "step": 10778
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5569643034487215,
      "learning_rate": 2.4753897667321446e-08,
      "loss": 0.1549,
      "step": 10779
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5963244874460196,
      "learning_rate": 2.4653130665353298e-08,
      "loss": 0.2332,
      "step": 10780
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.8626319964237843,
      "learning_rate": 2.4552568164407374e-08,
      "loss": 0.2265,
      "step": 10781
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.207136178201698,
      "learning_rate": 2.4452210172792035e-08,
      "loss": 0.38,
      "step": 10782
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.120399432281496,
      "learning_rate": 2.4352056698797867e-08,
      "loss": 0.2454,
      "step": 10783
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6970127498913863,
      "learning_rate": 2.425210775069936e-08,
      "loss": 0.1755,
      "step": 10784
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.143405543031485,
      "learning_rate": 2.4152363336753804e-08,
      "loss": 0.1663,
      "step": 10785
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.158671730839422,
      "learning_rate": 2.4052823465201824e-08,
      "loss": 0.3009,
      "step": 10786
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.36653679255349,
      "learning_rate": 2.395348814426629e-08,
      "loss": 0.4037,
      "step": 10787
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.259310140828304,
      "learning_rate": 2.3854357382154526e-08,
      "loss": 0.2199,
      "step": 10788
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.239180041815195,
      "learning_rate": 2.3755431187055532e-08,
      "loss": 0.3616,
      "step": 10789
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.893731135367643,
      "learning_rate": 2.3656709567142498e-08,
      "loss": 0.1795,
      "step": 10790
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.096308496557021,
      "learning_rate": 2.35581925305714e-08,
      "loss": 0.3295,
      "step": 10791
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.711732662172219,
      "learning_rate": 2.3459880085481e-08,
      "loss": 0.267,
      "step": 10792
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.8254939138719175,
      "learning_rate": 2.336177223999342e-08,
      "loss": 0.2518,
      "step": 10793
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.423588476534997,
      "learning_rate": 2.326386900221411e-08,
      "loss": 0.0411,
      "step": 10794
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.6628772738247513,
      "learning_rate": 2.3166170380231057e-08,
      "loss": 0.1212,
      "step": 10795
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.811952759308974,
      "learning_rate": 2.3068676382115573e-08,
      "loss": 0.2885,
      "step": 10796
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.6676809363443095,
      "learning_rate": 2.2971387015922607e-08,
      "loss": 0.4325,
      "step": 10797
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.160320743547501,
      "learning_rate": 2.287430228968934e-08,
      "loss": 0.3602,
      "step": 10798
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.6995333388977505,
      "learning_rate": 2.2777422211436306e-08,
      "loss": 0.2268,
      "step": 10799
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8518149517926283,
      "learning_rate": 2.268074678916765e-08,
      "loss": 0.1101,
      "step": 10800
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.267019647498581,
      "learning_rate": 2.2584276030870322e-08,
      "loss": 0.4118,
      "step": 10801
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.323479728031115,
      "learning_rate": 2.24880099445135e-08,
      "loss": 0.2662,
      "step": 10802
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.24032111406155,
      "learning_rate": 2.23919485380511e-08,
      "loss": 0.1855,
      "step": 10803
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.175551965944524,
      "learning_rate": 2.229609181941844e-08,
      "loss": 0.1268,
      "step": 10804
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.4654362910709278,
      "learning_rate": 2.2200439796535577e-08,
      "loss": 0.2796,
      "step": 10805
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8930862760350498,
      "learning_rate": 2.2104992477304243e-08,
      "loss": 0.1818,
      "step": 10806
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.21849773861259,
      "learning_rate": 2.20097498696098e-08,
      "loss": 0.3303,
      "step": 10807
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.033744848209427,
      "learning_rate": 2.1914711981321225e-08,
      "loss": 0.2468,
      "step": 10808
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.225399506857565,
      "learning_rate": 2.1819878820289464e-08,
      "loss": 0.2425,
      "step": 10809
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.4306280169232135,
      "learning_rate": 2.1725250394349362e-08,
      "loss": 0.2644,
      "step": 10810
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.355892757755594,
      "learning_rate": 2.163082671131911e-08,
      "loss": 0.335,
      "step": 10811
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.476243567592923,
      "learning_rate": 2.1536607778998854e-08,
      "loss": 0.2631,
      "step": 10812
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.062824228963911,
      "learning_rate": 2.1442593605172925e-08,
      "loss": 0.2576,
      "step": 10813
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.362746071227482,
      "learning_rate": 2.1348784197608163e-08,
      "loss": 0.0888,
      "step": 10814
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.126561958566438,
      "learning_rate": 2.1255179564054762e-08,
      "loss": 0.1578,
      "step": 10815
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.416824844217589,
      "learning_rate": 2.11617797122457e-08,
      "loss": 0.2217,
      "step": 10816
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5424382484141224,
      "learning_rate": 2.1068584649897306e-08,
      "loss": 0.2804,
      "step": 10817
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.1863848864691615,
      "learning_rate": 2.0975594384708976e-08,
      "loss": 0.3276,
      "step": 10818
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.998865128298915,
      "learning_rate": 2.0882808924362897e-08,
      "loss": 0.1758,
      "step": 10819
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.243281748447788,
      "learning_rate": 2.0790228276524883e-08,
      "loss": 0.2798,
      "step": 10820
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.154397777068416,
      "learning_rate": 2.069785244884326e-08,
      "loss": 0.3559,
      "step": 10821
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.229420062939478,
      "learning_rate": 2.06056814489497e-08,
      "loss": 0.1787,
      "step": 10822
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.152486907643382,
      "learning_rate": 2.0513715284458945e-08,
      "loss": 0.3287,
      "step": 10823
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.933827741993229,
      "learning_rate": 2.042195396296853e-08,
      "loss": 0.2942,
      "step": 10824
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.950608913851872,
      "learning_rate": 2.0330397492059893e-08,
      "loss": 0.3226,
      "step": 10825
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.966684002434575,
      "learning_rate": 2.0239045879296427e-08,
      "loss": 0.3653,
      "step": 10826
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.296453251546859,
      "learning_rate": 2.0147899132225423e-08,
      "loss": 0.3339,
      "step": 10827
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.268464814876758,
      "learning_rate": 2.0056957258376973e-08,
      "loss": 0.1226,
      "step": 10828
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.526263835096969,
      "learning_rate": 1.9966220265264235e-08,
      "loss": 0.3638,
      "step": 10829
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9853289394988416,
      "learning_rate": 1.9875688160383434e-08,
      "loss": 0.064,
      "step": 10830
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.165006990568199,
      "learning_rate": 1.978536095121386e-08,
      "loss": 0.3208,
      "step": 10831
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.18409175707986,
      "learning_rate": 1.969523864521761e-08,
      "loss": 0.1874,
      "step": 10832
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.784154796017188,
      "learning_rate": 1.9605321249840936e-08,
      "loss": 0.4009,
      "step": 10833
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.188085605264321,
      "learning_rate": 1.9515608772511798e-08,
      "loss": 0.4174,
      "step": 10834
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.2066138056597935,
      "learning_rate": 1.9426101220641757e-08,
      "loss": 0.1906,
      "step": 10835
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.775401952623943,
      "learning_rate": 1.933679860162574e-08,
      "loss": 0.2859,
      "step": 10836
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.6528737174958787,
      "learning_rate": 1.9247700922841173e-08,
      "loss": 0.2721,
      "step": 10837
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.581656891588215,
      "learning_rate": 1.915880819164939e-08,
      "loss": 0.2909,
      "step": 10838
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.4621577524011595,
      "learning_rate": 1.907012041539369e-08,
      "loss": 0.2164,
      "step": 10839
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.4425444898741215,
      "learning_rate": 1.8981637601401536e-08,
      "loss": 0.3479,
      "step": 10840
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.913081734269102,
      "learning_rate": 1.8893359756982642e-08,
      "loss": 0.5795,
      "step": 10841
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.152950789027525,
      "learning_rate": 1.8805286889430064e-08,
      "loss": 0.3595,
      "step": 10842
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.0016171845132815,
      "learning_rate": 1.87174190060202e-08,
      "loss": 0.37,
      "step": 10843
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.189215167336073,
      "learning_rate": 1.8629756114011965e-08,
      "loss": 0.383,
      "step": 10844
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.646725458895227,
      "learning_rate": 1.8542298220647626e-08,
      "loss": 0.2029,
      "step": 10845
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.487613406640546,
      "learning_rate": 1.8455045333153066e-08,
      "loss": 0.3877,
      "step": 10846
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.032614063540247,
      "learning_rate": 1.8367997458735854e-08,
      "loss": 0.3407,
      "step": 10847
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.192557560603383,
      "learning_rate": 1.828115460458829e-08,
      "loss": 0.2342,
      "step": 10848
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.389635485056706,
      "learning_rate": 1.8194516777884363e-08,
      "loss": 0.3278,
      "step": 10849
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.918552793938703,
      "learning_rate": 1.8108083985781956e-08,
      "loss": 0.4039,
      "step": 10850
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.621949998193301,
      "learning_rate": 1.8021856235421463e-08,
      "loss": 0.3742,
      "step": 10851
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.216727545724447,
      "learning_rate": 1.7935833533926638e-08,
      "loss": 0.3145,
      "step": 10852
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.1229160439049295,
      "learning_rate": 1.785001588840457e-08,
      "loss": 0.2219,
      "step": 10853
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.7383370879016615,
      "learning_rate": 1.7764403305944587e-08,
      "loss": 0.2317,
      "step": 10854
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.470925643635847,
      "learning_rate": 1.7678995793619924e-08,
      "loss": 0.2875,
      "step": 10855
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.124160968704077,
      "learning_rate": 1.75937933584866e-08,
      "loss": 0.2677,
      "step": 10856
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.859478053683501,
      "learning_rate": 1.750879600758343e-08,
      "loss": 0.3814,
      "step": 10857
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.754949712185222,
      "learning_rate": 1.7424003747932573e-08,
      "loss": 0.405,
      "step": 10858
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.819221215421265,
      "learning_rate": 1.7339416586538983e-08,
      "loss": 0.2605,
      "step": 10859
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.026272792735743,
      "learning_rate": 1.7255034530390957e-08,
      "loss": 0.2231,
      "step": 10860
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.415255339817023,
      "learning_rate": 1.7170857586460145e-08,
      "loss": 0.5411,
      "step": 10861
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.748624644201685,
      "learning_rate": 1.7086885761699866e-08,
      "loss": 0.2437,
      "step": 10862
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.8041620330911368,
      "learning_rate": 1.700311906304847e-08,
      "loss": 0.1232,
      "step": 10863
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.505543493096918,
      "learning_rate": 1.691955749742541e-08,
      "loss": 0.4529,
      "step": 10864
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.329283729947315,
      "learning_rate": 1.6836201071734893e-08,
      "loss": 0.5072,
      "step": 10865
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.175722819240125,
      "learning_rate": 1.675304979286335e-08,
      "loss": 0.2444,
      "step": 10866
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.5120597410019023,
      "learning_rate": 1.667010366767974e-08,
      "loss": 0.1179,
      "step": 10867
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.084790711473009,
      "learning_rate": 1.6587362703037457e-08,
      "loss": 0.3416,
      "step": 10868
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.311513797165033,
      "learning_rate": 1.6504826905771597e-08,
      "loss": 0.5871,
      "step": 10869
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.055930744655754,
      "learning_rate": 1.642249628270087e-08,
      "loss": 0.3935,
      "step": 10870
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.7045655994547824,
      "learning_rate": 1.6340370840627338e-08,
      "loss": 0.1629,
      "step": 10871
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.42871130826474,
      "learning_rate": 1.6258450586335294e-08,
      "loss": 0.3,
      "step": 10872
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.1403158515536767,
      "learning_rate": 1.617673552659349e-08,
      "loss": 0.1642,
      "step": 10873
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.4779774731280515,
      "learning_rate": 1.6095225668151803e-08,
      "loss": 0.369,
      "step": 10874
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.18117671906551,
      "learning_rate": 1.601392101774485e-08,
      "loss": 0.213,
      "step": 10875
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.5854128483407175,
      "learning_rate": 1.5932821582089476e-08,
      "loss": 0.3323,
      "step": 10876
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.635122212855707,
      "learning_rate": 1.58519273678856e-08,
      "loss": 0.3395,
      "step": 10877
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.1829941238971875,
      "learning_rate": 1.577123838181621e-08,
      "loss": 0.3744,
      "step": 10878
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.690004315479327,
      "learning_rate": 1.5690754630547644e-08,
      "loss": 0.2943,
      "step": 10879
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.5740564631087715,
      "learning_rate": 1.5610476120729024e-08,
      "loss": 0.1689,
      "step": 10880
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.609337749072169,
      "learning_rate": 1.553040285899282e-08,
      "loss": 0.3012,
      "step": 10881
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0437904541440073,
      "learning_rate": 1.5450534851953746e-08,
      "loss": 0.1248,
      "step": 10882
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.7973653140032493,
      "learning_rate": 1.537087210621041e-08,
      "loss": 0.206,
      "step": 10883
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0508165717481712,
      "learning_rate": 1.5291414628344493e-08,
      "loss": 0.0878,
      "step": 10884
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.807868069386936,
      "learning_rate": 1.521216242491963e-08,
      "loss": 0.4208,
      "step": 10885
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.159147102981391,
      "learning_rate": 1.513311550248392e-08,
      "loss": 0.2232,
      "step": 10886
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.246165644898821,
      "learning_rate": 1.5054273867567138e-08,
      "loss": 0.2781,
      "step": 10887
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.07180551386898,
      "learning_rate": 1.4975637526683516e-08,
      "loss": 0.2739,
      "step": 10888
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.1124907425923105,
      "learning_rate": 1.4897206486329253e-08,
      "loss": 0.2606,
      "step": 10889
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.804725307430654,
      "learning_rate": 1.481898075298388e-08,
      "loss": 0.2963,
      "step": 10890
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.558201077587531,
      "learning_rate": 1.4740960333110566e-08,
      "loss": 0.3429,
      "step": 10891
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.7163395373297456,
      "learning_rate": 1.4663145233154152e-08,
      "loss": 0.2002,
      "step": 10892
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.526056717410777,
      "learning_rate": 1.4585535459543665e-08,
      "loss": 0.487,
      "step": 10893
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.527430283803421,
      "learning_rate": 1.4508131018691197e-08,
      "loss": 0.1552,
      "step": 10894
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.829744548660376,
      "learning_rate": 1.4430931916991076e-08,
      "loss": 0.3295,
      "step": 10895
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.768676942171273,
      "learning_rate": 1.4353938160821256e-08,
      "loss": 0.3899,
      "step": 10896
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.206829930373474,
      "learning_rate": 1.4277149756542764e-08,
      "loss": 0.2631,
      "step": 10897
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.024198782078346,
      "learning_rate": 1.420056671049913e-08,
      "loss": 0.345,
      "step": 10898
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.7870369916574758,
      "learning_rate": 1.4124189029017521e-08,
      "loss": 0.1892,
      "step": 10899
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.154519637672756,
      "learning_rate": 1.4048016718407609e-08,
      "loss": 0.1486,
      "step": 10900
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.834723868128784,
      "learning_rate": 1.3972049784962972e-08,
      "loss": 0.1555,
      "step": 10901
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.483314343501201,
      "learning_rate": 1.3896288234958866e-08,
      "loss": 0.397,
      "step": 10902
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.231095392294292,
      "learning_rate": 1.3820732074655008e-08,
      "loss": 0.1802,
      "step": 10903
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.677957447361542,
      "learning_rate": 1.374538131029307e-08,
      "loss": 0.2513,
      "step": 10904
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.461018878916081,
      "learning_rate": 1.3670235948098354e-08,
      "loss": 0.2576,
      "step": 10905
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.316375524318658,
      "learning_rate": 1.3595295994278945e-08,
      "loss": 0.2718,
      "step": 10906
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.2270782608678745,
      "learning_rate": 1.3520561455025727e-08,
      "loss": 0.1542,
      "step": 10907
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.009659631094911,
      "learning_rate": 1.3446032336513482e-08,
      "loss": 0.4205,
      "step": 10908
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.418857896517115,
      "learning_rate": 1.337170864489895e-08,
      "loss": 0.2283,
      "step": 10909
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.194132608318951,
      "learning_rate": 1.32975903863225e-08,
      "loss": 0.3351,
      "step": 10910
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.222357681021923,
      "learning_rate": 1.3223677566907845e-08,
      "loss": 0.3731,
      "step": 10911
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0515846324849534,
      "learning_rate": 1.3149970192760653e-08,
      "loss": 0.2246,
      "step": 10912
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.472544012981003,
      "learning_rate": 1.30764682699705e-08,
      "loss": 0.3687,
      "step": 10913
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.647728158201586,
      "learning_rate": 1.3003171804610026e-08,
      "loss": 0.0684,
      "step": 10914
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.750966122639453,
      "learning_rate": 1.2930080802734113e-08,
      "loss": 0.3634,
      "step": 10915
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0088813390687896,
      "learning_rate": 1.2857195270381817e-08,
      "loss": 0.1833,
      "step": 10916
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.920808994895772,
      "learning_rate": 1.2784515213574156e-08,
      "loss": 0.2487,
      "step": 10917
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.191073088020696,
      "learning_rate": 1.2712040638315492e-08,
      "loss": 0.2233,
      "step": 10918
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.79081981560976,
      "learning_rate": 1.2639771550593815e-08,
      "loss": 0.5164,
      "step": 10919
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.80117780431693,
      "learning_rate": 1.2567707956379072e-08,
      "loss": 0.2272,
      "step": 10920
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.707794844908555,
      "learning_rate": 1.2495849861625386e-08,
      "loss": 0.3112,
      "step": 10921
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.432891728894957,
      "learning_rate": 1.2424197272268567e-08,
      "loss": 0.2049,
      "step": 10922
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.018031501744917,
      "learning_rate": 1.235275019422888e-08,
      "loss": 0.5286,
      "step": 10923
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.5739195096790564,
      "learning_rate": 1.2281508633408823e-08,
      "loss": 0.1708,
      "step": 10924
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.4854189345831683,
      "learning_rate": 1.221047259569369e-08,
      "loss": 0.2106,
      "step": 10925
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.7139998442347606,
      "learning_rate": 1.2139642086952397e-08,
      "loss": 0.1458,
      "step": 10926
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.439117037832681,
      "learning_rate": 1.2069017113036374e-08,
      "loss": 0.1928,
      "step": 10927
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.429088228200503,
      "learning_rate": 1.1998597679780676e-08,
      "loss": 0.2819,
      "step": 10928
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.164797758601316,
      "learning_rate": 1.1928383793002873e-08,
      "loss": 0.2292,
      "step": 10929
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.853228762558952,
      "learning_rate": 1.1858375458503601e-08,
      "loss": 0.4835,
      "step": 10930
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.595108324273858,
      "learning_rate": 1.1788572682066569e-08,
      "loss": 0.3061,
      "step": 10931
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.5069037340365155,
      "learning_rate": 1.1718975469458827e-08,
      "loss": 0.4905,
      "step": 10932
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.9523838483239313,
      "learning_rate": 1.1649583826429944e-08,
      "loss": 0.1332,
      "step": 10933
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.0651828314093,
      "learning_rate": 1.1580397758712558e-08,
      "loss": 0.2336,
      "step": 10934
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.827901488474064,
      "learning_rate": 1.1511417272022373e-08,
      "loss": 0.2931,
      "step": 10935
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.056780797687255,
      "learning_rate": 1.1442642372058999e-08,
      "loss": 0.1822,
      "step": 10936
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.6502250240182805,
      "learning_rate": 1.1374073064503443e-08,
      "loss": 0.2492,
      "step": 10937
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.7933049479561864,
      "learning_rate": 1.1305709355021178e-08,
      "loss": 0.5132,
      "step": 10938
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.626710772571281,
      "learning_rate": 1.1237551249259627e-08,
      "loss": 0.1717,
      "step": 10939
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.844184966953198,
      "learning_rate": 1.1169598752849841e-08,
      "loss": 0.1663,
      "step": 10940
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.538914186185981,
      "learning_rate": 1.1101851871405666e-08,
      "loss": 0.4141,
      "step": 10941
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.8331614831017102,
      "learning_rate": 1.1034310610524013e-08,
      "loss": 0.3035,
      "step": 10942
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.574339999934802,
      "learning_rate": 1.0966974975785138e-08,
      "loss": 0.4143,
      "step": 10943
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.445953397047577,
      "learning_rate": 1.089984497275154e-08,
      "loss": 0.3332,
      "step": 10944
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.357273902384948,
      "learning_rate": 1.0832920606969332e-08,
      "loss": 0.3776,
      "step": 10945
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.92415358434339,
      "learning_rate": 1.0766201883967708e-08,
      "loss": 0.2697,
      "step": 10946
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.3139152574414488,
      "learning_rate": 1.0699688809258368e-08,
      "loss": 0.1099,
      "step": 10947
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.020134034735494,
      "learning_rate": 1.0633381388336084e-08,
      "loss": 0.3161,
      "step": 10948
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.947002920604402,
      "learning_rate": 1.0567279626679527e-08,
      "loss": 0.3631,
      "step": 10949
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6203726480447536,
      "learning_rate": 1.0501383529749053e-08,
      "loss": 0.1406,
      "step": 10950
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.493980003743094,
      "learning_rate": 1.043569310298892e-08,
      "loss": 0.38,
      "step": 10951
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.5594182656276665,
      "learning_rate": 1.0370208351826172e-08,
      "loss": 0.3128,
      "step": 10952
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.50026615168927,
      "learning_rate": 1.030492928167065e-08,
      "loss": 0.2135,
      "step": 10953
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.02390170140192,
      "learning_rate": 1.0239855897915817e-08,
      "loss": 0.4107,
      "step": 10954
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.3619428528014375,
      "learning_rate": 1.0174988205937375e-08,
      "loss": 0.3933,
      "step": 10955
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.676445096328155,
      "learning_rate": 1.0110326211094367e-08,
      "loss": 0.292,
      "step": 10956
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.5059284269715154,
      "learning_rate": 1.0045869918728912e-08,
      "loss": 0.1711,
      "step": 10957
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.047587677167617,
      "learning_rate": 9.981619334166471e-09,
      "loss": 0.0564,
      "step": 10958
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.956799215371586,
      "learning_rate": 9.917574462714463e-09,
      "loss": 0.1653,
      "step": 10959
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.7414595910617776,
      "learning_rate": 9.853735309663936e-09,
      "loss": 0.1591,
      "step": 10960
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.676286661542599,
      "learning_rate": 9.79010188028956e-09,
      "loss": 0.2851,
      "step": 10961
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.2975736682083365,
      "learning_rate": 9.72667417984824e-09,
      "loss": 0.3137,
      "step": 10962
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.035370357134355,
      "learning_rate": 9.663452213579949e-09,
      "loss": 0.2146,
      "step": 10963
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.9378602983455306,
      "learning_rate": 9.600435986707457e-09,
      "loss": 0.2748,
      "step": 10964
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.035589416468447,
      "learning_rate": 9.53762550443743e-09,
      "loss": 0.166,
      "step": 10965
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3446621500312976,
      "learning_rate": 9.475020771958776e-09,
      "loss": 0.0359,
      "step": 10966
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.625993085469409,
      "learning_rate": 9.412621794443189e-09,
      "loss": 0.2278,
      "step": 10967
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.0030202434895426,
      "learning_rate": 9.350428577046266e-09,
      "loss": 0.3166,
      "step": 10968
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.56876840968009,
      "learning_rate": 9.288441124906122e-09,
      "loss": 0.1028,
      "step": 10969
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.04240153874382,
      "learning_rate": 9.226659443143383e-09,
      "loss": 0.2212,
      "step": 10970
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.264512562038982,
      "learning_rate": 9.165083536862574e-09,
      "loss": 0.2684,
      "step": 10971
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.09948996561205,
      "learning_rate": 9.103713411150738e-09,
      "loss": 0.2434,
      "step": 10972
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.265580086657933,
      "learning_rate": 9.042549071077711e-09,
      "loss": 0.2396,
      "step": 10973
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.216444746934983,
      "learning_rate": 8.981590521696948e-09,
      "loss": 0.222,
      "step": 10974
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.130206205476968,
      "learning_rate": 8.920837768044144e-09,
      "loss": 0.4669,
      "step": 10975
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.024251898701425,
      "learning_rate": 8.860290815138894e-09,
      "loss": 0.2092,
      "step": 10976
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.0757910280938465,
      "learning_rate": 8.799949667983031e-09,
      "loss": 0.1867,
      "step": 10977
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.2532779120576105,
      "learning_rate": 8.739814331561458e-09,
      "loss": 0.3758,
      "step": 10978
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.585392427026205,
      "learning_rate": 8.67988481084242e-09,
      "loss": 0.3289,
      "step": 10979
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.7501183043930872,
      "learning_rate": 8.62016111077696e-09,
      "loss": 0.1231,
      "step": 10980
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.3915180634584505,
      "learning_rate": 8.56064323629946e-09,
      "loss": 0.4547,
      "step": 10981
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.455977741588228,
      "learning_rate": 8.501331192326546e-09,
      "loss": 0.212,
      "step": 10982
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.729250803264852,
      "learning_rate": 8.442224983758462e-09,
      "loss": 0.4346,
      "step": 10983
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.153228683115469,
      "learning_rate": 8.38332461547825e-09,
      "loss": 0.2467,
      "step": 10984
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.66049271446571,
      "learning_rate": 8.324630092352293e-09,
      "loss": 0.3924,
      "step": 10985
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.316780160840544,
      "learning_rate": 8.266141419229213e-09,
      "loss": 0.2143,
      "step": 10986
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.0679731487241915,
      "learning_rate": 8.207858600941255e-09,
      "loss": 0.3765,
      "step": 10987
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.851187405873682,
      "learning_rate": 8.149781642303456e-09,
      "loss": 0.4221,
      "step": 10988
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.345262770552694,
      "learning_rate": 8.091910548113645e-09,
      "loss": 0.2584,
      "step": 10989
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.877047474909005,
      "learning_rate": 8.034245323153278e-09,
      "loss": 0.1821,
      "step": 10990
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9396122008638086,
      "learning_rate": 7.976785972185763e-09,
      "loss": 0.2214,
      "step": 10991
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.002010547750842,
      "learning_rate": 7.919532499958693e-09,
      "loss": 0.1855,
      "step": 10992
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.300591948191663,
      "learning_rate": 7.862484911201895e-09,
      "loss": 0.2984,
      "step": 10993
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9667246538598895,
      "learning_rate": 7.805643210628266e-09,
      "loss": 0.1763,
      "step": 10994
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.443803040620041,
      "learning_rate": 7.749007402933773e-09,
      "loss": 0.2488,
      "step": 10995
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.1209303228029635,
      "learning_rate": 7.692577492797448e-09,
      "loss": 0.3206,
      "step": 10996
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.653579898528025,
      "learning_rate": 7.63635348488112e-09,
      "loss": 0.0959,
      "step": 10997
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.5134237941362976,
      "learning_rate": 7.580335383829962e-09,
      "loss": 0.1226,
      "step": 10998
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.72891524137785,
      "learning_rate": 7.52452319427166e-09,
      "loss": 0.2837,
      "step": 10999
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3449629012208044,
      "learning_rate": 7.468916920817526e-09,
      "loss": 0.1367,
      "step": 11000
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.428725904633606,
      "learning_rate": 7.413516568061385e-09,
      "loss": 0.1874,
      "step": 11001
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2094641392835777,
      "learning_rate": 7.358322140579854e-09,
      "loss": 0.0366,
      "step": 11002
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.366623376319067,
      "learning_rate": 7.303333642932897e-09,
      "loss": 0.2487,
      "step": 11003
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.087316758057978,
      "learning_rate": 7.248551079663546e-09,
      "loss": 0.3028,
      "step": 11004
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.131321500057657,
      "learning_rate": 7.193974455297625e-09,
      "loss": 0.2003,
      "step": 11005
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.4847219876904285,
      "learning_rate": 7.139603774343751e-09,
      "loss": 0.558,
      "step": 11006
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.278892754310659,
      "learning_rate": 7.085439041294162e-09,
      "loss": 0.4359,
      "step": 11007
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.15078809920056,
      "learning_rate": 7.031480260623613e-09,
      "loss": 0.3676,
      "step": 11008
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.994948555842427,
      "learning_rate": 6.97772743678965e-09,
      "loss": 0.2172,
      "step": 11009
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.475090755058565,
      "learning_rate": 6.924180574233164e-09,
      "loss": 0.2128,
      "step": 11010
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.025513976571834,
      "learning_rate": 6.870839677378116e-09,
      "loss": 0.3436,
      "step": 11011
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.817449253146332,
      "learning_rate": 6.81770475063126e-09,
      "loss": 0.1591,
      "step": 11012
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.268568073121123,
      "learning_rate": 6.764775798382139e-09,
      "loss": 0.4913,
      "step": 11013
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.934499392714094,
      "learning_rate": 6.712052825003368e-09,
      "loss": 0.1922,
      "step": 11014
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.7541416899272604,
      "learning_rate": 6.659535834850906e-09,
      "loss": 0.2336,
      "step": 11015
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.683711279964092,
      "learning_rate": 6.607224832263504e-09,
      "loss": 0.3649,
      "step": 11016
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.361308530549506,
      "learning_rate": 6.555119821562705e-09,
      "loss": 0.3285,
      "step": 11017
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.70412162251953,
      "learning_rate": 6.503220807053401e-09,
      "loss": 0.4022,
      "step": 11018
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.6347797647432865,
      "learning_rate": 6.451527793022716e-09,
      "loss": 0.5024,
      "step": 11019
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.593095304543681,
      "learning_rate": 6.400040783741679e-09,
      "loss": 0.2922,
      "step": 11020
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.544825858214171,
      "learning_rate": 6.348759783463831e-09,
      "loss": 0.4356,
      "step": 11021
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.117830873093502,
      "learning_rate": 6.297684796425785e-09,
      "loss": 0.2209,
      "step": 11022
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.574839557761855,
      "learning_rate": 6.246815826847219e-09,
      "loss": 0.1724,
      "step": 11023
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.433510539736242,
      "learning_rate": 6.196152878930051e-09,
      "loss": 0.2878,
      "step": 11024
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.465510545059412,
      "learning_rate": 6.145695956860653e-09,
      "loss": 0.2636,
      "step": 11025
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.9733569788833805,
      "learning_rate": 6.095445064807082e-09,
      "loss": 0.2619,
      "step": 11026
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.360197836435571,
      "learning_rate": 6.045400206920737e-09,
      "loss": 0.3005,
      "step": 11027
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.4157461027991505,
      "learning_rate": 5.995561387336368e-09,
      "loss": 0.2629,
      "step": 11028
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.145124233173006,
      "learning_rate": 5.9459286101712365e-09,
      "loss": 0.306,
      "step": 11029
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.865551356749795,
      "learning_rate": 5.896501879525951e-09,
      "loss": 0.1354,
      "step": 11030
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.1179468886804615,
      "learning_rate": 5.847281199483635e-09,
      "loss": 0.2109,
      "step": 11031
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.079019447848224,
      "learning_rate": 5.798266574110478e-09,
      "loss": 0.0912,
      "step": 11032
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.685759084158897,
      "learning_rate": 5.749458007456576e-09,
      "loss": 0.4283,
      "step": 11033
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.674304644548105,
      "learning_rate": 5.700855503553704e-09,
      "loss": 0.2595,
      "step": 11034
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.318940513889842,
      "learning_rate": 5.652459066416982e-09,
      "loss": 0.3882,
      "step": 11035
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.174871307542799,
      "learning_rate": 5.604268700045157e-09,
      "loss": 0.2196,
      "step": 11036
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.692258339669711,
      "learning_rate": 5.5562844084192105e-09,
      "loss": 0.2807,
      "step": 11037
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.8812418008685605,
      "learning_rate": 5.508506195503749e-09,
      "loss": 0.1501,
      "step": 11038
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.755517097021527,
      "learning_rate": 5.460934065245338e-09,
      "loss": 0.1741,
      "step": 11039
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.565042323325527,
      "learning_rate": 5.4135680215747226e-09,
      "loss": 0.0797,
      "step": 11040
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.119333932327251,
      "learning_rate": 5.366408068404605e-09,
      "loss": 0.3716,
      "step": 11041
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.497467816750697,
      "learning_rate": 5.319454209631314e-09,
      "loss": 0.4147,
      "step": 11042
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3959121696234007,
      "learning_rate": 5.2727064491342465e-09,
      "loss": 0.2293,
      "step": 11043
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.701285101835407,
      "learning_rate": 5.226164790774757e-09,
      "loss": 0.2479,
      "step": 11044
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.572218552642889,
      "learning_rate": 5.17982923839866e-09,
      "loss": 0.1262,
      "step": 11045
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.049282961759825,
      "learning_rate": 5.133699795833447e-09,
      "loss": 0.368,
      "step": 11046
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.7665885648479085,
      "learning_rate": 5.087776466890237e-09,
      "loss": 0.3513,
      "step": 11047
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.306804415555556,
      "learning_rate": 5.042059255363219e-09,
      "loss": 0.1945,
      "step": 11048
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.156496939924493,
      "learning_rate": 4.996548165029092e-09,
      "loss": 0.2013,
      "step": 11049
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.956393073086782,
      "learning_rate": 4.951243199647904e-09,
      "loss": 0.4622,
      "step": 11050
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.235129785736757,
      "learning_rate": 4.906144362962217e-09,
      "loss": 0.2995,
      "step": 11051
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.608085561304136,
      "learning_rate": 4.861251658698218e-09,
      "loss": 0.2863,
      "step": 11052
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.526106160539211,
      "learning_rate": 4.816565090564607e-09,
      "loss": 0.2682,
      "step": 11053
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.996120884374608,
      "learning_rate": 4.7720846622531515e-09,
      "loss": 0.3186,
      "step": 11054
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.856464593213071,
      "learning_rate": 4.727810377438691e-09,
      "loss": 0.2889,
      "step": 11055
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.679977387320049,
      "learning_rate": 4.683742239778855e-09,
      "loss": 0.3776,
      "step": 11056
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3688164586695795,
      "learning_rate": 4.639880252914064e-09,
      "loss": 0.0569,
      "step": 11057
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.227162057040938,
      "learning_rate": 4.596224420468642e-09,
      "loss": 0.253,
      "step": 11058
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.65385890171314,
      "learning_rate": 4.552774746048871e-09,
      "loss": 0.3372,
      "step": 11059
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.055869485077352,
      "learning_rate": 4.5095312332441024e-09,
      "loss": 0.2561,
      "step": 11060
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.8119437665464257,
      "learning_rate": 4.4664938856273096e-09,
      "loss": 0.2365,
      "step": 11061
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.5408379548529205,
      "learning_rate": 4.423662706753706e-09,
      "loss": 0.2805,
      "step": 11062
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.2003326118817235,
      "learning_rate": 4.3810377001618496e-09,
      "loss": 0.1963,
      "step": 11063
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.028515018813822,
      "learning_rate": 4.3386188693733675e-09,
      "loss": 0.2553,
      "step": 11064
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.045234393162937,
      "learning_rate": 4.296406217892679e-09,
      "loss": 0.1904,
      "step": 11065
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.275467011494899,
      "learning_rate": 4.254399749207272e-09,
      "loss": 0.3433,
      "step": 11066
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.5173285281072744,
      "learning_rate": 4.212599466787148e-09,
      "loss": 0.1258,
      "step": 11067
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.1005163115479255,
      "learning_rate": 4.171005374086212e-09,
      "loss": 0.1778,
      "step": 11068
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.537085384753105,
      "learning_rate": 4.129617474540048e-09,
      "loss": 0.3133,
      "step": 11069
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.678545799007033,
      "learning_rate": 4.088435771568422e-09,
      "loss": 0.4378,
      "step": 11070
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.626762940982672,
      "learning_rate": 4.047460268573611e-09,
      "loss": 0.2857,
      "step": 11071
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.402150670969471,
      "learning_rate": 4.006690968940408e-09,
      "loss": 0.2411,
      "step": 11072
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.944470554322721,
      "learning_rate": 3.966127876037229e-09,
      "loss": 0.2934,
      "step": 11073
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.0774066347801226,
      "learning_rate": 3.925770993215283e-09,
      "loss": 0.3169,
      "step": 11074
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.82699123882394,
      "learning_rate": 3.885620323808293e-09,
      "loss": 0.155,
      "step": 11075
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.072924931185512,
      "learning_rate": 3.84567587113388e-09,
      "loss": 0.2973,
      "step": 11076
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.860906068567096,
      "learning_rate": 3.80593763849163e-09,
      "loss": 0.3585,
      "step": 11077
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.972412084639901,
      "learning_rate": 3.766405629164472e-09,
      "loss": 0.239,
      "step": 11078
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.9554512198847642,
      "learning_rate": 3.72707984641868e-09,
      "loss": 0.2098,
      "step": 11079
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.81762575147765,
      "learning_rate": 3.6879602935027683e-09,
      "loss": 0.2581,
      "step": 11080
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.789120511463075,
      "learning_rate": 3.6490469736491506e-09,
      "loss": 0.3218,
      "step": 11081
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.59258831719288,
      "learning_rate": 3.6103398900722008e-09,
      "loss": 0.3809,
      "step": 11082
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.652380736246634,
      "learning_rate": 3.571839045969638e-09,
      "loss": 0.4103,
      "step": 11083
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.689337922324584,
      "learning_rate": 3.53354444452253e-09,
      "loss": 0.1558,
      "step": 11084
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.742848971570284,
      "learning_rate": 3.4954560888944557e-09,
      "loss": 0.3058,
      "step": 11085
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.188553149239541,
      "learning_rate": 3.457573982232343e-09,
      "loss": 0.1148,
      "step": 11086
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.3851475727235605,
      "learning_rate": 3.4198981276653554e-09,
      "loss": 0.2255,
      "step": 11087
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.259470228110273,
      "learning_rate": 3.382428528306281e-09,
      "loss": 0.4074,
      "step": 11088
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.822863704431984,
      "learning_rate": 3.3451651872506984e-09,
      "loss": 0.1991,
      "step": 11089
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.579215689343271,
      "learning_rate": 3.308108107577257e-09,
      "loss": 0.1853,
      "step": 11090
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.4583091160415016,
      "learning_rate": 3.2712572923471186e-09,
      "loss": 0.1367,
      "step": 11091
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3174122709291156,
      "learning_rate": 3.23461274460507e-09,
      "loss": 0.1172,
      "step": 11092
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.528174965858616,
      "learning_rate": 3.198174467378412e-09,
      "loss": 0.1839,
      "step": 11093
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.428754556349996,
      "learning_rate": 3.1619424636772365e-09,
      "loss": 0.3489,
      "step": 11094
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.933244490526852,
      "learning_rate": 3.125916736494983e-09,
      "loss": 0.5211,
      "step": 11095
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.460182268239848,
      "learning_rate": 3.090097288808158e-09,
      "loss": 0.353,
      "step": 11096
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.688174007020816,
      "learning_rate": 3.0544841235755074e-09,
      "loss": 0.3932,
      "step": 11097
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.6938202322519755,
      "learning_rate": 3.019077243739954e-09,
      "loss": 0.2698,
      "step": 11098
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.693702175357452,
      "learning_rate": 2.9838766522258256e-09,
      "loss": 0.3514,
      "step": 11099
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.8758475593769193,
      "learning_rate": 2.948882351941629e-09,
      "loss": 0.2165,
      "step": 11100
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.0105154686999995,
      "learning_rate": 2.914094345778662e-09,
      "loss": 0.4832,
      "step": 11101
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.5744206562070637,
      "learning_rate": 2.87951263661046e-09,
      "loss": 0.1235,
      "step": 11102
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.9978604103414948,
      "learning_rate": 2.845137227294181e-09,
      "loss": 0.2594,
      "step": 11103
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.535910404418846,
      "learning_rate": 2.8109681206700545e-09,
      "loss": 0.2638,
      "step": 11104
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.660921639280941,
      "learning_rate": 2.7770053195602666e-09,
      "loss": 0.3245,
      "step": 11105
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4353182649446463,
      "learning_rate": 2.7432488267714607e-09,
      "loss": 0.248,
      "step": 11106
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.132288726493214,
      "learning_rate": 2.709698645091685e-09,
      "loss": 0.2073,
      "step": 11107
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.152970501799413,
      "learning_rate": 2.676354777293444e-09,
      "loss": 0.2147,
      "step": 11108
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.189993333656948,
      "learning_rate": 2.6432172261306453e-09,
      "loss": 0.2482,
      "step": 11109
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.562848951786321,
      "learning_rate": 2.610285994341655e-09,
      "loss": 0.5707,
      "step": 11110
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.781083770408325,
      "learning_rate": 2.5775610846467957e-09,
      "loss": 0.2232,
      "step": 11111
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.890423124430727,
      "learning_rate": 2.5450424997494616e-09,
      "loss": 0.2471,
      "step": 11112
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.36924554664695,
      "learning_rate": 2.512730242336392e-09,
      "loss": 0.3739,
      "step": 11113
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.92097787865467,
      "learning_rate": 2.480624315077118e-09,
      "loss": 0.3902,
      "step": 11114
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.084128966137284,
      "learning_rate": 2.4487247206239627e-09,
      "loss": 0.2396,
      "step": 11115
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.190212806188969,
      "learning_rate": 2.417031461612318e-09,
      "loss": 0.2987,
      "step": 11116
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.932460680551153,
      "learning_rate": 2.3855445406603672e-09,
      "loss": 0.064,
      "step": 11117
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.745164480100887,
      "learning_rate": 2.354263960369918e-09,
      "loss": 0.2868,
      "step": 11118
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.376765963308886,
      "learning_rate": 2.3231897233247368e-09,
      "loss": 0.0464,
      "step": 11119
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.295442253449569,
      "learning_rate": 2.292321832091937e-09,
      "loss": 0.2443,
      "step": 11120
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.5542717362380714,
      "learning_rate": 2.261660289222256e-09,
      "loss": 0.0462,
      "step": 11121
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.077383100843841,
      "learning_rate": 2.23120509724839e-09,
      "loss": 0.4169,
      "step": 11122
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.6419101110946226,
      "learning_rate": 2.200956258686382e-09,
      "loss": 0.1676,
      "step": 11123
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.324582930426615,
      "learning_rate": 2.1709137760353438e-09,
      "loss": 0.1359,
      "step": 11124
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.8526965646478235,
      "learning_rate": 2.1410776517769017e-09,
      "loss": 0.5523,
      "step": 11125
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.119845974414522,
      "learning_rate": 2.111447888376861e-09,
      "loss": 0.1373,
      "step": 11126
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7790211162042393,
      "learning_rate": 2.082024488282153e-09,
      "loss": 0.3694,
      "step": 11127
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.037211478937513,
      "learning_rate": 2.0528074539238885e-09,
      "loss": 0.3136,
      "step": 11128
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.883176374928471,
      "learning_rate": 2.0237967877159702e-09,
      "loss": 0.4158,
      "step": 11129
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.4207441003569565,
      "learning_rate": 1.9949924920550923e-09,
      "loss": 0.5016,
      "step": 11130
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.3498062288716115,
      "learning_rate": 1.9663945693210172e-09,
      "loss": 0.242,
      "step": 11131
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.777751333571241,
      "learning_rate": 1.938003021876023e-09,
      "loss": 0.2155,
      "step": 11132
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.305433804502702,
      "learning_rate": 1.9098178520660094e-09,
      "loss": 0.2637,
      "step": 11133
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.823336331120618,
      "learning_rate": 1.8818390622191152e-09,
      "loss": 0.469,
      "step": 11134
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.231664415435482,
      "learning_rate": 1.8540666546473796e-09,
      "loss": 0.134,
      "step": 11135
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.56620106858947,
      "learning_rate": 1.8265006316448008e-09,
      "loss": 0.2798,
      "step": 11136
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.2368115026467095,
      "learning_rate": 1.799140995488724e-09,
      "loss": 0.354,
      "step": 11137
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.286161522352478,
      "learning_rate": 1.7719877484398407e-09,
      "loss": 0.2447,
      "step": 11138
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9540421350732156,
      "learning_rate": 1.7450408927408014e-09,
      "loss": 0.0642,
      "step": 11139
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4687843692846974,
      "learning_rate": 1.7183004306184358e-09,
      "loss": 0.2458,
      "step": 11140
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.557626628935253,
      "learning_rate": 1.6917663642818104e-09,
      "loss": 0.2818,
      "step": 11141
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.35918934678696,
      "learning_rate": 1.6654386959227831e-09,
      "loss": 0.2495,
      "step": 11142
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.8825741776393023,
      "learning_rate": 1.639317427716558e-09,
      "loss": 0.2572,
      "step": 11143
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.058451075256056,
      "learning_rate": 1.6134025618208537e-09,
      "loss": 0.3603,
      "step": 11144
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.592332049824705,
      "learning_rate": 1.5876941003772905e-09,
      "loss": 0.1639,
      "step": 11145
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0079238278948774,
      "learning_rate": 1.5621920455091699e-09,
      "loss": 0.0632,
      "step": 11146
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.757436198407654,
      "learning_rate": 1.536896399323695e-09,
      "loss": 0.2743,
      "step": 11147
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.641781486555263,
      "learning_rate": 1.511807163910306e-09,
      "loss": 0.2773,
      "step": 11148
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.554489785784413,
      "learning_rate": 1.486924341342344e-09,
      "loss": 0.2731,
      "step": 11149
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.686520504759883,
      "learning_rate": 1.4622479336748318e-09,
      "loss": 0.3589,
      "step": 11150
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.402622379268168,
      "learning_rate": 1.4377779429466943e-09,
      "loss": 0.2182,
      "step": 11151
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.767398407457648,
      "learning_rate": 1.4135143711796473e-09,
      "loss": 0.2826,
      "step": 11152
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.176916191686973,
      "learning_rate": 1.3894572203781987e-09,
      "loss": 0.5001,
      "step": 11153
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.73900569864347,
      "learning_rate": 1.365606492529925e-09,
      "loss": 0.1827,
      "step": 11154
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.269959986683457,
      "learning_rate": 1.3419621896049173e-09,
      "loss": 0.1951,
      "step": 11155
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.498797011072601,
      "learning_rate": 1.3185243135566128e-09,
      "loss": 0.2258,
      "step": 11156
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.57777624264216,
      "learning_rate": 1.295292866321518e-09,
      "loss": 0.4545,
      "step": 11157
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.080568812387861,
      "learning_rate": 1.2722678498189311e-09,
      "loss": 0.2289,
      "step": 11158
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.50685142461918,
      "learning_rate": 1.2494492659509417e-09,
      "loss": 0.3559,
      "step": 11159
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.634282679949752,
      "learning_rate": 1.2268371166027083e-09,
      "loss": 0.242,
      "step": 11160
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.4217882890991556,
      "learning_rate": 1.2044314036424588e-09,
      "loss": 0.2727,
      "step": 11161
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.8271506616969906,
      "learning_rate": 1.1822321289209348e-09,
      "loss": 0.3246,
      "step": 11162
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.206780135846133,
      "learning_rate": 1.1602392942725027e-09,
      "loss": 0.1497,
      "step": 11163
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.267302367741353,
      "learning_rate": 1.1384529015140423e-09,
      "loss": 0.2416,
      "step": 11164
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.350017248243911,
      "learning_rate": 1.1168729524452249e-09,
      "loss": 0.2937,
      "step": 11165
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.082185249008869,
      "learning_rate": 1.095499448849069e-09,
      "loss": 0.2316,
      "step": 11166
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.286123927164004,
      "learning_rate": 1.0743323924911064e-09,
      "loss": 0.487,
      "step": 11167
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.294729803759721,
      "learning_rate": 1.0533717851204938e-09,
      "loss": 0.4014,
      "step": 11168
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7745412628387323,
      "learning_rate": 1.0326176284683466e-09,
      "loss": 0.2129,
      "step": 11169
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.7982254744796045,
      "learning_rate": 1.0120699242496812e-09,
      "loss": 0.139,
      "step": 11170
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7103414295944153,
      "learning_rate": 9.91728674162029e-10,
      "loss": 0.2838,
      "step": 11171
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7166073089291656,
      "learning_rate": 9.715938798857127e-10,
      "loss": 0.2569,
      "step": 11172
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.3895913790497008,
      "learning_rate": 9.516655430841238e-10,
      "loss": 0.1814,
      "step": 11173
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.8445696900747173,
      "learning_rate": 9.319436654037227e-10,
      "loss": 0.2843,
      "step": 11174
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.4705530336791135,
      "learning_rate": 9.124282484740398e-10,
      "loss": 0.229,
      "step": 11175
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.024921173043989,
      "learning_rate": 8.931192939071187e-10,
      "loss": 0.2958,
      "step": 11176
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.867646798053624,
      "learning_rate": 8.740168032980723e-10,
      "loss": 0.1576,
      "step": 11177
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.353722352523348,
      "learning_rate": 8.551207782253601e-10,
      "loss": 0.3805,
      "step": 11178
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.9312488685766445,
      "learning_rate": 8.364312202499558e-10,
      "loss": 0.2533,
      "step": 11179
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7377889946304093,
      "learning_rate": 8.17948130915902e-10,
      "loss": 0.304,
      "step": 11180
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.5252160936873094,
      "learning_rate": 7.996715117500331e-10,
      "loss": 0.3315,
      "step": 11181
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.15967923494784,
      "learning_rate": 7.816013642622522e-10,
      "loss": 0.2175,
      "step": 11182
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.173991999040336,
      "learning_rate": 7.637376899458094e-10,
      "loss": 0.2896,
      "step": 11183
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.251486105551536,
      "learning_rate": 7.460804902761909e-10,
      "loss": 0.2028,
      "step": 11184
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.743049578079321,
      "learning_rate": 7.286297667119524e-10,
      "loss": 0.3648,
      "step": 11185
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.2838651876548575,
      "learning_rate": 7.113855206952736e-10,
      "loss": 0.3516,
      "step": 11186
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.7030632924133875,
      "learning_rate": 6.943477536505705e-10,
      "loss": 0.1935,
      "step": 11187
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.56406788597246,
      "learning_rate": 6.77516466985606e-10,
      "loss": 0.2315,
      "step": 11188
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.512492888407406,
      "learning_rate": 6.608916620903793e-10,
      "loss": 0.2496,
      "step": 11189
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.089679199837506,
      "learning_rate": 6.44473340339069e-10,
      "loss": 0.4453,
      "step": 11190
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.560868646488458,
      "learning_rate": 6.282615030872574e-10,
      "loss": 0.2802,
      "step": 11191
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.384038625233684,
      "learning_rate": 6.122561516749837e-10,
      "loss": 0.3705,
      "step": 11192
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.4108538250478,
      "learning_rate": 5.964572874242458e-10,
      "loss": 0.411,
      "step": 11193
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.429155850416151,
      "learning_rate": 5.808649116403886e-10,
      "loss": 0.5711,
      "step": 11194
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.435826944801256,
      "learning_rate": 5.654790256112708e-10,
      "loss": 0.1788,
      "step": 11195
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.685536330181899,
      "learning_rate": 5.502996306083752e-10,
      "loss": 0.3297,
      "step": 11196
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.847364325593019,
      "learning_rate": 5.353267278854213e-10,
      "loss": 0.2426,
      "step": 11197
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.999196002086301,
      "learning_rate": 5.20560318679475e-10,
      "loss": 0.2693,
      "step": 11198
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.9209602348714956,
      "learning_rate": 5.06000404210949e-10,
      "loss": 0.2013,
      "step": 11199
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.847436763525256,
      "learning_rate": 4.916469856819373e-10,
      "loss": 0.3532,
      "step": 11200
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.268548382833926,
      "learning_rate": 4.775000642787131e-10,
      "loss": 0.3535,
      "step": 11201
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.375226262877211,
      "learning_rate": 4.635596411700638e-10,
      "loss": 0.2506,
      "step": 11202
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.6519922936245455,
      "learning_rate": 4.4982571750729065e-10,
      "loss": 0.2629,
      "step": 11203
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.345526462177946,
      "learning_rate": 4.3629829442559667e-10,
      "loss": 0.3754,
      "step": 11204
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.924165340060216,
      "learning_rate": 4.229773730421438e-10,
      "loss": 0.35,
      "step": 11205
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.6914655101841123,
      "learning_rate": 4.098629544574406e-10,
      "loss": 0.2032,
      "step": 11206
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.388914422461834,
      "learning_rate": 3.9695503975506476e-10,
      "loss": 0.2586,
      "step": 11207
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.959794390377846,
      "learning_rate": 3.842536300013855e-10,
      "loss": 0.184,
      "step": 11208
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.6180320432096087,
      "learning_rate": 3.717587262455635e-10,
      "loss": 0.1189,
      "step": 11209
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.747531311784399,
      "learning_rate": 3.5947032952038385e-10,
      "loss": 0.367,
      "step": 11210
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.565374129785444,
      "learning_rate": 3.473884408403128e-10,
      "loss": 0.2244,
      "step": 11211
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.709541600999806,
      "learning_rate": 3.3551306120399586e-10,
      "loss": 0.2727,
      "step": 11212
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.6157767338137665,
      "learning_rate": 3.2384419159259275e-10,
      "loss": 0.1945,
      "step": 11213
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.254093450728184,
      "learning_rate": 3.1238183296977697e-10,
      "loss": 0.1046,
      "step": 11214
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.037280198671992,
      "learning_rate": 3.0112598628256885e-10,
      "loss": 0.291,
      "step": 11215
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.22339088624665,
      "learning_rate": 2.9007665246105765e-10,
      "loss": 0.1252,
      "step": 11216
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.13218162694251,
      "learning_rate": 2.7923383241784674e-10,
      "loss": 0.3474,
      "step": 11217
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.788800046507825,
      "learning_rate": 2.6859752704916364e-10,
      "loss": 0.3335,
      "step": 11218
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.140851040102807,
      "learning_rate": 2.5816773723319476e-10,
      "loss": 0.2136,
      "step": 11219
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.2109744107802825,
      "learning_rate": 2.4794446383202833e-10,
      "loss": 0.1693,
      "step": 11220
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.980425921009694,
      "learning_rate": 2.379277076897113e-10,
      "loss": 0.1884,
      "step": 11221
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.088473424200426,
      "learning_rate": 2.281174696344701e-10,
      "loss": 0.1752,
      "step": 11222
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.745460784807124,
      "learning_rate": 2.1851375047649003e-10,
      "loss": 0.2621,
      "step": 11223
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.504846026382764,
      "learning_rate": 2.0911655100874783e-10,
      "loss": 0.2369,
      "step": 11224
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.166560061159065,
      "learning_rate": 1.9992587200839963e-10,
      "loss": 0.3394,
      "step": 11225
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.4769897420977616,
      "learning_rate": 1.9094171423400532e-10,
      "loss": 0.1816,
      "step": 11226
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9361548467634995,
      "learning_rate": 1.8216407842830407e-10,
      "loss": 0.1878,
      "step": 11227
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.389459309463192,
      "learning_rate": 1.7359296531627157e-10,
      "loss": 0.2488,
      "step": 11228
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.135568885730241,
      "learning_rate": 1.652283756059525e-10,
      "loss": 0.0875,
      "step": 11229
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.49203310009099,
      "learning_rate": 1.5707030998818317e-10,
      "loss": 0.4143,
      "step": 11230
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.368009168546814,
      "learning_rate": 1.4911876913742407e-10,
      "loss": 0.3777,
      "step": 11231
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.1966456649058435,
      "learning_rate": 1.413737537103721e-10,
      "loss": 0.4273,
      "step": 11232
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.15276477232086,
      "learning_rate": 1.3383526434707083e-10,
      "loss": 0.2189,
      "step": 11233
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.153062248359868,
      "learning_rate": 1.265033016698003e-10,
      "loss": 0.3827,
      "step": 11234
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.2504280430445,
      "learning_rate": 1.1937786628474224e-10,
      "loss": 0.4348,
      "step": 11235
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.703496373447613,
      "learning_rate": 1.1245895878031487e-10,
      "loss": 0.3161,
      "step": 11236
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.685033423781793,
      "learning_rate": 1.0574657972828306e-10,
      "loss": 0.1819,
      "step": 11237
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9505442370827653,
      "learning_rate": 9.92407296832032e-11,
      "loss": 0.2483,
      "step": 11238
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.311323949288005,
      "learning_rate": 9.294140918242323e-11,
      "loss": 0.2967,
      "step": 11239
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.4250753219715726,
      "learning_rate": 8.684861874663775e-11,
      "loss": 0.2185,
      "step": 11240
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.894352927233986,
      "learning_rate": 8.096235887877779e-11,
      "loss": 0.271,
      "step": 11241
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.557749860447504,
      "learning_rate": 7.52826300653986e-11,
      "loss": 0.2367,
      "step": 11242
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.726126504570686,
      "learning_rate": 6.980943277584696e-11,
      "loss": 0.291,
      "step": 11243
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.390017117118707,
      "learning_rate": 6.454276746198362e-11,
      "loss": 0.239,
      "step": 11244
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.7296206693727063,
      "learning_rate": 5.948263455929359e-11,
      "loss": 0.3548,
      "step": 11245
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.713372445673693,
      "learning_rate": 5.4629034485498275e-11,
      "loss": 0.2209,
      "step": 11246
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.26612072837066,
      "learning_rate": 4.9981967641388185e-11,
      "loss": 0.3019,
      "step": 11247
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.090270966672931,
      "learning_rate": 4.5541434411655595e-11,
      "loss": 0.1487,
      "step": 11248
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.938118596816308,
      "learning_rate": 4.130743516239655e-11,
      "loss": 0.0085,
      "step": 11249
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.778904271486067,
      "learning_rate": 3.7279970243608854e-11,
      "loss": 0.2519,
      "step": 11250
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.815793763048347,
      "learning_rate": 3.345903998808187e-11,
      "loss": 0.1917,
      "step": 11251
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5043163495918717,
      "learning_rate": 2.984464471167403e-11,
      "loss": 0.2127,
      "step": 11252
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.600406285680922,
      "learning_rate": 2.643678471275779e-11,
      "loss": 0.155,
      "step": 11253
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.239071521725727,
      "learning_rate": 2.3235460273052235e-11,
      "loss": 0.4328,
      "step": 11254
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.71801497818678,
      "learning_rate": 2.0240671656790446e-11,
      "loss": 0.3839,
      "step": 11255
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.315153441740194,
      "learning_rate": 1.74524191112746e-11,
      "loss": 0.2258,
      "step": 11256
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9791362306601816,
      "learning_rate": 1.4870702867431086e-11,
      "loss": 0.3473,
      "step": 11257
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.873156909353834,
      "learning_rate": 1.2495523137867615e-11,
      "loss": 0.1876,
      "step": 11258
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8866674469437428,
      "learning_rate": 1.032688011937122e-11,
      "loss": 0.0561,
      "step": 11259
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.4295098585808965,
      "learning_rate": 8.364773990965358e-12,
      "loss": 0.1999,
      "step": 11260
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.233060929211409,
      "learning_rate": 6.609204914465039e-12,
      "loss": 0.2556,
      "step": 11261
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.009414919953733,
      "learning_rate": 5.0601730353094795e-12,
      "loss": 0.3259,
      "step": 11262
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.036550230053096,
      "learning_rate": 3.717678481174325e-12,
      "loss": 0.4948,
      "step": 11263
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0671348905018427,
      "learning_rate": 2.5817213628043237e-12,
      "loss": 0.0681,
      "step": 11264
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.161755249477581,
      "learning_rate": 1.6523017745684322e-12,
      "loss": 0.5151,
      "step": 11265
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.2360435722797547,
      "learning_rate": 9.294197930720394e-13,
      "loss": 0.2183,
      "step": 11266
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.754278399666144,
      "learning_rate": 4.1307547771207714e-13,
      "loss": 0.2861,
      "step": 11267
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.280715106752542,
      "learning_rate": 1.0326887150968745e-13,
      "loss": 0.2185,
      "step": 11268
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.5458029549960175,
      "learning_rate": 0.0,
      "loss": 0.2329,
      "step": 11269
    },
    {
      "epoch": 1.0,
      "step": 11269,
      "total_flos": 25789582226560.0,
      "train_loss": 0.305097568407725,
      "train_runtime": 32921.6822,
      "train_samples_per_second": 1.369,
      "train_steps_per_second": 0.342
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 11269,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 25789582226560.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}