{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5118,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00019538882375928098,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 18.159,
      "step": 1
    },
    {
      "epoch": 0.00039077764751856197,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 18.2968,
      "step": 2
    },
    {
      "epoch": 0.0005861664712778429,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 18.881,
      "step": 3
    },
    {
      "epoch": 0.0007815552950371239,
      "grad_norm": 16.800939559936523,
      "learning_rate": 1.298701298701299e-07,
      "loss": 18.863,
      "step": 4
    },
    {
      "epoch": 0.0009769441187964048,
      "grad_norm": 16.903465270996094,
      "learning_rate": 2.597402597402598e-07,
      "loss": 19.0107,
      "step": 5
    },
    {
      "epoch": 0.0011723329425556857,
      "grad_norm": 17.979801177978516,
      "learning_rate": 3.896103896103896e-07,
      "loss": 18.9991,
      "step": 6
    },
    {
      "epoch": 0.0013677217663149667,
      "grad_norm": 17.979801177978516,
      "learning_rate": 3.896103896103896e-07,
      "loss": 18.9156,
      "step": 7
    },
    {
      "epoch": 0.0015631105900742479,
      "grad_norm": 19.2469425201416,
      "learning_rate": 5.194805194805196e-07,
      "loss": 19.1087,
      "step": 8
    },
    {
      "epoch": 0.0017584994138335288,
      "grad_norm": 19.234683990478516,
      "learning_rate": 6.493506493506493e-07,
      "loss": 19.2963,
      "step": 9
    },
    {
      "epoch": 0.0019538882375928096,
      "grad_norm": 15.376220703125,
      "learning_rate": 7.792207792207792e-07,
      "loss": 17.9463,
      "step": 10
    },
    {
      "epoch": 0.0021492770613520907,
      "grad_norm": 15.44509506225586,
      "learning_rate": 9.090909090909091e-07,
      "loss": 17.3,
      "step": 11
    },
    {
      "epoch": 0.0023446658851113715,
      "grad_norm": 17.092782974243164,
      "learning_rate": 1.0389610389610392e-06,
      "loss": 19.3398,
      "step": 12
    },
    {
      "epoch": 0.0025400547088706526,
      "grad_norm": 18.010583877563477,
      "learning_rate": 1.168831168831169e-06,
      "loss": 18.7621,
      "step": 13
    },
    {
      "epoch": 0.0027354435326299334,
      "grad_norm": 15.407889366149902,
      "learning_rate": 1.2987012987012986e-06,
      "loss": 18.2084,
      "step": 14
    },
    {
      "epoch": 0.0029308323563892145,
      "grad_norm": 15.01679801940918,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 17.2003,
      "step": 15
    },
    {
      "epoch": 0.0031262211801484957,
      "grad_norm": 16.848499298095703,
      "learning_rate": 1.5584415584415584e-06,
      "loss": 20.4405,
      "step": 16
    },
    {
      "epoch": 0.0033216100039077765,
      "grad_norm": 15.197953224182129,
      "learning_rate": 1.6883116883116885e-06,
      "loss": 16.384,
      "step": 17
    },
    {
      "epoch": 0.0035169988276670576,
      "grad_norm": 16.76958465576172,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 17.7668,
      "step": 18
    },
    {
      "epoch": 0.0037123876514263384,
      "grad_norm": 18.03081512451172,
      "learning_rate": 1.9480519480519483e-06,
      "loss": 19.181,
      "step": 19
    },
    {
      "epoch": 0.003907776475185619,
      "grad_norm": 17.887229919433594,
      "learning_rate": 2.0779220779220784e-06,
      "loss": 18.8617,
      "step": 20
    },
    {
      "epoch": 0.004103165298944901,
      "grad_norm": 17.887229919433594,
      "learning_rate": 2.0779220779220784e-06,
      "loss": 18.5894,
      "step": 21
    },
    {
      "epoch": 0.0042985541227041815,
      "grad_norm": 15.389242172241211,
      "learning_rate": 2.207792207792208e-06,
      "loss": 17.2314,
      "step": 22
    },
    {
      "epoch": 0.004493942946463462,
      "grad_norm": 14.811219215393066,
      "learning_rate": 2.337662337662338e-06,
      "loss": 16.7673,
      "step": 23
    },
    {
      "epoch": 0.004689331770222743,
      "grad_norm": 17.104415893554688,
      "learning_rate": 2.4675324675324676e-06,
      "loss": 17.2118,
      "step": 24
    },
    {
      "epoch": 0.0048847205939820245,
      "grad_norm": 17.192047119140625,
      "learning_rate": 2.597402597402597e-06,
      "loss": 15.7282,
      "step": 25
    },
    {
      "epoch": 0.005080109417741305,
      "grad_norm": 18.115428924560547,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 18.8801,
      "step": 26
    },
    {
      "epoch": 0.005275498241500586,
      "grad_norm": 16.415836334228516,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 16.1387,
      "step": 27
    },
    {
      "epoch": 0.005470887065259867,
      "grad_norm": 15.510481834411621,
      "learning_rate": 2.9870129870129873e-06,
      "loss": 15.881,
      "step": 28
    },
    {
      "epoch": 0.005666275889019148,
      "grad_norm": 14.883086204528809,
      "learning_rate": 3.116883116883117e-06,
      "loss": 17.2384,
      "step": 29
    },
    {
      "epoch": 0.005861664712778429,
      "grad_norm": 17.018898010253906,
      "learning_rate": 3.246753246753247e-06,
      "loss": 17.8932,
      "step": 30
    },
    {
      "epoch": 0.00605705353653771,
      "grad_norm": 17.309856414794922,
      "learning_rate": 3.376623376623377e-06,
      "loss": 17.3969,
      "step": 31
    },
    {
      "epoch": 0.0062524423602969914,
      "grad_norm": 16.117368698120117,
      "learning_rate": 3.506493506493507e-06,
      "loss": 17.9658,
      "step": 32
    },
    {
      "epoch": 0.006447831184056272,
      "grad_norm": 11.697212219238281,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 13.7439,
      "step": 33
    },
    {
      "epoch": 0.006643220007815553,
      "grad_norm": 12.093951225280762,
      "learning_rate": 3.7662337662337666e-06,
      "loss": 15.5621,
      "step": 34
    },
    {
      "epoch": 0.006838608831574834,
      "grad_norm": 15.0689058303833,
      "learning_rate": 3.896103896103897e-06,
      "loss": 17.4019,
      "step": 35
    },
    {
      "epoch": 0.007033997655334115,
      "grad_norm": 13.318926811218262,
      "learning_rate": 4.025974025974026e-06,
      "loss": 16.8068,
      "step": 36
    },
    {
      "epoch": 0.007229386479093396,
      "grad_norm": 15.324155807495117,
      "learning_rate": 4.155844155844157e-06,
      "loss": 16.4426,
      "step": 37
    },
    {
      "epoch": 0.007424775302852677,
      "grad_norm": 14.826874732971191,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 16.845,
      "step": 38
    },
    {
      "epoch": 0.0076201641266119575,
      "grad_norm": 13.863877296447754,
      "learning_rate": 4.415584415584416e-06,
      "loss": 17.7616,
      "step": 39
    },
    {
      "epoch": 0.007815552950371238,
      "grad_norm": 17.50107192993164,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 17.392,
      "step": 40
    },
    {
      "epoch": 0.008010941774130519,
      "grad_norm": 16.738174438476562,
      "learning_rate": 4.675324675324676e-06,
      "loss": 16.4532,
      "step": 41
    },
    {
      "epoch": 0.008206330597889801,
      "grad_norm": 9.62713623046875,
      "learning_rate": 4.805194805194806e-06,
      "loss": 15.1813,
      "step": 42
    },
    {
      "epoch": 0.008401719421649082,
      "grad_norm": 14.606769561767578,
      "learning_rate": 4.935064935064935e-06,
      "loss": 17.5544,
      "step": 43
    },
    {
      "epoch": 0.008597108245408363,
      "grad_norm": 14.058442115783691,
      "learning_rate": 5.064935064935065e-06,
      "loss": 15.967,
      "step": 44
    },
    {
      "epoch": 0.008792497069167644,
      "grad_norm": 15.973811149597168,
      "learning_rate": 5.194805194805194e-06,
      "loss": 17.6413,
      "step": 45
    },
    {
      "epoch": 0.008987885892926924,
      "grad_norm": 14.27064037322998,
      "learning_rate": 5.324675324675325e-06,
      "loss": 16.7137,
      "step": 46
    },
    {
      "epoch": 0.009183274716686205,
      "grad_norm": 12.279301643371582,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 14.5723,
      "step": 47
    },
    {
      "epoch": 0.009378663540445486,
      "grad_norm": 13.54649543762207,
      "learning_rate": 5.584415584415585e-06,
      "loss": 15.7591,
      "step": 48
    },
    {
      "epoch": 0.009574052364204768,
      "grad_norm": 11.76611042022705,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 15.5008,
      "step": 49
    },
    {
      "epoch": 0.009769441187964049,
      "grad_norm": 11.76115894317627,
      "learning_rate": 5.844155844155844e-06,
      "loss": 14.9208,
      "step": 50
    },
    {
      "epoch": 0.00996483001172333,
      "grad_norm": 12.934619903564453,
      "learning_rate": 5.9740259740259746e-06,
      "loss": 14.8215,
      "step": 51
    },
    {
      "epoch": 0.01016021883548261,
      "grad_norm": 11.615428924560547,
      "learning_rate": 6.103896103896104e-06,
      "loss": 14.7069,
      "step": 52
    },
    {
      "epoch": 0.010355607659241891,
      "grad_norm": 11.138904571533203,
      "learning_rate": 6.233766233766234e-06,
      "loss": 15.4979,
      "step": 53
    },
    {
      "epoch": 0.010550996483001172,
      "grad_norm": 11.6993408203125,
      "learning_rate": 6.363636363636364e-06,
      "loss": 15.0529,
      "step": 54
    },
    {
      "epoch": 0.010746385306760453,
      "grad_norm": 12.620027542114258,
      "learning_rate": 6.493506493506494e-06,
      "loss": 16.0377,
      "step": 55
    },
    {
      "epoch": 0.010941774130519734,
      "grad_norm": 11.459622383117676,
      "learning_rate": 6.623376623376624e-06,
      "loss": 14.9392,
      "step": 56
    },
    {
      "epoch": 0.011137162954279016,
      "grad_norm": 11.75425910949707,
      "learning_rate": 6.753246753246754e-06,
      "loss": 14.584,
      "step": 57
    },
    {
      "epoch": 0.011332551778038297,
      "grad_norm": 11.240898132324219,
      "learning_rate": 6.8831168831168835e-06,
      "loss": 12.9321,
      "step": 58
    },
    {
      "epoch": 0.011527940601797577,
      "grad_norm": 12.680749893188477,
      "learning_rate": 7.012987012987014e-06,
      "loss": 15.2367,
      "step": 59
    },
    {
      "epoch": 0.011723329425556858,
      "grad_norm": 12.971368789672852,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 14.4319,
      "step": 60
    },
    {
      "epoch": 0.011918718249316139,
      "grad_norm": 12.750476837158203,
      "learning_rate": 7.272727272727273e-06,
      "loss": 15.405,
      "step": 61
    },
    {
      "epoch": 0.01211410707307542,
      "grad_norm": 13.858943939208984,
      "learning_rate": 7.402597402597404e-06,
      "loss": 14.7354,
      "step": 62
    },
    {
      "epoch": 0.0123094958968347,
      "grad_norm": 11.960023880004883,
      "learning_rate": 7.532467532467533e-06,
      "loss": 14.515,
      "step": 63
    },
    {
      "epoch": 0.012504884720593983,
      "grad_norm": 13.708508491516113,
      "learning_rate": 7.662337662337663e-06,
      "loss": 14.8207,
      "step": 64
    },
    {
      "epoch": 0.012700273544353264,
      "grad_norm": 11.947013854980469,
      "learning_rate": 7.792207792207793e-06,
      "loss": 14.1906,
      "step": 65
    },
    {
      "epoch": 0.012895662368112544,
      "grad_norm": 13.315733909606934,
      "learning_rate": 7.922077922077924e-06,
      "loss": 13.5782,
      "step": 66
    },
    {
      "epoch": 0.013091051191871825,
      "grad_norm": 11.261037826538086,
      "learning_rate": 8.051948051948052e-06,
      "loss": 13.8077,
      "step": 67
    },
    {
      "epoch": 0.013286440015631106,
      "grad_norm": 11.723245620727539,
      "learning_rate": 8.181818181818183e-06,
      "loss": 12.6595,
      "step": 68
    },
    {
      "epoch": 0.013481828839390387,
      "grad_norm": 13.365504264831543,
      "learning_rate": 8.311688311688313e-06,
      "loss": 15.6787,
      "step": 69
    },
    {
      "epoch": 0.013677217663149667,
      "grad_norm": 13.995412826538086,
      "learning_rate": 8.441558441558442e-06,
      "loss": 14.7963,
      "step": 70
    },
    {
      "epoch": 0.013872606486908948,
      "grad_norm": 12.938873291015625,
      "learning_rate": 8.571428571428571e-06,
      "loss": 13.5083,
      "step": 71
    },
    {
      "epoch": 0.01406799531066823,
      "grad_norm": 12.961465835571289,
      "learning_rate": 8.701298701298701e-06,
      "loss": 15.0235,
      "step": 72
    },
    {
      "epoch": 0.014263384134427511,
      "grad_norm": 12.395155906677246,
      "learning_rate": 8.831168831168832e-06,
      "loss": 14.7144,
      "step": 73
    },
    {
      "epoch": 0.014458772958186792,
      "grad_norm": 12.128456115722656,
      "learning_rate": 8.96103896103896e-06,
      "loss": 13.4361,
      "step": 74
    },
    {
      "epoch": 0.014654161781946073,
      "grad_norm": 12.000514030456543,
      "learning_rate": 9.090909090909091e-06,
      "loss": 12.8412,
      "step": 75
    },
    {
      "epoch": 0.014849550605705353,
      "grad_norm": 14.545674324035645,
      "learning_rate": 9.220779220779221e-06,
      "loss": 14.0329,
      "step": 76
    },
    {
      "epoch": 0.015044939429464634,
      "grad_norm": 15.808619499206543,
      "learning_rate": 9.350649350649352e-06,
      "loss": 14.0651,
      "step": 77
    },
    {
      "epoch": 0.015240328253223915,
      "grad_norm": 14.302594184875488,
      "learning_rate": 9.48051948051948e-06,
      "loss": 14.3772,
      "step": 78
    },
    {
      "epoch": 0.015435717076983196,
      "grad_norm": 14.302594184875488,
      "learning_rate": 9.48051948051948e-06,
      "loss": 14.3133,
      "step": 79
    },
    {
      "epoch": 0.015631105900742476,
      "grad_norm": 14.733464241027832,
      "learning_rate": 9.610389610389611e-06,
      "loss": 13.9054,
      "step": 80
    },
    {
      "epoch": 0.015826494724501757,
      "grad_norm": 14.733464241027832,
      "learning_rate": 9.610389610389611e-06,
      "loss": 14.3721,
      "step": 81
    },
    {
      "epoch": 0.016021883548261038,
      "grad_norm": 13.027717590332031,
      "learning_rate": 9.740259740259742e-06,
      "loss": 13.9081,
      "step": 82
    },
    {
      "epoch": 0.016217272372020322,
      "grad_norm": 13.918047904968262,
      "learning_rate": 9.87012987012987e-06,
      "loss": 14.737,
      "step": 83
    },
    {
      "epoch": 0.016412661195779603,
      "grad_norm": 12.087900161743164,
      "learning_rate": 1e-05,
      "loss": 13.6766,
      "step": 84
    },
    {
      "epoch": 0.016608050019538884,
      "grad_norm": 12.087900161743164,
      "learning_rate": 1e-05,
      "loss": 14.2403,
      "step": 85
    },
    {
      "epoch": 0.016803438843298164,
      "grad_norm": 13.96923828125,
      "learning_rate": 1.012987012987013e-05,
      "loss": 13.2121,
      "step": 86
    },
    {
      "epoch": 0.016998827667057445,
      "grad_norm": 11.908881187438965,
      "learning_rate": 1.025974025974026e-05,
      "loss": 13.5275,
      "step": 87
    },
    {
      "epoch": 0.017194216490816726,
      "grad_norm": 12.391776084899902,
      "learning_rate": 1.0389610389610389e-05,
      "loss": 14.2606,
      "step": 88
    },
    {
      "epoch": 0.017389605314576007,
      "grad_norm": 12.35890007019043,
      "learning_rate": 1.0519480519480521e-05,
      "loss": 14.1348,
      "step": 89
    },
    {
      "epoch": 0.017584994138335287,
      "grad_norm": 12.73941421508789,
      "learning_rate": 1.064935064935065e-05,
      "loss": 13.7867,
      "step": 90
    },
    {
      "epoch": 0.017780382962094568,
      "grad_norm": 12.822901725769043,
      "learning_rate": 1.077922077922078e-05,
      "loss": 13.2284,
      "step": 91
    },
    {
      "epoch": 0.01797577178585385,
      "grad_norm": 11.797019004821777,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 12.2814,
      "step": 92
    },
    {
      "epoch": 0.01817116060961313,
      "grad_norm": 10.80223560333252,
      "learning_rate": 1.1038961038961041e-05,
      "loss": 12.0659,
      "step": 93
    },
    {
      "epoch": 0.01836654943337241,
      "grad_norm": 12.917786598205566,
      "learning_rate": 1.116883116883117e-05,
      "loss": 15.0527,
      "step": 94
    },
    {
      "epoch": 0.01856193825713169,
      "grad_norm": 11.387985229492188,
      "learning_rate": 1.12987012987013e-05,
      "loss": 13.0093,
      "step": 95
    },
    {
      "epoch": 0.01875732708089097,
      "grad_norm": 12.817272186279297,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 13.1882,
      "step": 96
    },
    {
      "epoch": 0.018952715904650252,
      "grad_norm": 13.179692268371582,
      "learning_rate": 1.155844155844156e-05,
      "loss": 13.6843,
      "step": 97
    },
    {
      "epoch": 0.019148104728409537,
      "grad_norm": 13.599560737609863,
      "learning_rate": 1.1688311688311688e-05,
      "loss": 13.4325,
      "step": 98
    },
    {
      "epoch": 0.019343493552168817,
      "grad_norm": 12.171883583068848,
      "learning_rate": 1.181818181818182e-05,
      "loss": 13.1249,
      "step": 99
    },
    {
      "epoch": 0.019538882375928098,
      "grad_norm": 12.852227210998535,
      "learning_rate": 1.1948051948051949e-05,
      "loss": 11.6582,
      "step": 100
    },
    {
      "epoch": 0.01973427119968738,
      "grad_norm": 14.056069374084473,
      "learning_rate": 1.207792207792208e-05,
      "loss": 14.5759,
      "step": 101
    },
    {
      "epoch": 0.01992966002344666,
      "grad_norm": 12.38291072845459,
      "learning_rate": 1.2207792207792208e-05,
      "loss": 13.5927,
      "step": 102
    },
    {
      "epoch": 0.02012504884720594,
      "grad_norm": 11.649094581604004,
      "learning_rate": 1.2337662337662339e-05,
      "loss": 13.3672,
      "step": 103
    },
    {
      "epoch": 0.02032043767096522,
      "grad_norm": 12.83110237121582,
      "learning_rate": 1.2467532467532468e-05,
      "loss": 14.0482,
      "step": 104
    },
    {
      "epoch": 0.020515826494724502,
      "grad_norm": 13.384620666503906,
      "learning_rate": 1.25974025974026e-05,
      "loss": 12.7853,
      "step": 105
    },
    {
      "epoch": 0.020711215318483783,
      "grad_norm": 17.137697219848633,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 13.875,
      "step": 106
    },
    {
      "epoch": 0.020906604142243063,
      "grad_norm": 11.776130676269531,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 12.762,
      "step": 107
    },
    {
      "epoch": 0.021101992966002344,
      "grad_norm": 13.227246284484863,
      "learning_rate": 1.2987012987012988e-05,
      "loss": 13.6374,
      "step": 108
    },
    {
      "epoch": 0.021297381789761625,
      "grad_norm": 11.719467163085938,
      "learning_rate": 1.311688311688312e-05,
      "loss": 13.027,
      "step": 109
    },
    {
      "epoch": 0.021492770613520906,
      "grad_norm": 11.77922248840332,
      "learning_rate": 1.3246753246753249e-05,
      "loss": 12.9044,
      "step": 110
    },
    {
      "epoch": 0.021688159437280186,
      "grad_norm": 13.76087760925293,
      "learning_rate": 1.3376623376623377e-05,
      "loss": 12.3403,
      "step": 111
    },
    {
      "epoch": 0.021883548261039467,
      "grad_norm": 10.988027572631836,
      "learning_rate": 1.3506493506493508e-05,
      "loss": 12.9417,
      "step": 112
    },
    {
      "epoch": 0.02207893708479875,
      "grad_norm": 13.056401252746582,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 12.7452,
      "step": 113
    },
    {
      "epoch": 0.022274325908558032,
      "grad_norm": 9.933206558227539,
      "learning_rate": 1.3766233766233767e-05,
      "loss": 13.4601,
      "step": 114
    },
    {
      "epoch": 0.022469714732317313,
      "grad_norm": 13.234012603759766,
      "learning_rate": 1.3896103896103896e-05,
      "loss": 13.0663,
      "step": 115
    },
    {
      "epoch": 0.022665103556076593,
      "grad_norm": 13.284384727478027,
      "learning_rate": 1.4025974025974028e-05,
      "loss": 12.7403,
      "step": 116
    },
    {
      "epoch": 0.022860492379835874,
      "grad_norm": 14.263381958007812,
      "learning_rate": 1.4155844155844157e-05,
      "loss": 14.07,
      "step": 117
    },
    {
      "epoch": 0.023055881203595155,
      "grad_norm": 17.6854248046875,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 13.6625,
      "step": 118
    },
    {
      "epoch": 0.023251270027354436,
      "grad_norm": 7.057964324951172,
      "learning_rate": 1.4415584415584416e-05,
      "loss": 11.2206,
      "step": 119
    },
    {
      "epoch": 0.023446658851113716,
      "grad_norm": 11.81092357635498,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 11.8661,
      "step": 120
    },
    {
      "epoch": 0.023642047674872997,
      "grad_norm": 10.942760467529297,
      "learning_rate": 1.4675324675324675e-05,
      "loss": 12.795,
      "step": 121
    },
    {
      "epoch": 0.023837436498632278,
      "grad_norm": 10.352446556091309,
      "learning_rate": 1.4805194805194807e-05,
      "loss": 12.6598,
      "step": 122
    },
    {
      "epoch": 0.02403282532239156,
      "grad_norm": 8.63533878326416,
      "learning_rate": 1.4935064935064936e-05,
      "loss": 12.3454,
      "step": 123
    },
    {
      "epoch": 0.02422821414615084,
      "grad_norm": 7.654079437255859,
      "learning_rate": 1.5064935064935066e-05,
      "loss": 14.233,
      "step": 124
    },
    {
      "epoch": 0.02442360296991012,
      "grad_norm": 7.536815166473389,
      "learning_rate": 1.5194805194805195e-05,
      "loss": 13.4265,
      "step": 125
    },
    {
      "epoch": 0.0246189917936694,
      "grad_norm": 7.602650165557861,
      "learning_rate": 1.5324675324675326e-05,
      "loss": 11.8791,
      "step": 126
    },
    {
      "epoch": 0.02481438061742868,
      "grad_norm": 10.716212272644043,
      "learning_rate": 1.5454545454545454e-05,
      "loss": 13.0185,
      "step": 127
    },
    {
      "epoch": 0.025009769441187966,
      "grad_norm": 9.813230514526367,
      "learning_rate": 1.5584415584415587e-05,
      "loss": 12.7345,
      "step": 128
    },
    {
      "epoch": 0.025205158264947247,
      "grad_norm": 11.557696342468262,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 11.8847,
      "step": 129
    },
    {
      "epoch": 0.025400547088706527,
      "grad_norm": 7.174719333648682,
      "learning_rate": 1.5844155844155847e-05,
      "loss": 13.8129,
      "step": 130
    },
    {
      "epoch": 0.025595935912465808,
      "grad_norm": 6.26338005065918,
      "learning_rate": 1.5974025974025976e-05,
      "loss": 13.1788,
      "step": 131
    },
    {
      "epoch": 0.02579132473622509,
      "grad_norm": 5.924908638000488,
      "learning_rate": 1.6103896103896105e-05,
      "loss": 12.8278,
      "step": 132
    },
    {
      "epoch": 0.02598671355998437,
      "grad_norm": 13.797770500183105,
      "learning_rate": 1.6233766233766234e-05,
      "loss": 11.9606,
      "step": 133
    },
    {
      "epoch": 0.02618210238374365,
      "grad_norm": 6.960059642791748,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 13.5888,
      "step": 134
    },
    {
      "epoch": 0.02637749120750293,
      "grad_norm": 11.013506889343262,
      "learning_rate": 1.6493506493506495e-05,
      "loss": 10.7678,
      "step": 135
    },
    {
      "epoch": 0.02657288003126221,
      "grad_norm": 14.11983871459961,
      "learning_rate": 1.6623376623376627e-05,
      "loss": 12.7964,
      "step": 136
    },
    {
      "epoch": 0.026768268855021492,
      "grad_norm": 16.5230712890625,
      "learning_rate": 1.6753246753246756e-05,
      "loss": 13.8711,
      "step": 137
    },
    {
      "epoch": 0.026963657678780773,
      "grad_norm": 9.295743942260742,
      "learning_rate": 1.6883116883116884e-05,
      "loss": 12.0177,
      "step": 138
    },
    {
      "epoch": 0.027159046502540054,
      "grad_norm": 17.73920249938965,
      "learning_rate": 1.7012987012987013e-05,
      "loss": 13.5454,
      "step": 139
    },
    {
      "epoch": 0.027354435326299335,
      "grad_norm": 11.929683685302734,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 11.7523,
      "step": 140
    },
    {
      "epoch": 0.027549824150058615,
      "grad_norm": 7.2855424880981445,
      "learning_rate": 1.7272727272727274e-05,
      "loss": 11.6333,
      "step": 141
    },
    {
      "epoch": 0.027745212973817896,
      "grad_norm": 7.469539165496826,
      "learning_rate": 1.7402597402597403e-05,
      "loss": 13.5007,
      "step": 142
    },
    {
      "epoch": 0.027940601797577177,
      "grad_norm": 7.750254154205322,
      "learning_rate": 1.7532467532467535e-05,
      "loss": 12.3127,
      "step": 143
    },
    {
      "epoch": 0.02813599062133646,
      "grad_norm": 8.031920433044434,
      "learning_rate": 1.7662337662337664e-05,
      "loss": 11.4355,
      "step": 144
    },
    {
      "epoch": 0.028331379445095742,
      "grad_norm": 12.954949378967285,
      "learning_rate": 1.7792207792207792e-05,
      "loss": 13.1203,
      "step": 145
    },
    {
      "epoch": 0.028526768268855023,
      "grad_norm": 8.291078567504883,
      "learning_rate": 1.792207792207792e-05,
      "loss": 13.5938,
      "step": 146
    },
    {
      "epoch": 0.028722157092614303,
      "grad_norm": 9.29495620727539,
      "learning_rate": 1.8051948051948053e-05,
      "loss": 13.3973,
      "step": 147
    },
    {
      "epoch": 0.028917545916373584,
      "grad_norm": 7.3590192794799805,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 12.2823,
      "step": 148
    },
    {
      "epoch": 0.029112934740132865,
      "grad_norm": 13.084951400756836,
      "learning_rate": 1.8311688311688314e-05,
      "loss": 11.847,
      "step": 149
    },
    {
      "epoch": 0.029308323563892145,
      "grad_norm": 7.148357391357422,
      "learning_rate": 1.8441558441558443e-05,
      "loss": 12.9952,
      "step": 150
    },
    {
      "epoch": 0.029503712387651426,
      "grad_norm": 6.886097431182861,
      "learning_rate": 1.8571428571428575e-05,
      "loss": 12.9661,
      "step": 151
    },
    {
      "epoch": 0.029699101211410707,
      "grad_norm": 7.425508499145508,
      "learning_rate": 1.8701298701298704e-05,
      "loss": 12.0623,
      "step": 152
    },
    {
      "epoch": 0.029894490035169988,
      "grad_norm": 6.25291633605957,
      "learning_rate": 1.8831168831168833e-05,
      "loss": 12.4557,
      "step": 153
    },
    {
      "epoch": 0.03008987885892927,
      "grad_norm": 10.292274475097656,
      "learning_rate": 1.896103896103896e-05,
      "loss": 13.0426,
      "step": 154
    },
    {
      "epoch": 0.03028526768268855,
      "grad_norm": 5.456650733947754,
      "learning_rate": 1.9090909090909094e-05,
      "loss": 11.7076,
      "step": 155
    },
    {
      "epoch": 0.03048065650644783,
      "grad_norm": 9.853564262390137,
      "learning_rate": 1.9220779220779222e-05,
      "loss": 13.1965,
      "step": 156
    },
    {
      "epoch": 0.03067604533020711,
      "grad_norm": 6.572847366333008,
      "learning_rate": 1.9350649350649354e-05,
      "loss": 10.9203,
      "step": 157
    },
    {
      "epoch": 0.03087143415396639,
      "grad_norm": 5.121306419372559,
      "learning_rate": 1.9480519480519483e-05,
      "loss": 12.3719,
      "step": 158
    },
    {
      "epoch": 0.031066822977725676,
      "grad_norm": 7.446605682373047,
      "learning_rate": 1.9610389610389612e-05,
      "loss": 12.1669,
      "step": 159
    },
    {
      "epoch": 0.03126221180148495,
      "grad_norm": 9.406105995178223,
      "learning_rate": 1.974025974025974e-05,
      "loss": 12.3783,
      "step": 160
    },
    {
      "epoch": 0.031457600625244234,
      "grad_norm": 7.835089683532715,
      "learning_rate": 1.9870129870129873e-05,
      "loss": 11.9766,
      "step": 161
    },
    {
      "epoch": 0.031652989449003514,
      "grad_norm": 9.111235618591309,
      "learning_rate": 2e-05,
      "loss": 13.4484,
      "step": 162
    },
    {
      "epoch": 0.031848378272762795,
      "grad_norm": 9.407170295715332,
      "learning_rate": 1.9999997997344768e-05,
      "loss": 13.429,
      "step": 163
    },
    {
      "epoch": 0.032043767096522076,
      "grad_norm": 13.125298500061035,
      "learning_rate": 1.9999991989379874e-05,
      "loss": 13.0715,
      "step": 164
    },
    {
      "epoch": 0.03223915592028136,
      "grad_norm": 6.865466117858887,
      "learning_rate": 1.9999981976107724e-05,
      "loss": 12.4914,
      "step": 165
    },
    {
      "epoch": 0.032434544744040644,
      "grad_norm": 4.316725730895996,
      "learning_rate": 1.999996795753233e-05,
      "loss": 11.7864,
      "step": 166
    },
    {
      "epoch": 0.032629933567799925,
      "grad_norm": 7.447556972503662,
      "learning_rate": 1.9999949933659307e-05,
      "loss": 13.1166,
      "step": 167
    },
    {
      "epoch": 0.032825322391559206,
      "grad_norm": 6.406036376953125,
      "learning_rate": 1.999992790449587e-05,
      "loss": 12.461,
      "step": 168
    },
    {
      "epoch": 0.033020711215318486,
      "grad_norm": 7.764867305755615,
      "learning_rate": 1.9999901870050848e-05,
      "loss": 11.5924,
      "step": 169
    },
    {
      "epoch": 0.03321610003907777,
      "grad_norm": 7.2458624839782715,
      "learning_rate": 1.9999871830334667e-05,
      "loss": 12.7767,
      "step": 170
    },
    {
      "epoch": 0.03341148886283705,
      "grad_norm": 5.0661396980285645,
      "learning_rate": 1.9999837785359356e-05,
      "loss": 11.0157,
      "step": 171
    },
    {
      "epoch": 0.03360687768659633,
      "grad_norm": 7.504276752471924,
      "learning_rate": 1.999979973513855e-05,
      "loss": 10.8756,
      "step": 172
    },
    {
      "epoch": 0.03380226651035561,
      "grad_norm": 9.193746566772461,
      "learning_rate": 1.9999757679687498e-05,
      "loss": 13.5181,
      "step": 173
    },
    {
      "epoch": 0.03399765533411489,
      "grad_norm": 13.39449405670166,
      "learning_rate": 1.9999711619023036e-05,
      "loss": 13.0867,
      "step": 174
    },
    {
      "epoch": 0.03419304415787417,
      "grad_norm": 10.203667640686035,
      "learning_rate": 1.9999661553163615e-05,
      "loss": 11.7621,
      "step": 175
    },
    {
      "epoch": 0.03438843298163345,
      "grad_norm": 6.521947383880615,
      "learning_rate": 1.999960748212929e-05,
      "loss": 12.5624,
      "step": 176
    },
    {
      "epoch": 0.03458382180539273,
      "grad_norm": 7.569502830505371,
      "learning_rate": 1.9999549405941714e-05,
      "loss": 11.277,
      "step": 177
    },
    {
      "epoch": 0.03477921062915201,
      "grad_norm": 6.592679023742676,
      "learning_rate": 1.9999487324624152e-05,
      "loss": 12.7629,
      "step": 178
    },
    {
      "epoch": 0.034974599452911294,
      "grad_norm": 8.883557319641113,
      "learning_rate": 1.9999421238201465e-05,
      "loss": 11.6271,
      "step": 179
    },
    {
      "epoch": 0.035169988276670575,
      "grad_norm": 6.424564361572266,
      "learning_rate": 1.9999351146700133e-05,
      "loss": 12.3532,
      "step": 180
    },
    {
      "epoch": 0.035365377100429855,
      "grad_norm": 7.161871910095215,
      "learning_rate": 1.9999277050148217e-05,
      "loss": 12.4987,
      "step": 181
    },
    {
      "epoch": 0.035560765924189136,
      "grad_norm": 6.069089412689209,
      "learning_rate": 1.9999198948575404e-05,
      "loss": 11.6968,
      "step": 182
    },
    {
      "epoch": 0.03575615474794842,
      "grad_norm": 8.944799423217773,
      "learning_rate": 1.9999116842012973e-05,
      "loss": 12.5904,
      "step": 183
    },
    {
      "epoch": 0.0359515435717077,
      "grad_norm": 18.52510643005371,
      "learning_rate": 1.999903073049381e-05,
      "loss": 12.4171,
      "step": 184
    },
    {
      "epoch": 0.03614693239546698,
      "grad_norm": 5.86504602432251,
      "learning_rate": 1.9998940614052404e-05,
      "loss": 12.2995,
      "step": 185
    },
    {
      "epoch": 0.03634232121922626,
      "grad_norm": 7.921477794647217,
      "learning_rate": 1.999884649272485e-05,
      "loss": 12.1127,
      "step": 186
    },
    {
      "epoch": 0.03653771004298554,
      "grad_norm": 5.69157600402832,
      "learning_rate": 1.999874836654885e-05,
      "loss": 11.7619,
      "step": 187
    },
    {
      "epoch": 0.03673309886674482,
      "grad_norm": 5.952302932739258,
      "learning_rate": 1.9998646235563702e-05,
      "loss": 11.5656,
      "step": 188
    },
    {
      "epoch": 0.0369284876905041,
      "grad_norm": 5.239631175994873,
      "learning_rate": 1.9998540099810318e-05,
      "loss": 11.8899,
      "step": 189
    },
    {
      "epoch": 0.03712387651426338,
      "grad_norm": 5.957993507385254,
      "learning_rate": 1.9998429959331204e-05,
      "loss": 12.6495,
      "step": 190
    },
    {
      "epoch": 0.03731926533802266,
      "grad_norm": 6.800102233886719,
      "learning_rate": 1.9998315814170473e-05,
      "loss": 12.0542,
      "step": 191
    },
    {
      "epoch": 0.03751465416178194,
      "grad_norm": 7.192645072937012,
      "learning_rate": 1.999819766437385e-05,
      "loss": 12.5862,
      "step": 192
    },
    {
      "epoch": 0.037710042985541224,
      "grad_norm": 6.849038124084473,
      "learning_rate": 1.999807550998865e-05,
      "loss": 11.8042,
      "step": 193
    },
    {
      "epoch": 0.037905431809300505,
      "grad_norm": 5.00752592086792,
      "learning_rate": 1.9997949351063807e-05,
      "loss": 12.224,
      "step": 194
    },
    {
      "epoch": 0.038100820633059786,
      "grad_norm": 8.420522689819336,
      "learning_rate": 1.999781918764985e-05,
      "loss": 12.8256,
      "step": 195
    },
    {
      "epoch": 0.03829620945681907,
      "grad_norm": 7.47088098526001,
      "learning_rate": 1.9997685019798913e-05,
      "loss": 12.2079,
      "step": 196
    },
    {
      "epoch": 0.038491598280578354,
      "grad_norm": 8.45734691619873,
      "learning_rate": 1.9997546847564727e-05,
      "loss": 11.7802,
      "step": 197
    },
    {
      "epoch": 0.038686987104337635,
      "grad_norm": 10.523774147033691,
      "learning_rate": 1.9997404671002645e-05,
      "loss": 11.5177,
      "step": 198
    },
    {
      "epoch": 0.038882375928096916,
      "grad_norm": 14.023902893066406,
      "learning_rate": 1.999725849016961e-05,
      "loss": 13.3077,
      "step": 199
    },
    {
      "epoch": 0.039077764751856196,
      "grad_norm": 6.061577320098877,
      "learning_rate": 1.9997108305124165e-05,
      "loss": 10.655,
      "step": 200
    },
    {
      "epoch": 0.03927315357561548,
      "grad_norm": 18.192489624023438,
      "learning_rate": 1.9996954115926477e-05,
      "loss": 12.3448,
      "step": 201
    },
    {
      "epoch": 0.03946854239937476,
      "grad_norm": 7.7355780601501465,
      "learning_rate": 1.9996795922638293e-05,
      "loss": 11.3258,
      "step": 202
    },
    {
      "epoch": 0.03966393122313404,
      "grad_norm": 8.186224937438965,
      "learning_rate": 1.9996633725322977e-05,
      "loss": 13.2872,
      "step": 203
    },
    {
      "epoch": 0.03985932004689332,
      "grad_norm": 8.051215171813965,
      "learning_rate": 1.9996467524045492e-05,
      "loss": 12.226,
      "step": 204
    },
    {
      "epoch": 0.0400547088706526,
      "grad_norm": 8.462077140808105,
      "learning_rate": 1.9996297318872413e-05,
      "loss": 12.6791,
      "step": 205
    },
    {
      "epoch": 0.04025009769441188,
      "grad_norm": 7.602774620056152,
      "learning_rate": 1.999612310987191e-05,
      "loss": 12.0686,
      "step": 206
    },
    {
      "epoch": 0.04044548651817116,
      "grad_norm": 5.274927139282227,
      "learning_rate": 1.9995944897113753e-05,
      "loss": 13.5675,
      "step": 207
    },
    {
      "epoch": 0.04064087534193044,
      "grad_norm": 6.607327461242676,
      "learning_rate": 1.9995762680669326e-05,
      "loss": 11.4844,
      "step": 208
    },
    {
      "epoch": 0.04083626416568972,
      "grad_norm": 12.272229194641113,
      "learning_rate": 1.9995576460611617e-05,
      "loss": 13.1514,
      "step": 209
    },
    {
      "epoch": 0.041031652989449004,
      "grad_norm": 11.912178993225098,
      "learning_rate": 1.9995386237015208e-05,
      "loss": 11.9987,
      "step": 210
    },
    {
      "epoch": 0.041227041813208284,
      "grad_norm": 8.682504653930664,
      "learning_rate": 1.999519200995629e-05,
      "loss": 12.7482,
      "step": 211
    },
    {
      "epoch": 0.041422430636967565,
      "grad_norm": 12.171260833740234,
      "learning_rate": 1.9994993779512656e-05,
      "loss": 11.9504,
      "step": 212
    },
    {
      "epoch": 0.041617819460726846,
      "grad_norm": 15.512253761291504,
      "learning_rate": 1.9994791545763707e-05,
      "loss": 13.1384,
      "step": 213
    },
    {
      "epoch": 0.04181320828448613,
      "grad_norm": 7.158082962036133,
      "learning_rate": 1.9994585308790437e-05,
      "loss": 10.9148,
      "step": 214
    },
    {
      "epoch": 0.04200859710824541,
      "grad_norm": 7.843011379241943,
      "learning_rate": 1.999437506867546e-05,
      "loss": 12.889,
      "step": 215
    },
    {
      "epoch": 0.04220398593200469,
      "grad_norm": 7.392290115356445,
      "learning_rate": 1.9994160825502975e-05,
      "loss": 13.2473,
      "step": 216
    },
    {
      "epoch": 0.04239937475576397,
      "grad_norm": 7.660642623901367,
      "learning_rate": 1.99939425793588e-05,
      "loss": 11.0385,
      "step": 217
    },
    {
      "epoch": 0.04259476357952325,
      "grad_norm": 7.692232131958008,
      "learning_rate": 1.9993720330330343e-05,
      "loss": 12.1109,
      "step": 218
    },
    {
      "epoch": 0.04279015240328253,
      "grad_norm": 5.706768989562988,
      "learning_rate": 1.9993494078506626e-05,
      "loss": 12.2178,
      "step": 219
    },
    {
      "epoch": 0.04298554122704181,
      "grad_norm": 5.357436180114746,
      "learning_rate": 1.9993263823978268e-05,
      "loss": 12.0862,
      "step": 220
    },
    {
      "epoch": 0.04318093005080109,
      "grad_norm": 7.361543655395508,
      "learning_rate": 1.9993029566837493e-05,
      "loss": 13.2949,
      "step": 221
    },
    {
      "epoch": 0.04337631887456037,
      "grad_norm": 8.009748458862305,
      "learning_rate": 1.9992791307178133e-05,
      "loss": 10.9137,
      "step": 222
    },
    {
      "epoch": 0.04357170769831965,
      "grad_norm": 6.979041576385498,
      "learning_rate": 1.9992549045095607e-05,
      "loss": 12.2136,
      "step": 223
    },
    {
      "epoch": 0.043767096522078934,
      "grad_norm": 8.310110092163086,
      "learning_rate": 1.999230278068696e-05,
      "loss": 11.1071,
      "step": 224
    },
    {
      "epoch": 0.043962485345838215,
      "grad_norm": 8.9794282913208,
      "learning_rate": 1.9992052514050824e-05,
      "loss": 11.9055,
      "step": 225
    },
    {
      "epoch": 0.0441578741695975,
      "grad_norm": 15.550240516662598,
      "learning_rate": 1.9991798245287438e-05,
      "loss": 12.0324,
      "step": 226
    },
    {
      "epoch": 0.04435326299335678,
      "grad_norm": 7.737309455871582,
      "learning_rate": 1.999153997449865e-05,
      "loss": 12.1904,
      "step": 227
    },
    {
      "epoch": 0.044548651817116064,
      "grad_norm": 7.448500633239746,
      "learning_rate": 1.9991277701787895e-05,
      "loss": 11.7476,
      "step": 228
    },
    {
      "epoch": 0.044744040640875345,
      "grad_norm": 11.323410034179688,
      "learning_rate": 1.999101142726023e-05,
      "loss": 12.7668,
      "step": 229
    },
    {
      "epoch": 0.044939429464634625,
      "grad_norm": 10.129905700683594,
      "learning_rate": 1.9990741151022302e-05,
      "loss": 10.9619,
      "step": 230
    },
    {
      "epoch": 0.045134818288393906,
      "grad_norm": 10.254401206970215,
      "learning_rate": 1.9990466873182367e-05,
      "loss": 10.8593,
      "step": 231
    },
    {
      "epoch": 0.04533020711215319,
      "grad_norm": 9.408255577087402,
      "learning_rate": 1.999018859385028e-05,
      "loss": 11.9756,
      "step": 232
    },
    {
      "epoch": 0.04552559593591247,
      "grad_norm": 12.784259796142578,
      "learning_rate": 1.9989906313137502e-05,
      "loss": 12.2801,
      "step": 233
    },
    {
      "epoch": 0.04572098475967175,
      "grad_norm": 10.667810440063477,
      "learning_rate": 1.9989620031157098e-05,
      "loss": 12.3217,
      "step": 234
    },
    {
      "epoch": 0.04591637358343103,
      "grad_norm": 8.796360969543457,
      "learning_rate": 1.9989329748023728e-05,
      "loss": 12.3541,
      "step": 235
    },
    {
      "epoch": 0.04611176240719031,
      "grad_norm": 9.64390754699707,
      "learning_rate": 1.998903546385366e-05,
      "loss": 12.063,
      "step": 236
    },
    {
      "epoch": 0.04630715123094959,
      "grad_norm": 10.169878959655762,
      "learning_rate": 1.9988737178764764e-05,
      "loss": 12.7145,
      "step": 237
    },
    {
      "epoch": 0.04650254005470887,
      "grad_norm": 10.228228569030762,
      "learning_rate": 1.9988434892876515e-05,
      "loss": 11.8427,
      "step": 238
    },
    {
      "epoch": 0.04669792887846815,
      "grad_norm": 13.552098274230957,
      "learning_rate": 1.9988128606309988e-05,
      "loss": 12.853,
      "step": 239
    },
    {
      "epoch": 0.04689331770222743,
      "grad_norm": 6.725545406341553,
      "learning_rate": 1.9987818319187857e-05,
      "loss": 11.8622,
      "step": 240
    },
    {
      "epoch": 0.047088706525986714,
      "grad_norm": 6.490736484527588,
      "learning_rate": 1.9987504031634403e-05,
      "loss": 12.8614,
      "step": 241
    },
    {
      "epoch": 0.047284095349745994,
      "grad_norm": 7.924977779388428,
      "learning_rate": 1.9987185743775507e-05,
      "loss": 12.0172,
      "step": 242
    },
    {
      "epoch": 0.047479484173505275,
      "grad_norm": 7.123000144958496,
      "learning_rate": 1.998686345573866e-05,
      "loss": 12.1521,
      "step": 243
    },
    {
      "epoch": 0.047674872997264556,
      "grad_norm": 6.6585845947265625,
      "learning_rate": 1.9986537167652937e-05,
      "loss": 11.3805,
      "step": 244
    },
    {
      "epoch": 0.047870261821023836,
      "grad_norm": 5.809377193450928,
      "learning_rate": 1.9986206879649033e-05,
      "loss": 11.1165,
      "step": 245
    },
    {
      "epoch": 0.04806565064478312,
      "grad_norm": 6.047052383422852,
      "learning_rate": 1.9985872591859242e-05,
      "loss": 10.7509,
      "step": 246
    },
    {
      "epoch": 0.0482610394685424,
      "grad_norm": 12.317660331726074,
      "learning_rate": 1.9985534304417448e-05,
      "loss": 11.8707,
      "step": 247
    },
    {
      "epoch": 0.04845642829230168,
      "grad_norm": 8.26644229888916,
      "learning_rate": 1.9985192017459153e-05,
      "loss": 12.3413,
      "step": 248
    },
    {
      "epoch": 0.04865181711606096,
      "grad_norm": 8.184246063232422,
      "learning_rate": 1.9984845731121447e-05,
      "loss": 10.454,
      "step": 249
    },
    {
      "epoch": 0.04884720593982024,
      "grad_norm": 6.725503921508789,
      "learning_rate": 1.9984495445543035e-05,
      "loss": 10.6297,
      "step": 250
    },
    {
      "epoch": 0.04904259476357952,
      "grad_norm": 14.417993545532227,
      "learning_rate": 1.9984141160864214e-05,
      "loss": 12.5653,
      "step": 251
    },
    {
      "epoch": 0.0492379835873388,
      "grad_norm": 7.391936302185059,
      "learning_rate": 1.9983782877226888e-05,
      "loss": 11.6914,
      "step": 252
    },
    {
      "epoch": 0.04943337241109808,
      "grad_norm": 9.729405403137207,
      "learning_rate": 1.9983420594774556e-05,
      "loss": 12.3673,
      "step": 253
    },
    {
      "epoch": 0.04962876123485736,
      "grad_norm": 13.310443878173828,
      "learning_rate": 1.998305431365233e-05,
      "loss": 12.5482,
      "step": 254
    },
    {
      "epoch": 0.049824150058616644,
      "grad_norm": 12.740357398986816,
      "learning_rate": 1.9982684034006906e-05,
      "loss": 11.3177,
      "step": 255
    },
    {
      "epoch": 0.05001953888237593,
      "grad_norm": 13.545549392700195,
      "learning_rate": 1.9982309755986607e-05,
      "loss": 12.0766,
      "step": 256
    },
    {
      "epoch": 0.05021492770613521,
      "grad_norm": 8.222143173217773,
      "learning_rate": 1.9981931479741336e-05,
      "loss": 10.2483,
      "step": 257
    },
    {
      "epoch": 0.05041031652989449,
      "grad_norm": 17.599727630615234,
      "learning_rate": 1.9981549205422602e-05,
      "loss": 12.1119,
      "step": 258
    },
    {
      "epoch": 0.050605705353653774,
      "grad_norm": 9.997603416442871,
      "learning_rate": 1.998116293318352e-05,
      "loss": 11.3142,
      "step": 259
    },
    {
      "epoch": 0.050801094177413054,
      "grad_norm": 15.584720611572266,
      "learning_rate": 1.99807726631788e-05,
      "loss": 10.2511,
      "step": 260
    },
    {
      "epoch": 0.050996483001172335,
      "grad_norm": 15.335906982421875,
      "learning_rate": 1.9980378395564772e-05,
      "loss": 11.6443,
      "step": 261
    },
    {
      "epoch": 0.051191871824931616,
      "grad_norm": 12.51326847076416,
      "learning_rate": 1.997998013049933e-05,
      "loss": 12.287,
      "step": 262
    },
    {
      "epoch": 0.0513872606486909,
      "grad_norm": 8.552204132080078,
      "learning_rate": 1.997957786814201e-05,
      "loss": 11.6877,
      "step": 263
    },
    {
      "epoch": 0.05158264947245018,
      "grad_norm": 6.9093241691589355,
      "learning_rate": 1.9979171608653926e-05,
      "loss": 12.0065,
      "step": 264
    },
    {
      "epoch": 0.05177803829620946,
      "grad_norm": 13.546882629394531,
      "learning_rate": 1.997876135219779e-05,
      "loss": 11.411,
      "step": 265
    },
    {
      "epoch": 0.05197342711996874,
      "grad_norm": 11.272443771362305,
      "learning_rate": 1.997834709893793e-05,
      "loss": 12.0057,
      "step": 266
    },
    {
      "epoch": 0.05216881594372802,
      "grad_norm": 8.439807891845703,
      "learning_rate": 1.9977928849040263e-05,
      "loss": 11.9195,
      "step": 267
    },
    {
      "epoch": 0.0523642047674873,
      "grad_norm": 10.880560874938965,
      "learning_rate": 1.9977506602672317e-05,
      "loss": 11.9902,
      "step": 268
    },
    {
      "epoch": 0.05255959359124658,
      "grad_norm": 7.07627010345459,
      "learning_rate": 1.9977080360003206e-05,
      "loss": 11.8873,
      "step": 269
    },
    {
      "epoch": 0.05275498241500586,
      "grad_norm": 14.284347534179688,
      "learning_rate": 1.9976650121203663e-05,
      "loss": 10.994,
      "step": 270
    },
    {
      "epoch": 0.05295037123876514,
      "grad_norm": 8.325899124145508,
      "learning_rate": 1.9976215886446005e-05,
      "loss": 12.2574,
      "step": 271
    },
    {
      "epoch": 0.05314576006252442,
      "grad_norm": 9.194845199584961,
      "learning_rate": 1.9975777655904157e-05,
      "loss": 12.5032,
      "step": 272
    },
    {
      "epoch": 0.053341148886283704,
      "grad_norm": 11.341529846191406,
      "learning_rate": 1.997533542975365e-05,
      "loss": 10.0915,
      "step": 273
    },
    {
      "epoch": 0.053536537710042985,
      "grad_norm": 9.99206256866455,
      "learning_rate": 1.9974889208171602e-05,
      "loss": 11.9477,
      "step": 274
    },
    {
      "epoch": 0.053731926533802266,
      "grad_norm": 8.00964069366455,
      "learning_rate": 1.9974438991336743e-05,
      "loss": 11.3129,
      "step": 275
    },
    {
      "epoch": 0.053927315357561546,
      "grad_norm": 9.96445083618164,
      "learning_rate": 1.9973984779429396e-05,
      "loss": 11.4211,
      "step": 276
    },
    {
      "epoch": 0.05412270418132083,
      "grad_norm": 11.463232040405273,
      "learning_rate": 1.997352657263149e-05,
      "loss": 12.0209,
      "step": 277
    },
    {
      "epoch": 0.05431809300508011,
      "grad_norm": 9.95913314819336,
      "learning_rate": 1.997306437112655e-05,
      "loss": 12.3077,
      "step": 278
    },
    {
      "epoch": 0.05451348182883939,
      "grad_norm": 13.20826244354248,
      "learning_rate": 1.9972598175099698e-05,
      "loss": 12.2697,
      "step": 279
    },
    {
      "epoch": 0.05470887065259867,
      "grad_norm": 9.180987358093262,
      "learning_rate": 1.9972127984737667e-05,
      "loss": 12.1211,
      "step": 280
    },
    {
      "epoch": 0.05490425947635795,
      "grad_norm": 9.276026725769043,
      "learning_rate": 1.9971653800228782e-05,
      "loss": 12.6417,
      "step": 281
    },
    {
      "epoch": 0.05509964830011723,
      "grad_norm": 11.767862319946289,
      "learning_rate": 1.997117562176296e-05,
      "loss": 12.017,
      "step": 282
    },
    {
      "epoch": 0.05529503712387651,
      "grad_norm": 9.630742073059082,
      "learning_rate": 1.9970693449531735e-05,
      "loss": 10.7546,
      "step": 283
    },
    {
      "epoch": 0.05549042594763579,
      "grad_norm": 10.929373741149902,
      "learning_rate": 1.9970207283728227e-05,
      "loss": 11.5425,
      "step": 284
    },
    {
      "epoch": 0.05568581477139507,
      "grad_norm": 12.832663536071777,
      "learning_rate": 1.9969717124547165e-05,
      "loss": 11.6781,
      "step": 285
    },
    {
      "epoch": 0.055881203595154354,
      "grad_norm": 20.87643051147461,
      "learning_rate": 1.9969222972184873e-05,
      "loss": 12.2216,
      "step": 286
    },
    {
      "epoch": 0.05607659241891364,
      "grad_norm": 10.373276710510254,
      "learning_rate": 1.9968724826839268e-05,
      "loss": 12.3443,
      "step": 287
    },
    {
      "epoch": 0.05627198124267292,
      "grad_norm": 8.58100414276123,
      "learning_rate": 1.996822268870988e-05,
      "loss": 10.6689,
      "step": 288
    },
    {
      "epoch": 0.0564673700664322,
      "grad_norm": 8.117618560791016,
      "learning_rate": 1.9967716557997824e-05,
      "loss": 12.3755,
      "step": 289
    },
    {
      "epoch": 0.056662758890191484,
      "grad_norm": 10.374201774597168,
      "learning_rate": 1.9967206434905832e-05,
      "loss": 12.1325,
      "step": 290
    },
    {
      "epoch": 0.056858147713950764,
      "grad_norm": 8.217313766479492,
      "learning_rate": 1.996669231963821e-05,
      "loss": 10.9547,
      "step": 291
    },
    {
      "epoch": 0.057053536537710045,
      "grad_norm": 7.501300811767578,
      "learning_rate": 1.9966174212400885e-05,
      "loss": 11.9509,
      "step": 292
    },
    {
      "epoch": 0.057248925361469326,
      "grad_norm": 6.74676513671875,
      "learning_rate": 1.9965652113401374e-05,
      "loss": 11.4133,
      "step": 293
    },
    {
      "epoch": 0.05744431418522861,
      "grad_norm": 11.078866958618164,
      "learning_rate": 1.9965126022848796e-05,
      "loss": 12.1601,
      "step": 294
    },
    {
      "epoch": 0.05763970300898789,
      "grad_norm": 7.624950885772705,
      "learning_rate": 1.9964595940953863e-05,
      "loss": 11.3853,
      "step": 295
    },
    {
      "epoch": 0.05783509183274717,
      "grad_norm": 7.737289905548096,
      "learning_rate": 1.9964061867928892e-05,
      "loss": 11.9907,
      "step": 296
    },
    {
      "epoch": 0.05803048065650645,
      "grad_norm": 9.63924503326416,
      "learning_rate": 1.9963523803987792e-05,
      "loss": 11.5848,
      "step": 297
    },
    {
      "epoch": 0.05822586948026573,
      "grad_norm": 7.582948207855225,
      "learning_rate": 1.996298174934608e-05,
      "loss": 11.4951,
      "step": 298
    },
    {
      "epoch": 0.05842125830402501,
      "grad_norm": 8.697115898132324,
      "learning_rate": 1.996243570422086e-05,
      "loss": 12.8351,
      "step": 299
    },
    {
      "epoch": 0.05861664712778429,
      "grad_norm": 9.946210861206055,
      "learning_rate": 1.9961885668830844e-05,
      "loss": 10.6673,
      "step": 300
    },
    {
      "epoch": 0.05881203595154357,
      "grad_norm": 6.278243064880371,
      "learning_rate": 1.9961331643396336e-05,
      "loss": 11.1935,
      "step": 301
    },
    {
      "epoch": 0.05900742477530285,
      "grad_norm": 8.534592628479004,
      "learning_rate": 1.9960773628139244e-05,
      "loss": 12.0799,
      "step": 302
    },
    {
      "epoch": 0.05920281359906213,
      "grad_norm": 6.725213050842285,
      "learning_rate": 1.9960211623283064e-05,
      "loss": 10.8297,
      "step": 303
    },
    {
      "epoch": 0.059398202422821414,
      "grad_norm": 11.384440422058105,
      "learning_rate": 1.9959645629052905e-05,
      "loss": 11.9136,
      "step": 304
    },
    {
      "epoch": 0.059593591246580695,
      "grad_norm": 11.529409408569336,
      "learning_rate": 1.9959075645675457e-05,
      "loss": 10.7427,
      "step": 305
    },
    {
      "epoch": 0.059788980070339975,
      "grad_norm": 12.019720077514648,
      "learning_rate": 1.9958501673379018e-05,
      "loss": 12.0401,
      "step": 306
    },
    {
      "epoch": 0.059984368894099256,
      "grad_norm": 8.388792037963867,
      "learning_rate": 1.9957923712393486e-05,
      "loss": 13.0926,
      "step": 307
    },
    {
      "epoch": 0.06017975771785854,
      "grad_norm": 12.770638465881348,
      "learning_rate": 1.9957341762950346e-05,
      "loss": 11.7429,
      "step": 308
    },
    {
      "epoch": 0.06037514654161782,
      "grad_norm": 5.92092227935791,
      "learning_rate": 1.995675582528269e-05,
      "loss": 10.5684,
      "step": 309
    },
    {
      "epoch": 0.0605705353653771,
      "grad_norm": 7.7345781326293945,
      "learning_rate": 1.995616589962521e-05,
      "loss": 12.3008,
      "step": 310
    },
    {
      "epoch": 0.06076592418913638,
      "grad_norm": 8.71921443939209,
      "learning_rate": 1.9955571986214182e-05,
      "loss": 13.1848,
      "step": 311
    },
    {
      "epoch": 0.06096131301289566,
      "grad_norm": 6.207778453826904,
      "learning_rate": 1.9954974085287487e-05,
      "loss": 12.2107,
      "step": 312
    },
    {
      "epoch": 0.06115670183665494,
      "grad_norm": 7.688232421875,
      "learning_rate": 1.9954372197084607e-05,
      "loss": 11.741,
      "step": 313
    },
    {
      "epoch": 0.06135209066041422,
      "grad_norm": 9.183834075927734,
      "learning_rate": 1.995376632184661e-05,
      "loss": 12.838,
      "step": 314
    },
    {
      "epoch": 0.0615474794841735,
      "grad_norm": 9.513407707214355,
      "learning_rate": 1.995315645981618e-05,
      "loss": 11.5654,
      "step": 315
    },
    {
      "epoch": 0.06174286830793278,
      "grad_norm": 8.319108009338379,
      "learning_rate": 1.995254261123757e-05,
      "loss": 12.2501,
      "step": 316
    },
    {
      "epoch": 0.06193825713169207,
      "grad_norm": 6.513503551483154,
      "learning_rate": 1.9951924776356657e-05,
      "loss": 11.0926,
      "step": 317
    },
    {
      "epoch": 0.06213364595545135,
      "grad_norm": 14.238517761230469,
      "learning_rate": 1.99513029554209e-05,
      "loss": 13.1257,
      "step": 318
    },
    {
      "epoch": 0.06232903477921063,
      "grad_norm": 11.35081672668457,
      "learning_rate": 1.9950677148679357e-05,
      "loss": 11.6198,
      "step": 319
    },
    {
      "epoch": 0.0625244236029699,
      "grad_norm": 7.610073566436768,
      "learning_rate": 1.9950047356382684e-05,
      "loss": 10.8325,
      "step": 320
    },
    {
      "epoch": 0.0627198124267292,
      "grad_norm": 7.52583646774292,
      "learning_rate": 1.9949413578783133e-05,
      "loss": 11.8231,
      "step": 321
    },
    {
      "epoch": 0.06291520125048847,
      "grad_norm": 12.769768714904785,
      "learning_rate": 1.9948775816134547e-05,
      "loss": 12.0816,
      "step": 322
    },
    {
      "epoch": 0.06311059007424775,
      "grad_norm": 13.12612533569336,
      "learning_rate": 1.9948134068692372e-05,
      "loss": 11.0729,
      "step": 323
    },
    {
      "epoch": 0.06330597889800703,
      "grad_norm": 8.223774909973145,
      "learning_rate": 1.994748833671365e-05,
      "loss": 10.8119,
      "step": 324
    },
    {
      "epoch": 0.06350136772176632,
      "grad_norm": 8.893241882324219,
      "learning_rate": 1.9946838620457015e-05,
      "loss": 11.7001,
      "step": 325
    },
    {
      "epoch": 0.06369675654552559,
      "grad_norm": 9.47516918182373,
      "learning_rate": 1.9946184920182704e-05,
      "loss": 12.1686,
      "step": 326
    },
    {
      "epoch": 0.06389214536928488,
      "grad_norm": 9.479887962341309,
      "learning_rate": 1.9945527236152533e-05,
      "loss": 11.6604,
      "step": 327
    },
    {
      "epoch": 0.06408753419304415,
      "grad_norm": 9.156216621398926,
      "learning_rate": 1.994486556862993e-05,
      "loss": 11.3712,
      "step": 328
    },
    {
      "epoch": 0.06428292301680344,
      "grad_norm": 8.44137191772461,
      "learning_rate": 1.994419991787992e-05,
      "loss": 11.162,
      "step": 329
    },
    {
      "epoch": 0.06447831184056271,
      "grad_norm": 8.18656063079834,
      "learning_rate": 1.9943530284169107e-05,
      "loss": 10.9821,
      "step": 330
    },
    {
      "epoch": 0.064673700664322,
      "grad_norm": 12.09242057800293,
      "learning_rate": 1.9942856667765706e-05,
      "loss": 10.8379,
      "step": 331
    },
    {
      "epoch": 0.06486908948808129,
      "grad_norm": 9.561943054199219,
      "learning_rate": 1.994217906893952e-05,
      "loss": 12.13,
      "step": 332
    },
    {
      "epoch": 0.06506447831184056,
      "grad_norm": 11.300248146057129,
      "learning_rate": 1.9941497487961946e-05,
      "loss": 11.8315,
      "step": 333
    },
    {
      "epoch": 0.06525986713559985,
      "grad_norm": 22.479005813598633,
      "learning_rate": 1.9940811925105984e-05,
      "loss": 12.5274,
      "step": 334
    },
    {
      "epoch": 0.06545525595935912,
      "grad_norm": 12.22573184967041,
      "learning_rate": 1.9940122380646218e-05,
      "loss": 12.6328,
      "step": 335
    },
    {
      "epoch": 0.06565064478311841,
      "grad_norm": 12.668004989624023,
      "learning_rate": 1.993942885485883e-05,
      "loss": 12.9096,
      "step": 336
    },
    {
      "epoch": 0.06584603360687769,
      "grad_norm": 14.226447105407715,
      "learning_rate": 1.9938731348021607e-05,
      "loss": 12.423,
      "step": 337
    },
    {
      "epoch": 0.06604142243063697,
      "grad_norm": 11.14017105102539,
      "learning_rate": 1.9938029860413914e-05,
      "loss": 11.2108,
      "step": 338
    },
    {
      "epoch": 0.06623681125439625,
      "grad_norm": 13.295299530029297,
      "learning_rate": 1.9937324392316725e-05,
      "loss": 11.8767,
      "step": 339
    },
    {
      "epoch": 0.06643220007815553,
      "grad_norm": 9.23791790008545,
      "learning_rate": 1.9936614944012595e-05,
      "loss": 11.8994,
      "step": 340
    },
    {
      "epoch": 0.06662758890191481,
      "grad_norm": 12.534093856811523,
      "learning_rate": 1.9935901515785686e-05,
      "loss": 11.687,
      "step": 341
    },
    {
      "epoch": 0.0668229777256741,
      "grad_norm": 8.44124984741211,
      "learning_rate": 1.9935184107921747e-05,
      "loss": 11.3491,
      "step": 342
    },
    {
      "epoch": 0.06701836654943337,
      "grad_norm": 8.509767532348633,
      "learning_rate": 1.993446272070812e-05,
      "loss": 10.6972,
      "step": 343
    },
    {
      "epoch": 0.06721375537319266,
      "grad_norm": 10.905438423156738,
      "learning_rate": 1.993373735443374e-05,
      "loss": 11.0463,
      "step": 344
    },
    {
      "epoch": 0.06740914419695193,
      "grad_norm": 8.30171012878418,
      "learning_rate": 1.993300800938915e-05,
      "loss": 10.7473,
      "step": 345
    },
    {
      "epoch": 0.06760453302071122,
      "grad_norm": 7.070130825042725,
      "learning_rate": 1.993227468586646e-05,
      "loss": 10.6074,
      "step": 346
    },
    {
      "epoch": 0.06779992184447049,
      "grad_norm": 10.217448234558105,
      "learning_rate": 1.99315373841594e-05,
      "loss": 10.8733,
      "step": 347
    },
    {
      "epoch": 0.06799531066822978,
      "grad_norm": 7.3030805587768555,
      "learning_rate": 1.993079610456328e-05,
      "loss": 11.4374,
      "step": 348
    },
    {
      "epoch": 0.06819069949198905,
      "grad_norm": 10.487565994262695,
      "learning_rate": 1.993005084737501e-05,
      "loss": 11.2645,
      "step": 349
    },
    {
      "epoch": 0.06838608831574834,
      "grad_norm": 24.83467674255371,
      "learning_rate": 1.9929301612893076e-05,
      "loss": 11.8643,
      "step": 350
    },
    {
      "epoch": 0.06858147713950762,
      "grad_norm": 10.412206649780273,
      "learning_rate": 1.992854840141758e-05,
      "loss": 10.8554,
      "step": 351
    },
    {
      "epoch": 0.0687768659632669,
      "grad_norm": 9.083630561828613,
      "learning_rate": 1.9927791213250202e-05,
      "loss": 11.3935,
      "step": 352
    },
    {
      "epoch": 0.06897225478702618,
      "grad_norm": 10.593680381774902,
      "learning_rate": 1.992703004869422e-05,
      "loss": 10.8293,
      "step": 353
    },
    {
      "epoch": 0.06916764361078546,
      "grad_norm": 33.345863342285156,
      "learning_rate": 1.9926264908054506e-05,
      "loss": 11.751,
      "step": 354
    },
    {
      "epoch": 0.06936303243454474,
      "grad_norm": 15.0076322555542,
      "learning_rate": 1.992549579163752e-05,
      "loss": 11.2289,
      "step": 355
    },
    {
      "epoch": 0.06955842125830403,
      "grad_norm": 12.340163230895996,
      "learning_rate": 1.992472269975132e-05,
      "loss": 10.9912,
      "step": 356
    },
    {
      "epoch": 0.0697538100820633,
      "grad_norm": 10.287991523742676,
      "learning_rate": 1.992394563270555e-05,
      "loss": 11.4063,
      "step": 357
    },
    {
      "epoch": 0.06994919890582259,
      "grad_norm": 19.07567596435547,
      "learning_rate": 1.9923164590811455e-05,
      "loss": 11.4794,
      "step": 358
    },
    {
      "epoch": 0.07014458772958186,
      "grad_norm": 8.652302742004395,
      "learning_rate": 1.992237957438186e-05,
      "loss": 11.6414,
      "step": 359
    },
    {
      "epoch": 0.07033997655334115,
      "grad_norm": 16.885883331298828,
      "learning_rate": 1.9921590583731195e-05,
      "loss": 10.9357,
      "step": 360
    },
    {
      "epoch": 0.07053536537710042,
      "grad_norm": 8.816641807556152,
      "learning_rate": 1.9920797619175466e-05,
      "loss": 10.4371,
      "step": 361
    },
    {
      "epoch": 0.07073075420085971,
      "grad_norm": 12.317264556884766,
      "learning_rate": 1.992000068103229e-05,
      "loss": 12.5566,
      "step": 362
    },
    {
      "epoch": 0.070926143024619,
      "grad_norm": 10.080653190612793,
      "learning_rate": 1.991919976962086e-05,
      "loss": 12.1561,
      "step": 363
    },
    {
      "epoch": 0.07112153184837827,
      "grad_norm": 10.67681884765625,
      "learning_rate": 1.9918394885261966e-05,
      "loss": 11.6402,
      "step": 364
    },
    {
      "epoch": 0.07131692067213756,
      "grad_norm": 9.851771354675293,
      "learning_rate": 1.9917586028277987e-05,
      "loss": 11.2789,
      "step": 365
    },
    {
      "epoch": 0.07151230949589683,
      "grad_norm": 16.777217864990234,
      "learning_rate": 1.99167731989929e-05,
      "loss": 11.3211,
      "step": 366
    },
    {
      "epoch": 0.07170769831965612,
      "grad_norm": 10.766368865966797,
      "learning_rate": 1.991595639773227e-05,
      "loss": 10.4509,
      "step": 367
    },
    {
      "epoch": 0.0719030871434154,
      "grad_norm": 9.359428405761719,
      "learning_rate": 1.9915135624823243e-05,
      "loss": 11.8863,
      "step": 368
    },
    {
      "epoch": 0.07209847596717468,
      "grad_norm": 10.939064025878906,
      "learning_rate": 1.9914310880594572e-05,
      "loss": 12.8301,
      "step": 369
    },
    {
      "epoch": 0.07229386479093396,
      "grad_norm": 15.838484764099121,
      "learning_rate": 1.991348216537659e-05,
      "loss": 11.0152,
      "step": 370
    },
    {
      "epoch": 0.07248925361469324,
      "grad_norm": 16.605043411254883,
      "learning_rate": 1.9912649479501222e-05,
      "loss": 12.3227,
      "step": 371
    },
    {
      "epoch": 0.07268464243845252,
      "grad_norm": 7.643439769744873,
      "learning_rate": 1.9911812823301985e-05,
      "loss": 10.9601,
      "step": 372
    },
    {
      "epoch": 0.0728800312622118,
      "grad_norm": 9.731760025024414,
      "learning_rate": 1.9910972197113986e-05,
      "loss": 11.597,
      "step": 373
    },
    {
      "epoch": 0.07307542008597108,
      "grad_norm": 11.740914344787598,
      "learning_rate": 1.9910127601273922e-05,
      "loss": 11.6196,
      "step": 374
    },
    {
      "epoch": 0.07327080890973037,
      "grad_norm": 10.138679504394531,
      "learning_rate": 1.9909279036120082e-05,
      "loss": 11.4332,
      "step": 375
    },
    {
      "epoch": 0.07346619773348964,
      "grad_norm": 8.679524421691895,
      "learning_rate": 1.9908426501992337e-05,
      "loss": 12.6449,
      "step": 376
    },
    {
      "epoch": 0.07366158655724893,
      "grad_norm": 13.47012710571289,
      "learning_rate": 1.9907569999232158e-05,
      "loss": 10.7538,
      "step": 377
    },
    {
      "epoch": 0.0738569753810082,
      "grad_norm": 7.856994152069092,
      "learning_rate": 1.9906709528182602e-05,
      "loss": 12.6015,
      "step": 378
    },
    {
      "epoch": 0.07405236420476749,
      "grad_norm": 9.596478462219238,
      "learning_rate": 1.990584508918831e-05,
      "loss": 11.9684,
      "step": 379
    },
    {
      "epoch": 0.07424775302852676,
      "grad_norm": 11.648253440856934,
      "learning_rate": 1.9904976682595518e-05,
      "loss": 12.2241,
      "step": 380
    },
    {
      "epoch": 0.07444314185228605,
      "grad_norm": 8.147643089294434,
      "learning_rate": 1.9904104308752053e-05,
      "loss": 11.0065,
      "step": 381
    },
    {
      "epoch": 0.07463853067604533,
      "grad_norm": 9.387744903564453,
      "learning_rate": 1.9903227968007324e-05,
      "loss": 11.2652,
      "step": 382
    },
    {
      "epoch": 0.07483391949980461,
      "grad_norm": 7.9869818687438965,
      "learning_rate": 1.9902347660712336e-05,
      "loss": 10.4613,
      "step": 383
    },
    {
      "epoch": 0.07502930832356389,
      "grad_norm": 12.57647705078125,
      "learning_rate": 1.9901463387219677e-05,
      "loss": 13.2274,
      "step": 384
    },
    {
      "epoch": 0.07522469714732317,
      "grad_norm": 16.30275535583496,
      "learning_rate": 1.9900575147883525e-05,
      "loss": 10.7593,
      "step": 385
    },
    {
      "epoch": 0.07542008597108245,
      "grad_norm": 8.52121353149414,
      "learning_rate": 1.9899682943059652e-05,
      "loss": 11.4779,
      "step": 386
    },
    {
      "epoch": 0.07561547479484174,
      "grad_norm": 9.90449047088623,
      "learning_rate": 1.9898786773105407e-05,
      "loss": 12.9545,
      "step": 387
    },
    {
      "epoch": 0.07581086361860101,
      "grad_norm": 11.800531387329102,
      "learning_rate": 1.989788663837974e-05,
      "loss": 12.4126,
      "step": 388
    },
    {
      "epoch": 0.0760062524423603,
      "grad_norm": 7.592154502868652,
      "learning_rate": 1.989698253924318e-05,
      "loss": 11.6059,
      "step": 389
    },
    {
      "epoch": 0.07620164126611957,
      "grad_norm": 13.224897384643555,
      "learning_rate": 1.9896074476057846e-05,
      "loss": 11.1759,
      "step": 390
    },
    {
      "epoch": 0.07639703008987886,
      "grad_norm": 7.516778469085693,
      "learning_rate": 1.9895162449187452e-05,
      "loss": 12.0758,
      "step": 391
    },
    {
      "epoch": 0.07659241891363815,
      "grad_norm": 10.208797454833984,
      "learning_rate": 1.989424645899728e-05,
      "loss": 11.1313,
      "step": 392
    },
    {
      "epoch": 0.07678780773739742,
      "grad_norm": 14.074166297912598,
      "learning_rate": 1.9893326505854228e-05,
      "loss": 11.9664,
      "step": 393
    },
    {
      "epoch": 0.07698319656115671,
      "grad_norm": 9.178504943847656,
      "learning_rate": 1.989240259012675e-05,
      "loss": 11.5146,
      "step": 394
    },
    {
      "epoch": 0.07717858538491598,
      "grad_norm": 10.253775596618652,
      "learning_rate": 1.9891474712184916e-05,
      "loss": 11.8304,
      "step": 395
    },
    {
      "epoch": 0.07737397420867527,
      "grad_norm": 11.380575180053711,
      "learning_rate": 1.9890542872400368e-05,
      "loss": 10.6585,
      "step": 396
    },
    {
      "epoch": 0.07756936303243454,
      "grad_norm": 13.956233024597168,
      "learning_rate": 1.988960707114633e-05,
      "loss": 11.6757,
      "step": 397
    },
    {
      "epoch": 0.07776475185619383,
      "grad_norm": 9.273847579956055,
      "learning_rate": 1.9888667308797622e-05,
      "loss": 11.9675,
      "step": 398
    },
    {
      "epoch": 0.0779601406799531,
      "grad_norm": 8.791987419128418,
      "learning_rate": 1.988772358573065e-05,
      "loss": 10.9511,
      "step": 399
    },
    {
      "epoch": 0.07815552950371239,
      "grad_norm": 10.47549057006836,
      "learning_rate": 1.9886775902323405e-05,
      "loss": 11.0095,
      "step": 400
    },
    {
      "epoch": 0.07835091832747167,
      "grad_norm": 9.35380744934082,
      "learning_rate": 1.9885824258955462e-05,
      "loss": 11.3787,
      "step": 401
    },
    {
      "epoch": 0.07854630715123095,
      "grad_norm": 9.344833374023438,
      "learning_rate": 1.9884868656007982e-05,
      "loss": 12.0148,
      "step": 402
    },
    {
      "epoch": 0.07874169597499023,
      "grad_norm": 10.613936424255371,
      "learning_rate": 1.988390909386372e-05,
      "loss": 11.4727,
      "step": 403
    },
    {
      "epoch": 0.07893708479874952,
      "grad_norm": 9.871330261230469,
      "learning_rate": 1.9882945572907002e-05,
      "loss": 11.2291,
      "step": 404
    },
    {
      "epoch": 0.07913247362250879,
      "grad_norm": 9.063941955566406,
      "learning_rate": 1.9881978093523754e-05,
      "loss": 11.2424,
      "step": 405
    },
    {
      "epoch": 0.07932786244626808,
      "grad_norm": 10.158225059509277,
      "learning_rate": 1.988100665610148e-05,
      "loss": 11.7028,
      "step": 406
    },
    {
      "epoch": 0.07952325127002735,
      "grad_norm": 20.984745025634766,
      "learning_rate": 1.988003126102927e-05,
      "loss": 12.7531,
      "step": 407
    },
    {
      "epoch": 0.07971864009378664,
      "grad_norm": 10.408671379089355,
      "learning_rate": 1.9879051908697803e-05,
      "loss": 9.6428,
      "step": 408
    },
    {
      "epoch": 0.07991402891754591,
      "grad_norm": 11.17059326171875,
      "learning_rate": 1.9878068599499333e-05,
      "loss": 11.5591,
      "step": 409
    },
    {
      "epoch": 0.0801094177413052,
      "grad_norm": 17.67782211303711,
      "learning_rate": 1.9877081333827714e-05,
      "loss": 12.1476,
      "step": 410
    },
    {
      "epoch": 0.08030480656506447,
      "grad_norm": 10.006389617919922,
      "learning_rate": 1.987609011207837e-05,
      "loss": 9.6199,
      "step": 411
    },
    {
      "epoch": 0.08050019538882376,
      "grad_norm": 11.344561576843262,
      "learning_rate": 1.987509493464832e-05,
      "loss": 11.0631,
      "step": 412
    },
    {
      "epoch": 0.08069558421258304,
      "grad_norm": 7.435112476348877,
      "learning_rate": 1.987409580193616e-05,
      "loss": 9.9161,
      "step": 413
    },
    {
      "epoch": 0.08089097303634232,
      "grad_norm": 11.36435317993164,
      "learning_rate": 1.987309271434208e-05,
      "loss": 12.5855,
      "step": 414
    },
    {
      "epoch": 0.0810863618601016,
      "grad_norm": 16.058595657348633,
      "learning_rate": 1.9872085672267842e-05,
      "loss": 11.5972,
      "step": 415
    },
    {
      "epoch": 0.08128175068386088,
      "grad_norm": 12.252034187316895,
      "learning_rate": 1.9871074676116802e-05,
      "loss": 9.6958,
      "step": 416
    },
    {
      "epoch": 0.08147713950762016,
      "grad_norm": 11.003941535949707,
      "learning_rate": 1.9870059726293892e-05,
      "loss": 10.7821,
      "step": 417
    },
    {
      "epoch": 0.08167252833137945,
      "grad_norm": 13.808038711547852,
      "learning_rate": 1.986904082320563e-05,
      "loss": 11.5391,
      "step": 418
    },
    {
      "epoch": 0.08186791715513872,
      "grad_norm": 12.557829856872559,
      "learning_rate": 1.986801796726012e-05,
      "loss": 12.2014,
      "step": 419
    },
    {
      "epoch": 0.08206330597889801,
      "grad_norm": 14.017550468444824,
      "learning_rate": 1.9866991158867047e-05,
      "loss": 10.9317,
      "step": 420
    },
    {
      "epoch": 0.08225869480265728,
      "grad_norm": 12.062165260314941,
      "learning_rate": 1.986596039843768e-05,
      "loss": 10.8878,
      "step": 421
    },
    {
      "epoch": 0.08245408362641657,
      "grad_norm": 10.640327453613281,
      "learning_rate": 1.9864925686384874e-05,
      "loss": 10.019,
      "step": 422
    },
    {
      "epoch": 0.08264947245017586,
      "grad_norm": 6.800312519073486,
      "learning_rate": 1.986388702312306e-05,
      "loss": 10.3276,
      "step": 423
    },
    {
      "epoch": 0.08284486127393513,
      "grad_norm": 8.374358177185059,
      "learning_rate": 1.9862844409068252e-05,
      "loss": 11.5054,
      "step": 424
    },
    {
      "epoch": 0.08304025009769442,
      "grad_norm": 11.760390281677246,
      "learning_rate": 1.9861797844638055e-05,
      "loss": 10.7713,
      "step": 425
    },
    {
      "epoch": 0.08323563892145369,
      "grad_norm": 15.597134590148926,
      "learning_rate": 1.9860747330251647e-05,
      "loss": 12.2257,
      "step": 426
    },
    {
      "epoch": 0.08343102774521298,
      "grad_norm": 11.854767799377441,
      "learning_rate": 1.985969286632979e-05,
      "loss": 12.0446,
      "step": 427
    },
    {
      "epoch": 0.08362641656897225,
      "grad_norm": 8.871596336364746,
      "learning_rate": 1.9858634453294835e-05,
      "loss": 11.391,
      "step": 428
    },
    {
      "epoch": 0.08382180539273154,
      "grad_norm": 6.248795032501221,
      "learning_rate": 1.9857572091570704e-05,
      "loss": 10.8528,
      "step": 429
    },
    {
      "epoch": 0.08401719421649081,
      "grad_norm": 10.798820495605469,
      "learning_rate": 1.985650578158291e-05,
      "loss": 10.1009,
      "step": 430
    },
    {
      "epoch": 0.0842125830402501,
      "grad_norm": 11.153851509094238,
      "learning_rate": 1.9855435523758542e-05,
      "loss": 10.4832,
      "step": 431
    },
    {
      "epoch": 0.08440797186400938,
      "grad_norm": 7.624647617340088,
      "learning_rate": 1.9854361318526273e-05,
      "loss": 12.0875,
      "step": 432
    },
    {
      "epoch": 0.08460336068776866,
      "grad_norm": 9.647201538085938,
      "learning_rate": 1.9853283166316348e-05,
      "loss": 11.6031,
      "step": 433
    },
    {
      "epoch": 0.08479874951152794,
      "grad_norm": 10.949299812316895,
      "learning_rate": 1.9852201067560607e-05,
      "loss": 11.5041,
      "step": 434
    },
    {
      "epoch": 0.08499413833528723,
      "grad_norm": 12.349892616271973,
      "learning_rate": 1.9851115022692468e-05,
      "loss": 11.2026,
      "step": 435
    },
    {
      "epoch": 0.0851895271590465,
      "grad_norm": 9.193042755126953,
      "learning_rate": 1.9850025032146916e-05,
      "loss": 10.3315,
      "step": 436
    },
    {
      "epoch": 0.08538491598280579,
      "grad_norm": 14.063774108886719,
      "learning_rate": 1.984893109636053e-05,
      "loss": 12.1384,
      "step": 437
    },
    {
      "epoch": 0.08558030480656506,
      "grad_norm": 13.098580360412598,
      "learning_rate": 1.984783321577147e-05,
      "loss": 10.7132,
      "step": 438
    },
    {
      "epoch": 0.08577569363032435,
      "grad_norm": 9.289295196533203,
      "learning_rate": 1.9846731390819463e-05,
      "loss": 11.0615,
      "step": 439
    },
    {
      "epoch": 0.08597108245408362,
      "grad_norm": 13.094406127929688,
      "learning_rate": 1.984562562194583e-05,
      "loss": 11.685,
      "step": 440
    },
    {
      "epoch": 0.08616647127784291,
      "grad_norm": 10.86340045928955,
      "learning_rate": 1.9844515909593466e-05,
      "loss": 11.3287,
      "step": 441
    },
    {
      "epoch": 0.08636186010160218,
      "grad_norm": 14.637931823730469,
      "learning_rate": 1.984340225420684e-05,
      "loss": 11.9579,
      "step": 442
    },
    {
      "epoch": 0.08655724892536147,
      "grad_norm": 16.27992820739746,
      "learning_rate": 1.984228465623201e-05,
      "loss": 11.3188,
      "step": 443
    },
    {
      "epoch": 0.08675263774912075,
      "grad_norm": 8.437994003295898,
      "learning_rate": 1.9841163116116607e-05,
      "loss": 10.107,
      "step": 444
    },
    {
      "epoch": 0.08694802657288003,
      "grad_norm": 12.490747451782227,
      "learning_rate": 1.9840037634309846e-05,
      "loss": 12.4261,
      "step": 445
    },
    {
      "epoch": 0.0871434153966393,
      "grad_norm": 9.785431861877441,
      "learning_rate": 1.9838908211262512e-05,
      "loss": 10.9658,
      "step": 446
    },
    {
      "epoch": 0.0873388042203986,
      "grad_norm": 9.814424514770508,
      "learning_rate": 1.983777484742698e-05,
      "loss": 12.4832,
      "step": 447
    },
    {
      "epoch": 0.08753419304415787,
      "grad_norm": 9.344550132751465,
      "learning_rate": 1.983663754325719e-05,
      "loss": 11.3044,
      "step": 448
    },
    {
      "epoch": 0.08772958186791716,
      "grad_norm": 12.776989936828613,
      "learning_rate": 1.9835496299208675e-05,
      "loss": 12.2846,
      "step": 449
    },
    {
      "epoch": 0.08792497069167643,
      "grad_norm": 6.569207668304443,
      "learning_rate": 1.9834351115738535e-05,
      "loss": 11.6907,
      "step": 450
    },
    {
      "epoch": 0.08812035951543572,
      "grad_norm": 8.714112281799316,
      "learning_rate": 1.9833201993305452e-05,
      "loss": 11.3617,
      "step": 451
    },
    {
      "epoch": 0.088315748339195,
      "grad_norm": 9.547415733337402,
      "learning_rate": 1.9832048932369684e-05,
      "loss": 12.0239,
      "step": 452
    },
    {
      "epoch": 0.08851113716295428,
      "grad_norm": 8.278121948242188,
      "learning_rate": 1.9830891933393068e-05,
      "loss": 11.1598,
      "step": 453
    },
    {
      "epoch": 0.08870652598671357,
      "grad_norm": 12.530584335327148,
      "learning_rate": 1.982973099683902e-05,
      "loss": 11.3444,
      "step": 454
    },
    {
      "epoch": 0.08890191481047284,
      "grad_norm": 7.621415615081787,
      "learning_rate": 1.982856612317253e-05,
      "loss": 10.6134,
      "step": 455
    },
    {
      "epoch": 0.08909730363423213,
      "grad_norm": 8.175232887268066,
      "learning_rate": 1.9827397312860165e-05,
      "loss": 12.4467,
      "step": 456
    },
    {
      "epoch": 0.0892926924579914,
      "grad_norm": 12.001221656799316,
      "learning_rate": 1.982622456637007e-05,
      "loss": 10.5231,
      "step": 457
    },
    {
      "epoch": 0.08948808128175069,
      "grad_norm": 10.777896881103516,
      "learning_rate": 1.982504788417197e-05,
      "loss": 11.8483,
      "step": 458
    },
    {
      "epoch": 0.08968347010550996,
      "grad_norm": 11.15634822845459,
      "learning_rate": 1.982386726673716e-05,
      "loss": 10.4165,
      "step": 459
    },
    {
      "epoch": 0.08987885892926925,
      "grad_norm": 15.486494064331055,
      "learning_rate": 1.9822682714538512e-05,
      "loss": 12.3944,
      "step": 460
    },
    {
      "epoch": 0.09007424775302852,
      "grad_norm": 17.380624771118164,
      "learning_rate": 1.982149422805048e-05,
      "loss": 11.0951,
      "step": 461
    },
    {
      "epoch": 0.09026963657678781,
      "grad_norm": 23.118669509887695,
      "learning_rate": 1.9820301807749084e-05,
      "loss": 11.6186,
      "step": 462
    },
    {
      "epoch": 0.09046502540054709,
      "grad_norm": 12.999380111694336,
      "learning_rate": 1.9819105454111932e-05,
      "loss": 11.1989,
      "step": 463
    },
    {
      "epoch": 0.09066041422430637,
      "grad_norm": 10.317048072814941,
      "learning_rate": 1.9817905167618194e-05,
      "loss": 11.1085,
      "step": 464
    },
    {
      "epoch": 0.09085580304806565,
      "grad_norm": 9.965431213378906,
      "learning_rate": 1.981670094874863e-05,
      "loss": 11.5608,
      "step": 465
    },
    {
      "epoch": 0.09105119187182494,
      "grad_norm": 8.059679985046387,
      "learning_rate": 1.981549279798556e-05,
      "loss": 10.2311,
      "step": 466
    },
    {
      "epoch": 0.09124658069558421,
      "grad_norm": 12.043992042541504,
      "learning_rate": 1.981428071581289e-05,
      "loss": 11.189,
      "step": 467
    },
    {
      "epoch": 0.0914419695193435,
      "grad_norm": 8.804059028625488,
      "learning_rate": 1.9813064702716094e-05,
      "loss": 12.1722,
      "step": 468
    },
    {
      "epoch": 0.09163735834310277,
      "grad_norm": 9.039908409118652,
      "learning_rate": 1.9811844759182226e-05,
      "loss": 9.9737,
      "step": 469
    },
    {
      "epoch": 0.09183274716686206,
      "grad_norm": 11.43051528930664,
      "learning_rate": 1.9810620885699908e-05,
      "loss": 11.2245,
      "step": 470
    },
    {
      "epoch": 0.09202813599062133,
      "grad_norm": 12.444183349609375,
      "learning_rate": 1.9809393082759342e-05,
      "loss": 13.1322,
      "step": 471
    },
    {
      "epoch": 0.09222352481438062,
      "grad_norm": 8.038171768188477,
      "learning_rate": 1.98081613508523e-05,
      "loss": 12.363,
      "step": 472
    },
    {
      "epoch": 0.0924189136381399,
      "grad_norm": 14.102073669433594,
      "learning_rate": 1.9806925690472128e-05,
      "loss": 11.5428,
      "step": 473
    },
    {
      "epoch": 0.09261430246189918,
      "grad_norm": 10.017738342285156,
      "learning_rate": 1.980568610211375e-05,
      "loss": 12.0738,
      "step": 474
    },
    {
      "epoch": 0.09280969128565845,
      "grad_norm": 8.079500198364258,
      "learning_rate": 1.980444258627365e-05,
      "loss": 10.8434,
      "step": 475
    },
    {
      "epoch": 0.09300508010941774,
      "grad_norm": 10.891678810119629,
      "learning_rate": 1.9803195143449905e-05,
      "loss": 11.4258,
      "step": 476
    },
    {
      "epoch": 0.09320046893317702,
      "grad_norm": 8.259817123413086,
      "learning_rate": 1.9801943774142152e-05,
      "loss": 11.2229,
      "step": 477
    },
    {
      "epoch": 0.0933958577569363,
      "grad_norm": 9.344334602355957,
      "learning_rate": 1.9800688478851603e-05,
      "loss": 11.9778,
      "step": 478
    },
    {
      "epoch": 0.09359124658069558,
      "grad_norm": 11.072267532348633,
      "learning_rate": 1.979942925808104e-05,
      "loss": 11.1355,
      "step": 479
    },
    {
      "epoch": 0.09378663540445487,
      "grad_norm": 11.740673065185547,
      "learning_rate": 1.9798166112334824e-05,
      "loss": 9.7471,
      "step": 480
    },
    {
      "epoch": 0.09398202422821414,
      "grad_norm": 8.792943954467773,
      "learning_rate": 1.9796899042118883e-05,
      "loss": 10.8433,
      "step": 481
    },
    {
      "epoch": 0.09417741305197343,
      "grad_norm": 9.371011734008789,
      "learning_rate": 1.9795628047940714e-05,
      "loss": 11.7686,
      "step": 482
    },
    {
      "epoch": 0.09437280187573271,
      "grad_norm": 11.211883544921875,
      "learning_rate": 1.9794353130309395e-05,
      "loss": 11.6966,
      "step": 483
    },
    {
      "epoch": 0.09456819069949199,
      "grad_norm": 16.014528274536133,
      "learning_rate": 1.9793074289735566e-05,
      "loss": 11.0515,
      "step": 484
    },
    {
      "epoch": 0.09476357952325128,
      "grad_norm": 8.87293815612793,
      "learning_rate": 1.9791791526731444e-05,
      "loss": 9.333,
      "step": 485
    },
    {
      "epoch": 0.09495896834701055,
      "grad_norm": 18.065631866455078,
      "learning_rate": 1.9790504841810817e-05,
      "loss": 10.9915,
      "step": 486
    },
    {
      "epoch": 0.09515435717076984,
      "grad_norm": 9.98987865447998,
      "learning_rate": 1.9789214235489042e-05,
      "loss": 10.7225,
      "step": 487
    },
    {
      "epoch": 0.09534974599452911,
      "grad_norm": 21.688825607299805,
      "learning_rate": 1.9787919708283044e-05,
      "loss": 11.0898,
      "step": 488
    },
    {
      "epoch": 0.0955451348182884,
      "grad_norm": 8.642053604125977,
      "learning_rate": 1.9786621260711322e-05,
      "loss": 9.916,
      "step": 489
    },
    {
      "epoch": 0.09574052364204767,
      "grad_norm": 10.877992630004883,
      "learning_rate": 1.978531889329395e-05,
      "loss": 12.2251,
      "step": 490
    },
    {
      "epoch": 0.09593591246580696,
      "grad_norm": 9.516289710998535,
      "learning_rate": 1.9784012606552557e-05,
      "loss": 12.1528,
      "step": 491
    },
    {
      "epoch": 0.09613130128956623,
      "grad_norm": 8.622530937194824,
      "learning_rate": 1.978270240101036e-05,
      "loss": 11.5102,
      "step": 492
    },
    {
      "epoch": 0.09632669011332552,
      "grad_norm": 8.162901878356934,
      "learning_rate": 1.9781388277192132e-05,
      "loss": 10.8012,
      "step": 493
    },
    {
      "epoch": 0.0965220789370848,
      "grad_norm": 10.565152168273926,
      "learning_rate": 1.9780070235624222e-05,
      "loss": 10.1863,
      "step": 494
    },
    {
      "epoch": 0.09671746776084408,
      "grad_norm": 7.8273749351501465,
      "learning_rate": 1.977874827683455e-05,
      "loss": 11.5141,
      "step": 495
    },
    {
      "epoch": 0.09691285658460336,
      "grad_norm": 9.111390113830566,
      "learning_rate": 1.9777422401352594e-05,
      "loss": 11.9115,
      "step": 496
    },
    {
      "epoch": 0.09710824540836265,
      "grad_norm": 8.689447402954102,
      "learning_rate": 1.9776092609709416e-05,
      "loss": 9.9112,
      "step": 497
    },
    {
      "epoch": 0.09730363423212192,
      "grad_norm": 9.937636375427246,
      "learning_rate": 1.9774758902437634e-05,
      "loss": 10.3904,
      "step": 498
    },
    {
      "epoch": 0.0974990230558812,
      "grad_norm": 12.398046493530273,
      "learning_rate": 1.9773421280071437e-05,
      "loss": 10.1024,
      "step": 499
    },
    {
      "epoch": 0.09769441187964048,
      "grad_norm": 9.194086074829102,
      "learning_rate": 1.977207974314659e-05,
      "loss": 11.3953,
      "step": 500
    },
    {
      "epoch": 0.09788980070339977,
      "grad_norm": 7.784787178039551,
      "learning_rate": 1.977073429220042e-05,
      "loss": 10.7072,
      "step": 501
    },
    {
      "epoch": 0.09808518952715904,
      "grad_norm": 15.563216209411621,
      "learning_rate": 1.976938492777182e-05,
      "loss": 10.7934,
      "step": 502
    },
    {
      "epoch": 0.09828057835091833,
      "grad_norm": 10.296627044677734,
      "learning_rate": 1.9768031650401248e-05,
      "loss": 11.5148,
      "step": 503
    },
    {
      "epoch": 0.0984759671746776,
      "grad_norm": 9.36056137084961,
      "learning_rate": 1.976667446063074e-05,
      "loss": 11.3866,
      "step": 504
    },
    {
      "epoch": 0.09867135599843689,
      "grad_norm": 15.350499153137207,
      "learning_rate": 1.976531335900389e-05,
      "loss": 13.5321,
      "step": 505
    },
    {
      "epoch": 0.09886674482219616,
      "grad_norm": 11.02067756652832,
      "learning_rate": 1.9763948346065865e-05,
      "loss": 9.5924,
      "step": 506
    },
    {
      "epoch": 0.09906213364595545,
      "grad_norm": 10.623998641967773,
      "learning_rate": 1.976257942236339e-05,
      "loss": 10.4124,
      "step": 507
    },
    {
      "epoch": 0.09925752246971473,
      "grad_norm": 11.29489517211914,
      "learning_rate": 1.976120658844476e-05,
      "loss": 10.2982,
      "step": 508
    },
    {
      "epoch": 0.09945291129347401,
      "grad_norm": 15.255050659179688,
      "learning_rate": 1.9759829844859843e-05,
      "loss": 11.665,
      "step": 509
    },
    {
      "epoch": 0.09964830011723329,
      "grad_norm": 10.669962882995605,
      "learning_rate": 1.9758449192160067e-05,
      "loss": 11.3428,
      "step": 510
    },
    {
      "epoch": 0.09984368894099258,
      "grad_norm": 11.840312004089355,
      "learning_rate": 1.9757064630898422e-05,
      "loss": 11.7798,
      "step": 511
    },
    {
      "epoch": 0.10003907776475186,
      "grad_norm": 9.912818908691406,
      "learning_rate": 1.9755676161629474e-05,
      "loss": 9.7258,
      "step": 512
    },
    {
      "epoch": 0.10023446658851114,
      "grad_norm": 9.993812561035156,
      "learning_rate": 1.9754283784909344e-05,
      "loss": 11.4899,
      "step": 513
    },
    {
      "epoch": 0.10042985541227042,
      "grad_norm": 8.480623245239258,
      "learning_rate": 1.975288750129572e-05,
      "loss": 10.9759,
      "step": 514
    },
    {
      "epoch": 0.1006252442360297,
      "grad_norm": 8.028963088989258,
      "learning_rate": 1.975148731134786e-05,
      "loss": 11.3543,
      "step": 515
    },
    {
      "epoch": 0.10082063305978899,
      "grad_norm": 9.241572380065918,
      "learning_rate": 1.9750083215626588e-05,
      "loss": 10.422,
      "step": 516
    },
    {
      "epoch": 0.10101602188354826,
      "grad_norm": 8.950488090515137,
      "learning_rate": 1.974867521469428e-05,
      "loss": 11.0477,
      "step": 517
    },
    {
      "epoch": 0.10121141070730755,
      "grad_norm": 9.833844184875488,
      "learning_rate": 1.9747263309114884e-05,
      "loss": 12.0173,
      "step": 518
    },
    {
      "epoch": 0.10140679953106682,
      "grad_norm": 8.743428230285645,
      "learning_rate": 1.9745847499453918e-05,
      "loss": 11.5385,
      "step": 519
    },
    {
      "epoch": 0.10160218835482611,
      "grad_norm": 7.573606014251709,
      "learning_rate": 1.9744427786278456e-05,
      "loss": 10.3587,
      "step": 520
    },
    {
      "epoch": 0.10179757717858538,
      "grad_norm": 11.663070678710938,
      "learning_rate": 1.9743004170157137e-05,
      "loss": 11.1453,
      "step": 521
    },
    {
      "epoch": 0.10199296600234467,
      "grad_norm": 9.460543632507324,
      "learning_rate": 1.9741576651660158e-05,
      "loss": 12.3356,
      "step": 522
    },
    {
      "epoch": 0.10218835482610394,
      "grad_norm": 11.570713996887207,
      "learning_rate": 1.9740145231359295e-05,
      "loss": 10.9296,
      "step": 523
    },
    {
      "epoch": 0.10238374364986323,
      "grad_norm": 11.01218318939209,
      "learning_rate": 1.9738709909827866e-05,
      "loss": 10.4656,
      "step": 524
    },
    {
      "epoch": 0.1025791324736225,
      "grad_norm": 10.380640029907227,
      "learning_rate": 1.9737270687640767e-05,
      "loss": 11.2092,
      "step": 525
    },
    {
      "epoch": 0.1027745212973818,
      "grad_norm": 12.392244338989258,
      "learning_rate": 1.9735827565374447e-05,
      "loss": 11.3003,
      "step": 526
    },
    {
      "epoch": 0.10296991012114107,
      "grad_norm": 13.215970993041992,
      "learning_rate": 1.9734380543606932e-05,
      "loss": 11.0708,
      "step": 527
    },
    {
      "epoch": 0.10316529894490035,
      "grad_norm": 8.894550323486328,
      "learning_rate": 1.9732929622917785e-05,
      "loss": 10.2508,
      "step": 528
    },
    {
      "epoch": 0.10336068776865963,
      "grad_norm": 9.56419849395752,
      "learning_rate": 1.9731474803888156e-05,
      "loss": 10.2722,
      "step": 529
    },
    {
      "epoch": 0.10355607659241892,
      "grad_norm": 17.47562026977539,
      "learning_rate": 1.973001608710074e-05,
      "loss": 12.7116,
      "step": 530
    },
    {
      "epoch": 0.10375146541617819,
      "grad_norm": 14.577072143554688,
      "learning_rate": 1.9728553473139795e-05,
      "loss": 11.1043,
      "step": 531
    },
    {
      "epoch": 0.10394685423993748,
      "grad_norm": 9.963520050048828,
      "learning_rate": 1.9727086962591153e-05,
      "loss": 12.1775,
      "step": 532
    },
    {
      "epoch": 0.10414224306369675,
      "grad_norm": 9.970325469970703,
      "learning_rate": 1.9725616556042188e-05,
      "loss": 11.3413,
      "step": 533
    },
    {
      "epoch": 0.10433763188745604,
      "grad_norm": 10.894572257995605,
      "learning_rate": 1.972414225408185e-05,
      "loss": 10.9785,
      "step": 534
    },
    {
      "epoch": 0.10453302071121531,
      "grad_norm": 10.209842681884766,
      "learning_rate": 1.9722664057300635e-05,
      "loss": 10.9789,
      "step": 535
    },
    {
      "epoch": 0.1047284095349746,
      "grad_norm": 8.99254322052002,
      "learning_rate": 1.9721181966290614e-05,
      "loss": 11.8011,
      "step": 536
    },
    {
      "epoch": 0.10492379835873387,
      "grad_norm": 9.370485305786133,
      "learning_rate": 1.971969598164541e-05,
      "loss": 9.4472,
      "step": 537
    },
    {
      "epoch": 0.10511918718249316,
      "grad_norm": 8.806024551391602,
      "learning_rate": 1.97182061039602e-05,
      "loss": 10.4979,
      "step": 538
    },
    {
      "epoch": 0.10531457600625244,
      "grad_norm": 9.16971492767334,
      "learning_rate": 1.971671233383173e-05,
      "loss": 11.3504,
      "step": 539
    },
    {
      "epoch": 0.10550996483001172,
      "grad_norm": 8.441150665283203,
      "learning_rate": 1.9715214671858303e-05,
      "loss": 10.4157,
      "step": 540
    },
    {
      "epoch": 0.105705353653771,
      "grad_norm": 12.984179496765137,
      "learning_rate": 1.9713713118639778e-05,
      "loss": 10.6053,
      "step": 541
    },
    {
      "epoch": 0.10590074247753029,
      "grad_norm": 13.540075302124023,
      "learning_rate": 1.9712207674777572e-05,
      "loss": 10.1771,
      "step": 542
    },
    {
      "epoch": 0.10609613130128957,
      "grad_norm": 13.014501571655273,
      "learning_rate": 1.9710698340874663e-05,
      "loss": 11.6918,
      "step": 543
    },
    {
      "epoch": 0.10629152012504885,
      "grad_norm": 13.211419105529785,
      "learning_rate": 1.9709185117535586e-05,
      "loss": 11.1228,
      "step": 544
    },
    {
      "epoch": 0.10648690894880813,
      "grad_norm": 13.059433937072754,
      "learning_rate": 1.9707668005366435e-05,
      "loss": 11.5204,
      "step": 545
    },
    {
      "epoch": 0.10668229777256741,
      "grad_norm": 12.231781959533691,
      "learning_rate": 1.9706147004974858e-05,
      "loss": 11.4718,
      "step": 546
    },
    {
      "epoch": 0.1068776865963267,
      "grad_norm": 9.824128150939941,
      "learning_rate": 1.9704622116970067e-05,
      "loss": 9.9623,
      "step": 547
    },
    {
      "epoch": 0.10707307542008597,
      "grad_norm": 8.887593269348145,
      "learning_rate": 1.9703093341962822e-05,
      "loss": 10.9046,
      "step": 548
    },
    {
      "epoch": 0.10726846424384526,
      "grad_norm": 9.651532173156738,
      "learning_rate": 1.970156068056545e-05,
      "loss": 11.2181,
      "step": 549
    },
    {
      "epoch": 0.10746385306760453,
      "grad_norm": 11.166255950927734,
      "learning_rate": 1.9700024133391826e-05,
      "loss": 12.0837,
      "step": 550
    },
    {
      "epoch": 0.10765924189136382,
      "grad_norm": 9.001782417297363,
      "learning_rate": 1.9698483701057384e-05,
      "loss": 10.9068,
      "step": 551
    },
    {
      "epoch": 0.10785463071512309,
      "grad_norm": 7.795221328735352,
      "learning_rate": 1.969693938417912e-05,
      "loss": 11.1963,
      "step": 552
    },
    {
      "epoch": 0.10805001953888238,
      "grad_norm": 11.027246475219727,
      "learning_rate": 1.9695391183375575e-05,
      "loss": 11.813,
      "step": 553
    },
    {
      "epoch": 0.10824540836264165,
      "grad_norm": 8.13861083984375,
      "learning_rate": 1.9693839099266855e-05,
      "loss": 11.1423,
      "step": 554
    },
    {
      "epoch": 0.10844079718640094,
      "grad_norm": 9.753663063049316,
      "learning_rate": 1.9692283132474616e-05,
      "loss": 11.9865,
      "step": 555
    },
    {
      "epoch": 0.10863618601016022,
      "grad_norm": 14.523602485656738,
      "learning_rate": 1.969072328362207e-05,
      "loss": 12.1644,
      "step": 556
    },
    {
      "epoch": 0.1088315748339195,
      "grad_norm": 9.290946006774902,
      "learning_rate": 1.968915955333399e-05,
      "loss": 10.767,
      "step": 557
    },
    {
      "epoch": 0.10902696365767878,
      "grad_norm": 8.2459077835083,
      "learning_rate": 1.9687591942236696e-05,
      "loss": 10.6439,
      "step": 558
    },
    {
      "epoch": 0.10922235248143806,
      "grad_norm": 9.10664176940918,
      "learning_rate": 1.968602045095806e-05,
      "loss": 10.5131,
      "step": 559
    },
    {
      "epoch": 0.10941774130519734,
      "grad_norm": 14.061135292053223,
      "learning_rate": 1.9684445080127518e-05,
      "loss": 10.7492,
      "step": 560
    },
    {
      "epoch": 0.10961313012895663,
      "grad_norm": 11.38359260559082,
      "learning_rate": 1.968286583037606e-05,
      "loss": 10.3915,
      "step": 561
    },
    {
      "epoch": 0.1098085189527159,
      "grad_norm": 7.756126403808594,
      "learning_rate": 1.9681282702336215e-05,
      "loss": 11.4694,
      "step": 562
    },
    {
      "epoch": 0.11000390777647519,
      "grad_norm": 9.486867904663086,
      "learning_rate": 1.9679695696642075e-05,
      "loss": 10.9713,
      "step": 563
    },
    {
      "epoch": 0.11019929660023446,
      "grad_norm": 7.7375264167785645,
      "learning_rate": 1.9678104813929292e-05,
      "loss": 11.4329,
      "step": 564
    },
    {
      "epoch": 0.11039468542399375,
      "grad_norm": 10.03549861907959,
      "learning_rate": 1.9676510054835064e-05,
      "loss": 12.6544,
      "step": 565
    },
    {
      "epoch": 0.11059007424775302,
      "grad_norm": 16.695438385009766,
      "learning_rate": 1.9674911419998133e-05,
      "loss": 11.2948,
      "step": 566
    },
    {
      "epoch": 0.11078546307151231,
      "grad_norm": 11.747823715209961,
      "learning_rate": 1.9673308910058807e-05,
      "loss": 10.7342,
      "step": 567
    },
    {
      "epoch": 0.11098085189527158,
      "grad_norm": 13.515654563903809,
      "learning_rate": 1.9671702525658946e-05,
      "loss": 11.5844,
      "step": 568
    },
    {
      "epoch": 0.11117624071903087,
      "grad_norm": 13.833075523376465,
      "learning_rate": 1.967009226744195e-05,
      "loss": 11.388,
      "step": 569
    },
    {
      "epoch": 0.11137162954279015,
      "grad_norm": 9.444323539733887,
      "learning_rate": 1.9668478136052776e-05,
      "loss": 10.6101,
      "step": 570
    },
    {
      "epoch": 0.11156701836654943,
      "grad_norm": 13.008130073547363,
      "learning_rate": 1.966686013213794e-05,
      "loss": 9.3365,
      "step": 571
    },
    {
      "epoch": 0.11176240719030871,
      "grad_norm": 11.551283836364746,
      "learning_rate": 1.96652382563455e-05,
      "loss": 10.8201,
      "step": 572
    },
    {
      "epoch": 0.111957796014068,
      "grad_norm": 9.139278411865234,
      "learning_rate": 1.9663612509325064e-05,
      "loss": 9.7352,
      "step": 573
    },
    {
      "epoch": 0.11215318483782728,
      "grad_norm": 9.821352005004883,
      "learning_rate": 1.9661982891727803e-05,
      "loss": 10.9528,
      "step": 574
    },
    {
      "epoch": 0.11234857366158656,
      "grad_norm": 16.140743255615234,
      "learning_rate": 1.966034940420642e-05,
      "loss": 12.4749,
      "step": 575
    },
    {
      "epoch": 0.11254396248534584,
      "grad_norm": 11.304898262023926,
      "learning_rate": 1.965871204741518e-05,
      "loss": 10.5936,
      "step": 576
    },
    {
      "epoch": 0.11273935130910512,
      "grad_norm": 10.936923027038574,
      "learning_rate": 1.9657070822009903e-05,
      "loss": 11.798,
      "step": 577
    },
    {
      "epoch": 0.1129347401328644,
      "grad_norm": 12.186397552490234,
      "learning_rate": 1.965542572864794e-05,
      "loss": 11.4177,
      "step": 578
    },
    {
      "epoch": 0.11313012895662368,
      "grad_norm": 12.100482940673828,
      "learning_rate": 1.9653776767988208e-05,
      "loss": 11.537,
      "step": 579
    },
    {
      "epoch": 0.11332551778038297,
      "grad_norm": 17.33932876586914,
      "learning_rate": 1.9652123940691163e-05,
      "loss": 11.5476,
      "step": 580
    },
    {
      "epoch": 0.11352090660414224,
      "grad_norm": 36.325687408447266,
      "learning_rate": 1.965046724741882e-05,
      "loss": 10.6124,
      "step": 581
    },
    {
      "epoch": 0.11371629542790153,
      "grad_norm": 12.444219589233398,
      "learning_rate": 1.964880668883473e-05,
      "loss": 10.8188,
      "step": 582
    },
    {
      "epoch": 0.1139116842516608,
      "grad_norm": 11.519243240356445,
      "learning_rate": 1.9647142265603998e-05,
      "loss": 10.4957,
      "step": 583
    },
    {
      "epoch": 0.11410707307542009,
      "grad_norm": 12.104283332824707,
      "learning_rate": 1.9645473978393283e-05,
      "loss": 11.0054,
      "step": 584
    },
    {
      "epoch": 0.11430246189917936,
      "grad_norm": 10.913578033447266,
      "learning_rate": 1.9643801827870778e-05,
      "loss": 10.4854,
      "step": 585
    },
    {
      "epoch": 0.11449785072293865,
      "grad_norm": 11.549110412597656,
      "learning_rate": 1.964212581470624e-05,
      "loss": 10.5053,
      "step": 586
    },
    {
      "epoch": 0.11469323954669793,
      "grad_norm": 13.377059936523438,
      "learning_rate": 1.964044593957096e-05,
      "loss": 11.7833,
      "step": 587
    },
    {
      "epoch": 0.11488862837045721,
      "grad_norm": 13.456280708312988,
      "learning_rate": 1.9638762203137774e-05,
      "loss": 11.9166,
      "step": 588
    },
    {
      "epoch": 0.11508401719421649,
      "grad_norm": 8.56250286102295,
      "learning_rate": 1.9637074606081084e-05,
      "loss": 10.6182,
      "step": 589
    },
    {
      "epoch": 0.11527940601797577,
      "grad_norm": 15.668073654174805,
      "learning_rate": 1.9635383149076818e-05,
      "loss": 9.1693,
      "step": 590
    },
    {
      "epoch": 0.11547479484173505,
      "grad_norm": 10.171055793762207,
      "learning_rate": 1.963368783280245e-05,
      "loss": 10.1066,
      "step": 591
    },
    {
      "epoch": 0.11567018366549434,
      "grad_norm": 8.605528831481934,
      "learning_rate": 1.9631988657937018e-05,
      "loss": 9.7485,
      "step": 592
    },
    {
      "epoch": 0.11586557248925361,
      "grad_norm": 10.60910415649414,
      "learning_rate": 1.963028562516109e-05,
      "loss": 11.2119,
      "step": 593
    },
    {
      "epoch": 0.1160609613130129,
      "grad_norm": 10.013952255249023,
      "learning_rate": 1.962857873515678e-05,
      "loss": 10.3981,
      "step": 594
    },
    {
      "epoch": 0.11625635013677217,
      "grad_norm": 11.97077465057373,
      "learning_rate": 1.9626867988607755e-05,
      "loss": 11.9388,
      "step": 595
    },
    {
      "epoch": 0.11645173896053146,
      "grad_norm": 9.654369354248047,
      "learning_rate": 1.9625153386199225e-05,
      "loss": 10.5881,
      "step": 596
    },
    {
      "epoch": 0.11664712778429073,
      "grad_norm": 8.684782028198242,
      "learning_rate": 1.9623434928617936e-05,
      "loss": 10.8913,
      "step": 597
    },
    {
      "epoch": 0.11684251660805002,
      "grad_norm": 9.611656188964844,
      "learning_rate": 1.962171261655218e-05,
      "loss": 10.6154,
      "step": 598
    },
    {
      "epoch": 0.1170379054318093,
      "grad_norm": 10.223007202148438,
      "learning_rate": 1.9619986450691808e-05,
      "loss": 10.6338,
      "step": 599
    },
    {
      "epoch": 0.11723329425556858,
      "grad_norm": 10.545039176940918,
      "learning_rate": 1.961825643172819e-05,
      "loss": 11.1157,
      "step": 600
    },
    {
      "epoch": 0.11742868307932786,
      "grad_norm": 21.339628219604492,
      "learning_rate": 1.9616522560354265e-05,
      "loss": 11.5241,
      "step": 601
    },
    {
      "epoch": 0.11762407190308714,
      "grad_norm": 10.725296020507812,
      "learning_rate": 1.9614784837264493e-05,
      "loss": 10.6171,
      "step": 602
    },
    {
      "epoch": 0.11781946072684643,
      "grad_norm": 11.286303520202637,
      "learning_rate": 1.961304326315489e-05,
      "loss": 11.0865,
      "step": 603
    },
    {
      "epoch": 0.1180148495506057,
      "grad_norm": 8.145142555236816,
      "learning_rate": 1.961129783872301e-05,
      "loss": 10.4172,
      "step": 604
    },
    {
      "epoch": 0.11821023837436499,
      "grad_norm": 22.039506912231445,
      "learning_rate": 1.960954856466795e-05,
      "loss": 10.4725,
      "step": 605
    },
    {
      "epoch": 0.11840562719812427,
      "grad_norm": 10.486015319824219,
      "learning_rate": 1.9607795441690343e-05,
      "loss": 11.1312,
      "step": 606
    },
    {
      "epoch": 0.11860101602188355,
      "grad_norm": 8.452276229858398,
      "learning_rate": 1.9606038470492378e-05,
      "loss": 9.6146,
      "step": 607
    },
    {
      "epoch": 0.11879640484564283,
      "grad_norm": 9.439180374145508,
      "learning_rate": 1.9604277651777774e-05,
      "loss": 10.1609,
      "step": 608
    },
    {
      "epoch": 0.11899179366940212,
      "grad_norm": 9.560111999511719,
      "learning_rate": 1.960251298625179e-05,
      "loss": 10.1234,
      "step": 609
    },
    {
      "epoch": 0.11918718249316139,
      "grad_norm": 11.915494918823242,
      "learning_rate": 1.960074447462123e-05,
      "loss": 11.364,
      "step": 610
    },
    {
      "epoch": 0.11938257131692068,
      "grad_norm": 10.901596069335938,
      "learning_rate": 1.9598972117594445e-05,
      "loss": 11.3737,
      "step": 611
    },
    {
      "epoch": 0.11957796014067995,
      "grad_norm": 8.30836296081543,
      "learning_rate": 1.959719591588131e-05,
      "loss": 11.3081,
      "step": 612
    },
    {
      "epoch": 0.11977334896443924,
      "grad_norm": 10.83170223236084,
      "learning_rate": 1.959541587019325e-05,
      "loss": 11.5524,
      "step": 613
    },
    {
      "epoch": 0.11996873778819851,
      "grad_norm": 12.46660041809082,
      "learning_rate": 1.959363198124323e-05,
      "loss": 10.2701,
      "step": 614
    },
    {
      "epoch": 0.1201641266119578,
      "grad_norm": 8.10671329498291,
      "learning_rate": 1.959184424974576e-05,
      "loss": 9.8735,
      "step": 615
    },
    {
      "epoch": 0.12035951543571707,
      "grad_norm": 25.261287689208984,
      "learning_rate": 1.9590052676416872e-05,
      "loss": 11.7366,
      "step": 616
    },
    {
      "epoch": 0.12055490425947636,
      "grad_norm": 14.915023803710938,
      "learning_rate": 1.9588257261974147e-05,
      "loss": 10.4955,
      "step": 617
    },
    {
      "epoch": 0.12075029308323564,
      "grad_norm": 17.870332717895508,
      "learning_rate": 1.958645800713671e-05,
      "loss": 11.134,
      "step": 618
    },
    {
      "epoch": 0.12094568190699492,
      "grad_norm": 8.873143196105957,
      "learning_rate": 1.958465491262522e-05,
      "loss": 9.836,
      "step": 619
    },
    {
      "epoch": 0.1211410707307542,
      "grad_norm": 10.999107360839844,
      "learning_rate": 1.9582847979161864e-05,
      "loss": 10.983,
      "step": 620
    },
    {
      "epoch": 0.12133645955451348,
      "grad_norm": 12.157037734985352,
      "learning_rate": 1.9581037207470382e-05,
      "loss": 10.6826,
      "step": 621
    },
    {
      "epoch": 0.12153184837827276,
      "grad_norm": 13.782594680786133,
      "learning_rate": 1.957922259827604e-05,
      "loss": 11.2176,
      "step": 622
    },
    {
      "epoch": 0.12172723720203205,
      "grad_norm": 11.599889755249023,
      "learning_rate": 1.957740415230565e-05,
      "loss": 11.4917,
      "step": 623
    },
    {
      "epoch": 0.12192262602579132,
      "grad_norm": 12.650994300842285,
      "learning_rate": 1.957558187028755e-05,
      "loss": 10.5263,
      "step": 624
    },
    {
      "epoch": 0.12211801484955061,
      "grad_norm": 8.949012756347656,
      "learning_rate": 1.9573755752951625e-05,
      "loss": 11.2604,
      "step": 625
    },
    {
      "epoch": 0.12231340367330988,
      "grad_norm": 9.102046966552734,
      "learning_rate": 1.957192580102929e-05,
      "loss": 10.1976,
      "step": 626
    },
    {
      "epoch": 0.12250879249706917,
      "grad_norm": 11.349761009216309,
      "learning_rate": 1.9570092015253497e-05,
      "loss": 10.8624,
      "step": 627
    },
    {
      "epoch": 0.12270418132082844,
      "grad_norm": 7.718044757843018,
      "learning_rate": 1.9568254396358735e-05,
      "loss": 10.177,
      "step": 628
    },
    {
      "epoch": 0.12289957014458773,
      "grad_norm": 11.953253746032715,
      "learning_rate": 1.9566412945081028e-05,
      "loss": 11.9735,
      "step": 629
    },
    {
      "epoch": 0.123094958968347,
      "grad_norm": 11.112428665161133,
      "learning_rate": 1.9564567662157932e-05,
      "loss": 10.7037,
      "step": 630
    },
    {
      "epoch": 0.12329034779210629,
      "grad_norm": 11.177035331726074,
      "learning_rate": 1.9562718548328544e-05,
      "loss": 11.331,
      "step": 631
    },
    {
      "epoch": 0.12348573661586557,
      "grad_norm": 17.98822021484375,
      "learning_rate": 1.9560865604333487e-05,
      "loss": 12.0022,
      "step": 632
    },
    {
      "epoch": 0.12368112543962485,
      "grad_norm": 21.346542358398438,
      "learning_rate": 1.955900883091493e-05,
      "loss": 11.8838,
      "step": 633
    },
    {
      "epoch": 0.12387651426338414,
      "grad_norm": 41.60587692260742,
      "learning_rate": 1.955714822881656e-05,
      "loss": 11.0619,
      "step": 634
    },
    {
      "epoch": 0.12407190308714341,
      "grad_norm": 11.272843360900879,
      "learning_rate": 1.9555283798783605e-05,
      "loss": 11.3297,
      "step": 635
    },
    {
      "epoch": 0.1242672919109027,
      "grad_norm": 19.868553161621094,
      "learning_rate": 1.9553415541562833e-05,
      "loss": 12.7203,
      "step": 636
    },
    {
      "epoch": 0.12446268073466198,
      "grad_norm": 16.160825729370117,
      "learning_rate": 1.9551543457902544e-05,
      "loss": 10.2777,
      "step": 637
    },
    {
      "epoch": 0.12465806955842126,
      "grad_norm": 9.786787986755371,
      "learning_rate": 1.9549667548552557e-05,
      "loss": 12.1408,
      "step": 638
    },
    {
      "epoch": 0.12485345838218054,
      "grad_norm": 9.837309837341309,
      "learning_rate": 1.9547787814264228e-05,
      "loss": 11.0809,
      "step": 639
    },
    {
      "epoch": 0.1250488472059398,
      "grad_norm": 12.483675003051758,
      "learning_rate": 1.954590425579046e-05,
      "loss": 10.9152,
      "step": 640
    },
    {
      "epoch": 0.1252442360296991,
      "grad_norm": 15.183236122131348,
      "learning_rate": 1.954401687388567e-05,
      "loss": 11.2427,
      "step": 641
    },
    {
      "epoch": 0.1254396248534584,
      "grad_norm": 12.204520225524902,
      "learning_rate": 1.9542125669305813e-05,
      "loss": 11.4033,
      "step": 642
    },
    {
      "epoch": 0.12563501367721766,
      "grad_norm": 26.315473556518555,
      "learning_rate": 1.954023064280838e-05,
      "loss": 10.4009,
      "step": 643
    },
    {
      "epoch": 0.12583040250097693,
      "grad_norm": 10.41172981262207,
      "learning_rate": 1.9538331795152383e-05,
      "loss": 10.8517,
      "step": 644
    },
    {
      "epoch": 0.12602579132473624,
      "grad_norm": 9.543086051940918,
      "learning_rate": 1.9536429127098372e-05,
      "loss": 11.2244,
      "step": 645
    },
    {
      "epoch": 0.1262211801484955,
      "grad_norm": 8.386434555053711,
      "learning_rate": 1.9534522639408422e-05,
      "loss": 10.9421,
      "step": 646
    },
    {
      "epoch": 0.12641656897225478,
      "grad_norm": 10.363856315612793,
      "learning_rate": 1.9532612332846143e-05,
      "loss": 11.924,
      "step": 647
    },
    {
      "epoch": 0.12661195779601406,
      "grad_norm": 17.130685806274414,
      "learning_rate": 1.9530698208176674e-05,
      "loss": 10.718,
      "step": 648
    },
    {
      "epoch": 0.12680734661977336,
      "grad_norm": 9.325024604797363,
      "learning_rate": 1.9528780266166674e-05,
      "loss": 10.0044,
      "step": 649
    },
    {
      "epoch": 0.12700273544353263,
      "grad_norm": 10.563580513000488,
      "learning_rate": 1.9526858507584347e-05,
      "loss": 10.5433,
      "step": 650
    },
    {
      "epoch": 0.1271981242672919,
      "grad_norm": 9.915545463562012,
      "learning_rate": 1.952493293319941e-05,
      "loss": 10.2272,
      "step": 651
    },
    {
      "epoch": 0.12739351309105118,
      "grad_norm": 9.59200668334961,
      "learning_rate": 1.9523003543783122e-05,
      "loss": 10.0584,
      "step": 652
    },
    {
      "epoch": 0.12758890191481048,
      "grad_norm": 20.589073181152344,
      "learning_rate": 1.9521070340108258e-05,
      "loss": 10.1349,
      "step": 653
    },
    {
      "epoch": 0.12778429073856976,
      "grad_norm": 13.821552276611328,
      "learning_rate": 1.9519133322949124e-05,
      "loss": 10.4477,
      "step": 654
    },
    {
      "epoch": 0.12797967956232903,
      "grad_norm": 13.572846412658691,
      "learning_rate": 1.9517192493081563e-05,
      "loss": 10.4236,
      "step": 655
    },
    {
      "epoch": 0.1281750683860883,
      "grad_norm": 29.654455184936523,
      "learning_rate": 1.9515247851282936e-05,
      "loss": 11.0006,
      "step": 656
    },
    {
      "epoch": 0.1283704572098476,
      "grad_norm": 18.17194175720215,
      "learning_rate": 1.9513299398332126e-05,
      "loss": 10.744,
      "step": 657
    },
    {
      "epoch": 0.12856584603360688,
      "grad_norm": 15.245993614196777,
      "learning_rate": 1.9511347135009556e-05,
      "loss": 10.9422,
      "step": 658
    },
    {
      "epoch": 0.12876123485736615,
      "grad_norm": 9.658390045166016,
      "learning_rate": 1.9509391062097163e-05,
      "loss": 10.7485,
      "step": 659
    },
    {
      "epoch": 0.12895662368112543,
      "grad_norm": 19.675369262695312,
      "learning_rate": 1.950743118037842e-05,
      "loss": 10.4852,
      "step": 660
    },
    {
      "epoch": 0.12915201250488473,
      "grad_norm": 16.983983993530273,
      "learning_rate": 1.9505467490638317e-05,
      "loss": 10.9365,
      "step": 661
    },
    {
      "epoch": 0.129347401328644,
      "grad_norm": 12.646660804748535,
      "learning_rate": 1.950349999366337e-05,
      "loss": 11.257,
      "step": 662
    },
    {
      "epoch": 0.12954279015240328,
      "grad_norm": 10.177793502807617,
      "learning_rate": 1.9501528690241632e-05,
      "loss": 10.5301,
      "step": 663
    },
    {
      "epoch": 0.12973817897616258,
      "grad_norm": 25.35265350341797,
      "learning_rate": 1.9499553581162658e-05,
      "loss": 10.7971,
      "step": 664
    },
    {
      "epoch": 0.12993356779992185,
      "grad_norm": 9.132247924804688,
      "learning_rate": 1.9497574667217553e-05,
      "loss": 10.6206,
      "step": 665
    },
    {
      "epoch": 0.13012895662368112,
      "grad_norm": 12.48060131072998,
      "learning_rate": 1.9495591949198926e-05,
      "loss": 11.35,
      "step": 666
    },
    {
      "epoch": 0.1303243454474404,
      "grad_norm": 18.60825538635254,
      "learning_rate": 1.9493605427900922e-05,
      "loss": 10.384,
      "step": 667
    },
    {
      "epoch": 0.1305197342711997,
      "grad_norm": 10.191056251525879,
      "learning_rate": 1.94916151041192e-05,
      "loss": 10.8634,
      "step": 668
    },
    {
      "epoch": 0.13071512309495897,
      "grad_norm": 13.43942928314209,
      "learning_rate": 1.9489620978650946e-05,
      "loss": 10.8465,
      "step": 669
    },
    {
      "epoch": 0.13091051191871825,
      "grad_norm": 9.228296279907227,
      "learning_rate": 1.9487623052294875e-05,
      "loss": 11.0102,
      "step": 670
    },
    {
      "epoch": 0.13110590074247752,
      "grad_norm": 10.576723098754883,
      "learning_rate": 1.9485621325851213e-05,
      "loss": 10.9432,
      "step": 671
    },
    {
      "epoch": 0.13130128956623682,
      "grad_norm": 12.026520729064941,
      "learning_rate": 1.9483615800121717e-05,
      "loss": 11.577,
      "step": 672
    },
    {
      "epoch": 0.1314966783899961,
      "grad_norm": 12.680463790893555,
      "learning_rate": 1.948160647590966e-05,
      "loss": 10.7402,
      "step": 673
    },
    {
      "epoch": 0.13169206721375537,
      "grad_norm": 9.338397026062012,
      "learning_rate": 1.9479593354019843e-05,
      "loss": 10.3435,
      "step": 674
    },
    {
      "epoch": 0.13188745603751464,
      "grad_norm": 9.28903865814209,
      "learning_rate": 1.9477576435258575e-05,
      "loss": 10.9857,
      "step": 675
    },
    {
      "epoch": 0.13208284486127395,
      "grad_norm": 14.938960075378418,
      "learning_rate": 1.94755557204337e-05,
      "loss": 9.6975,
      "step": 676
    },
    {
      "epoch": 0.13227823368503322,
      "grad_norm": 20.733600616455078,
      "learning_rate": 1.947353121035458e-05,
      "loss": 10.5668,
      "step": 677
    },
    {
      "epoch": 0.1324736225087925,
      "grad_norm": 16.638071060180664,
      "learning_rate": 1.9471502905832088e-05,
      "loss": 11.0326,
      "step": 678
    },
    {
      "epoch": 0.13266901133255177,
      "grad_norm": 11.815279960632324,
      "learning_rate": 1.946947080767863e-05,
      "loss": 10.4106,
      "step": 679
    },
    {
      "epoch": 0.13286440015631107,
      "grad_norm": 11.046647071838379,
      "learning_rate": 1.9467434916708118e-05,
      "loss": 10.8296,
      "step": 680
    },
    {
      "epoch": 0.13305978898007034,
      "grad_norm": 14.552698135375977,
      "learning_rate": 1.9465395233735992e-05,
      "loss": 12.1462,
      "step": 681
    },
    {
      "epoch": 0.13325517780382962,
      "grad_norm": 17.292787551879883,
      "learning_rate": 1.946335175957921e-05,
      "loss": 11.1024,
      "step": 682
    },
    {
      "epoch": 0.1334505666275889,
      "grad_norm": 19.84165382385254,
      "learning_rate": 1.9461304495056247e-05,
      "loss": 10.2571,
      "step": 683
    },
    {
      "epoch": 0.1336459554513482,
      "grad_norm": 10.883853912353516,
      "learning_rate": 1.945925344098709e-05,
      "loss": 9.9972,
      "step": 684
    },
    {
      "epoch": 0.13384134427510747,
      "grad_norm": 12.545206069946289,
      "learning_rate": 1.9457198598193257e-05,
      "loss": 10.76,
      "step": 685
    },
    {
      "epoch": 0.13403673309886674,
      "grad_norm": 8.704316139221191,
      "learning_rate": 1.9455139967497773e-05,
      "loss": 10.6959,
      "step": 686
    },
    {
      "epoch": 0.134232121922626,
      "grad_norm": 11.672131538391113,
      "learning_rate": 1.945307754972518e-05,
      "loss": 11.8341,
      "step": 687
    },
    {
      "epoch": 0.13442751074638531,
      "grad_norm": 10.522601127624512,
      "learning_rate": 1.945101134570155e-05,
      "loss": 10.5467,
      "step": 688
    },
    {
      "epoch": 0.1346228995701446,
      "grad_norm": 10.88532543182373,
      "learning_rate": 1.9448941356254453e-05,
      "loss": 11.1203,
      "step": 689
    },
    {
      "epoch": 0.13481828839390386,
      "grad_norm": 10.269328117370605,
      "learning_rate": 1.9446867582212987e-05,
      "loss": 10.3559,
      "step": 690
    },
    {
      "epoch": 0.13501367721766314,
      "grad_norm": 11.798625946044922,
      "learning_rate": 1.9444790024407766e-05,
      "loss": 9.7001,
      "step": 691
    },
    {
      "epoch": 0.13520906604142244,
      "grad_norm": 10.87350845336914,
      "learning_rate": 1.944270868367091e-05,
      "loss": 11.1203,
      "step": 692
    },
    {
      "epoch": 0.1354044548651817,
      "grad_norm": 15.473928451538086,
      "learning_rate": 1.9440623560836064e-05,
      "loss": 10.6878,
      "step": 693
    },
    {
      "epoch": 0.13559984368894099,
      "grad_norm": 10.6022310256958,
      "learning_rate": 1.9438534656738386e-05,
      "loss": 10.6721,
      "step": 694
    },
    {
      "epoch": 0.1357952325127003,
      "grad_norm": 9.145467758178711,
      "learning_rate": 1.9436441972214545e-05,
      "loss": 10.7177,
      "step": 695
    },
    {
      "epoch": 0.13599062133645956,
      "grad_norm": 10.879833221435547,
      "learning_rate": 1.9434345508102726e-05,
      "loss": 10.2057,
      "step": 696
    },
    {
      "epoch": 0.13618601016021883,
      "grad_norm": 9.55604362487793,
      "learning_rate": 1.943224526524263e-05,
      "loss": 10.0616,
      "step": 697
    },
    {
      "epoch": 0.1363813989839781,
      "grad_norm": 11.760533332824707,
      "learning_rate": 1.9430141244475468e-05,
      "loss": 10.8176,
      "step": 698
    },
    {
      "epoch": 0.1365767878077374,
      "grad_norm": 17.020631790161133,
      "learning_rate": 1.9428033446643964e-05,
      "loss": 10.1112,
      "step": 699
    },
    {
      "epoch": 0.13677217663149668,
      "grad_norm": 12.788175582885742,
      "learning_rate": 1.942592187259236e-05,
      "loss": 11.4064,
      "step": 700
    },
    {
      "epoch": 0.13696756545525596,
      "grad_norm": 8.893692016601562,
      "learning_rate": 1.94238065231664e-05,
      "loss": 10.5016,
      "step": 701
    },
    {
      "epoch": 0.13716295427901523,
      "grad_norm": 10.239310264587402,
      "learning_rate": 1.942168739921336e-05,
      "loss": 10.837,
      "step": 702
    },
    {
      "epoch": 0.13735834310277453,
      "grad_norm": 11.442367553710938,
      "learning_rate": 1.9419564501582e-05,
      "loss": 11.4701,
      "step": 703
    },
    {
      "epoch": 0.1375537319265338,
      "grad_norm": 19.924348831176758,
      "learning_rate": 1.9417437831122615e-05,
      "loss": 10.4355,
      "step": 704
    },
    {
      "epoch": 0.13774912075029308,
      "grad_norm": 11.848478317260742,
      "learning_rate": 1.9415307388687007e-05,
      "loss": 12.0589,
      "step": 705
    },
    {
      "epoch": 0.13794450957405235,
      "grad_norm": 10.013603210449219,
      "learning_rate": 1.9413173175128472e-05,
      "loss": 10.827,
      "step": 706
    },
    {
      "epoch": 0.13813989839781166,
      "grad_norm": 15.492804527282715,
      "learning_rate": 1.9411035191301836e-05,
      "loss": 8.7754,
      "step": 707
    },
    {
      "epoch": 0.13833528722157093,
      "grad_norm": 15.368468284606934,
      "learning_rate": 1.9408893438063428e-05,
      "loss": 11.1972,
      "step": 708
    },
    {
      "epoch": 0.1385306760453302,
      "grad_norm": 8.977986335754395,
      "learning_rate": 1.9406747916271088e-05,
      "loss": 10.1648,
      "step": 709
    },
    {
      "epoch": 0.13872606486908948,
      "grad_norm": 10.708306312561035,
      "learning_rate": 1.940459862678416e-05,
      "loss": 10.5425,
      "step": 710
    },
    {
      "epoch": 0.13892145369284878,
      "grad_norm": 9.737686157226562,
      "learning_rate": 1.9402445570463502e-05,
      "loss": 9.6175,
      "step": 711
    },
    {
      "epoch": 0.13911684251660805,
      "grad_norm": 11.807138442993164,
      "learning_rate": 1.940028874817148e-05,
      "loss": 11.2366,
      "step": 712
    },
    {
      "epoch": 0.13931223134036733,
      "grad_norm": 9.303187370300293,
      "learning_rate": 1.9398128160771973e-05,
      "loss": 9.8696,
      "step": 713
    },
    {
      "epoch": 0.1395076201641266,
      "grad_norm": 19.42157745361328,
      "learning_rate": 1.9395963809130355e-05,
      "loss": 10.5176,
      "step": 714
    },
    {
      "epoch": 0.1397030089878859,
      "grad_norm": 10.6793794631958,
      "learning_rate": 1.9393795694113526e-05,
      "loss": 10.3265,
      "step": 715
    },
    {
      "epoch": 0.13989839781164518,
      "grad_norm": 10.98534870147705,
      "learning_rate": 1.9391623816589874e-05,
      "loss": 10.4471,
      "step": 716
    },
    {
      "epoch": 0.14009378663540445,
      "grad_norm": 9.268878936767578,
      "learning_rate": 1.9389448177429305e-05,
      "loss": 9.6479,
      "step": 717
    },
    {
      "epoch": 0.14028917545916372,
      "grad_norm": 6.744013786315918,
      "learning_rate": 1.9387268777503237e-05,
      "loss": 9.1637,
      "step": 718
    },
    {
      "epoch": 0.14048456428292302,
      "grad_norm": 10.005885124206543,
      "learning_rate": 1.938508561768458e-05,
      "loss": 9.3408,
      "step": 719
    },
    {
      "epoch": 0.1406799531066823,
      "grad_norm": 11.842926979064941,
      "learning_rate": 1.938289869884776e-05,
      "loss": 9.6945,
      "step": 720
    },
    {
      "epoch": 0.14087534193044157,
      "grad_norm": 10.483861923217773,
      "learning_rate": 1.9380708021868708e-05,
      "loss": 10.5852,
      "step": 721
    },
    {
      "epoch": 0.14107073075420085,
      "grad_norm": 10.77122688293457,
      "learning_rate": 1.9378513587624855e-05,
      "loss": 10.0683,
      "step": 722
    },
    {
      "epoch": 0.14126611957796015,
      "grad_norm": 10.417200088500977,
      "learning_rate": 1.937631539699514e-05,
      "loss": 9.9922,
      "step": 723
    },
    {
      "epoch": 0.14146150840171942,
      "grad_norm": 11.36428165435791,
      "learning_rate": 1.9374113450860012e-05,
      "loss": 10.2745,
      "step": 724
    },
    {
      "epoch": 0.1416568972254787,
      "grad_norm": 10.181743621826172,
      "learning_rate": 1.937190775010141e-05,
      "loss": 11.0796,
      "step": 725
    },
    {
      "epoch": 0.141852286049238,
      "grad_norm": 10.130002975463867,
      "learning_rate": 1.936969829560279e-05,
      "loss": 10.2408,
      "step": 726
    },
    {
      "epoch": 0.14204767487299727,
      "grad_norm": 12.734686851501465,
      "learning_rate": 1.9367485088249106e-05,
      "loss": 11.2567,
      "step": 727
    },
    {
      "epoch": 0.14224306369675654,
      "grad_norm": 12.297183990478516,
      "learning_rate": 1.936526812892682e-05,
      "loss": 10.2194,
      "step": 728
    },
    {
      "epoch": 0.14243845252051582,
      "grad_norm": 13.363584518432617,
      "learning_rate": 1.936304741852389e-05,
      "loss": 11.4515,
      "step": 729
    },
    {
      "epoch": 0.14263384134427512,
      "grad_norm": 12.167123794555664,
      "learning_rate": 1.936082295792978e-05,
      "loss": 11.2041,
      "step": 730
    },
    {
      "epoch": 0.1428292301680344,
      "grad_norm": 8.581811904907227,
      "learning_rate": 1.9358594748035455e-05,
      "loss": 9.6672,
      "step": 731
    },
    {
      "epoch": 0.14302461899179367,
      "grad_norm": 28.514732360839844,
      "learning_rate": 1.9356362789733383e-05,
      "loss": 10.7176,
      "step": 732
    },
    {
      "epoch": 0.14322000781555294,
      "grad_norm": 9.339582443237305,
      "learning_rate": 1.9354127083917535e-05,
      "loss": 10.3786,
      "step": 733
    },
    {
      "epoch": 0.14341539663931224,
      "grad_norm": 9.966968536376953,
      "learning_rate": 1.9351887631483372e-05,
      "loss": 11.0023,
      "step": 734
    },
    {
      "epoch": 0.14361078546307152,
      "grad_norm": 9.983609199523926,
      "learning_rate": 1.9349644433327875e-05,
      "loss": 10.4865,
      "step": 735
    },
    {
      "epoch": 0.1438061742868308,
      "grad_norm": 8.336555480957031,
      "learning_rate": 1.9347397490349507e-05,
      "loss": 10.2784,
      "step": 736
    },
    {
      "epoch": 0.14400156311059006,
      "grad_norm": 10.444098472595215,
      "learning_rate": 1.934514680344824e-05,
      "loss": 10.4828,
      "step": 737
    },
    {
      "epoch": 0.14419695193434937,
      "grad_norm": 9.705633163452148,
      "learning_rate": 1.9342892373525547e-05,
      "loss": 10.3624,
      "step": 738
    },
    {
      "epoch": 0.14439234075810864,
      "grad_norm": 7.4516921043396,
      "learning_rate": 1.9340634201484397e-05,
      "loss": 10.4507,
      "step": 739
    },
    {
      "epoch": 0.1445877295818679,
      "grad_norm": 8.806925773620605,
      "learning_rate": 1.9338372288229253e-05,
      "loss": 9.7585,
      "step": 740
    },
    {
      "epoch": 0.1447831184056272,
      "grad_norm": 10.420880317687988,
      "learning_rate": 1.9336106634666086e-05,
      "loss": 10.4827,
      "step": 741
    },
    {
      "epoch": 0.1449785072293865,
      "grad_norm": 14.015484809875488,
      "learning_rate": 1.933383724170236e-05,
      "loss": 11.1924,
      "step": 742
    },
    {
      "epoch": 0.14517389605314576,
      "grad_norm": 12.910109519958496,
      "learning_rate": 1.9331564110247035e-05,
      "loss": 10.4148,
      "step": 743
    },
    {
      "epoch": 0.14536928487690504,
      "grad_norm": 38.42607116699219,
      "learning_rate": 1.9329287241210574e-05,
      "loss": 11.107,
      "step": 744
    },
    {
      "epoch": 0.1455646737006643,
      "grad_norm": 16.01734161376953,
      "learning_rate": 1.932700663550493e-05,
      "loss": 10.0358,
      "step": 745
    },
    {
      "epoch": 0.1457600625244236,
      "grad_norm": 12.651199340820312,
      "learning_rate": 1.932472229404356e-05,
      "loss": 10.2062,
      "step": 746
    },
    {
      "epoch": 0.14595545134818289,
      "grad_norm": 13.449583053588867,
      "learning_rate": 1.9322434217741413e-05,
      "loss": 9.2374,
      "step": 747
    },
    {
      "epoch": 0.14615084017194216,
      "grad_norm": 15.523850440979004,
      "learning_rate": 1.932014240751493e-05,
      "loss": 11.7652,
      "step": 748
    },
    {
      "epoch": 0.14634622899570143,
      "grad_norm": 16.57215690612793,
      "learning_rate": 1.931784686428206e-05,
      "loss": 10.8459,
      "step": 749
    },
    {
      "epoch": 0.14654161781946073,
      "grad_norm": 11.21127700805664,
      "learning_rate": 1.9315547588962232e-05,
      "loss": 11.843,
      "step": 750
    },
    {
      "epoch": 0.14673700664322,
      "grad_norm": 9.245950698852539,
      "learning_rate": 1.9313244582476383e-05,
      "loss": 11.2157,
      "step": 751
    },
    {
      "epoch": 0.14693239546697928,
      "grad_norm": 9.136964797973633,
      "learning_rate": 1.9310937845746934e-05,
      "loss": 9.6121,
      "step": 752
    },
    {
      "epoch": 0.14712778429073858,
      "grad_norm": 10.128212928771973,
      "learning_rate": 1.930862737969781e-05,
      "loss": 10.1339,
      "step": 753
    },
    {
      "epoch": 0.14732317311449786,
      "grad_norm": 13.177711486816406,
      "learning_rate": 1.9306313185254417e-05,
      "loss": 10.2268,
      "step": 754
    },
    {
      "epoch": 0.14751856193825713,
      "grad_norm": 12.620972633361816,
      "learning_rate": 1.9303995263343665e-05,
      "loss": 10.7249,
      "step": 755
    },
    {
      "epoch": 0.1477139507620164,
      "grad_norm": 10.05113697052002,
      "learning_rate": 1.9301673614893957e-05,
      "loss": 10.9795,
      "step": 756
    },
    {
      "epoch": 0.1479093395857757,
      "grad_norm": 9.155972480773926,
      "learning_rate": 1.9299348240835182e-05,
      "loss": 9.1259,
      "step": 757
    },
    {
      "epoch": 0.14810472840953498,
      "grad_norm": 10.70669174194336,
      "learning_rate": 1.9297019142098725e-05,
      "loss": 10.718,
      "step": 758
    },
    {
      "epoch": 0.14830011723329425,
      "grad_norm": 13.88758659362793,
      "learning_rate": 1.9294686319617463e-05,
      "loss": 10.4523,
      "step": 759
    },
    {
      "epoch": 0.14849550605705353,
      "grad_norm": 17.19888687133789,
      "learning_rate": 1.9292349774325766e-05,
      "loss": 11.894,
      "step": 760
    },
    {
      "epoch": 0.14869089488081283,
      "grad_norm": 21.77732276916504,
      "learning_rate": 1.9290009507159488e-05,
      "loss": 12.2076,
      "step": 761
    },
    {
      "epoch": 0.1488862837045721,
      "grad_norm": 11.067423820495605,
      "learning_rate": 1.928766551905598e-05,
      "loss": 9.8428,
      "step": 762
    },
    {
      "epoch": 0.14908167252833138,
      "grad_norm": 16.04343605041504,
      "learning_rate": 1.9285317810954083e-05,
      "loss": 10.0406,
      "step": 763
    },
    {
      "epoch": 0.14927706135209065,
      "grad_norm": 10.960137367248535,
      "learning_rate": 1.9282966383794127e-05,
      "loss": 10.2735,
      "step": 764
    },
    {
      "epoch": 0.14947245017584995,
      "grad_norm": 10.08834171295166,
      "learning_rate": 1.9280611238517932e-05,
      "loss": 10.0519,
      "step": 765
    },
    {
      "epoch": 0.14966783899960923,
      "grad_norm": 14.802245140075684,
      "learning_rate": 1.9278252376068805e-05,
      "loss": 11.5211,
      "step": 766
    },
    {
      "epoch": 0.1498632278233685,
      "grad_norm": 19.65907859802246,
      "learning_rate": 1.927588979739155e-05,
      "loss": 11.1272,
      "step": 767
    },
    {
      "epoch": 0.15005861664712777,
      "grad_norm": 10.54307746887207,
      "learning_rate": 1.9273523503432444e-05,
      "loss": 10.9068,
      "step": 768
    },
    {
      "epoch": 0.15025400547088708,
      "grad_norm": 14.132508277893066,
      "learning_rate": 1.9271153495139266e-05,
      "loss": 10.1415,
      "step": 769
    },
    {
      "epoch": 0.15044939429464635,
      "grad_norm": 11.745526313781738,
      "learning_rate": 1.926877977346128e-05,
      "loss": 11.2422,
      "step": 770
    },
    {
      "epoch": 0.15064478311840562,
      "grad_norm": 9.170060157775879,
      "learning_rate": 1.9266402339349227e-05,
      "loss": 10.488,
      "step": 771
    },
    {
      "epoch": 0.1508401719421649,
      "grad_norm": 10.40644645690918,
      "learning_rate": 1.926402119375535e-05,
      "loss": 9.9418,
      "step": 772
    },
    {
      "epoch": 0.1510355607659242,
      "grad_norm": 10.796622276306152,
      "learning_rate": 1.9261636337633374e-05,
      "loss": 9.6721,
      "step": 773
    },
    {
      "epoch": 0.15123094958968347,
      "grad_norm": 12.15422534942627,
      "learning_rate": 1.92592477719385e-05,
      "loss": 10.767,
      "step": 774
    },
    {
      "epoch": 0.15142633841344275,
      "grad_norm": 8.324102401733398,
      "learning_rate": 1.925685549762743e-05,
      "loss": 10.0255,
      "step": 775
    },
    {
      "epoch": 0.15162172723720202,
      "grad_norm": 10.961838722229004,
      "learning_rate": 1.925445951565834e-05,
      "loss": 10.8336,
      "step": 776
    },
    {
      "epoch": 0.15181711606096132,
      "grad_norm": 11.413140296936035,
      "learning_rate": 1.9252059826990893e-05,
      "loss": 9.7799,
      "step": 777
    },
    {
      "epoch": 0.1520125048847206,
      "grad_norm": 11.092761993408203,
      "learning_rate": 1.9249656432586248e-05,
      "loss": 10.4188,
      "step": 778
    },
    {
      "epoch": 0.15220789370847987,
      "grad_norm": 13.403634071350098,
      "learning_rate": 1.924724933340703e-05,
      "loss": 10.6561,
      "step": 779
    },
    {
      "epoch": 0.15240328253223914,
      "grad_norm": 13.839494705200195,
      "learning_rate": 1.924483853041736e-05,
      "loss": 10.3124,
      "step": 780
    },
    {
      "epoch": 0.15259867135599844,
      "grad_norm": 16.102264404296875,
      "learning_rate": 1.924242402458284e-05,
      "loss": 10.6515,
      "step": 781
    },
    {
      "epoch": 0.15279406017975772,
      "grad_norm": 10.754448890686035,
      "learning_rate": 1.9240005816870554e-05,
      "loss": 10.3905,
      "step": 782
    },
    {
      "epoch": 0.152989449003517,
      "grad_norm": 14.738258361816406,
      "learning_rate": 1.9237583908249072e-05,
      "loss": 9.8646,
      "step": 783
    },
    {
      "epoch": 0.1531848378272763,
      "grad_norm": 17.20542335510254,
      "learning_rate": 1.9235158299688436e-05,
      "loss": 11.8768,
      "step": 784
    },
    {
      "epoch": 0.15338022665103557,
      "grad_norm": 12.346606254577637,
      "learning_rate": 1.9232728992160187e-05,
      "loss": 10.3553,
      "step": 785
    },
    {
      "epoch": 0.15357561547479484,
      "grad_norm": 12.346830368041992,
      "learning_rate": 1.923029598663733e-05,
      "loss": 10.2946,
      "step": 786
    },
    {
      "epoch": 0.15377100429855411,
      "grad_norm": 9.510976791381836,
      "learning_rate": 1.9227859284094367e-05,
      "loss": 9.9759,
      "step": 787
    },
    {
      "epoch": 0.15396639312231342,
      "grad_norm": 7.273228645324707,
      "learning_rate": 1.9225418885507264e-05,
      "loss": 9.3106,
      "step": 788
    },
    {
      "epoch": 0.1541617819460727,
      "grad_norm": 10.699312210083008,
      "learning_rate": 1.9222974791853485e-05,
      "loss": 10.2201,
      "step": 789
    },
    {
      "epoch": 0.15435717076983196,
      "grad_norm": 10.579397201538086,
      "learning_rate": 1.922052700411196e-05,
      "loss": 10.8552,
      "step": 790
    },
    {
      "epoch": 0.15455255959359124,
      "grad_norm": 8.557092666625977,
      "learning_rate": 1.9218075523263104e-05,
      "loss": 9.5428,
      "step": 791
    },
    {
      "epoch": 0.15474794841735054,
      "grad_norm": 9.219961166381836,
      "learning_rate": 1.9215620350288816e-05,
      "loss": 11.1206,
      "step": 792
    },
    {
      "epoch": 0.1549433372411098,
      "grad_norm": 9.186723709106445,
      "learning_rate": 1.9213161486172464e-05,
      "loss": 10.8128,
      "step": 793
    },
    {
      "epoch": 0.1551387260648691,
      "grad_norm": 14.034444808959961,
      "learning_rate": 1.9210698931898903e-05,
      "loss": 10.4753,
      "step": 794
    },
    {
      "epoch": 0.15533411488862836,
      "grad_norm": 13.460503578186035,
      "learning_rate": 1.920823268845446e-05,
      "loss": 10.4392,
      "step": 795
    },
    {
      "epoch": 0.15552950371238766,
      "grad_norm": 17.306602478027344,
      "learning_rate": 1.920576275682694e-05,
      "loss": 10.153,
      "step": 796
    },
    {
      "epoch": 0.15572489253614694,
      "grad_norm": 14.931540489196777,
      "learning_rate": 1.9203289138005634e-05,
      "loss": 11.0139,
      "step": 797
    },
    {
      "epoch": 0.1559202813599062,
      "grad_norm": 11.157635688781738,
      "learning_rate": 1.92008118329813e-05,
      "loss": 10.3161,
      "step": 798
    },
    {
      "epoch": 0.15611567018366548,
      "grad_norm": 16.768131256103516,
      "learning_rate": 1.919833084274617e-05,
      "loss": 9.3059,
      "step": 799
    },
    {
      "epoch": 0.15631105900742479,
      "grad_norm": 26.313011169433594,
      "learning_rate": 1.9195846168293967e-05,
      "loss": 9.0071,
      "step": 800
    },
    {
      "epoch": 0.15650644783118406,
      "grad_norm": 10.458858489990234,
      "learning_rate": 1.919335781061987e-05,
      "loss": 10.5491,
      "step": 801
    },
    {
      "epoch": 0.15670183665494333,
      "grad_norm": 8.418832778930664,
      "learning_rate": 1.919086577072055e-05,
      "loss": 9.9025,
      "step": 802
    },
    {
      "epoch": 0.1568972254787026,
      "grad_norm": 13.979361534118652,
      "learning_rate": 1.918837004959415e-05,
      "loss": 10.9538,
      "step": 803
    },
    {
      "epoch": 0.1570926143024619,
      "grad_norm": 8.292892456054688,
      "learning_rate": 1.9185870648240278e-05,
      "loss": 9.9292,
      "step": 804
    },
    {
      "epoch": 0.15728800312622118,
      "grad_norm": 18.740543365478516,
      "learning_rate": 1.9183367567660024e-05,
      "loss": 9.2841,
      "step": 805
    },
    {
      "epoch": 0.15748339194998046,
      "grad_norm": 11.974870681762695,
      "learning_rate": 1.9180860808855944e-05,
      "loss": 9.4875,
      "step": 806
    },
    {
      "epoch": 0.15767878077373973,
      "grad_norm": 11.385255813598633,
      "learning_rate": 1.917835037283208e-05,
      "loss": 10.3209,
      "step": 807
    },
    {
      "epoch": 0.15787416959749903,
      "grad_norm": 8.198461532592773,
      "learning_rate": 1.9175836260593937e-05,
      "loss": 9.3359,
      "step": 808
    },
    {
      "epoch": 0.1580695584212583,
      "grad_norm": 23.158065795898438,
      "learning_rate": 1.9173318473148495e-05,
      "loss": 10.8434,
      "step": 809
    },
    {
      "epoch": 0.15826494724501758,
      "grad_norm": 9.892696380615234,
      "learning_rate": 1.9170797011504206e-05,
      "loss": 10.5046,
      "step": 810
    },
    {
      "epoch": 0.15846033606877685,
      "grad_norm": 23.545766830444336,
      "learning_rate": 1.9168271876670994e-05,
      "loss": 9.549,
      "step": 811
    },
    {
      "epoch": 0.15865572489253615,
      "grad_norm": 10.899819374084473,
      "learning_rate": 1.9165743069660253e-05,
      "loss": 10.2441,
      "step": 812
    },
    {
      "epoch": 0.15885111371629543,
      "grad_norm": 22.16189193725586,
      "learning_rate": 1.9163210591484852e-05,
      "loss": 10.6823,
      "step": 813
    },
    {
      "epoch": 0.1590465025400547,
      "grad_norm": 10.303801536560059,
      "learning_rate": 1.916067444315912e-05,
      "loss": 10.7417,
      "step": 814
    },
    {
      "epoch": 0.159241891363814,
      "grad_norm": 11.283565521240234,
      "learning_rate": 1.915813462569887e-05,
      "loss": 9.949,
      "step": 815
    },
    {
      "epoch": 0.15943728018757328,
      "grad_norm": 15.853772163391113,
      "learning_rate": 1.9155591140121373e-05,
      "loss": 11.3669,
      "step": 816
    },
    {
      "epoch": 0.15963266901133255,
      "grad_norm": 16.68556022644043,
      "learning_rate": 1.9153043987445377e-05,
      "loss": 10.1564,
      "step": 817
    },
    {
      "epoch": 0.15982805783509182,
      "grad_norm": 7.7740654945373535,
      "learning_rate": 1.9150493168691097e-05,
      "loss": 11.4006,
      "step": 818
    },
    {
      "epoch": 0.16002344665885113,
      "grad_norm": 9.202394485473633,
      "learning_rate": 1.9147938684880213e-05,
      "loss": 11.4376,
      "step": 819
    },
    {
      "epoch": 0.1602188354826104,
      "grad_norm": 8.620976448059082,
      "learning_rate": 1.914538053703587e-05,
      "loss": 10.9465,
      "step": 820
    },
    {
      "epoch": 0.16041422430636967,
      "grad_norm": 8.079070091247559,
      "learning_rate": 1.9142818726182696e-05,
      "loss": 10.0092,
      "step": 821
    },
    {
      "epoch": 0.16060961313012895,
      "grad_norm": 8.546801567077637,
      "learning_rate": 1.9140253253346767e-05,
      "loss": 9.9966,
      "step": 822
    },
    {
      "epoch": 0.16080500195388825,
      "grad_norm": 8.872438430786133,
      "learning_rate": 1.9137684119555642e-05,
      "loss": 10.4949,
      "step": 823
    },
    {
      "epoch": 0.16100039077764752,
      "grad_norm": 8.324982643127441,
      "learning_rate": 1.913511132583833e-05,
      "loss": 10.6894,
      "step": 824
    },
    {
      "epoch": 0.1611957796014068,
      "grad_norm": 11.035906791687012,
      "learning_rate": 1.9132534873225323e-05,
      "loss": 10.7705,
      "step": 825
    },
    {
      "epoch": 0.16139116842516607,
      "grad_norm": 14.182964324951172,
      "learning_rate": 1.9129954762748567e-05,
      "loss": 11.0233,
      "step": 826
    },
    {
      "epoch": 0.16158655724892537,
      "grad_norm": 11.811083793640137,
      "learning_rate": 1.9127370995441475e-05,
      "loss": 10.7647,
      "step": 827
    },
    {
      "epoch": 0.16178194607268465,
      "grad_norm": 10.472794532775879,
      "learning_rate": 1.912478357233893e-05,
      "loss": 10.7073,
      "step": 828
    },
    {
      "epoch": 0.16197733489644392,
      "grad_norm": 12.881394386291504,
      "learning_rate": 1.9122192494477268e-05,
      "loss": 10.2255,
      "step": 829
    },
    {
      "epoch": 0.1621727237202032,
      "grad_norm": 11.292678833007812,
      "learning_rate": 1.9119597762894302e-05,
      "loss": 11.345,
      "step": 830
    },
    {
      "epoch": 0.1623681125439625,
      "grad_norm": 10.249587059020996,
      "learning_rate": 1.9116999378629305e-05,
      "loss": 10.1139,
      "step": 831
    },
    {
      "epoch": 0.16256350136772177,
      "grad_norm": 14.024880409240723,
      "learning_rate": 1.9114397342723005e-05,
      "loss": 10.1549,
      "step": 832
    },
    {
      "epoch": 0.16275889019148104,
      "grad_norm": 13.929058074951172,
      "learning_rate": 1.9111791656217598e-05,
      "loss": 10.9506,
      "step": 833
    },
    {
      "epoch": 0.16295427901524032,
      "grad_norm": 14.429847717285156,
      "learning_rate": 1.9109182320156744e-05,
      "loss": 9.965,
      "step": 834
    },
    {
      "epoch": 0.16314966783899962,
      "grad_norm": 10.850347518920898,
      "learning_rate": 1.9106569335585566e-05,
      "loss": 10.1431,
      "step": 835
    },
    {
      "epoch": 0.1633450566627589,
      "grad_norm": 9.337977409362793,
      "learning_rate": 1.9103952703550642e-05,
      "loss": 11.1465,
      "step": 836
    },
    {
      "epoch": 0.16354044548651817,
      "grad_norm": 8.037405014038086,
      "learning_rate": 1.9101332425100016e-05,
      "loss": 10.2202,
      "step": 837
    },
    {
      "epoch": 0.16373583431027744,
      "grad_norm": 10.364066123962402,
      "learning_rate": 1.9098708501283188e-05,
      "loss": 9.844,
      "step": 838
    },
    {
      "epoch": 0.16393122313403674,
      "grad_norm": 7.588009834289551,
      "learning_rate": 1.9096080933151123e-05,
      "loss": 9.8606,
      "step": 839
    },
    {
      "epoch": 0.16412661195779601,
      "grad_norm": 16.311908721923828,
      "learning_rate": 1.9093449721756246e-05,
      "loss": 10.1571,
      "step": 840
    },
    {
      "epoch": 0.1643220007815553,
      "grad_norm": 9.902872085571289,
      "learning_rate": 1.9090814868152435e-05,
      "loss": 8.7677,
      "step": 841
    },
    {
      "epoch": 0.16451738960531456,
      "grad_norm": 9.914548873901367,
      "learning_rate": 1.908817637339503e-05,
      "loss": 9.726,
      "step": 842
    },
    {
      "epoch": 0.16471277842907386,
      "grad_norm": 10.072443962097168,
      "learning_rate": 1.9085534238540834e-05,
      "loss": 10.4707,
      "step": 843
    },
    {
      "epoch": 0.16490816725283314,
      "grad_norm": 9.244873046875,
      "learning_rate": 1.9082888464648103e-05,
      "loss": 10.3142,
      "step": 844
    },
    {
      "epoch": 0.1651035560765924,
      "grad_norm": 8.684638023376465,
      "learning_rate": 1.9080239052776554e-05,
      "loss": 10.3544,
      "step": 845
    },
    {
      "epoch": 0.1652989449003517,
      "grad_norm": 12.886144638061523,
      "learning_rate": 1.907758600398735e-05,
      "loss": 10.8514,
      "step": 846
    },
    {
      "epoch": 0.165494333724111,
      "grad_norm": 13.057008743286133,
      "learning_rate": 1.9074929319343126e-05,
      "loss": 9.6809,
      "step": 847
    },
    {
      "epoch": 0.16568972254787026,
      "grad_norm": 11.642423629760742,
      "learning_rate": 1.9072268999907967e-05,
      "loss": 10.628,
      "step": 848
    },
    {
      "epoch": 0.16588511137162953,
      "grad_norm": 13.759044647216797,
      "learning_rate": 1.9069605046747414e-05,
      "loss": 10.9427,
      "step": 849
    },
    {
      "epoch": 0.16608050019538884,
      "grad_norm": 9.928781509399414,
      "learning_rate": 1.906693746092846e-05,
      "loss": 9.9875,
      "step": 850
    },
    {
      "epoch": 0.1662758890191481,
      "grad_norm": 15.753911018371582,
      "learning_rate": 1.9064266243519553e-05,
      "loss": 10.0349,
      "step": 851
    },
    {
      "epoch": 0.16647127784290738,
      "grad_norm": 10.317689895629883,
      "learning_rate": 1.9061591395590606e-05,
      "loss": 9.6778,
      "step": 852
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 9.685894012451172,
      "learning_rate": 1.9058912918212976e-05,
      "loss": 9.3574,
      "step": 853
    },
    {
      "epoch": 0.16686205549042596,
      "grad_norm": 12.898737907409668,
      "learning_rate": 1.9056230812459476e-05,
      "loss": 10.8164,
      "step": 854
    },
    {
      "epoch": 0.16705744431418523,
      "grad_norm": 11.205578804016113,
      "learning_rate": 1.905354507940437e-05,
      "loss": 9.3245,
      "step": 855
    },
    {
      "epoch": 0.1672528331379445,
      "grad_norm": 13.233284950256348,
      "learning_rate": 1.905085572012338e-05,
      "loss": 10.7563,
      "step": 856
    },
    {
      "epoch": 0.16744822196170378,
      "grad_norm": 12.376396179199219,
      "learning_rate": 1.904816273569368e-05,
      "loss": 10.7316,
      "step": 857
    },
    {
      "epoch": 0.16764361078546308,
      "grad_norm": 16.27195930480957,
      "learning_rate": 1.904546612719389e-05,
      "loss": 10.638,
      "step": 858
    },
    {
      "epoch": 0.16783899960922236,
      "grad_norm": 7.701115131378174,
      "learning_rate": 1.9042765895704085e-05,
      "loss": 9.1598,
      "step": 859
    },
    {
      "epoch": 0.16803438843298163,
      "grad_norm": 36.823429107666016,
      "learning_rate": 1.9040062042305796e-05,
      "loss": 10.7289,
      "step": 860
    },
    {
      "epoch": 0.1682297772567409,
      "grad_norm": 9.800819396972656,
      "learning_rate": 1.9037354568081995e-05,
      "loss": 10.1007,
      "step": 861
    },
    {
      "epoch": 0.1684251660805002,
      "grad_norm": 12.288325309753418,
      "learning_rate": 1.9034643474117114e-05,
      "loss": 9.1896,
      "step": 862
    },
    {
      "epoch": 0.16862055490425948,
      "grad_norm": 10.411764144897461,
      "learning_rate": 1.9031928761497026e-05,
      "loss": 10.3171,
      "step": 863
    },
    {
      "epoch": 0.16881594372801875,
      "grad_norm": 10.256525993347168,
      "learning_rate": 1.9029210431309063e-05,
      "loss": 9.6621,
      "step": 864
    },
    {
      "epoch": 0.16901133255177803,
      "grad_norm": 13.477582931518555,
      "learning_rate": 1.9026488484641995e-05,
      "loss": 9.3424,
      "step": 865
    },
    {
      "epoch": 0.16920672137553733,
      "grad_norm": 11.819972038269043,
      "learning_rate": 1.9023762922586052e-05,
      "loss": 9.5409,
      "step": 866
    },
    {
      "epoch": 0.1694021101992966,
      "grad_norm": 12.532838821411133,
      "learning_rate": 1.9021033746232903e-05,
      "loss": 9.0871,
      "step": 867
    },
    {
      "epoch": 0.16959749902305588,
      "grad_norm": 15.850306510925293,
      "learning_rate": 1.9018300956675667e-05,
      "loss": 9.578,
      "step": 868
    },
    {
      "epoch": 0.16979288784681515,
      "grad_norm": 46.099002838134766,
      "learning_rate": 1.9015564555008912e-05,
      "loss": 9.3759,
      "step": 869
    },
    {
      "epoch": 0.16998827667057445,
      "grad_norm": 20.86892318725586,
      "learning_rate": 1.901282454232865e-05,
      "loss": 11.2573,
      "step": 870
    },
    {
      "epoch": 0.17018366549433372,
      "grad_norm": 14.856077194213867,
      "learning_rate": 1.9010080919732347e-05,
      "loss": 10.2307,
      "step": 871
    },
    {
      "epoch": 0.170379054318093,
      "grad_norm": 11.684233665466309,
      "learning_rate": 1.9007333688318902e-05,
      "loss": 9.5953,
      "step": 872
    },
    {
      "epoch": 0.17057444314185227,
      "grad_norm": 21.605602264404297,
      "learning_rate": 1.9004582849188668e-05,
      "loss": 11.3202,
      "step": 873
    },
    {
      "epoch": 0.17076983196561157,
      "grad_norm": 20.927377700805664,
      "learning_rate": 1.9001828403443444e-05,
      "loss": 11.4154,
      "step": 874
    },
    {
      "epoch": 0.17096522078937085,
      "grad_norm": 13.724664688110352,
      "learning_rate": 1.8999070352186474e-05,
      "loss": 10.2794,
      "step": 875
    },
    {
      "epoch": 0.17116060961313012,
      "grad_norm": 11.638361930847168,
      "learning_rate": 1.8996308696522435e-05,
      "loss": 10.5548,
      "step": 876
    },
    {
      "epoch": 0.17135599843688942,
      "grad_norm": 23.183778762817383,
      "learning_rate": 1.8993543437557462e-05,
      "loss": 11.002,
      "step": 877
    },
    {
      "epoch": 0.1715513872606487,
      "grad_norm": 30.77211570739746,
      "learning_rate": 1.8990774576399124e-05,
      "loss": 11.6167,
      "step": 878
    },
    {
      "epoch": 0.17174677608440797,
      "grad_norm": 113.89714813232422,
      "learning_rate": 1.898800211415644e-05,
      "loss": 11.3287,
      "step": 879
    },
    {
      "epoch": 0.17194216490816724,
      "grad_norm": 8.955188751220703,
      "learning_rate": 1.898522605193986e-05,
      "loss": 10.0213,
      "step": 880
    },
    {
      "epoch": 0.17213755373192655,
      "grad_norm": 14.04103946685791,
      "learning_rate": 1.8982446390861286e-05,
      "loss": 10.7902,
      "step": 881
    },
    {
      "epoch": 0.17233294255568582,
      "grad_norm": 10.359882354736328,
      "learning_rate": 1.8979663132034062e-05,
      "loss": 10.6917,
      "step": 882
    },
    {
      "epoch": 0.1725283313794451,
      "grad_norm": 8.169754981994629,
      "learning_rate": 1.897687627657297e-05,
      "loss": 10.1715,
      "step": 883
    },
    {
      "epoch": 0.17272372020320437,
      "grad_norm": 10.486644744873047,
      "learning_rate": 1.8974085825594226e-05,
      "loss": 10.3867,
      "step": 884
    },
    {
      "epoch": 0.17291910902696367,
      "grad_norm": 34.91582489013672,
      "learning_rate": 1.8971291780215496e-05,
      "loss": 9.8048,
      "step": 885
    },
    {
      "epoch": 0.17311449785072294,
      "grad_norm": 11.902545928955078,
      "learning_rate": 1.8968494141555882e-05,
      "loss": 10.4441,
      "step": 886
    },
    {
      "epoch": 0.17330988667448222,
      "grad_norm": 11.266779899597168,
      "learning_rate": 1.8965692910735925e-05,
      "loss": 10.7083,
      "step": 887
    },
    {
      "epoch": 0.1735052754982415,
      "grad_norm": 25.55230712890625,
      "learning_rate": 1.8962888088877607e-05,
      "loss": 9.981,
      "step": 888
    },
    {
      "epoch": 0.1737006643220008,
      "grad_norm": 10.482330322265625,
      "learning_rate": 1.8960079677104342e-05,
      "loss": 9.7049,
      "step": 889
    },
    {
      "epoch": 0.17389605314576007,
      "grad_norm": 9.570494651794434,
      "learning_rate": 1.895726767654099e-05,
      "loss": 9.6493,
      "step": 890
    },
    {
      "epoch": 0.17409144196951934,
      "grad_norm": 10.879228591918945,
      "learning_rate": 1.895445208831384e-05,
      "loss": 10.1414,
      "step": 891
    },
    {
      "epoch": 0.1742868307932786,
      "grad_norm": 14.008447647094727,
      "learning_rate": 1.8951632913550625e-05,
      "loss": 10.9662,
      "step": 892
    },
    {
      "epoch": 0.17448221961703791,
      "grad_norm": 10.85938549041748,
      "learning_rate": 1.8948810153380514e-05,
      "loss": 9.9632,
      "step": 893
    },
    {
      "epoch": 0.1746776084407972,
      "grad_norm": 10.80557918548584,
      "learning_rate": 1.8945983808934105e-05,
      "loss": 10.1232,
      "step": 894
    },
    {
      "epoch": 0.17487299726455646,
      "grad_norm": 14.553106307983398,
      "learning_rate": 1.894315388134344e-05,
      "loss": 11.7757,
      "step": 895
    },
    {
      "epoch": 0.17506838608831574,
      "grad_norm": 10.305713653564453,
      "learning_rate": 1.8940320371741995e-05,
      "loss": 10.6498,
      "step": 896
    },
    {
      "epoch": 0.17526377491207504,
      "grad_norm": 19.3458194732666,
      "learning_rate": 1.893748328126468e-05,
      "loss": 10.2667,
      "step": 897
    },
    {
      "epoch": 0.1754591637358343,
      "grad_norm": 10.901649475097656,
      "learning_rate": 1.8934642611047828e-05,
      "loss": 9.6728,
      "step": 898
    },
    {
      "epoch": 0.17565455255959359,
      "grad_norm": 7.951071262359619,
      "learning_rate": 1.8931798362229227e-05,
      "loss": 9.3648,
      "step": 899
    },
    {
      "epoch": 0.17584994138335286,
      "grad_norm": 12.390594482421875,
      "learning_rate": 1.8928950535948077e-05,
      "loss": 9.9003,
      "step": 900
    },
    {
      "epoch": 0.17604533020711216,
      "grad_norm": 10.91873836517334,
      "learning_rate": 1.8926099133345025e-05,
      "loss": 10.0891,
      "step": 901
    },
    {
      "epoch": 0.17624071903087143,
      "grad_norm": 10.91873836517334,
      "learning_rate": 1.8926099133345025e-05,
      "loss": 11.0351,
      "step": 902
    },
    {
      "epoch": 0.1764361078546307,
      "grad_norm": 12.28889274597168,
      "learning_rate": 1.8923244155562147e-05,
      "loss": 10.2997,
      "step": 903
    },
    {
      "epoch": 0.17663149667839,
      "grad_norm": 13.323399543762207,
      "learning_rate": 1.8920385603742955e-05,
      "loss": 10.6953,
      "step": 904
    },
    {
      "epoch": 0.17682688550214928,
      "grad_norm": 9.011679649353027,
      "learning_rate": 1.8917523479032378e-05,
      "loss": 10.1096,
      "step": 905
    },
    {
      "epoch": 0.17702227432590856,
      "grad_norm": 13.647565841674805,
      "learning_rate": 1.8914657782576795e-05,
      "loss": 9.576,
      "step": 906
    },
    {
      "epoch": 0.17721766314966783,
      "grad_norm": 18.961706161499023,
      "learning_rate": 1.8911788515524e-05,
      "loss": 10.1923,
      "step": 907
    },
    {
      "epoch": 0.17741305197342713,
      "grad_norm": 15.747967720031738,
      "learning_rate": 1.8908915679023222e-05,
      "loss": 10.5809,
      "step": 908
    },
    {
      "epoch": 0.1776084407971864,
      "grad_norm": 14.569042205810547,
      "learning_rate": 1.8906039274225126e-05,
      "loss": 10.8621,
      "step": 909
    },
    {
      "epoch": 0.17780382962094568,
      "grad_norm": 14.774585723876953,
      "learning_rate": 1.8903159302281803e-05,
      "loss": 10.6481,
      "step": 910
    },
    {
      "epoch": 0.17799921844470495,
      "grad_norm": 11.765643119812012,
      "learning_rate": 1.890027576434677e-05,
      "loss": 10.7216,
      "step": 911
    },
    {
      "epoch": 0.17819460726846426,
      "grad_norm": 9.366642951965332,
      "learning_rate": 1.889738866157497e-05,
      "loss": 10.1954,
      "step": 912
    },
    {
      "epoch": 0.17838999609222353,
      "grad_norm": 10.075114250183105,
      "learning_rate": 1.8894497995122775e-05,
      "loss": 10.356,
      "step": 913
    },
    {
      "epoch": 0.1785853849159828,
      "grad_norm": 12.980185508728027,
      "learning_rate": 1.8891603766147994e-05,
      "loss": 10.7433,
      "step": 914
    },
    {
      "epoch": 0.17878077373974208,
      "grad_norm": 27.603477478027344,
      "learning_rate": 1.888870597580985e-05,
      "loss": 9.8225,
      "step": 915
    },
    {
      "epoch": 0.17897616256350138,
      "grad_norm": 9.369039535522461,
      "learning_rate": 1.8885804625269004e-05,
      "loss": 9.6697,
      "step": 916
    },
    {
      "epoch": 0.17917155138726065,
      "grad_norm": 9.285225868225098,
      "learning_rate": 1.888289971568753e-05,
      "loss": 11.0268,
      "step": 917
    },
    {
      "epoch": 0.17936694021101993,
      "grad_norm": 7.665126800537109,
      "learning_rate": 1.8879991248228942e-05,
      "loss": 10.105,
      "step": 918
    },
    {
      "epoch": 0.1795623290347792,
      "grad_norm": 21.36392593383789,
      "learning_rate": 1.8877079224058167e-05,
      "loss": 10.8853,
      "step": 919
    },
    {
      "epoch": 0.1797577178585385,
      "grad_norm": 8.818321228027344,
      "learning_rate": 1.8874163644341556e-05,
      "loss": 10.3977,
      "step": 920
    },
    {
      "epoch": 0.17995310668229778,
      "grad_norm": 21.660226821899414,
      "learning_rate": 1.8871244510246898e-05,
      "loss": 10.5013,
      "step": 921
    },
    {
      "epoch": 0.18014849550605705,
      "grad_norm": 15.823149681091309,
      "learning_rate": 1.8868321822943396e-05,
      "loss": 10.5462,
      "step": 922
    },
    {
      "epoch": 0.18034388432981632,
      "grad_norm": 8.88244342803955,
      "learning_rate": 1.886539558360167e-05,
      "loss": 10.4773,
      "step": 923
    },
    {
      "epoch": 0.18053927315357562,
      "grad_norm": 9.925248146057129,
      "learning_rate": 1.8862465793393775e-05,
      "loss": 10.5078,
      "step": 924
    },
    {
      "epoch": 0.1807346619773349,
      "grad_norm": 16.658496856689453,
      "learning_rate": 1.8859532453493182e-05,
      "loss": 10.6751,
      "step": 925
    },
    {
      "epoch": 0.18093005080109417,
      "grad_norm": 8.997674942016602,
      "learning_rate": 1.8856595565074786e-05,
      "loss": 9.6502,
      "step": 926
    },
    {
      "epoch": 0.18112543962485345,
      "grad_norm": 14.015752792358398,
      "learning_rate": 1.88536551293149e-05,
      "loss": 11.0057,
      "step": 927
    },
    {
      "epoch": 0.18132082844861275,
      "grad_norm": 9.927313804626465,
      "learning_rate": 1.8850711147391257e-05,
      "loss": 9.5504,
      "step": 928
    },
    {
      "epoch": 0.18151621727237202,
      "grad_norm": 9.687064170837402,
      "learning_rate": 1.8847763620483023e-05,
      "loss": 9.8367,
      "step": 929
    },
    {
      "epoch": 0.1817116060961313,
      "grad_norm": 11.51573371887207,
      "learning_rate": 1.8844812549770765e-05,
      "loss": 10.774,
      "step": 930
    },
    {
      "epoch": 0.18190699491989057,
      "grad_norm": 14.399081230163574,
      "learning_rate": 1.884185793643648e-05,
      "loss": 10.9734,
      "step": 931
    },
    {
      "epoch": 0.18210238374364987,
      "grad_norm": 11.186507225036621,
      "learning_rate": 1.8838899781663587e-05,
      "loss": 10.5601,
      "step": 932
    },
    {
      "epoch": 0.18229777256740914,
      "grad_norm": 8.207215309143066,
      "learning_rate": 1.8835938086636913e-05,
      "loss": 11.1318,
      "step": 933
    },
    {
      "epoch": 0.18249316139116842,
      "grad_norm": 10.036437034606934,
      "learning_rate": 1.883297285254271e-05,
      "loss": 9.1535,
      "step": 934
    },
    {
      "epoch": 0.18268855021492772,
      "grad_norm": 8.014676094055176,
      "learning_rate": 1.8830004080568648e-05,
      "loss": 9.3372,
      "step": 935
    },
    {
      "epoch": 0.182883939038687,
      "grad_norm": 11.679019927978516,
      "learning_rate": 1.8827031771903814e-05,
      "loss": 9.9121,
      "step": 936
    },
    {
      "epoch": 0.18307932786244627,
      "grad_norm": 8.254961013793945,
      "learning_rate": 1.8824055927738706e-05,
      "loss": 10.4003,
      "step": 937
    },
    {
      "epoch": 0.18327471668620554,
      "grad_norm": 10.134276390075684,
      "learning_rate": 1.8821076549265246e-05,
      "loss": 11.0081,
      "step": 938
    },
    {
      "epoch": 0.18347010550996484,
      "grad_norm": 8.73877239227295,
      "learning_rate": 1.8818093637676763e-05,
      "loss": 9.1944,
      "step": 939
    },
    {
      "epoch": 0.18366549433372412,
      "grad_norm": 33.710693359375,
      "learning_rate": 1.8815107194168007e-05,
      "loss": 10.5759,
      "step": 940
    },
    {
      "epoch": 0.1838608831574834,
      "grad_norm": 7.068522930145264,
      "learning_rate": 1.8812117219935143e-05,
      "loss": 10.0031,
      "step": 941
    },
    {
      "epoch": 0.18405627198124266,
      "grad_norm": 16.002485275268555,
      "learning_rate": 1.880912371617575e-05,
      "loss": 9.6456,
      "step": 942
    },
    {
      "epoch": 0.18425166080500197,
      "grad_norm": 32.332855224609375,
      "learning_rate": 1.8806126684088813e-05,
      "loss": 11.6795,
      "step": 943
    },
    {
      "epoch": 0.18444704962876124,
      "grad_norm": 9.968116760253906,
      "learning_rate": 1.8803126124874748e-05,
      "loss": 9.8382,
      "step": 944
    },
    {
      "epoch": 0.1846424384525205,
      "grad_norm": 11.146265983581543,
      "learning_rate": 1.880012203973536e-05,
      "loss": 10.6834,
      "step": 945
    },
    {
      "epoch": 0.1848378272762798,
      "grad_norm": 11.863654136657715,
      "learning_rate": 1.879711442987388e-05,
      "loss": 10.1275,
      "step": 946
    },
    {
      "epoch": 0.1850332161000391,
      "grad_norm": 11.573152542114258,
      "learning_rate": 1.8794103296494954e-05,
      "loss": 12.0389,
      "step": 947
    },
    {
      "epoch": 0.18522860492379836,
      "grad_norm": 10.724802017211914,
      "learning_rate": 1.8791088640804634e-05,
      "loss": 10.1915,
      "step": 948
    },
    {
      "epoch": 0.18542399374755764,
      "grad_norm": 11.720141410827637,
      "learning_rate": 1.878807046401038e-05,
      "loss": 9.1108,
      "step": 949
    },
    {
      "epoch": 0.1856193825713169,
      "grad_norm": 11.617509841918945,
      "learning_rate": 1.878504876732107e-05,
      "loss": 10.0505,
      "step": 950
    },
    {
      "epoch": 0.1858147713950762,
      "grad_norm": 12.604987144470215,
      "learning_rate": 1.878202355194698e-05,
      "loss": 9.9923,
      "step": 951
    },
    {
      "epoch": 0.18601016021883549,
      "grad_norm": 9.539042472839355,
      "learning_rate": 1.877899481909981e-05,
      "loss": 9.6927,
      "step": 952
    },
    {
      "epoch": 0.18620554904259476,
      "grad_norm": 9.913569450378418,
      "learning_rate": 1.8775962569992662e-05,
      "loss": 10.0365,
      "step": 953
    },
    {
      "epoch": 0.18640093786635403,
      "grad_norm": 35.884490966796875,
      "learning_rate": 1.8772926805840036e-05,
      "loss": 10.3549,
      "step": 954
    },
    {
      "epoch": 0.18659632669011333,
      "grad_norm": 16.634611129760742,
      "learning_rate": 1.876988752785786e-05,
      "loss": 10.5787,
      "step": 955
    },
    {
      "epoch": 0.1867917155138726,
      "grad_norm": 51.59297180175781,
      "learning_rate": 1.8766844737263456e-05,
      "loss": 10.6785,
      "step": 956
    },
    {
      "epoch": 0.18698710433763188,
      "grad_norm": 10.21894645690918,
      "learning_rate": 1.8763798435275554e-05,
      "loss": 10.3058,
      "step": 957
    },
    {
      "epoch": 0.18718249316139116,
      "grad_norm": 10.180886268615723,
      "learning_rate": 1.8760748623114295e-05,
      "loss": 10.3053,
      "step": 958
    },
    {
      "epoch": 0.18737788198515046,
      "grad_norm": 19.828392028808594,
      "learning_rate": 1.8757695302001223e-05,
      "loss": 10.1661,
      "step": 959
    },
    {
      "epoch": 0.18757327080890973,
      "grad_norm": 10.160892486572266,
      "learning_rate": 1.8754638473159285e-05,
      "loss": 11.049,
      "step": 960
    },
    {
      "epoch": 0.187768659632669,
      "grad_norm": 12.746784210205078,
      "learning_rate": 1.875157813781284e-05,
      "loss": 9.86,
      "step": 961
    },
    {
      "epoch": 0.18796404845642828,
      "grad_norm": 14.9076509475708,
      "learning_rate": 1.874851429718765e-05,
      "loss": 10.0313,
      "step": 962
    },
    {
      "epoch": 0.18815943728018758,
      "grad_norm": 8.683619499206543,
      "learning_rate": 1.8745446952510868e-05,
      "loss": 10.4593,
      "step": 963
    },
    {
      "epoch": 0.18835482610394685,
      "grad_norm": 17.066213607788086,
      "learning_rate": 1.8742376105011072e-05,
      "loss": 10.2991,
      "step": 964
    },
    {
      "epoch": 0.18855021492770613,
      "grad_norm": 17.554244995117188,
      "learning_rate": 1.873930175591822e-05,
      "loss": 10.2592,
      "step": 965
    },
    {
      "epoch": 0.18874560375146543,
      "grad_norm": 11.156131744384766,
      "learning_rate": 1.8736223906463698e-05,
      "loss": 9.3706,
      "step": 966
    },
    {
      "epoch": 0.1889409925752247,
      "grad_norm": 8.87385368347168,
      "learning_rate": 1.873314255788027e-05,
      "loss": 8.2635,
      "step": 967
    },
    {
      "epoch": 0.18913638139898398,
      "grad_norm": 11.072525978088379,
      "learning_rate": 1.8730057711402113e-05,
      "loss": 9.8821,
      "step": 968
    },
    {
      "epoch": 0.18933177022274325,
      "grad_norm": 10.544384002685547,
      "learning_rate": 1.872696936826481e-05,
      "loss": 9.6465,
      "step": 969
    },
    {
      "epoch": 0.18952715904650255,
      "grad_norm": 18.760879516601562,
      "learning_rate": 1.872387752970533e-05,
      "loss": 9.3924,
      "step": 970
    },
    {
      "epoch": 0.18972254787026183,
      "grad_norm": 10.454094886779785,
      "learning_rate": 1.8720782196962055e-05,
      "loss": 10.16,
      "step": 971
    },
    {
      "epoch": 0.1899179366940211,
      "grad_norm": 13.4886474609375,
      "learning_rate": 1.8717683371274763e-05,
      "loss": 11.129,
      "step": 972
    },
    {
      "epoch": 0.19011332551778037,
      "grad_norm": 8.18844223022461,
      "learning_rate": 1.8714581053884628e-05,
      "loss": 9.3449,
      "step": 973
    },
    {
      "epoch": 0.19030871434153968,
      "grad_norm": 7.980138301849365,
      "learning_rate": 1.871147524603422e-05,
      "loss": 10.3902,
      "step": 974
    },
    {
      "epoch": 0.19050410316529895,
      "grad_norm": 11.241415977478027,
      "learning_rate": 1.870836594896752e-05,
      "loss": 9.4598,
      "step": 975
    },
    {
      "epoch": 0.19069949198905822,
      "grad_norm": 8.223668098449707,
      "learning_rate": 1.8705253163929895e-05,
      "loss": 8.8009,
      "step": 976
    },
    {
      "epoch": 0.1908948808128175,
      "grad_norm": 8.505722999572754,
      "learning_rate": 1.870213689216811e-05,
      "loss": 9.5131,
      "step": 977
    },
    {
      "epoch": 0.1910902696365768,
      "grad_norm": 13.652040481567383,
      "learning_rate": 1.8699017134930327e-05,
      "loss": 9.904,
      "step": 978
    },
    {
      "epoch": 0.19128565846033607,
      "grad_norm": 8.11439037322998,
      "learning_rate": 1.869589389346611e-05,
      "loss": 9.4733,
      "step": 979
    },
    {
      "epoch": 0.19148104728409535,
      "grad_norm": 11.425436019897461,
      "learning_rate": 1.8692767169026415e-05,
      "loss": 10.8716,
      "step": 980
    },
    {
      "epoch": 0.19167643610785462,
      "grad_norm": 10.17133617401123,
      "learning_rate": 1.8689636962863588e-05,
      "loss": 9.3576,
      "step": 981
    },
    {
      "epoch": 0.19187182493161392,
      "grad_norm": 9.900402069091797,
      "learning_rate": 1.8686503276231372e-05,
      "loss": 10.6097,
      "step": 982
    },
    {
      "epoch": 0.1920672137553732,
      "grad_norm": 27.443452835083008,
      "learning_rate": 1.8683366110384912e-05,
      "loss": 11.2632,
      "step": 983
    },
    {
      "epoch": 0.19226260257913247,
      "grad_norm": 11.592450141906738,
      "learning_rate": 1.868022546658074e-05,
      "loss": 8.5889,
      "step": 984
    },
    {
      "epoch": 0.19245799140289174,
      "grad_norm": 16.061792373657227,
      "learning_rate": 1.8677081346076777e-05,
      "loss": 10.4353,
      "step": 985
    },
    {
      "epoch": 0.19265338022665104,
      "grad_norm": 19.611604690551758,
      "learning_rate": 1.8673933750132342e-05,
      "loss": 10.8018,
      "step": 986
    },
    {
      "epoch": 0.19284876905041032,
      "grad_norm": 10.408438682556152,
      "learning_rate": 1.8670782680008145e-05,
      "loss": 10.5335,
      "step": 987
    },
    {
      "epoch": 0.1930441578741696,
      "grad_norm": 6.8224873542785645,
      "learning_rate": 1.866762813696629e-05,
      "loss": 8.9648,
      "step": 988
    },
    {
      "epoch": 0.19323954669792887,
      "grad_norm": 8.705338478088379,
      "learning_rate": 1.866447012227026e-05,
      "loss": 10.8897,
      "step": 989
    },
    {
      "epoch": 0.19343493552168817,
      "grad_norm": 9.13908863067627,
      "learning_rate": 1.8661308637184955e-05,
      "loss": 10.2841,
      "step": 990
    },
    {
      "epoch": 0.19363032434544744,
      "grad_norm": 11.744199752807617,
      "learning_rate": 1.8658143682976636e-05,
      "loss": 10.8832,
      "step": 991
    },
    {
      "epoch": 0.19382571316920671,
      "grad_norm": 10.985271453857422,
      "learning_rate": 1.8654975260912963e-05,
      "loss": 11.1635,
      "step": 992
    },
    {
      "epoch": 0.194021101992966,
      "grad_norm": 31.188379287719727,
      "learning_rate": 1.8651803372263e-05,
      "loss": 10.509,
      "step": 993
    },
    {
      "epoch": 0.1942164908167253,
      "grad_norm": 22.462879180908203,
      "learning_rate": 1.864862801829717e-05,
      "loss": 11.0881,
      "step": 994
    },
    {
      "epoch": 0.19441187964048456,
      "grad_norm": 7.340974807739258,
      "learning_rate": 1.8645449200287317e-05,
      "loss": 10.0884,
      "step": 995
    },
    {
      "epoch": 0.19460726846424384,
      "grad_norm": 27.147672653198242,
      "learning_rate": 1.8642266919506644e-05,
      "loss": 9.2816,
      "step": 996
    },
    {
      "epoch": 0.19480265728800314,
      "grad_norm": 12.003215789794922,
      "learning_rate": 1.8639081177229763e-05,
      "loss": 11.5175,
      "step": 997
    },
    {
      "epoch": 0.1949980461117624,
      "grad_norm": 8.73083782196045,
      "learning_rate": 1.8635891974732655e-05,
      "loss": 10.0267,
      "step": 998
    },
    {
      "epoch": 0.1951934349355217,
      "grad_norm": 8.434202194213867,
      "learning_rate": 1.8632699313292702e-05,
      "loss": 10.004,
      "step": 999
    },
    {
      "epoch": 0.19538882375928096,
      "grad_norm": 14.637370109558105,
      "learning_rate": 1.8629503194188652e-05,
      "loss": 9.7757,
      "step": 1000
    },
    {
      "epoch": 0.19558421258304026,
      "grad_norm": 10.80057430267334,
      "learning_rate": 1.8626303618700664e-05,
      "loss": 9.355,
      "step": 1001
    },
    {
      "epoch": 0.19577960140679954,
      "grad_norm": 9.21177864074707,
      "learning_rate": 1.862310058811026e-05,
      "loss": 10.9338,
      "step": 1002
    },
    {
      "epoch": 0.1959749902305588,
      "grad_norm": 7.865182876586914,
      "learning_rate": 1.8619894103700354e-05,
      "loss": 8.2683,
      "step": 1003
    },
    {
      "epoch": 0.19617037905431808,
      "grad_norm": 13.538627624511719,
      "learning_rate": 1.861668416675524e-05,
      "loss": 11.2537,
      "step": 1004
    },
    {
      "epoch": 0.19636576787807739,
      "grad_norm": 9.453564643859863,
      "learning_rate": 1.86134707785606e-05,
      "loss": 9.258,
      "step": 1005
    },
    {
      "epoch": 0.19656115670183666,
      "grad_norm": 8.552481651306152,
      "learning_rate": 1.86102539404035e-05,
      "loss": 10.1384,
      "step": 1006
    },
    {
      "epoch": 0.19675654552559593,
      "grad_norm": 27.59398651123047,
      "learning_rate": 1.860703365357238e-05,
      "loss": 10.9349,
      "step": 1007
    },
    {
      "epoch": 0.1969519343493552,
      "grad_norm": 9.849193572998047,
      "learning_rate": 1.8603809919357063e-05,
      "loss": 11.5077,
      "step": 1008
    },
    {
      "epoch": 0.1971473231731145,
      "grad_norm": 8.214253425598145,
      "learning_rate": 1.8600582739048752e-05,
      "loss": 9.8551,
      "step": 1009
    },
    {
      "epoch": 0.19734271199687378,
      "grad_norm": 14.345209121704102,
      "learning_rate": 1.859735211394004e-05,
      "loss": 8.9469,
      "step": 1010
    },
    {
      "epoch": 0.19753810082063306,
      "grad_norm": 24.646177291870117,
      "learning_rate": 1.859411804532489e-05,
      "loss": 11.3968,
      "step": 1011
    },
    {
      "epoch": 0.19773348964439233,
      "grad_norm": 12.425567626953125,
      "learning_rate": 1.8590880534498644e-05,
      "loss": 10.8658,
      "step": 1012
    },
    {
      "epoch": 0.19792887846815163,
      "grad_norm": 10.923027038574219,
      "learning_rate": 1.8587639582758032e-05,
      "loss": 10.8557,
      "step": 1013
    },
    {
      "epoch": 0.1981242672919109,
      "grad_norm": 8.746227264404297,
      "learning_rate": 1.858439519140115e-05,
      "loss": 9.8136,
      "step": 1014
    },
    {
      "epoch": 0.19831965611567018,
      "grad_norm": 9.470844268798828,
      "learning_rate": 1.858114736172748e-05,
      "loss": 9.5208,
      "step": 1015
    },
    {
      "epoch": 0.19851504493942945,
      "grad_norm": 10.059723854064941,
      "learning_rate": 1.8577896095037875e-05,
      "loss": 9.477,
      "step": 1016
    },
    {
      "epoch": 0.19871043376318875,
      "grad_norm": 11.267123222351074,
      "learning_rate": 1.8574641392634573e-05,
      "loss": 10.8301,
      "step": 1017
    },
    {
      "epoch": 0.19890582258694803,
      "grad_norm": 10.173998832702637,
      "learning_rate": 1.857138325582118e-05,
      "loss": 9.8479,
      "step": 1018
    },
    {
      "epoch": 0.1991012114107073,
      "grad_norm": 9.12119197845459,
      "learning_rate": 1.8568121685902683e-05,
      "loss": 10.9387,
      "step": 1019
    },
    {
      "epoch": 0.19929660023446658,
      "grad_norm": 15.177370071411133,
      "learning_rate": 1.856485668418544e-05,
      "loss": 9.1175,
      "step": 1020
    },
    {
      "epoch": 0.19949198905822588,
      "grad_norm": 10.732454299926758,
      "learning_rate": 1.856158825197719e-05,
      "loss": 10.0042,
      "step": 1021
    },
    {
      "epoch": 0.19968737788198515,
      "grad_norm": 7.855399131774902,
      "learning_rate": 1.8558316390587036e-05,
      "loss": 9.8825,
      "step": 1022
    },
    {
      "epoch": 0.19988276670574442,
      "grad_norm": 11.231322288513184,
      "learning_rate": 1.8555041101325466e-05,
      "loss": 10.0247,
      "step": 1023
    },
    {
      "epoch": 0.20007815552950373,
      "grad_norm": 10.086041450500488,
      "learning_rate": 1.855176238550433e-05,
      "loss": 9.5155,
      "step": 1024
    },
    {
      "epoch": 0.200273544353263,
      "grad_norm": 11.300803184509277,
      "learning_rate": 1.8548480244436855e-05,
      "loss": 10.328,
      "step": 1025
    },
    {
      "epoch": 0.20046893317702227,
      "grad_norm": 12.238363265991211,
      "learning_rate": 1.8545194679437646e-05,
      "loss": 10.4964,
      "step": 1026
    },
    {
      "epoch": 0.20066432200078155,
      "grad_norm": 10.1900053024292,
      "learning_rate": 1.8541905691822667e-05,
      "loss": 8.9627,
      "step": 1027
    },
    {
      "epoch": 0.20085971082454085,
      "grad_norm": 14.579061508178711,
      "learning_rate": 1.8538613282909263e-05,
      "loss": 10.9295,
      "step": 1028
    },
    {
      "epoch": 0.20105509964830012,
      "grad_norm": 11.488454818725586,
      "learning_rate": 1.853531745401615e-05,
      "loss": 9.7865,
      "step": 1029
    },
    {
      "epoch": 0.2012504884720594,
      "grad_norm": 11.466203689575195,
      "learning_rate": 1.8532018206463404e-05,
      "loss": 9.4745,
      "step": 1030
    },
    {
      "epoch": 0.20144587729581867,
      "grad_norm": 12.232128143310547,
      "learning_rate": 1.8528715541572475e-05,
      "loss": 10.396,
      "step": 1031
    },
    {
      "epoch": 0.20164126611957797,
      "grad_norm": 10.786626815795898,
      "learning_rate": 1.852540946066619e-05,
      "loss": 10.9871,
      "step": 1032
    },
    {
      "epoch": 0.20183665494333725,
      "grad_norm": 18.612485885620117,
      "learning_rate": 1.852209996506873e-05,
      "loss": 11.799,
      "step": 1033
    },
    {
      "epoch": 0.20203204376709652,
      "grad_norm": 12.24704360961914,
      "learning_rate": 1.851878705610565e-05,
      "loss": 10.1001,
      "step": 1034
    },
    {
      "epoch": 0.2022274325908558,
      "grad_norm": 10.10387134552002,
      "learning_rate": 1.8515470735103883e-05,
      "loss": 9.901,
      "step": 1035
    },
    {
      "epoch": 0.2024228214146151,
      "grad_norm": 18.21666145324707,
      "learning_rate": 1.8512151003391708e-05,
      "loss": 10.8576,
      "step": 1036
    },
    {
      "epoch": 0.20261821023837437,
      "grad_norm": 9.849905014038086,
      "learning_rate": 1.8508827862298783e-05,
      "loss": 10.298,
      "step": 1037
    },
    {
      "epoch": 0.20281359906213364,
      "grad_norm": 9.479056358337402,
      "learning_rate": 1.8505501313156135e-05,
      "loss": 10.3479,
      "step": 1038
    },
    {
      "epoch": 0.20300898788589292,
      "grad_norm": 75.90926361083984,
      "learning_rate": 1.8502171357296144e-05,
      "loss": 11.1973,
      "step": 1039
    },
    {
      "epoch": 0.20320437670965222,
      "grad_norm": 9.197609901428223,
      "learning_rate": 1.8498837996052563e-05,
      "loss": 9.2738,
      "step": 1040
    },
    {
      "epoch": 0.2033997655334115,
      "grad_norm": 13.858948707580566,
      "learning_rate": 1.8495501230760505e-05,
      "loss": 11.1317,
      "step": 1041
    },
    {
      "epoch": 0.20359515435717077,
      "grad_norm": 10.339825630187988,
      "learning_rate": 1.849216106275645e-05,
      "loss": 8.5552,
      "step": 1042
    },
    {
      "epoch": 0.20379054318093004,
      "grad_norm": 9.709257125854492,
      "learning_rate": 1.8488817493378238e-05,
      "loss": 10.1478,
      "step": 1043
    },
    {
      "epoch": 0.20398593200468934,
      "grad_norm": 9.045165061950684,
      "learning_rate": 1.8485470523965073e-05,
      "loss": 11.3759,
      "step": 1044
    },
    {
      "epoch": 0.20418132082844861,
      "grad_norm": 7.252753734588623,
      "learning_rate": 1.8482120155857517e-05,
      "loss": 10.4401,
      "step": 1045
    },
    {
      "epoch": 0.2043767096522079,
      "grad_norm": 7.973372936248779,
      "learning_rate": 1.8478766390397503e-05,
      "loss": 11.0516,
      "step": 1046
    },
    {
      "epoch": 0.20457209847596716,
      "grad_norm": 18.249378204345703,
      "learning_rate": 1.8475409228928314e-05,
      "loss": 10.5308,
      "step": 1047
    },
    {
      "epoch": 0.20476748729972646,
      "grad_norm": 10.62024211883545,
      "learning_rate": 1.8472048672794597e-05,
      "loss": 10.3611,
      "step": 1048
    },
    {
      "epoch": 0.20496287612348574,
      "grad_norm": 8.392167091369629,
      "learning_rate": 1.846868472334236e-05,
      "loss": 9.6377,
      "step": 1049
    },
    {
      "epoch": 0.205158264947245,
      "grad_norm": 12.937665939331055,
      "learning_rate": 1.8465317381918968e-05,
      "loss": 9.0667,
      "step": 1050
    },
    {
      "epoch": 0.20535365377100429,
      "grad_norm": 7.9081573486328125,
      "learning_rate": 1.8461946649873146e-05,
      "loss": 10.0356,
      "step": 1051
    },
    {
      "epoch": 0.2055490425947636,
      "grad_norm": 9.666654586791992,
      "learning_rate": 1.845857252855498e-05,
      "loss": 10.3346,
      "step": 1052
    },
    {
      "epoch": 0.20574443141852286,
      "grad_norm": 9.820070266723633,
      "learning_rate": 1.845519501931591e-05,
      "loss": 10.1557,
      "step": 1053
    },
    {
      "epoch": 0.20593982024228213,
      "grad_norm": 8.45248031616211,
      "learning_rate": 1.845181412350873e-05,
      "loss": 10.1601,
      "step": 1054
    },
    {
      "epoch": 0.20613520906604144,
      "grad_norm": 13.481633186340332,
      "learning_rate": 1.8448429842487593e-05,
      "loss": 10.1516,
      "step": 1055
    },
    {
      "epoch": 0.2063305978898007,
      "grad_norm": 9.514289855957031,
      "learning_rate": 1.844504217760801e-05,
      "loss": 9.6983,
      "step": 1056
    },
    {
      "epoch": 0.20652598671355998,
      "grad_norm": 8.149691581726074,
      "learning_rate": 1.844165113022685e-05,
      "loss": 10.0595,
      "step": 1057
    },
    {
      "epoch": 0.20672137553731926,
      "grad_norm": 44.551353454589844,
      "learning_rate": 1.8438256701702326e-05,
      "loss": 11.4602,
      "step": 1058
    },
    {
      "epoch": 0.20691676436107856,
      "grad_norm": 11.108083724975586,
      "learning_rate": 1.8434858893394015e-05,
      "loss": 9.0455,
      "step": 1059
    },
    {
      "epoch": 0.20711215318483783,
      "grad_norm": 10.105135917663574,
      "learning_rate": 1.843145770666285e-05,
      "loss": 9.4722,
      "step": 1060
    },
    {
      "epoch": 0.2073075420085971,
      "grad_norm": 14.592355728149414,
      "learning_rate": 1.8428053142871106e-05,
      "loss": 9.2339,
      "step": 1061
    },
    {
      "epoch": 0.20750293083235638,
      "grad_norm": 10.019697189331055,
      "learning_rate": 1.8424645203382415e-05,
      "loss": 9.3722,
      "step": 1062
    },
    {
      "epoch": 0.20769831965611568,
      "grad_norm": 8.274886131286621,
      "learning_rate": 1.8421233889561766e-05,
      "loss": 8.479,
      "step": 1063
    },
    {
      "epoch": 0.20789370847987496,
      "grad_norm": 9.821599006652832,
      "learning_rate": 1.8417819202775497e-05,
      "loss": 9.8521,
      "step": 1064
    },
    {
      "epoch": 0.20808909730363423,
      "grad_norm": 8.140969276428223,
      "learning_rate": 1.8414401144391288e-05,
      "loss": 8.7278,
      "step": 1065
    },
    {
      "epoch": 0.2082844861273935,
      "grad_norm": 10.504135131835938,
      "learning_rate": 1.841097971577819e-05,
      "loss": 9.6927,
      "step": 1066
    },
    {
      "epoch": 0.2084798749511528,
      "grad_norm": 10.516091346740723,
      "learning_rate": 1.8407554918306583e-05,
      "loss": 9.8247,
      "step": 1067
    },
    {
      "epoch": 0.20867526377491208,
      "grad_norm": 10.204903602600098,
      "learning_rate": 1.840412675334821e-05,
      "loss": 10.1618,
      "step": 1068
    },
    {
      "epoch": 0.20887065259867135,
      "grad_norm": 10.61201286315918,
      "learning_rate": 1.840069522227615e-05,
      "loss": 10.2239,
      "step": 1069
    },
    {
      "epoch": 0.20906604142243063,
      "grad_norm": 22.232080459594727,
      "learning_rate": 1.8397260326464843e-05,
      "loss": 10.6251,
      "step": 1070
    },
    {
      "epoch": 0.20926143024618993,
      "grad_norm": 11.604631423950195,
      "learning_rate": 1.839382206729007e-05,
      "loss": 10.1518,
      "step": 1071
    },
    {
      "epoch": 0.2094568190699492,
      "grad_norm": 12.9887113571167,
      "learning_rate": 1.8390380446128965e-05,
      "loss": 9.9893,
      "step": 1072
    },
    {
      "epoch": 0.20965220789370848,
      "grad_norm": 10.097661972045898,
      "learning_rate": 1.8386935464359997e-05,
      "loss": 10.3341,
      "step": 1073
    },
    {
      "epoch": 0.20984759671746775,
      "grad_norm": 9.482799530029297,
      "learning_rate": 1.8383487123362992e-05,
      "loss": 9.2421,
      "step": 1074
    },
    {
      "epoch": 0.21004298554122705,
      "grad_norm": 11.762171745300293,
      "learning_rate": 1.8380035424519116e-05,
      "loss": 9.2298,
      "step": 1075
    },
    {
      "epoch": 0.21023837436498632,
      "grad_norm": 17.297704696655273,
      "learning_rate": 1.8376580369210882e-05,
      "loss": 9.5459,
      "step": 1076
    },
    {
      "epoch": 0.2104337631887456,
      "grad_norm": 17.030080795288086,
      "learning_rate": 1.837312195882215e-05,
      "loss": 8.9085,
      "step": 1077
    },
    {
      "epoch": 0.21062915201250487,
      "grad_norm": 80.03797912597656,
      "learning_rate": 1.8369660194738118e-05,
      "loss": 10.4275,
      "step": 1078
    },
    {
      "epoch": 0.21082454083626417,
      "grad_norm": 10.00363826751709,
      "learning_rate": 1.8366195078345327e-05,
      "loss": 8.8138,
      "step": 1079
    },
    {
      "epoch": 0.21101992966002345,
      "grad_norm": 10.246661186218262,
      "learning_rate": 1.836272661103167e-05,
      "loss": 10.1692,
      "step": 1080
    },
    {
      "epoch": 0.21121531848378272,
      "grad_norm": 14.411211967468262,
      "learning_rate": 1.8359254794186368e-05,
      "loss": 10.7699,
      "step": 1081
    },
    {
      "epoch": 0.211410707307542,
      "grad_norm": 9.111126899719238,
      "learning_rate": 1.83557796292e-05,
      "loss": 8.5393,
      "step": 1082
    },
    {
      "epoch": 0.2116060961313013,
      "grad_norm": 43.198951721191406,
      "learning_rate": 1.8352301117464476e-05,
      "loss": 10.8663,
      "step": 1083
    },
    {
      "epoch": 0.21180148495506057,
      "grad_norm": 12.209263801574707,
      "learning_rate": 1.834881926037304e-05,
      "loss": 9.4677,
      "step": 1084
    },
    {
      "epoch": 0.21199687377881984,
      "grad_norm": 6.728679656982422,
      "learning_rate": 1.8345334059320294e-05,
      "loss": 9.0522,
      "step": 1085
    },
    {
      "epoch": 0.21219226260257915,
      "grad_norm": 9.289886474609375,
      "learning_rate": 1.8341845515702164e-05,
      "loss": 10.0314,
      "step": 1086
    },
    {
      "epoch": 0.21238765142633842,
      "grad_norm": 11.61264419555664,
      "learning_rate": 1.8338353630915918e-05,
      "loss": 11.3864,
      "step": 1087
    },
    {
      "epoch": 0.2125830402500977,
      "grad_norm": 10.6533842086792,
      "learning_rate": 1.8334858406360168e-05,
      "loss": 10.0839,
      "step": 1088
    },
    {
      "epoch": 0.21277842907385697,
      "grad_norm": 15.192487716674805,
      "learning_rate": 1.833135984343486e-05,
      "loss": 10.1023,
      "step": 1089
    },
    {
      "epoch": 0.21297381789761627,
      "grad_norm": 23.368717193603516,
      "learning_rate": 1.8327857943541273e-05,
      "loss": 10.8994,
      "step": 1090
    },
    {
      "epoch": 0.21316920672137554,
      "grad_norm": 12.789835929870605,
      "learning_rate": 1.8324352708082033e-05,
      "loss": 10.3748,
      "step": 1091
    },
    {
      "epoch": 0.21336459554513482,
      "grad_norm": 52.898136138916016,
      "learning_rate": 1.832084413846109e-05,
      "loss": 10.3177,
      "step": 1092
    },
    {
      "epoch": 0.2135599843688941,
      "grad_norm": 11.188353538513184,
      "learning_rate": 1.8317332236083738e-05,
      "loss": 10.2827,
      "step": 1093
    },
    {
      "epoch": 0.2137553731926534,
      "grad_norm": 10.416289329528809,
      "learning_rate": 1.8313817002356602e-05,
      "loss": 10.7478,
      "step": 1094
    },
    {
      "epoch": 0.21395076201641267,
      "grad_norm": 10.080784797668457,
      "learning_rate": 1.8310298438687645e-05,
      "loss": 10.2522,
      "step": 1095
    },
    {
      "epoch": 0.21414615084017194,
      "grad_norm": 9.515165328979492,
      "learning_rate": 1.8306776546486155e-05,
      "loss": 9.1797,
      "step": 1096
    },
    {
      "epoch": 0.2143415396639312,
      "grad_norm": 8.99358081817627,
      "learning_rate": 1.8303251327162763e-05,
      "loss": 9.541,
      "step": 1097
    },
    {
      "epoch": 0.21453692848769051,
      "grad_norm": 8.59024429321289,
      "learning_rate": 1.8299722782129428e-05,
      "loss": 9.7474,
      "step": 1098
    },
    {
      "epoch": 0.2147323173114498,
      "grad_norm": 11.45220947265625,
      "learning_rate": 1.8296190912799444e-05,
      "loss": 9.8606,
      "step": 1099
    },
    {
      "epoch": 0.21492770613520906,
      "grad_norm": 8.631582260131836,
      "learning_rate": 1.829265572058743e-05,
      "loss": 10.3564,
      "step": 1100
    },
    {
      "epoch": 0.21512309495896834,
      "grad_norm": 76.1949691772461,
      "learning_rate": 1.8289117206909347e-05,
      "loss": 10.4504,
      "step": 1101
    },
    {
      "epoch": 0.21531848378272764,
      "grad_norm": 20.741443634033203,
      "learning_rate": 1.8285575373182474e-05,
      "loss": 11.4318,
      "step": 1102
    },
    {
      "epoch": 0.2155138726064869,
      "grad_norm": 14.55085277557373,
      "learning_rate": 1.8282030220825424e-05,
      "loss": 9.5378,
      "step": 1103
    },
    {
      "epoch": 0.21570926143024619,
      "grad_norm": 8.860641479492188,
      "learning_rate": 1.8278481751258145e-05,
      "loss": 9.9277,
      "step": 1104
    },
    {
      "epoch": 0.21590465025400546,
      "grad_norm": 11.90178394317627,
      "learning_rate": 1.8274929965901905e-05,
      "loss": 10.8726,
      "step": 1105
    },
    {
      "epoch": 0.21610003907776476,
      "grad_norm": 10.75643539428711,
      "learning_rate": 1.827137486617931e-05,
      "loss": 10.498,
      "step": 1106
    },
    {
      "epoch": 0.21629542790152403,
      "grad_norm": 10.311312675476074,
      "learning_rate": 1.8267816453514283e-05,
      "loss": 9.9612,
      "step": 1107
    },
    {
      "epoch": 0.2164908167252833,
      "grad_norm": 8.801495552062988,
      "learning_rate": 1.8264254729332083e-05,
      "loss": 9.3258,
      "step": 1108
    },
    {
      "epoch": 0.21668620554904258,
      "grad_norm": 25.568994522094727,
      "learning_rate": 1.8260689695059285e-05,
      "loss": 10.9771,
      "step": 1109
    },
    {
      "epoch": 0.21688159437280188,
      "grad_norm": 11.811704635620117,
      "learning_rate": 1.8257121352123803e-05,
      "loss": 9.777,
      "step": 1110
    },
    {
      "epoch": 0.21707698319656116,
      "grad_norm": 10.467910766601562,
      "learning_rate": 1.8253549701954863e-05,
      "loss": 8.6843,
      "step": 1111
    },
    {
      "epoch": 0.21727237202032043,
      "grad_norm": 16.515195846557617,
      "learning_rate": 1.8249974745983023e-05,
      "loss": 10.9219,
      "step": 1112
    },
    {
      "epoch": 0.2174677608440797,
      "grad_norm": 9.21031665802002,
      "learning_rate": 1.8246396485640168e-05,
      "loss": 10.6956,
      "step": 1113
    },
    {
      "epoch": 0.217663149667839,
      "grad_norm": 11.476834297180176,
      "learning_rate": 1.82428149223595e-05,
      "loss": 10.8747,
      "step": 1114
    },
    {
      "epoch": 0.21785853849159828,
      "grad_norm": 10.673362731933594,
      "learning_rate": 1.8239230057575542e-05,
      "loss": 10.5956,
      "step": 1115
    },
    {
      "epoch": 0.21805392731535755,
      "grad_norm": 12.2440824508667,
      "learning_rate": 1.823564189272415e-05,
      "loss": 10.2206,
      "step": 1116
    },
    {
      "epoch": 0.21824931613911686,
      "grad_norm": 9.135146141052246,
      "learning_rate": 1.8232050429242495e-05,
      "loss": 10.1133,
      "step": 1117
    },
    {
      "epoch": 0.21844470496287613,
      "grad_norm": 8.728495597839355,
      "learning_rate": 1.8228455668569065e-05,
      "loss": 9.8545,
      "step": 1118
    },
    {
      "epoch": 0.2186400937866354,
      "grad_norm": 9.4779052734375,
      "learning_rate": 1.8224857612143675e-05,
      "loss": 10.7018,
      "step": 1119
    },
    {
      "epoch": 0.21883548261039468,
      "grad_norm": 18.96925163269043,
      "learning_rate": 1.822125626140746e-05,
      "loss": 10.017,
      "step": 1120
    },
    {
      "epoch": 0.21903087143415398,
      "grad_norm": 10.221548080444336,
      "learning_rate": 1.8217651617802873e-05,
      "loss": 11.3653,
      "step": 1121
    },
    {
      "epoch": 0.21922626025791325,
      "grad_norm": 13.457391738891602,
      "learning_rate": 1.8214043682773685e-05,
      "loss": 10.3343,
      "step": 1122
    },
    {
      "epoch": 0.21942164908167253,
      "grad_norm": 11.418845176696777,
      "learning_rate": 1.8210432457764982e-05,
      "loss": 9.7316,
      "step": 1123
    },
    {
      "epoch": 0.2196170379054318,
      "grad_norm": 8.411755561828613,
      "learning_rate": 1.8206817944223178e-05,
      "loss": 10.5238,
      "step": 1124
    },
    {
      "epoch": 0.2198124267291911,
      "grad_norm": 9.965346336364746,
      "learning_rate": 1.8203200143595993e-05,
      "loss": 10.655,
      "step": 1125
    },
    {
      "epoch": 0.22000781555295038,
      "grad_norm": 14.216445922851562,
      "learning_rate": 1.8199579057332472e-05,
      "loss": 9.4912,
      "step": 1126
    },
    {
      "epoch": 0.22020320437670965,
      "grad_norm": 18.472694396972656,
      "learning_rate": 1.8195954686882972e-05,
      "loss": 10.7054,
      "step": 1127
    },
    {
      "epoch": 0.22039859320046892,
      "grad_norm": 10.878667831420898,
      "learning_rate": 1.8192327033699168e-05,
      "loss": 10.9642,
      "step": 1128
    },
    {
      "epoch": 0.22059398202422822,
      "grad_norm": 9.860835075378418,
      "learning_rate": 1.8188696099234036e-05,
      "loss": 10.7126,
      "step": 1129
    },
    {
      "epoch": 0.2207893708479875,
      "grad_norm": 10.548471450805664,
      "learning_rate": 1.8185061884941893e-05,
      "loss": 10.1515,
      "step": 1130
    },
    {
      "epoch": 0.22098475967174677,
      "grad_norm": 18.488536834716797,
      "learning_rate": 1.8181424392278347e-05,
      "loss": 9.2223,
      "step": 1131
    },
    {
      "epoch": 0.22118014849550605,
      "grad_norm": 12.526021003723145,
      "learning_rate": 1.8177783622700328e-05,
      "loss": 10.6216,
      "step": 1132
    },
    {
      "epoch": 0.22137553731926535,
      "grad_norm": 9.050341606140137,
      "learning_rate": 1.8174139577666078e-05,
      "loss": 8.8958,
      "step": 1133
    },
    {
      "epoch": 0.22157092614302462,
      "grad_norm": 14.065995216369629,
      "learning_rate": 1.8170492258635147e-05,
      "loss": 10.3226,
      "step": 1134
    },
    {
      "epoch": 0.2217663149667839,
      "grad_norm": 8.310761451721191,
      "learning_rate": 1.8166841667068403e-05,
      "loss": 9.8635,
      "step": 1135
    },
    {
      "epoch": 0.22196170379054317,
      "grad_norm": 11.958816528320312,
      "learning_rate": 1.816318780442802e-05,
      "loss": 11.0582,
      "step": 1136
    },
    {
      "epoch": 0.22215709261430247,
      "grad_norm": 23.47361183166504,
      "learning_rate": 1.8159530672177486e-05,
      "loss": 10.164,
      "step": 1137
    },
    {
      "epoch": 0.22235248143806174,
      "grad_norm": 18.067411422729492,
      "learning_rate": 1.815587027178159e-05,
      "loss": 12.0752,
      "step": 1138
    },
    {
      "epoch": 0.22254787026182102,
      "grad_norm": 11.176780700683594,
      "learning_rate": 1.8152206604706445e-05,
      "loss": 10.1291,
      "step": 1139
    },
    {
      "epoch": 0.2227432590855803,
      "grad_norm": 12.644645690917969,
      "learning_rate": 1.8148539672419457e-05,
      "loss": 9.5169,
      "step": 1140
    },
    {
      "epoch": 0.2229386479093396,
      "grad_norm": 13.717794418334961,
      "learning_rate": 1.8144869476389342e-05,
      "loss": 10.9828,
      "step": 1141
    },
    {
      "epoch": 0.22313403673309887,
      "grad_norm": 8.076347351074219,
      "learning_rate": 1.8141196018086136e-05,
      "loss": 9.2707,
      "step": 1142
    },
    {
      "epoch": 0.22332942555685814,
      "grad_norm": 10.27889633178711,
      "learning_rate": 1.8137519298981172e-05,
      "loss": 11.1033,
      "step": 1143
    },
    {
      "epoch": 0.22352481438061741,
      "grad_norm": 10.209543228149414,
      "learning_rate": 1.8133839320547085e-05,
      "loss": 9.5142,
      "step": 1144
    },
    {
      "epoch": 0.22372020320437672,
      "grad_norm": 11.779301643371582,
      "learning_rate": 1.8130156084257824e-05,
      "loss": 9.8682,
      "step": 1145
    },
    {
      "epoch": 0.223915592028136,
      "grad_norm": 8.995259284973145,
      "learning_rate": 1.8126469591588643e-05,
      "loss": 9.0289,
      "step": 1146
    },
    {
      "epoch": 0.22411098085189526,
      "grad_norm": 8.813530921936035,
      "learning_rate": 1.8122779844016088e-05,
      "loss": 9.8038,
      "step": 1147
    },
    {
      "epoch": 0.22430636967565457,
      "grad_norm": 26.275314331054688,
      "learning_rate": 1.8119086843018024e-05,
      "loss": 10.352,
      "step": 1148
    },
    {
      "epoch": 0.22450175849941384,
      "grad_norm": 8.708663940429688,
      "learning_rate": 1.8115390590073612e-05,
      "loss": 9.6875,
      "step": 1149
    },
    {
      "epoch": 0.2246971473231731,
      "grad_norm": 7.769646167755127,
      "learning_rate": 1.8111691086663316e-05,
      "loss": 9.5119,
      "step": 1150
    },
    {
      "epoch": 0.2248925361469324,
      "grad_norm": 7.051853656768799,
      "learning_rate": 1.81079883342689e-05,
      "loss": 10.1524,
      "step": 1151
    },
    {
      "epoch": 0.2250879249706917,
      "grad_norm": 11.214014053344727,
      "learning_rate": 1.8104282334373428e-05,
      "loss": 10.8635,
      "step": 1152
    },
    {
      "epoch": 0.22528331379445096,
      "grad_norm": 8.857877731323242,
      "learning_rate": 1.8100573088461275e-05,
      "loss": 8.7015,
      "step": 1153
    },
    {
      "epoch": 0.22547870261821024,
      "grad_norm": 9.525921821594238,
      "learning_rate": 1.809686059801811e-05,
      "loss": 8.728,
      "step": 1154
    },
    {
      "epoch": 0.2256740914419695,
      "grad_norm": 9.54394817352295,
      "learning_rate": 1.809314486453089e-05,
      "loss": 9.559,
      "step": 1155
    },
    {
      "epoch": 0.2258694802657288,
      "grad_norm": 13.451454162597656,
      "learning_rate": 1.8089425889487894e-05,
      "loss": 10.4519,
      "step": 1156
    },
    {
      "epoch": 0.22606486908948809,
      "grad_norm": 67.2001724243164,
      "learning_rate": 1.808570367437868e-05,
      "loss": 10.6649,
      "step": 1157
    },
    {
      "epoch": 0.22626025791324736,
      "grad_norm": 8.727229118347168,
      "learning_rate": 1.8081978220694113e-05,
      "loss": 9.2873,
      "step": 1158
    },
    {
      "epoch": 0.22645564673700663,
      "grad_norm": 9.808565139770508,
      "learning_rate": 1.8078249529926353e-05,
      "loss": 10.2171,
      "step": 1159
    },
    {
      "epoch": 0.22665103556076593,
      "grad_norm": 9.564712524414062,
      "learning_rate": 1.8074517603568855e-05,
      "loss": 9.0959,
      "step": 1160
    },
    {
      "epoch": 0.2268464243845252,
      "grad_norm": 13.291114807128906,
      "learning_rate": 1.807078244311637e-05,
      "loss": 9.4776,
      "step": 1161
    },
    {
      "epoch": 0.22704181320828448,
      "grad_norm": 22.615671157836914,
      "learning_rate": 1.806704405006495e-05,
      "loss": 9.9774,
      "step": 1162
    },
    {
      "epoch": 0.22723720203204376,
      "grad_norm": 7.402246952056885,
      "learning_rate": 1.8063302425911936e-05,
      "loss": 9.3268,
      "step": 1163
    },
    {
      "epoch": 0.22743259085580306,
      "grad_norm": 24.33492660522461,
      "learning_rate": 1.805955757215596e-05,
      "loss": 9.1528,
      "step": 1164
    },
    {
      "epoch": 0.22762797967956233,
      "grad_norm": 8.760977745056152,
      "learning_rate": 1.805580949029696e-05,
      "loss": 10.8391,
      "step": 1165
    },
    {
      "epoch": 0.2278233685033216,
      "grad_norm": 11.081182479858398,
      "learning_rate": 1.8052058181836152e-05,
      "loss": 11.2433,
      "step": 1166
    },
    {
      "epoch": 0.22801875732708088,
      "grad_norm": 12.74425983428955,
      "learning_rate": 1.804830364827606e-05,
      "loss": 9.6439,
      "step": 1167
    },
    {
      "epoch": 0.22821414615084018,
      "grad_norm": 14.619860649108887,
      "learning_rate": 1.804454589112048e-05,
      "loss": 10.2618,
      "step": 1168
    },
    {
      "epoch": 0.22840953497459945,
      "grad_norm": 10.31884765625,
      "learning_rate": 1.8040784911874517e-05,
      "loss": 10.6417,
      "step": 1169
    },
    {
      "epoch": 0.22860492379835873,
      "grad_norm": 7.929812908172607,
      "learning_rate": 1.803702071204456e-05,
      "loss": 9.5073,
      "step": 1170
    },
    {
      "epoch": 0.228800312622118,
      "grad_norm": 8.551639556884766,
      "learning_rate": 1.803325329313829e-05,
      "loss": 10.1904,
      "step": 1171
    },
    {
      "epoch": 0.2289957014458773,
      "grad_norm": 9.057095527648926,
      "learning_rate": 1.8029482656664672e-05,
      "loss": 10.4331,
      "step": 1172
    },
    {
      "epoch": 0.22919109026963658,
      "grad_norm": 9.363543510437012,
      "learning_rate": 1.8025708804133962e-05,
      "loss": 9.9363,
      "step": 1173
    },
    {
      "epoch": 0.22938647909339585,
      "grad_norm": 10.03451156616211,
      "learning_rate": 1.802193173705771e-05,
      "loss": 10.2171,
      "step": 1174
    },
    {
      "epoch": 0.22958186791715515,
      "grad_norm": 9.010741233825684,
      "learning_rate": 1.8018151456948738e-05,
      "loss": 10.2098,
      "step": 1175
    },
    {
      "epoch": 0.22977725674091443,
      "grad_norm": 8.350069999694824,
      "learning_rate": 1.8014367965321178e-05,
      "loss": 9.5197,
      "step": 1176
    },
    {
      "epoch": 0.2299726455646737,
      "grad_norm": 9.755274772644043,
      "learning_rate": 1.8010581263690428e-05,
      "loss": 10.6485,
      "step": 1177
    },
    {
      "epoch": 0.23016803438843297,
      "grad_norm": 15.555088996887207,
      "learning_rate": 1.8006791353573186e-05,
      "loss": 9.9917,
      "step": 1178
    },
    {
      "epoch": 0.23036342321219228,
      "grad_norm": 13.880017280578613,
      "learning_rate": 1.8002998236487423e-05,
      "loss": 10.0376,
      "step": 1179
    },
    {
      "epoch": 0.23055881203595155,
      "grad_norm": 11.880322456359863,
      "learning_rate": 1.79992019139524e-05,
      "loss": 11.0284,
      "step": 1180
    },
    {
      "epoch": 0.23075420085971082,
      "grad_norm": 12.753463745117188,
      "learning_rate": 1.7995402387488665e-05,
      "loss": 11.3415,
      "step": 1181
    },
    {
      "epoch": 0.2309495896834701,
      "grad_norm": 10.973676681518555,
      "learning_rate": 1.7991599658618044e-05,
      "loss": 10.3221,
      "step": 1182
    },
    {
      "epoch": 0.2311449785072294,
      "grad_norm": 8.763169288635254,
      "learning_rate": 1.798779372886365e-05,
      "loss": 9.3201,
      "step": 1183
    },
    {
      "epoch": 0.23134036733098867,
      "grad_norm": 9.328156471252441,
      "learning_rate": 1.7983984599749876e-05,
      "loss": 9.4876,
      "step": 1184
    },
    {
      "epoch": 0.23153575615474795,
      "grad_norm": 14.645464897155762,
      "learning_rate": 1.7980172272802398e-05,
      "loss": 9.8452,
      "step": 1185
    },
    {
      "epoch": 0.23173114497850722,
      "grad_norm": 9.281057357788086,
      "learning_rate": 1.797635674954817e-05,
      "loss": 10.4572,
      "step": 1186
    },
    {
      "epoch": 0.23192653380226652,
      "grad_norm": 9.923699378967285,
      "learning_rate": 1.7972538031515418e-05,
      "loss": 8.9601,
      "step": 1187
    },
    {
      "epoch": 0.2321219226260258,
      "grad_norm": 11.165024757385254,
      "learning_rate": 1.796871612023367e-05,
      "loss": 9.1202,
      "step": 1188
    },
    {
      "epoch": 0.23231731144978507,
      "grad_norm": 16.26569938659668,
      "learning_rate": 1.796489101723372e-05,
      "loss": 10.9923,
      "step": 1189
    },
    {
      "epoch": 0.23251270027354434,
      "grad_norm": 13.922295570373535,
      "learning_rate": 1.7961062724047632e-05,
      "loss": 10.104,
      "step": 1190
    },
    {
      "epoch": 0.23270808909730364,
      "grad_norm": 10.160478591918945,
      "learning_rate": 1.7957231242208763e-05,
      "loss": 9.7135,
      "step": 1191
    },
    {
      "epoch": 0.23290347792106292,
      "grad_norm": 12.048813819885254,
      "learning_rate": 1.7953396573251738e-05,
      "loss": 9.2826,
      "step": 1192
    },
    {
      "epoch": 0.2330988667448222,
      "grad_norm": 10.418081283569336,
      "learning_rate": 1.794955871871246e-05,
      "loss": 9.1642,
      "step": 1193
    },
    {
      "epoch": 0.23329425556858147,
      "grad_norm": 12.009995460510254,
      "learning_rate": 1.7945717680128107e-05,
      "loss": 10.4241,
      "step": 1194
    },
    {
      "epoch": 0.23348964439234077,
      "grad_norm": 10.573904991149902,
      "learning_rate": 1.7941873459037144e-05,
      "loss": 10.4035,
      "step": 1195
    },
    {
      "epoch": 0.23368503321610004,
      "grad_norm": 8.724102973937988,
      "learning_rate": 1.7938026056979288e-05,
      "loss": 9.8846,
      "step": 1196
    },
    {
      "epoch": 0.23388042203985931,
      "grad_norm": 11.088909149169922,
      "learning_rate": 1.7934175475495552e-05,
      "loss": 11.0432,
      "step": 1197
    },
    {
      "epoch": 0.2340758108636186,
      "grad_norm": 11.072525978088379,
      "learning_rate": 1.793032171612821e-05,
      "loss": 9.5168,
      "step": 1198
    },
    {
      "epoch": 0.2342711996873779,
      "grad_norm": 14.48965072631836,
      "learning_rate": 1.7926464780420813e-05,
      "loss": 9.6476,
      "step": 1199
    },
    {
      "epoch": 0.23446658851113716,
      "grad_norm": 25.64013671875,
      "learning_rate": 1.792260466991818e-05,
      "loss": 11.0423,
      "step": 1200
    },
    {
      "epoch": 0.23466197733489644,
      "grad_norm": 15.797342300415039,
      "learning_rate": 1.791874138616641e-05,
      "loss": 10.7754,
      "step": 1201
    },
    {
      "epoch": 0.2348573661586557,
      "grad_norm": 9.64517593383789,
      "learning_rate": 1.791487493071287e-05,
      "loss": 9.4815,
      "step": 1202
    },
    {
      "epoch": 0.235052754982415,
      "grad_norm": 9.780092239379883,
      "learning_rate": 1.791100530510619e-05,
      "loss": 9.5045,
      "step": 1203
    },
    {
      "epoch": 0.2352481438061743,
      "grad_norm": 10.812311172485352,
      "learning_rate": 1.7907132510896275e-05,
      "loss": 9.8301,
      "step": 1204
    },
    {
      "epoch": 0.23544353262993356,
      "grad_norm": 22.109315872192383,
      "learning_rate": 1.7903256549634304e-05,
      "loss": 9.8362,
      "step": 1205
    },
    {
      "epoch": 0.23563892145369286,
      "grad_norm": 13.416193008422852,
      "learning_rate": 1.7899377422872716e-05,
      "loss": 9.0167,
      "step": 1206
    },
    {
      "epoch": 0.23583431027745214,
      "grad_norm": 10.781249046325684,
      "learning_rate": 1.789549513216522e-05,
      "loss": 9.0485,
      "step": 1207
    },
    {
      "epoch": 0.2360296991012114,
      "grad_norm": 16.785064697265625,
      "learning_rate": 1.78916096790668e-05,
      "loss": 10.2705,
      "step": 1208
    },
    {
      "epoch": 0.23622508792497068,
      "grad_norm": 16.255413055419922,
      "learning_rate": 1.7887721065133695e-05,
      "loss": 10.4822,
      "step": 1209
    },
    {
      "epoch": 0.23642047674872999,
      "grad_norm": 13.382121086120605,
      "learning_rate": 1.7883829291923423e-05,
      "loss": 10.0025,
      "step": 1210
    },
    {
      "epoch": 0.23661586557248926,
      "grad_norm": 10.204498291015625,
      "learning_rate": 1.787993436099475e-05,
      "loss": 9.5152,
      "step": 1211
    },
    {
      "epoch": 0.23681125439624853,
      "grad_norm": 7.8883562088012695,
      "learning_rate": 1.7876036273907726e-05,
      "loss": 10.241,
      "step": 1212
    },
    {
      "epoch": 0.2370066432200078,
      "grad_norm": 22.64422607421875,
      "learning_rate": 1.7872135032223648e-05,
      "loss": 11.2415,
      "step": 1213
    },
    {
      "epoch": 0.2372020320437671,
      "grad_norm": 9.7103853225708,
      "learning_rate": 1.786823063750509e-05,
      "loss": 10.1178,
      "step": 1214
    },
    {
      "epoch": 0.23739742086752638,
      "grad_norm": 11.876055717468262,
      "learning_rate": 1.786432309131588e-05,
      "loss": 10.7425,
      "step": 1215
    },
    {
      "epoch": 0.23759280969128566,
      "grad_norm": 21.693655014038086,
      "learning_rate": 1.7860412395221113e-05,
      "loss": 9.4291,
      "step": 1216
    },
    {
      "epoch": 0.23778819851504493,
      "grad_norm": 15.459426879882812,
      "learning_rate": 1.7856498550787144e-05,
      "loss": 10.1343,
      "step": 1217
    },
    {
      "epoch": 0.23798358733880423,
      "grad_norm": 8.189459800720215,
      "learning_rate": 1.7852581559581594e-05,
      "loss": 9.7895,
      "step": 1218
    },
    {
      "epoch": 0.2381789761625635,
      "grad_norm": 10.463160514831543,
      "learning_rate": 1.7848661423173328e-05,
      "loss": 10.3121,
      "step": 1219
    },
    {
      "epoch": 0.23837436498632278,
      "grad_norm": 13.797211647033691,
      "learning_rate": 1.7844738143132494e-05,
      "loss": 10.7717,
      "step": 1220
    },
    {
      "epoch": 0.23856975381008205,
      "grad_norm": 9.269835472106934,
      "learning_rate": 1.784081172103048e-05,
      "loss": 9.1177,
      "step": 1221
    },
    {
      "epoch": 0.23876514263384135,
      "grad_norm": 9.249794006347656,
      "learning_rate": 1.7836882158439945e-05,
      "loss": 9.1639,
      "step": 1222
    },
    {
      "epoch": 0.23896053145760063,
      "grad_norm": 8.747522354125977,
      "learning_rate": 1.7832949456934797e-05,
      "loss": 9.1398,
      "step": 1223
    },
    {
      "epoch": 0.2391559202813599,
      "grad_norm": 9.711735725402832,
      "learning_rate": 1.7829013618090208e-05,
      "loss": 9.5402,
      "step": 1224
    },
    {
      "epoch": 0.23935130910511918,
      "grad_norm": 12.270694732666016,
      "learning_rate": 1.78250746434826e-05,
      "loss": 9.57,
      "step": 1225
    },
    {
      "epoch": 0.23954669792887848,
      "grad_norm": 10.28706169128418,
      "learning_rate": 1.782113253468966e-05,
      "loss": 9.8925,
      "step": 1226
    },
    {
      "epoch": 0.23974208675263775,
      "grad_norm": 11.051925659179688,
      "learning_rate": 1.781718729329032e-05,
      "loss": 10.4482,
      "step": 1227
    },
    {
      "epoch": 0.23993747557639702,
      "grad_norm": 10.1842679977417,
      "learning_rate": 1.7813238920864772e-05,
      "loss": 10.0631,
      "step": 1228
    },
    {
      "epoch": 0.2401328644001563,
      "grad_norm": 12.414724349975586,
      "learning_rate": 1.7809287418994463e-05,
      "loss": 9.2048,
      "step": 1229
    },
    {
      "epoch": 0.2403282532239156,
      "grad_norm": 13.164472579956055,
      "learning_rate": 1.7805332789262096e-05,
      "loss": 9.9888,
      "step": 1230
    },
    {
      "epoch": 0.24052364204767487,
      "grad_norm": 10.315644264221191,
      "learning_rate": 1.7801375033251617e-05,
      "loss": 9.5832,
      "step": 1231
    },
    {
      "epoch": 0.24071903087143415,
      "grad_norm": 10.495373725891113,
      "learning_rate": 1.7797414152548233e-05,
      "loss": 9.2239,
      "step": 1232
    },
    {
      "epoch": 0.24091441969519342,
      "grad_norm": 9.411508560180664,
      "learning_rate": 1.7793450148738398e-05,
      "loss": 9.907,
      "step": 1233
    },
    {
      "epoch": 0.24110980851895272,
      "grad_norm": 11.17962646484375,
      "learning_rate": 1.7789483023409822e-05,
      "loss": 9.591,
      "step": 1234
    },
    {
      "epoch": 0.241305197342712,
      "grad_norm": 8.910758972167969,
      "learning_rate": 1.778551277815146e-05,
      "loss": 8.7132,
      "step": 1235
    },
    {
      "epoch": 0.24150058616647127,
      "grad_norm": 19.328405380249023,
      "learning_rate": 1.7781539414553516e-05,
      "loss": 9.9229,
      "step": 1236
    },
    {
      "epoch": 0.24169597499023057,
      "grad_norm": 10.288520812988281,
      "learning_rate": 1.7777562934207447e-05,
      "loss": 9.5067,
      "step": 1237
    },
    {
      "epoch": 0.24189136381398985,
      "grad_norm": 14.588144302368164,
      "learning_rate": 1.777358333870596e-05,
      "loss": 9.5383,
      "step": 1238
    },
    {
      "epoch": 0.24208675263774912,
      "grad_norm": 13.42604923248291,
      "learning_rate": 1.7769600629643005e-05,
      "loss": 10.9858,
      "step": 1239
    },
    {
      "epoch": 0.2422821414615084,
      "grad_norm": 10.372969627380371,
      "learning_rate": 1.7765614808613775e-05,
      "loss": 9.9459,
      "step": 1240
    },
    {
      "epoch": 0.2424775302852677,
      "grad_norm": 11.277698516845703,
      "learning_rate": 1.7761625877214725e-05,
      "loss": 9.0366,
      "step": 1241
    },
    {
      "epoch": 0.24267291910902697,
      "grad_norm": 10.354175567626953,
      "learning_rate": 1.7757633837043536e-05,
      "loss": 10.0163,
      "step": 1242
    },
    {
      "epoch": 0.24286830793278624,
      "grad_norm": 11.91468620300293,
      "learning_rate": 1.7753638689699152e-05,
      "loss": 9.5083,
      "step": 1243
    },
    {
      "epoch": 0.24306369675654552,
      "grad_norm": 10.422640800476074,
      "learning_rate": 1.7749640436781748e-05,
      "loss": 9.8963,
      "step": 1244
    },
    {
      "epoch": 0.24325908558030482,
      "grad_norm": 10.919748306274414,
      "learning_rate": 1.7745639079892752e-05,
      "loss": 10.2643,
      "step": 1245
    },
    {
      "epoch": 0.2434544744040641,
      "grad_norm": 11.70514965057373,
      "learning_rate": 1.774163462063483e-05,
      "loss": 10.4714,
      "step": 1246
    },
    {
      "epoch": 0.24364986322782337,
      "grad_norm": 11.730195045471191,
      "learning_rate": 1.773762706061189e-05,
      "loss": 10.9369,
      "step": 1247
    },
    {
      "epoch": 0.24384525205158264,
      "grad_norm": 10.884676933288574,
      "learning_rate": 1.773361640142909e-05,
      "loss": 9.495,
      "step": 1248
    },
    {
      "epoch": 0.24404064087534194,
      "grad_norm": 11.580656051635742,
      "learning_rate": 1.7729602644692818e-05,
      "loss": 11.6008,
      "step": 1249
    },
    {
      "epoch": 0.24423602969910121,
      "grad_norm": 10.542391777038574,
      "learning_rate": 1.7725585792010714e-05,
      "loss": 10.0219,
      "step": 1250
    },
    {
      "epoch": 0.2444314185228605,
      "grad_norm": 12.857534408569336,
      "learning_rate": 1.7721565844991643e-05,
      "loss": 10.5635,
      "step": 1251
    },
    {
      "epoch": 0.24462680734661976,
      "grad_norm": 10.392827987670898,
      "learning_rate": 1.7717542805245728e-05,
      "loss": 9.2779,
      "step": 1252
    },
    {
      "epoch": 0.24482219617037906,
      "grad_norm": 9.28751277923584,
      "learning_rate": 1.771351667438432e-05,
      "loss": 10.2842,
      "step": 1253
    },
    {
      "epoch": 0.24501758499413834,
      "grad_norm": 14.696070671081543,
      "learning_rate": 1.770948745402e-05,
      "loss": 10.1721,
      "step": 1254
    },
    {
      "epoch": 0.2452129738178976,
      "grad_norm": 10.734374046325684,
      "learning_rate": 1.7705455145766607e-05,
      "loss": 9.8766,
      "step": 1255
    },
    {
      "epoch": 0.24540836264165689,
      "grad_norm": 13.675962448120117,
      "learning_rate": 1.77014197512392e-05,
      "loss": 9.8617,
      "step": 1256
    },
    {
      "epoch": 0.2456037514654162,
      "grad_norm": 8.64175796508789,
      "learning_rate": 1.7697381272054083e-05,
      "loss": 10.7436,
      "step": 1257
    },
    {
      "epoch": 0.24579914028917546,
      "grad_norm": 15.539191246032715,
      "learning_rate": 1.769333970982879e-05,
      "loss": 8.8873,
      "step": 1258
    },
    {
      "epoch": 0.24599452911293473,
      "grad_norm": 13.888678550720215,
      "learning_rate": 1.7689295066182092e-05,
      "loss": 10.2032,
      "step": 1259
    },
    {
      "epoch": 0.246189917936694,
      "grad_norm": 9.026545524597168,
      "learning_rate": 1.7685247342733996e-05,
      "loss": 10.028,
      "step": 1260
    },
    {
      "epoch": 0.2463853067604533,
      "grad_norm": 7.562590599060059,
      "learning_rate": 1.768119654110574e-05,
      "loss": 8.2881,
      "step": 1261
    },
    {
      "epoch": 0.24658069558421258,
      "grad_norm": 8.620518684387207,
      "learning_rate": 1.7677142662919794e-05,
      "loss": 9.7138,
      "step": 1262
    },
    {
      "epoch": 0.24677608440797186,
      "grad_norm": 9.384614944458008,
      "learning_rate": 1.7673085709799865e-05,
      "loss": 9.9614,
      "step": 1263
    },
    {
      "epoch": 0.24697147323173113,
      "grad_norm": 9.137187004089355,
      "learning_rate": 1.766902568337089e-05,
      "loss": 10.3298,
      "step": 1264
    },
    {
      "epoch": 0.24716686205549043,
      "grad_norm": 13.199517250061035,
      "learning_rate": 1.7664962585259028e-05,
      "loss": 9.4269,
      "step": 1265
    },
    {
      "epoch": 0.2473622508792497,
      "grad_norm": 7.940667629241943,
      "learning_rate": 1.7660896417091686e-05,
      "loss": 9.6282,
      "step": 1266
    },
    {
      "epoch": 0.24755763970300898,
      "grad_norm": 11.642074584960938,
      "learning_rate": 1.7656827180497484e-05,
      "loss": 10.6028,
      "step": 1267
    },
    {
      "epoch": 0.24775302852676828,
      "grad_norm": 9.434550285339355,
      "learning_rate": 1.7652754877106275e-05,
      "loss": 9.8025,
      "step": 1268
    },
    {
      "epoch": 0.24794841735052756,
      "grad_norm": 11.158660888671875,
      "learning_rate": 1.764867950854915e-05,
      "loss": 10.3788,
      "step": 1269
    },
    {
      "epoch": 0.24814380617428683,
      "grad_norm": 10.036881446838379,
      "learning_rate": 1.764460107645842e-05,
      "loss": 9.859,
      "step": 1270
    },
    {
      "epoch": 0.2483391949980461,
      "grad_norm": 8.490141868591309,
      "learning_rate": 1.764051958246762e-05,
      "loss": 9.2948,
      "step": 1271
    },
    {
      "epoch": 0.2485345838218054,
      "grad_norm": 10.146963119506836,
      "learning_rate": 1.763643502821152e-05,
      "loss": 9.7118,
      "step": 1272
    },
    {
      "epoch": 0.24872997264556468,
      "grad_norm": 7.648308277130127,
      "learning_rate": 1.7632347415326105e-05,
      "loss": 8.9883,
      "step": 1273
    },
    {
      "epoch": 0.24892536146932395,
      "grad_norm": 13.093494415283203,
      "learning_rate": 1.762825674544859e-05,
      "loss": 10.2064,
      "step": 1274
    },
    {
      "epoch": 0.24912075029308323,
      "grad_norm": 8.967086791992188,
      "learning_rate": 1.7624163020217418e-05,
      "loss": 9.5987,
      "step": 1275
    },
    {
      "epoch": 0.24931613911684253,
      "grad_norm": 8.7538480758667,
      "learning_rate": 1.7620066241272257e-05,
      "loss": 9.1515,
      "step": 1276
    },
    {
      "epoch": 0.2495115279406018,
      "grad_norm": 8.5569429397583,
      "learning_rate": 1.761596641025399e-05,
      "loss": 10.0197,
      "step": 1277
    },
    {
      "epoch": 0.24970691676436108,
      "grad_norm": 10.225385665893555,
      "learning_rate": 1.7611863528804728e-05,
      "loss": 9.0034,
      "step": 1278
    },
    {
      "epoch": 0.24990230558812035,
      "grad_norm": 9.1556978225708,
      "learning_rate": 1.76077575985678e-05,
      "loss": 8.9998,
      "step": 1279
    },
    {
      "epoch": 0.2500976944118796,
      "grad_norm": 8.018303871154785,
      "learning_rate": 1.760364862118776e-05,
      "loss": 9.8576,
      "step": 1280
    },
    {
      "epoch": 0.2502930832356389,
      "grad_norm": 13.046024322509766,
      "learning_rate": 1.7599536598310384e-05,
      "loss": 10.976,
      "step": 1281
    },
    {
      "epoch": 0.2504884720593982,
      "grad_norm": 11.197158813476562,
      "learning_rate": 1.759542153158266e-05,
      "loss": 10.0729,
      "step": 1282
    },
    {
      "epoch": 0.25068386088315747,
      "grad_norm": 8.969386100769043,
      "learning_rate": 1.7591303422652802e-05,
      "loss": 9.7935,
      "step": 1283
    },
    {
      "epoch": 0.2508792497069168,
      "grad_norm": 8.279151916503906,
      "learning_rate": 1.7587182273170244e-05,
      "loss": 9.8509,
      "step": 1284
    },
    {
      "epoch": 0.251074638530676,
      "grad_norm": 8.590933799743652,
      "learning_rate": 1.7583058084785626e-05,
      "loss": 8.6509,
      "step": 1285
    },
    {
      "epoch": 0.2512700273544353,
      "grad_norm": 8.488905906677246,
      "learning_rate": 1.757893085915082e-05,
      "loss": 8.9383,
      "step": 1286
    },
    {
      "epoch": 0.2514654161781946,
      "grad_norm": 19.685009002685547,
      "learning_rate": 1.7574800597918907e-05,
      "loss": 9.711,
      "step": 1287
    },
    {
      "epoch": 0.25166080500195387,
      "grad_norm": 17.11168098449707,
      "learning_rate": 1.7570667302744186e-05,
      "loss": 10.0746,
      "step": 1288
    },
    {
      "epoch": 0.25185619382571317,
      "grad_norm": 9.783052444458008,
      "learning_rate": 1.7566530975282166e-05,
      "loss": 10.2021,
      "step": 1289
    },
    {
      "epoch": 0.25205158264947247,
      "grad_norm": 7.9754767417907715,
      "learning_rate": 1.7562391617189576e-05,
      "loss": 9.676,
      "step": 1290
    },
    {
      "epoch": 0.2522469714732317,
      "grad_norm": 8.846024513244629,
      "learning_rate": 1.755824923012436e-05,
      "loss": 9.3899,
      "step": 1291
    },
    {
      "epoch": 0.252442360296991,
      "grad_norm": 13.300436019897461,
      "learning_rate": 1.755410381574567e-05,
      "loss": 10.2287,
      "step": 1292
    },
    {
      "epoch": 0.25263774912075027,
      "grad_norm": 10.40530776977539,
      "learning_rate": 1.7549955375713874e-05,
      "loss": 9.9376,
      "step": 1293
    },
    {
      "epoch": 0.25283313794450957,
      "grad_norm": 8.713654518127441,
      "learning_rate": 1.7545803911690552e-05,
      "loss": 8.6172,
      "step": 1294
    },
    {
      "epoch": 0.25302852676826887,
      "grad_norm": 8.503681182861328,
      "learning_rate": 1.754164942533849e-05,
      "loss": 10.3257,
      "step": 1295
    },
    {
      "epoch": 0.2532239155920281,
      "grad_norm": 9.206268310546875,
      "learning_rate": 1.7537491918321695e-05,
      "loss": 9.8505,
      "step": 1296
    },
    {
      "epoch": 0.2534193044157874,
      "grad_norm": 10.204298973083496,
      "learning_rate": 1.7533331392305375e-05,
      "loss": 10.1417,
      "step": 1297
    },
    {
      "epoch": 0.2536146932395467,
      "grad_norm": 43.227294921875,
      "learning_rate": 1.7529167848955947e-05,
      "loss": 11.0423,
      "step": 1298
    },
    {
      "epoch": 0.25381008206330596,
      "grad_norm": 8.194405555725098,
      "learning_rate": 1.7525001289941043e-05,
      "loss": 9.587,
      "step": 1299
    },
    {
      "epoch": 0.25400547088706527,
      "grad_norm": 8.818645477294922,
      "learning_rate": 1.75208317169295e-05,
      "loss": 9.5864,
      "step": 1300
    },
    {
      "epoch": 0.25420085971082457,
      "grad_norm": 13.115413665771484,
      "learning_rate": 1.7516659131591358e-05,
      "loss": 9.8418,
      "step": 1301
    },
    {
      "epoch": 0.2543962485345838,
      "grad_norm": 11.651979446411133,
      "learning_rate": 1.7512483535597868e-05,
      "loss": 10.203,
      "step": 1302
    },
    {
      "epoch": 0.2545916373583431,
      "grad_norm": 7.130728244781494,
      "learning_rate": 1.7508304930621486e-05,
      "loss": 8.2777,
      "step": 1303
    },
    {
      "epoch": 0.25478702618210236,
      "grad_norm": 9.03930377960205,
      "learning_rate": 1.7504123318335876e-05,
      "loss": 10.2767,
      "step": 1304
    },
    {
      "epoch": 0.25498241500586166,
      "grad_norm": 10.161637306213379,
      "learning_rate": 1.74999387004159e-05,
      "loss": 10.3646,
      "step": 1305
    },
    {
      "epoch": 0.25517780382962096,
      "grad_norm": 9.122706413269043,
      "learning_rate": 1.749575107853763e-05,
      "loss": 10.1747,
      "step": 1306
    },
    {
      "epoch": 0.2553731926533802,
      "grad_norm": 7.738844394683838,
      "learning_rate": 1.7491560454378333e-05,
      "loss": 9.664,
      "step": 1307
    },
    {
      "epoch": 0.2555685814771395,
      "grad_norm": 10.311731338500977,
      "learning_rate": 1.748736682961649e-05,
      "loss": 10.4507,
      "step": 1308
    },
    {
      "epoch": 0.2557639703008988,
      "grad_norm": 10.760847091674805,
      "learning_rate": 1.7483170205931775e-05,
      "loss": 9.6038,
      "step": 1309
    },
    {
      "epoch": 0.25595935912465806,
      "grad_norm": 15.414864540100098,
      "learning_rate": 1.7478970585005068e-05,
      "loss": 10.4599,
      "step": 1310
    },
    {
      "epoch": 0.25615474794841736,
      "grad_norm": 10.18409538269043,
      "learning_rate": 1.747476796851845e-05,
      "loss": 9.9979,
      "step": 1311
    },
    {
      "epoch": 0.2563501367721766,
      "grad_norm": 8.863759994506836,
      "learning_rate": 1.747056235815519e-05,
      "loss": 9.6182,
      "step": 1312
    },
    {
      "epoch": 0.2565455255959359,
      "grad_norm": 41.86937713623047,
      "learning_rate": 1.7466353755599773e-05,
      "loss": 9.6356,
      "step": 1313
    },
    {
      "epoch": 0.2567409144196952,
      "grad_norm": 8.577277183532715,
      "learning_rate": 1.7462142162537876e-05,
      "loss": 9.5527,
      "step": 1314
    },
    {
      "epoch": 0.25693630324345446,
      "grad_norm": 10.327160835266113,
      "learning_rate": 1.7457927580656367e-05,
      "loss": 9.6522,
      "step": 1315
    },
    {
      "epoch": 0.25713169206721376,
      "grad_norm": 18.54293441772461,
      "learning_rate": 1.7453710011643318e-05,
      "loss": 9.6847,
      "step": 1316
    },
    {
      "epoch": 0.25732708089097306,
      "grad_norm": 8.194198608398438,
      "learning_rate": 1.7449489457188005e-05,
      "loss": 9.6983,
      "step": 1317
    },
    {
      "epoch": 0.2575224697147323,
      "grad_norm": 10.148292541503906,
      "learning_rate": 1.744526591898088e-05,
      "loss": 9.8361,
      "step": 1318
    },
    {
      "epoch": 0.2577178585384916,
      "grad_norm": 9.461729049682617,
      "learning_rate": 1.744103939871361e-05,
      "loss": 9.5904,
      "step": 1319
    },
    {
      "epoch": 0.25791324736225085,
      "grad_norm": 8.193283081054688,
      "learning_rate": 1.743680989807904e-05,
      "loss": 9.0122,
      "step": 1320
    },
    {
      "epoch": 0.25810863618601015,
      "grad_norm": 11.62541675567627,
      "learning_rate": 1.743257741877122e-05,
      "loss": 10.3077,
      "step": 1321
    },
    {
      "epoch": 0.25830402500976946,
      "grad_norm": 10.485865592956543,
      "learning_rate": 1.7428341962485387e-05,
      "loss": 9.9068,
      "step": 1322
    },
    {
      "epoch": 0.2584994138335287,
      "grad_norm": 27.449573516845703,
      "learning_rate": 1.7424103530917978e-05,
      "loss": 9.5062,
      "step": 1323
    },
    {
      "epoch": 0.258694802657288,
      "grad_norm": 10.768765449523926,
      "learning_rate": 1.741986212576661e-05,
      "loss": 9.3951,
      "step": 1324
    },
    {
      "epoch": 0.2588901914810473,
      "grad_norm": 8.962034225463867,
      "learning_rate": 1.74156177487301e-05,
      "loss": 9.4387,
      "step": 1325
    },
    {
      "epoch": 0.25908558030480655,
      "grad_norm": 8.598007202148438,
      "learning_rate": 1.741137040150846e-05,
      "loss": 9.9422,
      "step": 1326
    },
    {
      "epoch": 0.25928096912856585,
      "grad_norm": 11.843433380126953,
      "learning_rate": 1.740712008580287e-05,
      "loss": 10.5862,
      "step": 1327
    },
    {
      "epoch": 0.25947635795232515,
      "grad_norm": 8.559103012084961,
      "learning_rate": 1.7402866803315726e-05,
      "loss": 9.9117,
      "step": 1328
    },
    {
      "epoch": 0.2596717467760844,
      "grad_norm": 7.808802127838135,
      "learning_rate": 1.7398610555750596e-05,
      "loss": 8.8458,
      "step": 1329
    },
    {
      "epoch": 0.2598671355998437,
      "grad_norm": 7.943463325500488,
      "learning_rate": 1.7394351344812236e-05,
      "loss": 9.1846,
      "step": 1330
    },
    {
      "epoch": 0.26006252442360295,
      "grad_norm": 8.736802101135254,
      "learning_rate": 1.7390089172206594e-05,
      "loss": 9.6133,
      "step": 1331
    },
    {
      "epoch": 0.26025791324736225,
      "grad_norm": 10.777621269226074,
      "learning_rate": 1.73858240396408e-05,
      "loss": 9.2166,
      "step": 1332
    },
    {
      "epoch": 0.26045330207112155,
      "grad_norm": 12.80964469909668,
      "learning_rate": 1.738155594882318e-05,
      "loss": 10.6826,
      "step": 1333
    },
    {
      "epoch": 0.2606486908948808,
      "grad_norm": 11.080780982971191,
      "learning_rate": 1.737728490146323e-05,
      "loss": 10.1382,
      "step": 1334
    },
    {
      "epoch": 0.2608440797186401,
      "grad_norm": 9.645936012268066,
      "learning_rate": 1.7373010899271636e-05,
      "loss": 10.6446,
      "step": 1335
    },
    {
      "epoch": 0.2610394685423994,
      "grad_norm": 13.024704933166504,
      "learning_rate": 1.7368733943960278e-05,
      "loss": 10.8037,
      "step": 1336
    },
    {
      "epoch": 0.26123485736615865,
      "grad_norm": 9.12859058380127,
      "learning_rate": 1.7364454037242196e-05,
      "loss": 9.665,
      "step": 1337
    },
    {
      "epoch": 0.26143024618991795,
      "grad_norm": 12.750452041625977,
      "learning_rate": 1.736017118083163e-05,
      "loss": 10.8155,
      "step": 1338
    },
    {
      "epoch": 0.2616256350136772,
      "grad_norm": 22.314653396606445,
      "learning_rate": 1.7355885376444007e-05,
      "loss": 10.0157,
      "step": 1339
    },
    {
      "epoch": 0.2618210238374365,
      "grad_norm": 9.28205680847168,
      "learning_rate": 1.7351596625795914e-05,
      "loss": 10.2323,
      "step": 1340
    },
    {
      "epoch": 0.2620164126611958,
      "grad_norm": 9.391312599182129,
      "learning_rate": 1.7347304930605135e-05,
      "loss": 8.8491,
      "step": 1341
    },
    {
      "epoch": 0.26221180148495504,
      "grad_norm": 20.81572723388672,
      "learning_rate": 1.734301029259062e-05,
      "loss": 10.0474,
      "step": 1342
    },
    {
      "epoch": 0.26240719030871434,
      "grad_norm": 8.251876831054688,
      "learning_rate": 1.7338712713472508e-05,
      "loss": 10.0025,
      "step": 1343
    },
    {
      "epoch": 0.26260257913247365,
      "grad_norm": 9.096527099609375,
      "learning_rate": 1.733441219497211e-05,
      "loss": 10.1042,
      "step": 1344
    },
    {
      "epoch": 0.2627979679562329,
      "grad_norm": 9.74231243133545,
      "learning_rate": 1.733010873881193e-05,
      "loss": 9.763,
      "step": 1345
    },
    {
      "epoch": 0.2629933567799922,
      "grad_norm": 7.231594562530518,
      "learning_rate": 1.7325802346715622e-05,
      "loss": 8.3591,
      "step": 1346
    },
    {
      "epoch": 0.26318874560375144,
      "grad_norm": 7.799582481384277,
      "learning_rate": 1.7321493020408033e-05,
      "loss": 9.7482,
      "step": 1347
    },
    {
      "epoch": 0.26338413442751074,
      "grad_norm": 14.953970909118652,
      "learning_rate": 1.7317180761615183e-05,
      "loss": 9.7535,
      "step": 1348
    },
    {
      "epoch": 0.26357952325127004,
      "grad_norm": 10.339173316955566,
      "learning_rate": 1.731286557206427e-05,
      "loss": 9.9535,
      "step": 1349
    },
    {
      "epoch": 0.2637749120750293,
      "grad_norm": 16.204267501831055,
      "learning_rate": 1.7308547453483653e-05,
      "loss": 9.771,
      "step": 1350
    },
    {
      "epoch": 0.2639703008987886,
      "grad_norm": 8.685306549072266,
      "learning_rate": 1.730422640760288e-05,
      "loss": 9.1089,
      "step": 1351
    },
    {
      "epoch": 0.2641656897225479,
      "grad_norm": 11.356358528137207,
      "learning_rate": 1.729990243615266e-05,
      "loss": 9.6862,
      "step": 1352
    },
    {
      "epoch": 0.26436107854630714,
      "grad_norm": 8.473872184753418,
      "learning_rate": 1.7295575540864878e-05,
      "loss": 9.07,
      "step": 1353
    },
    {
      "epoch": 0.26455646737006644,
      "grad_norm": 16.833782196044922,
      "learning_rate": 1.729124572347259e-05,
      "loss": 10.145,
      "step": 1354
    },
    {
      "epoch": 0.26475185619382574,
      "grad_norm": 25.658811569213867,
      "learning_rate": 1.7286912985710023e-05,
      "loss": 9.8756,
      "step": 1355
    },
    {
      "epoch": 0.264947245017585,
      "grad_norm": 8.838557243347168,
      "learning_rate": 1.7282577329312573e-05,
      "loss": 9.8892,
      "step": 1356
    },
    {
      "epoch": 0.2651426338413443,
      "grad_norm": 9.928974151611328,
      "learning_rate": 1.7278238756016805e-05,
      "loss": 9.5268,
      "step": 1357
    },
    {
      "epoch": 0.26533802266510353,
      "grad_norm": 8.144798278808594,
      "learning_rate": 1.7273897267560448e-05,
      "loss": 9.2465,
      "step": 1358
    },
    {
      "epoch": 0.26553341148886284,
      "grad_norm": 10.5440092086792,
      "learning_rate": 1.7269552865682414e-05,
      "loss": 9.5009,
      "step": 1359
    },
    {
      "epoch": 0.26572880031262214,
      "grad_norm": 9.936861991882324,
      "learning_rate": 1.726520555212276e-05,
      "loss": 10.153,
      "step": 1360
    },
    {
      "epoch": 0.2659241891363814,
      "grad_norm": 9.90738296508789,
      "learning_rate": 1.7260855328622725e-05,
      "loss": 10.5246,
      "step": 1361
    },
    {
      "epoch": 0.2661195779601407,
      "grad_norm": 9.371439933776855,
      "learning_rate": 1.7256502196924707e-05,
      "loss": 9.0179,
      "step": 1362
    },
    {
      "epoch": 0.2663149667839,
      "grad_norm": 9.719748497009277,
      "learning_rate": 1.725214615877227e-05,
      "loss": 10.1903,
      "step": 1363
    },
    {
      "epoch": 0.26651035560765923,
      "grad_norm": 11.304025650024414,
      "learning_rate": 1.7247787215910144e-05,
      "loss": 10.3357,
      "step": 1364
    },
    {
      "epoch": 0.26670574443141853,
      "grad_norm": 9.024648666381836,
      "learning_rate": 1.724342537008422e-05,
      "loss": 9.9289,
      "step": 1365
    },
    {
      "epoch": 0.2669011332551778,
      "grad_norm": 8.572917938232422,
      "learning_rate": 1.7239060623041552e-05,
      "loss": 8.4542,
      "step": 1366
    },
    {
      "epoch": 0.2670965220789371,
      "grad_norm": 9.70483112335205,
      "learning_rate": 1.7234692976530357e-05,
      "loss": 9.5483,
      "step": 1367
    },
    {
      "epoch": 0.2672919109026964,
      "grad_norm": 9.806533813476562,
      "learning_rate": 1.723032243230002e-05,
      "loss": 9.6311,
      "step": 1368
    },
    {
      "epoch": 0.26748729972645563,
      "grad_norm": 18.847232818603516,
      "learning_rate": 1.7225948992101067e-05,
      "loss": 9.3228,
      "step": 1369
    },
    {
      "epoch": 0.26768268855021493,
      "grad_norm": 12.063652038574219,
      "learning_rate": 1.7221572657685205e-05,
      "loss": 8.5508,
      "step": 1370
    },
    {
      "epoch": 0.26787807737397423,
      "grad_norm": 12.360920906066895,
      "learning_rate": 1.7217193430805286e-05,
      "loss": 9.7002,
      "step": 1371
    },
    {
      "epoch": 0.2680734661977335,
      "grad_norm": 9.31651782989502,
      "learning_rate": 1.7212811313215337e-05,
      "loss": 8.3971,
      "step": 1372
    },
    {
      "epoch": 0.2682688550214928,
      "grad_norm": 10.297537803649902,
      "learning_rate": 1.7208426306670522e-05,
      "loss": 9.3896,
      "step": 1373
    },
    {
      "epoch": 0.268464243845252,
      "grad_norm": 9.430858612060547,
      "learning_rate": 1.7204038412927177e-05,
      "loss": 9.7238,
      "step": 1374
    },
    {
      "epoch": 0.26865963266901133,
      "grad_norm": 9.866272926330566,
      "learning_rate": 1.7199647633742783e-05,
      "loss": 9.8938,
      "step": 1375
    },
    {
      "epoch": 0.26885502149277063,
      "grad_norm": 13.442451477050781,
      "learning_rate": 1.7195253970875994e-05,
      "loss": 9.7914,
      "step": 1376
    },
    {
      "epoch": 0.2690504103165299,
      "grad_norm": 8.995190620422363,
      "learning_rate": 1.71908574260866e-05,
      "loss": 9.7674,
      "step": 1377
    },
    {
      "epoch": 0.2692457991402892,
      "grad_norm": 10.069921493530273,
      "learning_rate": 1.7186458001135557e-05,
      "loss": 10.0635,
      "step": 1378
    },
    {
      "epoch": 0.2694411879640485,
      "grad_norm": 7.587112903594971,
      "learning_rate": 1.7182055697784972e-05,
      "loss": 8.6594,
      "step": 1379
    },
    {
      "epoch": 0.2696365767878077,
      "grad_norm": 9.20854663848877,
      "learning_rate": 1.71776505177981e-05,
      "loss": 9.5423,
      "step": 1380
    },
    {
      "epoch": 0.269831965611567,
      "grad_norm": 9.881065368652344,
      "learning_rate": 1.717324246293936e-05,
      "loss": 10.2872,
      "step": 1381
    },
    {
      "epoch": 0.27002735443532627,
      "grad_norm": 9.00755500793457,
      "learning_rate": 1.7168831534974306e-05,
      "loss": 9.0091,
      "step": 1382
    },
    {
      "epoch": 0.2702227432590856,
      "grad_norm": 31.67610740661621,
      "learning_rate": 1.7164417735669657e-05,
      "loss": 11.0973,
      "step": 1383
    },
    {
      "epoch": 0.2704181320828449,
      "grad_norm": 18.58461570739746,
      "learning_rate": 1.7160001066793275e-05,
      "loss": 8.8906,
      "step": 1384
    },
    {
      "epoch": 0.2706135209066041,
      "grad_norm": 8.737338066101074,
      "learning_rate": 1.7155581530114174e-05,
      "loss": 9.6321,
      "step": 1385
    },
    {
      "epoch": 0.2708089097303634,
      "grad_norm": 8.752185821533203,
      "learning_rate": 1.7151159127402518e-05,
      "loss": 9.5535,
      "step": 1386
    },
    {
      "epoch": 0.2710042985541227,
      "grad_norm": 14.492600440979004,
      "learning_rate": 1.7146733860429614e-05,
      "loss": 9.779,
      "step": 1387
    },
    {
      "epoch": 0.27119968737788197,
      "grad_norm": 7.780018329620361,
      "learning_rate": 1.7142305730967913e-05,
      "loss": 10.2408,
      "step": 1388
    },
    {
      "epoch": 0.27139507620164127,
      "grad_norm": 9.412854194641113,
      "learning_rate": 1.7137874740791027e-05,
      "loss": 10.4512,
      "step": 1389
    },
    {
      "epoch": 0.2715904650254006,
      "grad_norm": 10.296211242675781,
      "learning_rate": 1.71334408916737e-05,
      "loss": 11.2377,
      "step": 1390
    },
    {
      "epoch": 0.2717858538491598,
      "grad_norm": 14.402783393859863,
      "learning_rate": 1.712900418539183e-05,
      "loss": 10.3899,
      "step": 1391
    },
    {
      "epoch": 0.2719812426729191,
      "grad_norm": 11.80636215209961,
      "learning_rate": 1.7124564623722455e-05,
      "loss": 9.8263,
      "step": 1392
    },
    {
      "epoch": 0.27217663149667837,
      "grad_norm": 7.489226341247559,
      "learning_rate": 1.7120122208443755e-05,
      "loss": 10.4005,
      "step": 1393
    },
    {
      "epoch": 0.27237202032043767,
      "grad_norm": 9.502429008483887,
      "learning_rate": 1.7115676941335057e-05,
      "loss": 10.4095,
      "step": 1394
    },
    {
      "epoch": 0.27256740914419697,
      "grad_norm": 12.124082565307617,
      "learning_rate": 1.7111228824176827e-05,
      "loss": 10.1853,
      "step": 1395
    },
    {
      "epoch": 0.2727627979679562,
      "grad_norm": 11.134305000305176,
      "learning_rate": 1.7106777858750673e-05,
      "loss": 9.9589,
      "step": 1396
    },
    {
      "epoch": 0.2729581867917155,
      "grad_norm": 10.44892692565918,
      "learning_rate": 1.7102324046839346e-05,
      "loss": 9.6726,
      "step": 1397
    },
    {
      "epoch": 0.2731535756154748,
      "grad_norm": 14.50195598602295,
      "learning_rate": 1.709786739022674e-05,
      "loss": 10.07,
      "step": 1398
    },
    {
      "epoch": 0.27334896443923407,
      "grad_norm": 13.216436386108398,
      "learning_rate": 1.709340789069788e-05,
      "loss": 8.6138,
      "step": 1399
    },
    {
      "epoch": 0.27354435326299337,
      "grad_norm": 10.18933391571045,
      "learning_rate": 1.7088945550038935e-05,
      "loss": 9.2372,
      "step": 1400
    },
    {
      "epoch": 0.2737397420867526,
      "grad_norm": 10.000385284423828,
      "learning_rate": 1.708448037003721e-05,
      "loss": 9.677,
      "step": 1401
    },
    {
      "epoch": 0.2739351309105119,
      "grad_norm": 11.34453296661377,
      "learning_rate": 1.708001235248115e-05,
      "loss": 9.6858,
      "step": 1402
    },
    {
      "epoch": 0.2741305197342712,
      "grad_norm": 9.784234046936035,
      "learning_rate": 1.7075541499160336e-05,
      "loss": 10.3645,
      "step": 1403
    },
    {
      "epoch": 0.27432590855803046,
      "grad_norm": 8.841297149658203,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 9.9277,
      "step": 1404
    },
    {
      "epoch": 0.27452129738178976,
      "grad_norm": 8.425395011901855,
      "learning_rate": 1.706659129238843e-05,
      "loss": 9.4929,
      "step": 1405
    },
    {
      "epoch": 0.27471668620554907,
      "grad_norm": 8.213043212890625,
      "learning_rate": 1.7062111942522177e-05,
      "loss": 9.5276,
      "step": 1406
    },
    {
      "epoch": 0.2749120750293083,
      "grad_norm": 12.275148391723633,
      "learning_rate": 1.7057629764060842e-05,
      "loss": 9.3295,
      "step": 1407
    },
    {
      "epoch": 0.2751074638530676,
      "grad_norm": 10.934953689575195,
      "learning_rate": 1.7053144758799668e-05,
      "loss": 8.5463,
      "step": 1408
    },
    {
      "epoch": 0.27530285267682686,
      "grad_norm": 7.493126392364502,
      "learning_rate": 1.7048656928535045e-05,
      "loss": 9.9411,
      "step": 1409
    },
    {
      "epoch": 0.27549824150058616,
      "grad_norm": 10.218108177185059,
      "learning_rate": 1.704416627506449e-05,
      "loss": 10.0827,
      "step": 1410
    },
    {
      "epoch": 0.27569363032434546,
      "grad_norm": 10.551949501037598,
      "learning_rate": 1.703967280018664e-05,
      "loss": 10.771,
      "step": 1411
    },
    {
      "epoch": 0.2758890191481047,
      "grad_norm": 10.218538284301758,
      "learning_rate": 1.703517650570128e-05,
      "loss": 10.8535,
      "step": 1412
    },
    {
      "epoch": 0.276084407971864,
      "grad_norm": 10.639214515686035,
      "learning_rate": 1.703067739340931e-05,
      "loss": 9.3686,
      "step": 1413
    },
    {
      "epoch": 0.2762797967956233,
      "grad_norm": 10.4620361328125,
      "learning_rate": 1.702617546511277e-05,
      "loss": 9.0547,
      "step": 1414
    },
    {
      "epoch": 0.27647518561938256,
      "grad_norm": 9.750692367553711,
      "learning_rate": 1.7021670722614817e-05,
      "loss": 9.6319,
      "step": 1415
    },
    {
      "epoch": 0.27667057444314186,
      "grad_norm": 9.28245735168457,
      "learning_rate": 1.7017163167719743e-05,
      "loss": 9.6182,
      "step": 1416
    },
    {
      "epoch": 0.27686596326690116,
      "grad_norm": 14.629377365112305,
      "learning_rate": 1.701265280223296e-05,
      "loss": 9.664,
      "step": 1417
    },
    {
      "epoch": 0.2770613520906604,
      "grad_norm": 10.068516731262207,
      "learning_rate": 1.7008139627961012e-05,
      "loss": 9.9891,
      "step": 1418
    },
    {
      "epoch": 0.2772567409144197,
      "grad_norm": 13.83179759979248,
      "learning_rate": 1.700362364671157e-05,
      "loss": 10.4564,
      "step": 1419
    },
    {
      "epoch": 0.27745212973817895,
      "grad_norm": 10.409890174865723,
      "learning_rate": 1.699910486029342e-05,
      "loss": 9.4946,
      "step": 1420
    },
    {
      "epoch": 0.27764751856193826,
      "grad_norm": 9.16124439239502,
      "learning_rate": 1.699458327051647e-05,
      "loss": 9.3257,
      "step": 1421
    },
    {
      "epoch": 0.27784290738569756,
      "grad_norm": 7.58655309677124,
      "learning_rate": 1.699005887919177e-05,
      "loss": 9.4247,
      "step": 1422
    },
    {
      "epoch": 0.2780382962094568,
      "grad_norm": 8.163040161132812,
      "learning_rate": 1.698553168813147e-05,
      "loss": 10.097,
      "step": 1423
    },
    {
      "epoch": 0.2782336850332161,
      "grad_norm": 11.548359870910645,
      "learning_rate": 1.6981001699148853e-05,
      "loss": 8.9478,
      "step": 1424
    },
    {
      "epoch": 0.2784290738569754,
      "grad_norm": 16.783096313476562,
      "learning_rate": 1.697646891405832e-05,
      "loss": 10.8209,
      "step": 1425
    },
    {
      "epoch": 0.27862446268073465,
      "grad_norm": 12.144583702087402,
      "learning_rate": 1.6971933334675395e-05,
      "loss": 9.2343,
      "step": 1426
    },
    {
      "epoch": 0.27881985150449395,
      "grad_norm": 13.720064163208008,
      "learning_rate": 1.6967394962816713e-05,
      "loss": 9.8624,
      "step": 1427
    },
    {
      "epoch": 0.2790152403282532,
      "grad_norm": 10.508441925048828,
      "learning_rate": 1.696285380030004e-05,
      "loss": 9.7736,
      "step": 1428
    },
    {
      "epoch": 0.2792106291520125,
      "grad_norm": 10.424206733703613,
      "learning_rate": 1.6958309848944244e-05,
      "loss": 10.8513,
      "step": 1429
    },
    {
      "epoch": 0.2794060179757718,
      "grad_norm": 10.918111801147461,
      "learning_rate": 1.6953763110569325e-05,
      "loss": 9.6206,
      "step": 1430
    },
    {
      "epoch": 0.27960140679953105,
      "grad_norm": 7.589666366577148,
      "learning_rate": 1.694921358699639e-05,
      "loss": 10.1118,
      "step": 1431
    },
    {
      "epoch": 0.27979679562329035,
      "grad_norm": 10.684234619140625,
      "learning_rate": 1.6944661280047667e-05,
      "loss": 9.1125,
      "step": 1432
    },
    {
      "epoch": 0.27999218444704965,
      "grad_norm": 8.044968605041504,
      "learning_rate": 1.6940106191546493e-05,
      "loss": 9.7331,
      "step": 1433
    },
    {
      "epoch": 0.2801875732708089,
      "grad_norm": 7.673303604125977,
      "learning_rate": 1.6935548323317326e-05,
      "loss": 10.1433,
      "step": 1434
    },
    {
      "epoch": 0.2803829620945682,
      "grad_norm": 9.639922142028809,
      "learning_rate": 1.693098767718573e-05,
      "loss": 9.7468,
      "step": 1435
    },
    {
      "epoch": 0.28057835091832745,
      "grad_norm": 7.536942005157471,
      "learning_rate": 1.6926424254978388e-05,
      "loss": 9.557,
      "step": 1436
    },
    {
      "epoch": 0.28077373974208675,
      "grad_norm": 9.589834213256836,
      "learning_rate": 1.692185805852309e-05,
      "loss": 8.855,
      "step": 1437
    },
    {
      "epoch": 0.28096912856584605,
      "grad_norm": 8.762582778930664,
      "learning_rate": 1.6917289089648742e-05,
      "loss": 8.8605,
      "step": 1438
    },
    {
      "epoch": 0.2811645173896053,
      "grad_norm": 9.100251197814941,
      "learning_rate": 1.6912717350185353e-05,
      "loss": 9.8335,
      "step": 1439
    },
    {
      "epoch": 0.2813599062133646,
      "grad_norm": 8.672943115234375,
      "learning_rate": 1.6908142841964055e-05,
      "loss": 9.9712,
      "step": 1440
    },
    {
      "epoch": 0.2815552950371239,
      "grad_norm": 8.250885963439941,
      "learning_rate": 1.6903565566817073e-05,
      "loss": 10.1313,
      "step": 1441
    },
    {
      "epoch": 0.28175068386088314,
      "grad_norm": 67.72570037841797,
      "learning_rate": 1.689898552657775e-05,
      "loss": 7.4254,
      "step": 1442
    },
    {
      "epoch": 0.28194607268464245,
      "grad_norm": 12.001933097839355,
      "learning_rate": 1.689440272308054e-05,
      "loss": 9.6222,
      "step": 1443
    },
    {
      "epoch": 0.2821414615084017,
      "grad_norm": 8.292913436889648,
      "learning_rate": 1.688981715816099e-05,
      "loss": 9.6154,
      "step": 1444
    },
    {
      "epoch": 0.282336850332161,
      "grad_norm": 7.754610061645508,
      "learning_rate": 1.688522883365576e-05,
      "loss": 10.2804,
      "step": 1445
    },
    {
      "epoch": 0.2825322391559203,
      "grad_norm": 11.714853286743164,
      "learning_rate": 1.6880637751402623e-05,
      "loss": 10.2299,
      "step": 1446
    },
    {
      "epoch": 0.28272762797967954,
      "grad_norm": 9.187288284301758,
      "learning_rate": 1.687604391324045e-05,
      "loss": 10.3888,
      "step": 1447
    },
    {
      "epoch": 0.28292301680343884,
      "grad_norm": 13.724797248840332,
      "learning_rate": 1.6871447321009215e-05,
      "loss": 10.1753,
      "step": 1448
    },
    {
      "epoch": 0.28311840562719814,
      "grad_norm": 8.328743934631348,
      "learning_rate": 1.6866847976549988e-05,
      "loss": 9.6432,
      "step": 1449
    },
    {
      "epoch": 0.2833137944509574,
      "grad_norm": 10.776053428649902,
      "learning_rate": 1.686224588170496e-05,
      "loss": 10.8265,
      "step": 1450
    },
    {
      "epoch": 0.2835091832747167,
      "grad_norm": 9.196435928344727,
      "learning_rate": 1.6857641038317404e-05,
      "loss": 9.3719,
      "step": 1451
    },
    {
      "epoch": 0.283704572098476,
      "grad_norm": 11.425103187561035,
      "learning_rate": 1.6853033448231712e-05,
      "loss": 10.7306,
      "step": 1452
    },
    {
      "epoch": 0.28389996092223524,
      "grad_norm": 9.222018241882324,
      "learning_rate": 1.684842311329336e-05,
      "loss": 9.7921,
      "step": 1453
    },
    {
      "epoch": 0.28409534974599454,
      "grad_norm": 10.456576347351074,
      "learning_rate": 1.684381003534893e-05,
      "loss": 10.2656,
      "step": 1454
    },
    {
      "epoch": 0.2842907385697538,
      "grad_norm": 8.699284553527832,
      "learning_rate": 1.683919421624611e-05,
      "loss": 9.208,
      "step": 1455
    },
    {
      "epoch": 0.2844861273935131,
      "grad_norm": 10.197521209716797,
      "learning_rate": 1.683457565783367e-05,
      "loss": 9.3122,
      "step": 1456
    },
    {
      "epoch": 0.2846815162172724,
      "grad_norm": 15.487138748168945,
      "learning_rate": 1.6829954361961492e-05,
      "loss": 9.501,
      "step": 1457
    },
    {
      "epoch": 0.28487690504103164,
      "grad_norm": 10.236659049987793,
      "learning_rate": 1.6825330330480545e-05,
      "loss": 9.9833,
      "step": 1458
    },
    {
      "epoch": 0.28507229386479094,
      "grad_norm": 9.559532165527344,
      "learning_rate": 1.68207035652429e-05,
      "loss": 9.6962,
      "step": 1459
    },
    {
      "epoch": 0.28526768268855024,
      "grad_norm": 10.54056453704834,
      "learning_rate": 1.6816074068101715e-05,
      "loss": 8.1149,
      "step": 1460
    },
    {
      "epoch": 0.2854630715123095,
      "grad_norm": 8.070180892944336,
      "learning_rate": 1.6811441840911252e-05,
      "loss": 8.8212,
      "step": 1461
    },
    {
      "epoch": 0.2856584603360688,
      "grad_norm": 11.812846183776855,
      "learning_rate": 1.6806806885526864e-05,
      "loss": 10.4092,
      "step": 1462
    },
    {
      "epoch": 0.28585384915982803,
      "grad_norm": 11.560187339782715,
      "learning_rate": 1.6802169203804992e-05,
      "loss": 9.4563,
      "step": 1463
    },
    {
      "epoch": 0.28604923798358733,
      "grad_norm": 14.845040321350098,
      "learning_rate": 1.679752879760317e-05,
      "loss": 9.9109,
      "step": 1464
    },
    {
      "epoch": 0.28624462680734664,
      "grad_norm": 17.191116333007812,
      "learning_rate": 1.6792885668780026e-05,
      "loss": 10.3424,
      "step": 1465
    },
    {
      "epoch": 0.2864400156311059,
      "grad_norm": 9.194401741027832,
      "learning_rate": 1.6788239819195277e-05,
      "loss": 9.9862,
      "step": 1466
    },
    {
      "epoch": 0.2866354044548652,
      "grad_norm": 7.765769958496094,
      "learning_rate": 1.6783591250709726e-05,
      "loss": 9.7139,
      "step": 1467
    },
    {
      "epoch": 0.2868307932786245,
      "grad_norm": 14.478752136230469,
      "learning_rate": 1.6778939965185278e-05,
      "loss": 10.2629,
      "step": 1468
    },
    {
      "epoch": 0.28702618210238373,
      "grad_norm": 9.576632499694824,
      "learning_rate": 1.677428596448491e-05,
      "loss": 10.4502,
      "step": 1469
    },
    {
      "epoch": 0.28722157092614303,
      "grad_norm": 11.754566192626953,
      "learning_rate": 1.6769629250472696e-05,
      "loss": 9.6511,
      "step": 1470
    },
    {
      "epoch": 0.2874169597499023,
      "grad_norm": 10.358704566955566,
      "learning_rate": 1.6764969825013795e-05,
      "loss": 9.386,
      "step": 1471
    },
    {
      "epoch": 0.2876123485736616,
      "grad_norm": 10.96502685546875,
      "learning_rate": 1.676030768997445e-05,
      "loss": 9.974,
      "step": 1472
    },
    {
      "epoch": 0.2878077373974209,
      "grad_norm": 12.029976844787598,
      "learning_rate": 1.6755642847221992e-05,
      "loss": 10.6669,
      "step": 1473
    },
    {
      "epoch": 0.28800312622118013,
      "grad_norm": 12.324685096740723,
      "learning_rate": 1.6750975298624838e-05,
      "loss": 9.6188,
      "step": 1474
    },
    {
      "epoch": 0.28819851504493943,
      "grad_norm": 10.851834297180176,
      "learning_rate": 1.674630504605248e-05,
      "loss": 9.612,
      "step": 1475
    },
    {
      "epoch": 0.28839390386869873,
      "grad_norm": 11.32132339477539,
      "learning_rate": 1.6741632091375505e-05,
      "loss": 10.2633,
      "step": 1476
    },
    {
      "epoch": 0.288589292692458,
      "grad_norm": 10.077210426330566,
      "learning_rate": 1.6736956436465573e-05,
      "loss": 10.4579,
      "step": 1477
    },
    {
      "epoch": 0.2887846815162173,
      "grad_norm": 12.926339149475098,
      "learning_rate": 1.673227808319543e-05,
      "loss": 11.0451,
      "step": 1478
    },
    {
      "epoch": 0.2889800703399766,
      "grad_norm": 16.012737274169922,
      "learning_rate": 1.6727597033438903e-05,
      "loss": 9.7502,
      "step": 1479
    },
    {
      "epoch": 0.2891754591637358,
      "grad_norm": 13.25388240814209,
      "learning_rate": 1.6722913289070896e-05,
      "loss": 9.355,
      "step": 1480
    },
    {
      "epoch": 0.28937084798749513,
      "grad_norm": 9.597887992858887,
      "learning_rate": 1.671822685196739e-05,
      "loss": 8.222,
      "step": 1481
    },
    {
      "epoch": 0.2895662368112544,
      "grad_norm": 9.557008743286133,
      "learning_rate": 1.6713537724005457e-05,
      "loss": 9.1877,
      "step": 1482
    },
    {
      "epoch": 0.2897616256350137,
      "grad_norm": 12.110699653625488,
      "learning_rate": 1.6708845907063234e-05,
      "loss": 10.3352,
      "step": 1483
    },
    {
      "epoch": 0.289957014458773,
      "grad_norm": 10.078930854797363,
      "learning_rate": 1.6704151403019938e-05,
      "loss": 10.5232,
      "step": 1484
    },
    {
      "epoch": 0.2901524032825322,
      "grad_norm": 11.133780479431152,
      "learning_rate": 1.6699454213755868e-05,
      "loss": 10.0798,
      "step": 1485
    },
    {
      "epoch": 0.2903477921062915,
      "grad_norm": 11.386680603027344,
      "learning_rate": 1.669475434115239e-05,
      "loss": 10.3686,
      "step": 1486
    },
    {
      "epoch": 0.2905431809300508,
      "grad_norm": 12.639415740966797,
      "learning_rate": 1.669005178709195e-05,
      "loss": 9.2262,
      "step": 1487
    },
    {
      "epoch": 0.29073856975381007,
      "grad_norm": 12.360422134399414,
      "learning_rate": 1.6685346553458067e-05,
      "loss": 9.796,
      "step": 1488
    },
    {
      "epoch": 0.2909339585775694,
      "grad_norm": 9.941367149353027,
      "learning_rate": 1.6680638642135337e-05,
      "loss": 8.8607,
      "step": 1489
    },
    {
      "epoch": 0.2911293474013286,
      "grad_norm": 8.1162109375,
      "learning_rate": 1.6675928055009417e-05,
      "loss": 8.7807,
      "step": 1490
    },
    {
      "epoch": 0.2913247362250879,
      "grad_norm": 9.627894401550293,
      "learning_rate": 1.667121479396705e-05,
      "loss": 8.5564,
      "step": 1491
    },
    {
      "epoch": 0.2915201250488472,
      "grad_norm": 8.931618690490723,
      "learning_rate": 1.6666498860896036e-05,
      "loss": 9.3509,
      "step": 1492
    },
    {
      "epoch": 0.29171551387260647,
      "grad_norm": 11.936795234680176,
      "learning_rate": 1.666178025768526e-05,
      "loss": 10.2361,
      "step": 1493
    },
    {
      "epoch": 0.29191090269636577,
      "grad_norm": 10.509285926818848,
      "learning_rate": 1.6657058986224667e-05,
      "loss": 9.7758,
      "step": 1494
    },
    {
      "epoch": 0.29210629152012507,
      "grad_norm": 6.613710403442383,
      "learning_rate": 1.6652335048405275e-05,
      "loss": 9.2565,
      "step": 1495
    },
    {
      "epoch": 0.2923016803438843,
      "grad_norm": 27.692249298095703,
      "learning_rate": 1.6647608446119163e-05,
      "loss": 10.6953,
      "step": 1496
    },
    {
      "epoch": 0.2924970691676436,
      "grad_norm": 12.839340209960938,
      "learning_rate": 1.6642879181259483e-05,
      "loss": 9.8403,
      "step": 1497
    },
    {
      "epoch": 0.29269245799140287,
      "grad_norm": 8.90999984741211,
      "learning_rate": 1.663814725572045e-05,
      "loss": 10.0489,
      "step": 1498
    },
    {
      "epoch": 0.29288784681516217,
      "grad_norm": 10.096855163574219,
      "learning_rate": 1.6633412671397357e-05,
      "loss": 10.1652,
      "step": 1499
    },
    {
      "epoch": 0.29308323563892147,
      "grad_norm": 8.625805854797363,
      "learning_rate": 1.6628675430186543e-05,
      "loss": 9.3864,
      "step": 1500
    },
    {
      "epoch": 0.2932786244626807,
      "grad_norm": 12.385692596435547,
      "learning_rate": 1.662393553398542e-05,
      "loss": 8.7917,
      "step": 1501
    },
    {
      "epoch": 0.29347401328644,
      "grad_norm": 10.390081405639648,
      "learning_rate": 1.661919298469247e-05,
      "loss": 10.0085,
      "step": 1502
    },
    {
      "epoch": 0.2936694021101993,
      "grad_norm": 16.05337142944336,
      "learning_rate": 1.6614447784207226e-05,
      "loss": 10.1085,
      "step": 1503
    },
    {
      "epoch": 0.29386479093395856,
      "grad_norm": 10.438437461853027,
      "learning_rate": 1.660969993443029e-05,
      "loss": 9.9805,
      "step": 1504
    },
    {
      "epoch": 0.29406017975771787,
      "grad_norm": 11.209931373596191,
      "learning_rate": 1.6604949437263322e-05,
      "loss": 10.66,
      "step": 1505
    },
    {
      "epoch": 0.29425556858147717,
      "grad_norm": 11.702326774597168,
      "learning_rate": 1.6600196294609047e-05,
      "loss": 9.4944,
      "step": 1506
    },
    {
      "epoch": 0.2944509574052364,
      "grad_norm": 9.846290588378906,
      "learning_rate": 1.6595440508371238e-05,
      "loss": 9.8013,
      "step": 1507
    },
    {
      "epoch": 0.2946463462289957,
      "grad_norm": 8.696939468383789,
      "learning_rate": 1.6590682080454748e-05,
      "loss": 9.4621,
      "step": 1508
    },
    {
      "epoch": 0.29484173505275496,
      "grad_norm": 9.543524742126465,
      "learning_rate": 1.6585921012765465e-05,
      "loss": 9.3651,
      "step": 1509
    },
    {
      "epoch": 0.29503712387651426,
      "grad_norm": 11.260859489440918,
      "learning_rate": 1.6581157307210345e-05,
      "loss": 9.5992,
      "step": 1510
    },
    {
      "epoch": 0.29523251270027356,
      "grad_norm": 10.757518768310547,
      "learning_rate": 1.6576390965697403e-05,
      "loss": 10.2075,
      "step": 1511
    },
    {
      "epoch": 0.2954279015240328,
      "grad_norm": 9.129809379577637,
      "learning_rate": 1.6571621990135708e-05,
      "loss": 9.8907,
      "step": 1512
    },
    {
      "epoch": 0.2956232903477921,
      "grad_norm": 8.124305725097656,
      "learning_rate": 1.6566850382435378e-05,
      "loss": 7.8761,
      "step": 1513
    },
    {
      "epoch": 0.2958186791715514,
      "grad_norm": 8.545839309692383,
      "learning_rate": 1.6562076144507595e-05,
      "loss": 9.7005,
      "step": 1514
    },
    {
      "epoch": 0.29601406799531066,
      "grad_norm": 8.617210388183594,
      "learning_rate": 1.6557299278264586e-05,
      "loss": 9.6294,
      "step": 1515
    },
    {
      "epoch": 0.29620945681906996,
      "grad_norm": 10.131733894348145,
      "learning_rate": 1.6552519785619633e-05,
      "loss": 8.6846,
      "step": 1516
    },
    {
      "epoch": 0.2964048456428292,
      "grad_norm": 12.793623924255371,
      "learning_rate": 1.6547737668487078e-05,
      "loss": 9.1373,
      "step": 1517
    },
    {
      "epoch": 0.2966002344665885,
      "grad_norm": 10.047884941101074,
      "learning_rate": 1.6542952928782302e-05,
      "loss": 10.5081,
      "step": 1518
    },
    {
      "epoch": 0.2967956232903478,
      "grad_norm": 11.640722274780273,
      "learning_rate": 1.6538165568421742e-05,
      "loss": 8.9963,
      "step": 1519
    },
    {
      "epoch": 0.29699101211410706,
      "grad_norm": 9.68021297454834,
      "learning_rate": 1.6533375589322883e-05,
      "loss": 10.3106,
      "step": 1520
    },
    {
      "epoch": 0.29718640093786636,
      "grad_norm": 29.985857009887695,
      "learning_rate": 1.6528582993404266e-05,
      "loss": 10.0603,
      "step": 1521
    },
    {
      "epoch": 0.29738178976162566,
      "grad_norm": 10.854527473449707,
      "learning_rate": 1.652378778258547e-05,
      "loss": 9.6001,
      "step": 1522
    },
    {
      "epoch": 0.2975771785853849,
      "grad_norm": 11.029958724975586,
      "learning_rate": 1.6518989958787126e-05,
      "loss": 9.0629,
      "step": 1523
    },
    {
      "epoch": 0.2977725674091442,
      "grad_norm": 8.962169647216797,
      "learning_rate": 1.6514189523930913e-05,
      "loss": 8.4699,
      "step": 1524
    },
    {
      "epoch": 0.29796795623290345,
      "grad_norm": 8.509246826171875,
      "learning_rate": 1.6509386479939554e-05,
      "loss": 9.2693,
      "step": 1525
    },
    {
      "epoch": 0.29816334505666275,
      "grad_norm": 10.364874839782715,
      "learning_rate": 1.650458082873682e-05,
      "loss": 8.625,
      "step": 1526
    },
    {
      "epoch": 0.29835873388042206,
      "grad_norm": 10.83619499206543,
      "learning_rate": 1.6499772572247518e-05,
      "loss": 10.3062,
      "step": 1527
    },
    {
      "epoch": 0.2985541227041813,
      "grad_norm": 13.112933158874512,
      "learning_rate": 1.64949617123975e-05,
      "loss": 9.9893,
      "step": 1528
    },
    {
      "epoch": 0.2987495115279406,
      "grad_norm": 7.441256046295166,
      "learning_rate": 1.649014825111368e-05,
      "loss": 9.1927,
      "step": 1529
    },
    {
      "epoch": 0.2989449003516999,
      "grad_norm": 14.765359878540039,
      "learning_rate": 1.6485332190323984e-05,
      "loss": 10.6637,
      "step": 1530
    },
    {
      "epoch": 0.29914028917545915,
      "grad_norm": 11.117351531982422,
      "learning_rate": 1.64805135319574e-05,
      "loss": 9.4884,
      "step": 1531
    },
    {
      "epoch": 0.29933567799921845,
      "grad_norm": 6.486770153045654,
      "learning_rate": 1.647569227794395e-05,
      "loss": 9.4596,
      "step": 1532
    },
    {
      "epoch": 0.2995310668229777,
      "grad_norm": 9.962038040161133,
      "learning_rate": 1.64708684302147e-05,
      "loss": 9.6684,
      "step": 1533
    },
    {
      "epoch": 0.299726455646737,
      "grad_norm": 8.842194557189941,
      "learning_rate": 1.6466041990701743e-05,
      "loss": 8.9051,
      "step": 1534
    },
    {
      "epoch": 0.2999218444704963,
      "grad_norm": 10.530966758728027,
      "learning_rate": 1.646121296133822e-05,
      "loss": 10.2824,
      "step": 1535
    },
    {
      "epoch": 0.30011723329425555,
      "grad_norm": 10.525800704956055,
      "learning_rate": 1.645638134405831e-05,
      "loss": 9.6589,
      "step": 1536
    },
    {
      "epoch": 0.30031262211801485,
      "grad_norm": 8.37814998626709,
      "learning_rate": 1.6451547140797223e-05,
      "loss": 9.7822,
      "step": 1537
    },
    {
      "epoch": 0.30050801094177415,
      "grad_norm": 8.279181480407715,
      "learning_rate": 1.6446710353491213e-05,
      "loss": 9.5246,
      "step": 1538
    },
    {
      "epoch": 0.3007033997655334,
      "grad_norm": 10.936617851257324,
      "learning_rate": 1.6441870984077556e-05,
      "loss": 9.5018,
      "step": 1539
    },
    {
      "epoch": 0.3008987885892927,
      "grad_norm": 8.712897300720215,
      "learning_rate": 1.6437029034494574e-05,
      "loss": 10.3973,
      "step": 1540
    },
    {
      "epoch": 0.301094177413052,
      "grad_norm": 24.52730369567871,
      "learning_rate": 1.6432184506681615e-05,
      "loss": 9.4271,
      "step": 1541
    },
    {
      "epoch": 0.30128956623681125,
      "grad_norm": 11.265439987182617,
      "learning_rate": 1.6427337402579065e-05,
      "loss": 8.8542,
      "step": 1542
    },
    {
      "epoch": 0.30148495506057055,
      "grad_norm": 8.999545097351074,
      "learning_rate": 1.6422487724128343e-05,
      "loss": 8.8798,
      "step": 1543
    },
    {
      "epoch": 0.3016803438843298,
      "grad_norm": 10.456027030944824,
      "learning_rate": 1.6417635473271893e-05,
      "loss": 9.0947,
      "step": 1544
    },
    {
      "epoch": 0.3018757327080891,
      "grad_norm": 15.720943450927734,
      "learning_rate": 1.6412780651953187e-05,
      "loss": 9.244,
      "step": 1545
    },
    {
      "epoch": 0.3020711215318484,
      "grad_norm": 11.093077659606934,
      "learning_rate": 1.6407923262116736e-05,
      "loss": 10.4142,
      "step": 1546
    },
    {
      "epoch": 0.30226651035560764,
      "grad_norm": 10.72560977935791,
      "learning_rate": 1.6403063305708076e-05,
      "loss": 9.9932,
      "step": 1547
    },
    {
      "epoch": 0.30246189917936694,
      "grad_norm": 10.00912094116211,
      "learning_rate": 1.6398200784673773e-05,
      "loss": 10.841,
      "step": 1548
    },
    {
      "epoch": 0.30265728800312625,
      "grad_norm": 11.052425384521484,
      "learning_rate": 1.639333570096141e-05,
      "loss": 9.2576,
      "step": 1549
    },
    {
      "epoch": 0.3028526768268855,
      "grad_norm": 8.651049613952637,
      "learning_rate": 1.638846805651961e-05,
      "loss": 9.9265,
      "step": 1550
    },
    {
      "epoch": 0.3030480656506448,
      "grad_norm": 7.818670272827148,
      "learning_rate": 1.638359785329802e-05,
      "loss": 8.814,
      "step": 1551
    },
    {
      "epoch": 0.30324345447440404,
      "grad_norm": 9.49549674987793,
      "learning_rate": 1.6378725093247297e-05,
      "loss": 9.3743,
      "step": 1552
    },
    {
      "epoch": 0.30343884329816334,
      "grad_norm": 7.494236469268799,
      "learning_rate": 1.637384977831914e-05,
      "loss": 10.0042,
      "step": 1553
    },
    {
      "epoch": 0.30363423212192264,
      "grad_norm": 10.093852043151855,
      "learning_rate": 1.636897191046626e-05,
      "loss": 9.2741,
      "step": 1554
    },
    {
      "epoch": 0.3038296209456819,
      "grad_norm": 9.646723747253418,
      "learning_rate": 1.63640914916424e-05,
      "loss": 10.0743,
      "step": 1555
    },
    {
      "epoch": 0.3040250097694412,
      "grad_norm": 14.061182975769043,
      "learning_rate": 1.6359208523802308e-05,
      "loss": 9.7071,
      "step": 1556
    },
    {
      "epoch": 0.3042203985932005,
      "grad_norm": 7.9713664054870605,
      "learning_rate": 1.6354323008901774e-05,
      "loss": 8.947,
      "step": 1557
    },
    {
      "epoch": 0.30441578741695974,
      "grad_norm": 7.878183364868164,
      "learning_rate": 1.63494349488976e-05,
      "loss": 9.4146,
      "step": 1558
    },
    {
      "epoch": 0.30461117624071904,
      "grad_norm": 9.49204158782959,
      "learning_rate": 1.63445443457476e-05,
      "loss": 10.5473,
      "step": 1559
    },
    {
      "epoch": 0.3048065650644783,
      "grad_norm": 9.174206733703613,
      "learning_rate": 1.6339651201410613e-05,
      "loss": 9.1709,
      "step": 1560
    },
    {
      "epoch": 0.3050019538882376,
      "grad_norm": 8.276881217956543,
      "learning_rate": 1.6334755517846496e-05,
      "loss": 9.2224,
      "step": 1561
    },
    {
      "epoch": 0.3051973427119969,
      "grad_norm": 11.268220901489258,
      "learning_rate": 1.632985729701612e-05,
      "loss": 9.8205,
      "step": 1562
    },
    {
      "epoch": 0.30539273153575613,
      "grad_norm": 9.72193717956543,
      "learning_rate": 1.6324956540881386e-05,
      "loss": 8.4075,
      "step": 1563
    },
    {
      "epoch": 0.30558812035951544,
      "grad_norm": 10.148341178894043,
      "learning_rate": 1.6320053251405187e-05,
      "loss": 9.7546,
      "step": 1564
    },
    {
      "epoch": 0.30578350918327474,
      "grad_norm": 9.791842460632324,
      "learning_rate": 1.6315147430551442e-05,
      "loss": 9.1544,
      "step": 1565
    },
    {
      "epoch": 0.305978898007034,
      "grad_norm": 27.13488006591797,
      "learning_rate": 1.6310239080285088e-05,
      "loss": 9.4843,
      "step": 1566
    },
    {
      "epoch": 0.3061742868307933,
      "grad_norm": 8.317152976989746,
      "learning_rate": 1.6305328202572078e-05,
      "loss": 9.1695,
      "step": 1567
    },
    {
      "epoch": 0.3063696756545526,
      "grad_norm": 13.713216781616211,
      "learning_rate": 1.630041479937936e-05,
      "loss": 9.5583,
      "step": 1568
    },
    {
      "epoch": 0.30656506447831183,
      "grad_norm": 27.447023391723633,
      "learning_rate": 1.629549887267491e-05,
      "loss": 10.8052,
      "step": 1569
    },
    {
      "epoch": 0.30676045330207113,
      "grad_norm": 11.13930892944336,
      "learning_rate": 1.6290580424427713e-05,
      "loss": 9.2008,
      "step": 1570
    },
    {
      "epoch": 0.3069558421258304,
      "grad_norm": 12.243010520935059,
      "learning_rate": 1.628565945660775e-05,
      "loss": 8.071,
      "step": 1571
    },
    {
      "epoch": 0.3071512309495897,
      "grad_norm": 9.27982234954834,
      "learning_rate": 1.6280735971186033e-05,
      "loss": 9.7449,
      "step": 1572
    },
    {
      "epoch": 0.307346619773349,
      "grad_norm": 10.390607833862305,
      "learning_rate": 1.6275809970134565e-05,
      "loss": 9.8616,
      "step": 1573
    },
    {
      "epoch": 0.30754200859710823,
      "grad_norm": 12.398528099060059,
      "learning_rate": 1.6270881455426358e-05,
      "loss": 9.521,
      "step": 1574
    },
    {
      "epoch": 0.30773739742086753,
      "grad_norm": 11.779969215393066,
      "learning_rate": 1.6265950429035445e-05,
      "loss": 10.1166,
      "step": 1575
    },
    {
      "epoch": 0.30793278624462683,
      "grad_norm": 10.970566749572754,
      "learning_rate": 1.6261016892936844e-05,
      "loss": 10.4658,
      "step": 1576
    },
    {
      "epoch": 0.3081281750683861,
      "grad_norm": 7.940483093261719,
      "learning_rate": 1.6256080849106603e-05,
      "loss": 9.0797,
      "step": 1577
    },
    {
      "epoch": 0.3083235638921454,
      "grad_norm": 11.488015174865723,
      "learning_rate": 1.6251142299521746e-05,
      "loss": 8.9782,
      "step": 1578
    },
    {
      "epoch": 0.3085189527159046,
      "grad_norm": 8.953866958618164,
      "learning_rate": 1.6246201246160327e-05,
      "loss": 8.9,
      "step": 1579
    },
    {
      "epoch": 0.30871434153966393,
      "grad_norm": 12.140908241271973,
      "learning_rate": 1.6241257691001386e-05,
      "loss": 11.0306,
      "step": 1580
    },
    {
      "epoch": 0.30890973036342323,
      "grad_norm": 10.783838272094727,
      "learning_rate": 1.623631163602497e-05,
      "loss": 9.4288,
      "step": 1581
    },
    {
      "epoch": 0.3091051191871825,
      "grad_norm": 10.663506507873535,
      "learning_rate": 1.6231363083212128e-05,
      "loss": 9.1419,
      "step": 1582
    },
    {
      "epoch": 0.3093005080109418,
      "grad_norm": 28.735572814941406,
      "learning_rate": 1.622641203454491e-05,
      "loss": 9.655,
      "step": 1583
    },
    {
      "epoch": 0.3094958968347011,
      "grad_norm": 9.377638816833496,
      "learning_rate": 1.6221458492006367e-05,
      "loss": 9.819,
      "step": 1584
    },
    {
      "epoch": 0.3096912856584603,
      "grad_norm": 16.58684539794922,
      "learning_rate": 1.6216502457580542e-05,
      "loss": 10.2696,
      "step": 1585
    },
    {
      "epoch": 0.3098866744822196,
      "grad_norm": 11.1240816116333,
      "learning_rate": 1.621154393325248e-05,
      "loss": 9.5811,
      "step": 1586
    },
    {
      "epoch": 0.31008206330597887,
      "grad_norm": 8.305037498474121,
      "learning_rate": 1.6206582921008233e-05,
      "loss": 9.3075,
      "step": 1587
    },
    {
      "epoch": 0.3102774521297382,
      "grad_norm": 10.81719970703125,
      "learning_rate": 1.620161942283483e-05,
      "loss": 10.3475,
      "step": 1588
    },
    {
      "epoch": 0.3104728409534975,
      "grad_norm": 14.510613441467285,
      "learning_rate": 1.619665344072031e-05,
      "loss": 9.751,
      "step": 1589
    },
    {
      "epoch": 0.3106682297772567,
      "grad_norm": 9.30660629272461,
      "learning_rate": 1.6191684976653707e-05,
      "loss": 9.7671,
      "step": 1590
    },
    {
      "epoch": 0.310863618601016,
      "grad_norm": 8.766847610473633,
      "learning_rate": 1.6186714032625036e-05,
      "loss": 9.5391,
      "step": 1591
    },
    {
      "epoch": 0.3110590074247753,
      "grad_norm": 8.718366622924805,
      "learning_rate": 1.6181740610625324e-05,
      "loss": 9.3201,
      "step": 1592
    },
    {
      "epoch": 0.31125439624853457,
      "grad_norm": 7.148858547210693,
      "learning_rate": 1.6176764712646575e-05,
      "loss": 9.3771,
      "step": 1593
    },
    {
      "epoch": 0.31144978507229387,
      "grad_norm": 14.913329124450684,
      "learning_rate": 1.617178634068179e-05,
      "loss": 10.153,
      "step": 1594
    },
    {
      "epoch": 0.3116451738960531,
      "grad_norm": 9.894980430603027,
      "learning_rate": 1.6166805496724965e-05,
      "loss": 9.9227,
      "step": 1595
    },
    {
      "epoch": 0.3118405627198124,
      "grad_norm": 8.090249061584473,
      "learning_rate": 1.6161822182771084e-05,
      "loss": 10.1949,
      "step": 1596
    },
    {
      "epoch": 0.3120359515435717,
      "grad_norm": 10.747549057006836,
      "learning_rate": 1.6156836400816112e-05,
      "loss": 9.2133,
      "step": 1597
    },
    {
      "epoch": 0.31223134036733097,
      "grad_norm": 8.165460586547852,
      "learning_rate": 1.6151848152857017e-05,
      "loss": 7.9304,
      "step": 1598
    },
    {
      "epoch": 0.31242672919109027,
      "grad_norm": 9.286834716796875,
      "learning_rate": 1.6146857440891744e-05,
      "loss": 9.6123,
      "step": 1599
    },
    {
      "epoch": 0.31262211801484957,
      "grad_norm": 10.36357307434082,
      "learning_rate": 1.6141864266919228e-05,
      "loss": 9.3243,
      "step": 1600
    },
    {
      "epoch": 0.3128175068386088,
      "grad_norm": 13.824201583862305,
      "learning_rate": 1.613686863293939e-05,
      "loss": 9.7002,
      "step": 1601
    },
    {
      "epoch": 0.3130128956623681,
      "grad_norm": 8.38644027709961,
      "learning_rate": 1.6131870540953135e-05,
      "loss": 9.3296,
      "step": 1602
    },
    {
      "epoch": 0.3132082844861274,
      "grad_norm": 39.436885833740234,
      "learning_rate": 1.6126869992962357e-05,
      "loss": 9.4538,
      "step": 1603
    },
    {
      "epoch": 0.31340367330988667,
      "grad_norm": 21.05567741394043,
      "learning_rate": 1.6121866990969926e-05,
      "loss": 9.7832,
      "step": 1604
    },
    {
      "epoch": 0.31359906213364597,
      "grad_norm": 9.28034496307373,
      "learning_rate": 1.611686153697971e-05,
      "loss": 9.8811,
      "step": 1605
    },
    {
      "epoch": 0.3137944509574052,
      "grad_norm": 7.713963985443115,
      "learning_rate": 1.611185363299654e-05,
      "loss": 9.5874,
      "step": 1606
    },
    {
      "epoch": 0.3139898397811645,
      "grad_norm": 8.904071807861328,
      "learning_rate": 1.6106843281026237e-05,
      "loss": 10.1699,
      "step": 1607
    },
    {
      "epoch": 0.3141852286049238,
      "grad_norm": 11.779389381408691,
      "learning_rate": 1.6101830483075603e-05,
      "loss": 9.974,
      "step": 1608
    },
    {
      "epoch": 0.31438061742868306,
      "grad_norm": 10.87238597869873,
      "learning_rate": 1.6096815241152427e-05,
      "loss": 10.2095,
      "step": 1609
    },
    {
      "epoch": 0.31457600625244236,
      "grad_norm": 8.63390827178955,
      "learning_rate": 1.6091797557265455e-05,
      "loss": 9.8568,
      "step": 1610
    },
    {
      "epoch": 0.31477139507620167,
      "grad_norm": 13.952505111694336,
      "learning_rate": 1.6086777433424435e-05,
      "loss": 9.9195,
      "step": 1611
    },
    {
      "epoch": 0.3149667838999609,
      "grad_norm": 7.961946964263916,
      "learning_rate": 1.6081754871640084e-05,
      "loss": 8.5242,
      "step": 1612
    },
    {
      "epoch": 0.3151621727237202,
      "grad_norm": 9.592717170715332,
      "learning_rate": 1.607672987392409e-05,
      "loss": 9.3546,
      "step": 1613
    },
    {
      "epoch": 0.31535756154747946,
      "grad_norm": 6.381619930267334,
      "learning_rate": 1.6071702442289115e-05,
      "loss": 9.2855,
      "step": 1614
    },
    {
      "epoch": 0.31555295037123876,
      "grad_norm": 10.898557662963867,
      "learning_rate": 1.6066672578748814e-05,
      "loss": 9.8498,
      "step": 1615
    },
    {
      "epoch": 0.31574833919499806,
      "grad_norm": 13.637084007263184,
      "learning_rate": 1.6061640285317795e-05,
      "loss": 10.5978,
      "step": 1616
    },
    {
      "epoch": 0.3159437280187573,
      "grad_norm": 10.367321014404297,
      "learning_rate": 1.605660556401165e-05,
      "loss": 8.6832,
      "step": 1617
    },
    {
      "epoch": 0.3161391168425166,
      "grad_norm": 9.48491382598877,
      "learning_rate": 1.6051568416846936e-05,
      "loss": 9.1167,
      "step": 1618
    },
    {
      "epoch": 0.3163345056662759,
      "grad_norm": 12.643638610839844,
      "learning_rate": 1.6046528845841195e-05,
      "loss": 9.4473,
      "step": 1619
    },
    {
      "epoch": 0.31652989449003516,
      "grad_norm": 9.137619972229004,
      "learning_rate": 1.604148685301293e-05,
      "loss": 9.1735,
      "step": 1620
    },
    {
      "epoch": 0.31672528331379446,
      "grad_norm": 13.785964012145996,
      "learning_rate": 1.603644244038161e-05,
      "loss": 10.3451,
      "step": 1621
    },
    {
      "epoch": 0.3169206721375537,
      "grad_norm": 10.2887544631958,
      "learning_rate": 1.6031395609967683e-05,
      "loss": 8.9395,
      "step": 1622
    },
    {
      "epoch": 0.317116060961313,
      "grad_norm": 12.658092498779297,
      "learning_rate": 1.6026346363792565e-05,
      "loss": 8.7574,
      "step": 1623
    },
    {
      "epoch": 0.3173114497850723,
      "grad_norm": 9.074075698852539,
      "learning_rate": 1.6021294703878632e-05,
      "loss": 9.1438,
      "step": 1624
    },
    {
      "epoch": 0.31750683860883155,
      "grad_norm": 9.075652122497559,
      "learning_rate": 1.6016240632249224e-05,
      "loss": 9.7798,
      "step": 1625
    },
    {
      "epoch": 0.31770222743259086,
      "grad_norm": 9.15244197845459,
      "learning_rate": 1.6011184150928667e-05,
      "loss": 8.891,
      "step": 1626
    },
    {
      "epoch": 0.31789761625635016,
      "grad_norm": 8.605587005615234,
      "learning_rate": 1.600612526194223e-05,
      "loss": 9.6025,
      "step": 1627
    },
    {
      "epoch": 0.3180930050801094,
      "grad_norm": 8.151103973388672,
      "learning_rate": 1.6001063967316156e-05,
      "loss": 9.012,
      "step": 1628
    },
    {
      "epoch": 0.3182883939038687,
      "grad_norm": 9.278593063354492,
      "learning_rate": 1.599600026907765e-05,
      "loss": 9.6449,
      "step": 1629
    },
    {
      "epoch": 0.318483782727628,
      "grad_norm": 9.308077812194824,
      "learning_rate": 1.5990934169254885e-05,
      "loss": 10.2807,
      "step": 1630
    },
    {
      "epoch": 0.31867917155138725,
      "grad_norm": 10.449454307556152,
      "learning_rate": 1.598586566987699e-05,
      "loss": 10.107,
      "step": 1631
    },
    {
      "epoch": 0.31887456037514655,
      "grad_norm": 11.943887710571289,
      "learning_rate": 1.598079477297405e-05,
      "loss": 9.5064,
      "step": 1632
    },
    {
      "epoch": 0.3190699491989058,
      "grad_norm": 7.746683120727539,
      "learning_rate": 1.5975721480577124e-05,
      "loss": 8.9351,
      "step": 1633
    },
    {
      "epoch": 0.3192653380226651,
      "grad_norm": 14.373797416687012,
      "learning_rate": 1.597064579471822e-05,
      "loss": 9.4663,
      "step": 1634
    },
    {
      "epoch": 0.3194607268464244,
      "grad_norm": 8.865236282348633,
      "learning_rate": 1.5965567717430304e-05,
      "loss": 9.718,
      "step": 1635
    },
    {
      "epoch": 0.31965611567018365,
      "grad_norm": 11.387980461120605,
      "learning_rate": 1.5960487250747314e-05,
      "loss": 9.918,
      "step": 1636
    },
    {
      "epoch": 0.31985150449394295,
      "grad_norm": 9.649009704589844,
      "learning_rate": 1.5955404396704127e-05,
      "loss": 8.6794,
      "step": 1637
    },
    {
      "epoch": 0.32004689331770225,
      "grad_norm": 10.171775817871094,
      "learning_rate": 1.595031915733658e-05,
      "loss": 10.2739,
      "step": 1638
    },
    {
      "epoch": 0.3202422821414615,
      "grad_norm": 17.069379806518555,
      "learning_rate": 1.5945231534681483e-05,
      "loss": 8.3431,
      "step": 1639
    },
    {
      "epoch": 0.3204376709652208,
      "grad_norm": 9.645978927612305,
      "learning_rate": 1.5940141530776573e-05,
      "loss": 9.5865,
      "step": 1640
    },
    {
      "epoch": 0.32063305978898005,
      "grad_norm": 9.110332489013672,
      "learning_rate": 1.593504914766056e-05,
      "loss": 9.1653,
      "step": 1641
    },
    {
      "epoch": 0.32082844861273935,
      "grad_norm": 21.16898536682129,
      "learning_rate": 1.5929954387373103e-05,
      "loss": 10.0758,
      "step": 1642
    },
    {
      "epoch": 0.32102383743649865,
      "grad_norm": 13.377840042114258,
      "learning_rate": 1.5924857251954805e-05,
      "loss": 9.098,
      "step": 1643
    },
    {
      "epoch": 0.3212192262602579,
      "grad_norm": 9.212141036987305,
      "learning_rate": 1.5919757743447237e-05,
      "loss": 9.5257,
      "step": 1644
    },
    {
      "epoch": 0.3214146150840172,
      "grad_norm": 10.903090476989746,
      "learning_rate": 1.5914655863892907e-05,
      "loss": 10.3237,
      "step": 1645
    },
    {
      "epoch": 0.3216100039077765,
      "grad_norm": 10.470623016357422,
      "learning_rate": 1.5909551615335274e-05,
      "loss": 9.6061,
      "step": 1646
    },
    {
      "epoch": 0.32180539273153574,
      "grad_norm": 19.114530563354492,
      "learning_rate": 1.5904444999818748e-05,
      "loss": 9.5387,
      "step": 1647
    },
    {
      "epoch": 0.32200078155529505,
      "grad_norm": 10.143712997436523,
      "learning_rate": 1.5899336019388686e-05,
      "loss": 9.6485,
      "step": 1648
    },
    {
      "epoch": 0.3221961703790543,
      "grad_norm": 12.910961151123047,
      "learning_rate": 1.5894224676091395e-05,
      "loss": 9.614,
      "step": 1649
    },
    {
      "epoch": 0.3223915592028136,
      "grad_norm": 9.364129066467285,
      "learning_rate": 1.588911097197413e-05,
      "loss": 10.3305,
      "step": 1650
    },
    {
      "epoch": 0.3225869480265729,
      "grad_norm": 11.014893531799316,
      "learning_rate": 1.588399490908508e-05,
      "loss": 10.3429,
      "step": 1651
    },
    {
      "epoch": 0.32278233685033214,
      "grad_norm": 11.558199882507324,
      "learning_rate": 1.5878876489473397e-05,
      "loss": 9.7504,
      "step": 1652
    },
    {
      "epoch": 0.32297772567409144,
      "grad_norm": 11.256964683532715,
      "learning_rate": 1.587375571518916e-05,
      "loss": 9.6539,
      "step": 1653
    },
    {
      "epoch": 0.32317311449785074,
      "grad_norm": 10.854321479797363,
      "learning_rate": 1.58686325882834e-05,
      "loss": 9.5224,
      "step": 1654
    },
    {
      "epoch": 0.32336850332161,
      "grad_norm": 12.20022201538086,
      "learning_rate": 1.586350711080809e-05,
      "loss": 9.9883,
      "step": 1655
    },
    {
      "epoch": 0.3235638921453693,
      "grad_norm": 9.082717895507812,
      "learning_rate": 1.5858379284816136e-05,
      "loss": 10.2387,
      "step": 1656
    },
    {
      "epoch": 0.3237592809691286,
      "grad_norm": 9.35047721862793,
      "learning_rate": 1.58532491123614e-05,
      "loss": 9.0864,
      "step": 1657
    },
    {
      "epoch": 0.32395466979288784,
      "grad_norm": 9.99332332611084,
      "learning_rate": 1.5848116595498676e-05,
      "loss": 9.0237,
      "step": 1658
    },
    {
      "epoch": 0.32415005861664714,
      "grad_norm": 8.102620124816895,
      "learning_rate": 1.5842981736283686e-05,
      "loss": 9.7664,
      "step": 1659
    },
    {
      "epoch": 0.3243454474404064,
      "grad_norm": 9.926274299621582,
      "learning_rate": 1.583784453677311e-05,
      "loss": 8.8169,
      "step": 1660
    },
    {
      "epoch": 0.3245408362641657,
      "grad_norm": 8.649129867553711,
      "learning_rate": 1.5832704999024552e-05,
      "loss": 8.4021,
      "step": 1661
    },
    {
      "epoch": 0.324736225087925,
      "grad_norm": 10.611956596374512,
      "learning_rate": 1.5827563125096557e-05,
      "loss": 10.3733,
      "step": 1662
    },
    {
      "epoch": 0.32493161391168424,
      "grad_norm": 8.192151069641113,
      "learning_rate": 1.582241891704861e-05,
      "loss": 9.454,
      "step": 1663
    },
    {
      "epoch": 0.32512700273544354,
      "grad_norm": 10.12881851196289,
      "learning_rate": 1.581727237694112e-05,
      "loss": 10.0913,
      "step": 1664
    },
    {
      "epoch": 0.32532239155920284,
      "grad_norm": 9.004416465759277,
      "learning_rate": 1.5812123506835435e-05,
      "loss": 9.4404,
      "step": 1665
    },
    {
      "epoch": 0.3255177803829621,
      "grad_norm": 13.841079711914062,
      "learning_rate": 1.580697230879384e-05,
      "loss": 9.7292,
      "step": 1666
    },
    {
      "epoch": 0.3257131692067214,
      "grad_norm": 12.088377952575684,
      "learning_rate": 1.580181878487955e-05,
      "loss": 9.7977,
      "step": 1667
    },
    {
      "epoch": 0.32590855803048063,
      "grad_norm": 7.993840217590332,
      "learning_rate": 1.5796662937156714e-05,
      "loss": 8.9673,
      "step": 1668
    },
    {
      "epoch": 0.32610394685423993,
      "grad_norm": 8.977736473083496,
      "learning_rate": 1.5791504767690403e-05,
      "loss": 9.04,
      "step": 1669
    },
    {
      "epoch": 0.32629933567799924,
      "grad_norm": 13.686468124389648,
      "learning_rate": 1.5786344278546627e-05,
      "loss": 10.4661,
      "step": 1670
    },
    {
      "epoch": 0.3264947245017585,
      "grad_norm": 7.914223670959473,
      "learning_rate": 1.5781181471792326e-05,
      "loss": 9.5493,
      "step": 1671
    },
    {
      "epoch": 0.3266901133255178,
      "grad_norm": 7.73126745223999,
      "learning_rate": 1.5776016349495355e-05,
      "loss": 10.191,
      "step": 1672
    },
    {
      "epoch": 0.3268855021492771,
      "grad_norm": 12.177936553955078,
      "learning_rate": 1.5770848913724516e-05,
      "loss": 10.3791,
      "step": 1673
    },
    {
      "epoch": 0.32708089097303633,
      "grad_norm": 20.568925857543945,
      "learning_rate": 1.576567916654952e-05,
      "loss": 9.4601,
      "step": 1674
    },
    {
      "epoch": 0.32727627979679563,
      "grad_norm": 12.07807731628418,
      "learning_rate": 1.5760507110041017e-05,
      "loss": 10.3394,
      "step": 1675
    },
    {
      "epoch": 0.3274716686205549,
      "grad_norm": 7.828347206115723,
      "learning_rate": 1.5755332746270574e-05,
      "loss": 9.282,
      "step": 1676
    },
    {
      "epoch": 0.3276670574443142,
      "grad_norm": 6.870009422302246,
      "learning_rate": 1.575015607731068e-05,
      "loss": 9.4014,
      "step": 1677
    },
    {
      "epoch": 0.3278624462680735,
      "grad_norm": 9.491081237792969,
      "learning_rate": 1.5744977105234756e-05,
      "loss": 10.2289,
      "step": 1678
    },
    {
      "epoch": 0.32805783509183273,
      "grad_norm": 8.813817977905273,
      "learning_rate": 1.5739795832117142e-05,
      "loss": 8.1194,
      "step": 1679
    },
    {
      "epoch": 0.32825322391559203,
      "grad_norm": 8.379867553710938,
      "learning_rate": 1.5734612260033094e-05,
      "loss": 9.8672,
      "step": 1680
    },
    {
      "epoch": 0.32844861273935133,
      "grad_norm": 13.194073677062988,
      "learning_rate": 1.57294263910588e-05,
      "loss": 9.9284,
      "step": 1681
    },
    {
      "epoch": 0.3286440015631106,
      "grad_norm": 6.913089275360107,
      "learning_rate": 1.572423822727136e-05,
      "loss": 9.1499,
      "step": 1682
    },
    {
      "epoch": 0.3288393903868699,
      "grad_norm": 11.585538864135742,
      "learning_rate": 1.571904777074879e-05,
      "loss": 9.3277,
      "step": 1683
    },
    {
      "epoch": 0.3290347792106291,
      "grad_norm": 10.432095527648926,
      "learning_rate": 1.5713855023570032e-05,
      "loss": 9.8317,
      "step": 1684
    },
    {
      "epoch": 0.3292301680343884,
      "grad_norm": 8.566866874694824,
      "learning_rate": 1.5708659987814945e-05,
      "loss": 8.5996,
      "step": 1685
    },
    {
      "epoch": 0.32942555685814773,
      "grad_norm": 18.440052032470703,
      "learning_rate": 1.5703462665564296e-05,
      "loss": 10.0485,
      "step": 1686
    },
    {
      "epoch": 0.329620945681907,
      "grad_norm": 65.54505920410156,
      "learning_rate": 1.5698263058899778e-05,
      "loss": 8.6768,
      "step": 1687
    },
    {
      "epoch": 0.3298163345056663,
      "grad_norm": 10.525182723999023,
      "learning_rate": 1.5693061169903998e-05,
      "loss": 9.9954,
      "step": 1688
    },
    {
      "epoch": 0.3300117233294256,
      "grad_norm": 11.578959465026855,
      "learning_rate": 1.5687857000660467e-05,
      "loss": 10.4881,
      "step": 1689
    },
    {
      "epoch": 0.3302071121531848,
      "grad_norm": 11.128236770629883,
      "learning_rate": 1.568265055325362e-05,
      "loss": 10.3662,
      "step": 1690
    },
    {
      "epoch": 0.3304025009769441,
      "grad_norm": 9.159039497375488,
      "learning_rate": 1.5677441829768802e-05,
      "loss": 9.982,
      "step": 1691
    },
    {
      "epoch": 0.3305978898007034,
      "grad_norm": 15.565377235412598,
      "learning_rate": 1.5672230832292264e-05,
      "loss": 10.0356,
      "step": 1692
    },
    {
      "epoch": 0.33079327862446267,
      "grad_norm": 12.125045776367188,
      "learning_rate": 1.566701756291118e-05,
      "loss": 9.6776,
      "step": 1693
    },
    {
      "epoch": 0.330988667448222,
      "grad_norm": 9.87791633605957,
      "learning_rate": 1.5661802023713618e-05,
      "loss": 10.6344,
      "step": 1694
    },
    {
      "epoch": 0.3311840562719812,
      "grad_norm": 8.393369674682617,
      "learning_rate": 1.5656584216788567e-05,
      "loss": 9.7636,
      "step": 1695
    },
    {
      "epoch": 0.3313794450957405,
      "grad_norm": 11.450444221496582,
      "learning_rate": 1.565136414422592e-05,
      "loss": 11.0457,
      "step": 1696
    },
    {
      "epoch": 0.3315748339194998,
      "grad_norm": 10.343286514282227,
      "learning_rate": 1.564614180811648e-05,
      "loss": 9.8457,
      "step": 1697
    },
    {
      "epoch": 0.33177022274325907,
      "grad_norm": 8.537278175354004,
      "learning_rate": 1.5640917210551955e-05,
      "loss": 9.4423,
      "step": 1698
    },
    {
      "epoch": 0.33196561156701837,
      "grad_norm": 16.415647506713867,
      "learning_rate": 1.5635690353624955e-05,
      "loss": 9.2602,
      "step": 1699
    },
    {
      "epoch": 0.33216100039077767,
      "grad_norm": 12.417532920837402,
      "learning_rate": 1.5630461239429e-05,
      "loss": 10.9,
      "step": 1700
    },
    {
      "epoch": 0.3323563892145369,
      "grad_norm": 6.117668628692627,
      "learning_rate": 1.5625229870058515e-05,
      "loss": 8.3,
      "step": 1701
    },
    {
      "epoch": 0.3325517780382962,
      "grad_norm": 8.208422660827637,
      "learning_rate": 1.561999624760882e-05,
      "loss": 9.441,
      "step": 1702
    },
    {
      "epoch": 0.33274716686205547,
      "grad_norm": 7.063049793243408,
      "learning_rate": 1.5614760374176153e-05,
      "loss": 9.227,
      "step": 1703
    },
    {
      "epoch": 0.33294255568581477,
      "grad_norm": 27.353660583496094,
      "learning_rate": 1.5609522251857634e-05,
      "loss": 10.7016,
      "step": 1704
    },
    {
      "epoch": 0.33313794450957407,
      "grad_norm": 8.364476203918457,
      "learning_rate": 1.5604281882751297e-05,
      "loss": 9.6392,
      "step": 1705
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 12.525715827941895,
      "learning_rate": 1.5599039268956072e-05,
      "loss": 10.7313,
      "step": 1706
    },
    {
      "epoch": 0.3335287221570926,
      "grad_norm": 10.193623542785645,
      "learning_rate": 1.5593794412571798e-05,
      "loss": 9.5809,
      "step": 1707
    },
    {
      "epoch": 0.3337241109808519,
      "grad_norm": 12.494501113891602,
      "learning_rate": 1.5588547315699188e-05,
      "loss": 9.9022,
      "step": 1708
    },
    {
      "epoch": 0.33391949980461116,
      "grad_norm": 13.320052146911621,
      "learning_rate": 1.5583297980439873e-05,
      "loss": 10.4568,
      "step": 1709
    },
    {
      "epoch": 0.33411488862837047,
      "grad_norm": 8.454721450805664,
      "learning_rate": 1.5578046408896378e-05,
      "loss": 9.1103,
      "step": 1710
    },
    {
      "epoch": 0.3343102774521297,
      "grad_norm": 10.274154663085938,
      "learning_rate": 1.5572792603172117e-05,
      "loss": 9.895,
      "step": 1711
    },
    {
      "epoch": 0.334505666275889,
      "grad_norm": 9.209165573120117,
      "learning_rate": 1.5567536565371404e-05,
      "loss": 10.0295,
      "step": 1712
    },
    {
      "epoch": 0.3347010550996483,
      "grad_norm": 57.516273498535156,
      "learning_rate": 1.5562278297599444e-05,
      "loss": 9.9245,
      "step": 1713
    },
    {
      "epoch": 0.33489644392340756,
      "grad_norm": 8.862438201904297,
      "learning_rate": 1.5557017801962334e-05,
      "loss": 9.4419,
      "step": 1714
    },
    {
      "epoch": 0.33509183274716686,
      "grad_norm": 8.271980285644531,
      "learning_rate": 1.5551755080567076e-05,
      "loss": 9.4185,
      "step": 1715
    },
    {
      "epoch": 0.33528722157092616,
      "grad_norm": 9.476627349853516,
      "learning_rate": 1.554649013552154e-05,
      "loss": 9.6879,
      "step": 1716
    },
    {
      "epoch": 0.3354826103946854,
      "grad_norm": 10.101987838745117,
      "learning_rate": 1.554122296893451e-05,
      "loss": 8.4631,
      "step": 1717
    },
    {
      "epoch": 0.3356779992184447,
      "grad_norm": 9.779273986816406,
      "learning_rate": 1.5535953582915643e-05,
      "loss": 9.814,
      "step": 1718
    },
    {
      "epoch": 0.335873388042204,
      "grad_norm": 12.370697021484375,
      "learning_rate": 1.5530681979575496e-05,
      "loss": 10.0133,
      "step": 1719
    },
    {
      "epoch": 0.33606877686596326,
      "grad_norm": 7.9293694496154785,
      "learning_rate": 1.5525408161025507e-05,
      "loss": 9.9495,
      "step": 1720
    },
    {
      "epoch": 0.33626416568972256,
      "grad_norm": 19.309507369995117,
      "learning_rate": 1.5520132129378006e-05,
      "loss": 9.9442,
      "step": 1721
    },
    {
      "epoch": 0.3364595545134818,
      "grad_norm": 7.228127956390381,
      "learning_rate": 1.551485388674621e-05,
      "loss": 9.0594,
      "step": 1722
    },
    {
      "epoch": 0.3366549433372411,
      "grad_norm": 18.292306900024414,
      "learning_rate": 1.5509573435244215e-05,
      "loss": 9.2468,
      "step": 1723
    },
    {
      "epoch": 0.3368503321610004,
      "grad_norm": 14.835461616516113,
      "learning_rate": 1.5504290776987005e-05,
      "loss": 10.1081,
      "step": 1724
    },
    {
      "epoch": 0.33704572098475966,
      "grad_norm": 9.334471702575684,
      "learning_rate": 1.5499005914090447e-05,
      "loss": 9.1263,
      "step": 1725
    },
    {
      "epoch": 0.33724110980851896,
      "grad_norm": 12.06333065032959,
      "learning_rate": 1.5493718848671302e-05,
      "loss": 8.5533,
      "step": 1726
    },
    {
      "epoch": 0.33743649863227826,
      "grad_norm": 10.127058029174805,
      "learning_rate": 1.5488429582847194e-05,
      "loss": 8.5659,
      "step": 1727
    },
    {
      "epoch": 0.3376318874560375,
      "grad_norm": 9.882468223571777,
      "learning_rate": 1.5483138118736643e-05,
      "loss": 8.3786,
      "step": 1728
    },
    {
      "epoch": 0.3378272762797968,
      "grad_norm": 9.601874351501465,
      "learning_rate": 1.5477844458459046e-05,
      "loss": 10.2354,
      "step": 1729
    },
    {
      "epoch": 0.33802266510355605,
      "grad_norm": 11.464741706848145,
      "learning_rate": 1.5472548604134675e-05,
      "loss": 8.6515,
      "step": 1730
    },
    {
      "epoch": 0.33821805392731535,
      "grad_norm": 9.604881286621094,
      "learning_rate": 1.5467250557884685e-05,
      "loss": 10.6031,
      "step": 1731
    },
    {
      "epoch": 0.33841344275107466,
      "grad_norm": 8.280776023864746,
      "learning_rate": 1.5461950321831105e-05,
      "loss": 10.2086,
      "step": 1732
    },
    {
      "epoch": 0.3386088315748339,
      "grad_norm": 10.27190113067627,
      "learning_rate": 1.545664789809685e-05,
      "loss": 9.7946,
      "step": 1733
    },
    {
      "epoch": 0.3388042203985932,
      "grad_norm": 8.053266525268555,
      "learning_rate": 1.54513432888057e-05,
      "loss": 9.3896,
      "step": 1734
    },
    {
      "epoch": 0.3389996092223525,
      "grad_norm": 9.985084533691406,
      "learning_rate": 1.544603649608232e-05,
      "loss": 10.0078,
      "step": 1735
    },
    {
      "epoch": 0.33919499804611175,
      "grad_norm": 10.035849571228027,
      "learning_rate": 1.5440727522052247e-05,
      "loss": 10.6757,
      "step": 1736
    },
    {
      "epoch": 0.33939038686987105,
      "grad_norm": 11.540842056274414,
      "learning_rate": 1.543541636884188e-05,
      "loss": 9.3436,
      "step": 1737
    },
    {
      "epoch": 0.3395857756936303,
      "grad_norm": 7.954980373382568,
      "learning_rate": 1.5430103038578513e-05,
      "loss": 9.0339,
      "step": 1738
    },
    {
      "epoch": 0.3397811645173896,
      "grad_norm": 19.364994049072266,
      "learning_rate": 1.542478753339029e-05,
      "loss": 8.4139,
      "step": 1739
    },
    {
      "epoch": 0.3399765533411489,
      "grad_norm": 8.921440124511719,
      "learning_rate": 1.5419469855406243e-05,
      "loss": 8.6741,
      "step": 1740
    },
    {
      "epoch": 0.34017194216490815,
      "grad_norm": 26.762006759643555,
      "learning_rate": 1.5414150006756262e-05,
      "loss": 10.0279,
      "step": 1741
    },
    {
      "epoch": 0.34036733098866745,
      "grad_norm": 7.822473526000977,
      "learning_rate": 1.540882798957111e-05,
      "loss": 9.2955,
      "step": 1742
    },
    {
      "epoch": 0.34056271981242675,
      "grad_norm": 9.464631080627441,
      "learning_rate": 1.5403503805982428e-05,
      "loss": 9.8347,
      "step": 1743
    },
    {
      "epoch": 0.340758108636186,
      "grad_norm": 8.8168363571167,
      "learning_rate": 1.539817745812271e-05,
      "loss": 10.2843,
      "step": 1744
    },
    {
      "epoch": 0.3409534974599453,
      "grad_norm": 9.104181289672852,
      "learning_rate": 1.539284894812533e-05,
      "loss": 9.0267,
      "step": 1745
    },
    {
      "epoch": 0.34114888628370454,
      "grad_norm": 9.172795295715332,
      "learning_rate": 1.5387518278124512e-05,
      "loss": 10.059,
      "step": 1746
    },
    {
      "epoch": 0.34134427510746385,
      "grad_norm": 8.707510948181152,
      "learning_rate": 1.5382185450255365e-05,
      "loss": 9.7586,
      "step": 1747
    },
    {
      "epoch": 0.34153966393122315,
      "grad_norm": 9.747553825378418,
      "learning_rate": 1.5376850466653844e-05,
      "loss": 9.5557,
      "step": 1748
    },
    {
      "epoch": 0.3417350527549824,
      "grad_norm": 10.045906066894531,
      "learning_rate": 1.537151332945678e-05,
      "loss": 9.9312,
      "step": 1749
    },
    {
      "epoch": 0.3419304415787417,
      "grad_norm": 11.068819046020508,
      "learning_rate": 1.5366174040801866e-05,
      "loss": 9.7978,
      "step": 1750
    },
    {
      "epoch": 0.342125830402501,
      "grad_norm": 11.23110580444336,
      "learning_rate": 1.5360832602827644e-05,
      "loss": 10.2003,
      "step": 1751
    },
    {
      "epoch": 0.34232121922626024,
      "grad_norm": 8.130640029907227,
      "learning_rate": 1.5355489017673534e-05,
      "loss": 9.0259,
      "step": 1752
    },
    {
      "epoch": 0.34251660805001954,
      "grad_norm": 9.70206069946289,
      "learning_rate": 1.5350143287479796e-05,
      "loss": 9.0877,
      "step": 1753
    },
    {
      "epoch": 0.34271199687377885,
      "grad_norm": 10.980212211608887,
      "learning_rate": 1.5344795414387575e-05,
      "loss": 10.7845,
      "step": 1754
    },
    {
      "epoch": 0.3429073856975381,
      "grad_norm": 9.177943229675293,
      "learning_rate": 1.5339445400538852e-05,
      "loss": 9.9836,
      "step": 1755
    },
    {
      "epoch": 0.3431027745212974,
      "grad_norm": 8.501229286193848,
      "learning_rate": 1.5334093248076472e-05,
      "loss": 9.2696,
      "step": 1756
    },
    {
      "epoch": 0.34329816334505664,
      "grad_norm": 10.649053573608398,
      "learning_rate": 1.5328738959144146e-05,
      "loss": 8.7284,
      "step": 1757
    },
    {
      "epoch": 0.34349355216881594,
      "grad_norm": 9.538617134094238,
      "learning_rate": 1.532338253588643e-05,
      "loss": 9.8119,
      "step": 1758
    },
    {
      "epoch": 0.34368894099257524,
      "grad_norm": 7.793221473693848,
      "learning_rate": 1.5318023980448735e-05,
      "loss": 9.6133,
      "step": 1759
    },
    {
      "epoch": 0.3438843298163345,
      "grad_norm": 7.8072590827941895,
      "learning_rate": 1.531266329497733e-05,
      "loss": 9.7504,
      "step": 1760
    },
    {
      "epoch": 0.3440797186400938,
      "grad_norm": 7.636564254760742,
      "learning_rate": 1.5307300481619334e-05,
      "loss": 9.3218,
      "step": 1761
    },
    {
      "epoch": 0.3442751074638531,
      "grad_norm": 6.859006404876709,
      "learning_rate": 1.5301935542522725e-05,
      "loss": 9.4971,
      "step": 1762
    },
    {
      "epoch": 0.34447049628761234,
      "grad_norm": 7.057887554168701,
      "learning_rate": 1.5296568479836326e-05,
      "loss": 8.7375,
      "step": 1763
    },
    {
      "epoch": 0.34466588511137164,
      "grad_norm": 10.694537162780762,
      "learning_rate": 1.5291199295709808e-05,
      "loss": 9.8011,
      "step": 1764
    },
    {
      "epoch": 0.3448612739351309,
      "grad_norm": 7.027894973754883,
      "learning_rate": 1.5285827992293703e-05,
      "loss": 9.0979,
      "step": 1765
    },
    {
      "epoch": 0.3450566627588902,
      "grad_norm": 9.455521583557129,
      "learning_rate": 1.528045457173938e-05,
      "loss": 9.0439,
      "step": 1766
    },
    {
      "epoch": 0.3452520515826495,
      "grad_norm": 8.299240112304688,
      "learning_rate": 1.5275079036199062e-05,
      "loss": 9.5298,
      "step": 1767
    },
    {
      "epoch": 0.34544744040640873,
      "grad_norm": 192.6204071044922,
      "learning_rate": 1.5269701387825815e-05,
      "loss": 9.5424,
      "step": 1768
    },
    {
      "epoch": 0.34564282923016804,
      "grad_norm": 35.3787956237793,
      "learning_rate": 1.526432162877356e-05,
      "loss": 9.8712,
      "step": 1769
    },
    {
      "epoch": 0.34583821805392734,
      "grad_norm": 12.05440616607666,
      "learning_rate": 1.525893976119705e-05,
      "loss": 9.401,
      "step": 1770
    },
    {
      "epoch": 0.3460336068776866,
      "grad_norm": 16.06743621826172,
      "learning_rate": 1.5253555787251899e-05,
      "loss": 9.285,
      "step": 1771
    },
    {
      "epoch": 0.3462289957014459,
      "grad_norm": 9.969664573669434,
      "learning_rate": 1.5248169709094551e-05,
      "loss": 9.3563,
      "step": 1772
    },
    {
      "epoch": 0.34642438452520513,
      "grad_norm": 43.8010139465332,
      "learning_rate": 1.5242781528882296e-05,
      "loss": 10.163,
      "step": 1773
    },
    {
      "epoch": 0.34661977334896443,
      "grad_norm": 10.185831069946289,
      "learning_rate": 1.5237391248773273e-05,
      "loss": 10.1241,
      "step": 1774
    },
    {
      "epoch": 0.34681516217272373,
      "grad_norm": 58.117881774902344,
      "learning_rate": 1.5231998870926448e-05,
      "loss": 10.2288,
      "step": 1775
    },
    {
      "epoch": 0.347010550996483,
      "grad_norm": 9.52348804473877,
      "learning_rate": 1.522660439750164e-05,
      "loss": 8.9495,
      "step": 1776
    },
    {
      "epoch": 0.3472059398202423,
      "grad_norm": 18.378759384155273,
      "learning_rate": 1.5221207830659506e-05,
      "loss": 10.8928,
      "step": 1777
    },
    {
      "epoch": 0.3474013286440016,
      "grad_norm": 12.922347068786621,
      "learning_rate": 1.521580917256154e-05,
      "loss": 10.3628,
      "step": 1778
    },
    {
      "epoch": 0.34759671746776083,
      "grad_norm": 16.34457778930664,
      "learning_rate": 1.5210408425370062e-05,
      "loss": 10.6221,
      "step": 1779
    },
    {
      "epoch": 0.34779210629152013,
      "grad_norm": 7.653688907623291,
      "learning_rate": 1.5205005591248248e-05,
      "loss": 8.9949,
      "step": 1780
    },
    {
      "epoch": 0.34798749511527943,
      "grad_norm": 10.004982948303223,
      "learning_rate": 1.5199600672360096e-05,
      "loss": 8.9597,
      "step": 1781
    },
    {
      "epoch": 0.3481828839390387,
      "grad_norm": 12.164095878601074,
      "learning_rate": 1.5194193670870447e-05,
      "loss": 9.9517,
      "step": 1782
    },
    {
      "epoch": 0.348378272762798,
      "grad_norm": 10.798413276672363,
      "learning_rate": 1.5188784588944972e-05,
      "loss": 9.9719,
      "step": 1783
    },
    {
      "epoch": 0.3485736615865572,
      "grad_norm": 10.619060516357422,
      "learning_rate": 1.5183373428750177e-05,
      "loss": 8.9537,
      "step": 1784
    },
    {
      "epoch": 0.34876905041031653,
      "grad_norm": 27.77444839477539,
      "learning_rate": 1.5177960192453399e-05,
      "loss": 10.713,
      "step": 1785
    },
    {
      "epoch": 0.34896443923407583,
      "grad_norm": 12.40777587890625,
      "learning_rate": 1.5172544882222804e-05,
      "loss": 9.9726,
      "step": 1786
    },
    {
      "epoch": 0.3491598280578351,
      "grad_norm": 8.204813003540039,
      "learning_rate": 1.5167127500227397e-05,
      "loss": 9.753,
      "step": 1787
    },
    {
      "epoch": 0.3493552168815944,
      "grad_norm": 10.18585205078125,
      "learning_rate": 1.5161708048637002e-05,
      "loss": 9.7292,
      "step": 1788
    },
    {
      "epoch": 0.3495506057053537,
      "grad_norm": 9.334511756896973,
      "learning_rate": 1.5156286529622284e-05,
      "loss": 9.4344,
      "step": 1789
    },
    {
      "epoch": 0.3497459945291129,
      "grad_norm": 12.067798614501953,
      "learning_rate": 1.5150862945354728e-05,
      "loss": 10.4729,
      "step": 1790
    },
    {
      "epoch": 0.3499413833528722,
      "grad_norm": 9.017687797546387,
      "learning_rate": 1.5145437298006642e-05,
      "loss": 9.6782,
      "step": 1791
    },
    {
      "epoch": 0.35013677217663147,
      "grad_norm": 8.172805786132812,
      "learning_rate": 1.5140009589751174e-05,
      "loss": 9.9983,
      "step": 1792
    },
    {
      "epoch": 0.3503321610003908,
      "grad_norm": 13.06108570098877,
      "learning_rate": 1.5134579822762287e-05,
      "loss": 9.3318,
      "step": 1793
    },
    {
      "epoch": 0.3505275498241501,
      "grad_norm": 8.837092399597168,
      "learning_rate": 1.512914799921477e-05,
      "loss": 10.2875,
      "step": 1794
    },
    {
      "epoch": 0.3507229386479093,
      "grad_norm": 7.7178263664245605,
      "learning_rate": 1.512371412128424e-05,
      "loss": 9.5769,
      "step": 1795
    },
    {
      "epoch": 0.3509183274716686,
      "grad_norm": 7.829312324523926,
      "learning_rate": 1.5118278191147129e-05,
      "loss": 9.539,
      "step": 1796
    },
    {
      "epoch": 0.3511137162954279,
      "grad_norm": 9.15108585357666,
      "learning_rate": 1.5112840210980698e-05,
      "loss": 9.4839,
      "step": 1797
    },
    {
      "epoch": 0.35130910511918717,
      "grad_norm": 9.459131240844727,
      "learning_rate": 1.5107400182963025e-05,
      "loss": 10.9973,
      "step": 1798
    },
    {
      "epoch": 0.35150449394294647,
      "grad_norm": 25.22171401977539,
      "learning_rate": 1.5101958109273016e-05,
      "loss": 10.1825,
      "step": 1799
    },
    {
      "epoch": 0.3516998827667057,
      "grad_norm": 10.554967880249023,
      "learning_rate": 1.5096513992090386e-05,
      "loss": 9.6381,
      "step": 1800
    },
    {
      "epoch": 0.351895271590465,
      "grad_norm": 9.855436325073242,
      "learning_rate": 1.5091067833595672e-05,
      "loss": 10.3401,
      "step": 1801
    },
    {
      "epoch": 0.3520906604142243,
      "grad_norm": 7.670804500579834,
      "learning_rate": 1.5085619635970235e-05,
      "loss": 8.7629,
      "step": 1802
    },
    {
      "epoch": 0.35228604923798357,
      "grad_norm": 12.818094253540039,
      "learning_rate": 1.508016940139624e-05,
      "loss": 10.2233,
      "step": 1803
    },
    {
      "epoch": 0.35248143806174287,
      "grad_norm": 9.649227142333984,
      "learning_rate": 1.5074717132056678e-05,
      "loss": 9.2624,
      "step": 1804
    },
    {
      "epoch": 0.35267682688550217,
      "grad_norm": 9.296341896057129,
      "learning_rate": 1.5069262830135356e-05,
      "loss": 9.6492,
      "step": 1805
    },
    {
      "epoch": 0.3528722157092614,
      "grad_norm": 16.60502815246582,
      "learning_rate": 1.5063806497816887e-05,
      "loss": 9.9349,
      "step": 1806
    },
    {
      "epoch": 0.3530676045330207,
      "grad_norm": 23.677478790283203,
      "learning_rate": 1.5058348137286704e-05,
      "loss": 10.1734,
      "step": 1807
    },
    {
      "epoch": 0.35326299335678,
      "grad_norm": 9.695001602172852,
      "learning_rate": 1.5052887750731046e-05,
      "loss": 9.4951,
      "step": 1808
    },
    {
      "epoch": 0.35345838218053927,
      "grad_norm": 8.7704439163208,
      "learning_rate": 1.5047425340336972e-05,
      "loss": 8.7867,
      "step": 1809
    },
    {
      "epoch": 0.35365377100429857,
      "grad_norm": 8.951241493225098,
      "learning_rate": 1.5041960908292343e-05,
      "loss": 8.2299,
      "step": 1810
    },
    {
      "epoch": 0.3538491598280578,
      "grad_norm": 9.119094848632812,
      "learning_rate": 1.5036494456785837e-05,
      "loss": 9.7607,
      "step": 1811
    },
    {
      "epoch": 0.3540445486518171,
      "grad_norm": 13.40417194366455,
      "learning_rate": 1.5031025988006935e-05,
      "loss": 10.2292,
      "step": 1812
    },
    {
      "epoch": 0.3542399374755764,
      "grad_norm": 19.00713348388672,
      "learning_rate": 1.502555550414593e-05,
      "loss": 10.1842,
      "step": 1813
    },
    {
      "epoch": 0.35443532629933566,
      "grad_norm": 12.991162300109863,
      "learning_rate": 1.5020083007393919e-05,
      "loss": 10.2614,
      "step": 1814
    },
    {
      "epoch": 0.35463071512309496,
      "grad_norm": 9.391922950744629,
      "learning_rate": 1.5014608499942808e-05,
      "loss": 9.3905,
      "step": 1815
    },
    {
      "epoch": 0.35482610394685427,
      "grad_norm": 9.280734062194824,
      "learning_rate": 1.5009131983985308e-05,
      "loss": 9.9366,
      "step": 1816
    },
    {
      "epoch": 0.3550214927706135,
      "grad_norm": 7.751325607299805,
      "learning_rate": 1.5003653461714934e-05,
      "loss": 9.3776,
      "step": 1817
    },
    {
      "epoch": 0.3552168815943728,
      "grad_norm": 7.636277198791504,
      "learning_rate": 1.4998172935326003e-05,
      "loss": 9.1454,
      "step": 1818
    },
    {
      "epoch": 0.35541227041813206,
      "grad_norm": 8.686695098876953,
      "learning_rate": 1.4992690407013635e-05,
      "loss": 9.4679,
      "step": 1819
    },
    {
      "epoch": 0.35560765924189136,
      "grad_norm": 9.64791488647461,
      "learning_rate": 1.4987205878973755e-05,
      "loss": 9.3872,
      "step": 1820
    },
    {
      "epoch": 0.35580304806565066,
      "grad_norm": 11.299792289733887,
      "learning_rate": 1.4981719353403085e-05,
      "loss": 10.114,
      "step": 1821
    },
    {
      "epoch": 0.3559984368894099,
      "grad_norm": 12.038385391235352,
      "learning_rate": 1.497623083249915e-05,
      "loss": 8.477,
      "step": 1822
    },
    {
      "epoch": 0.3561938257131692,
      "grad_norm": 9.453662872314453,
      "learning_rate": 1.4970740318460271e-05,
      "loss": 9.853,
      "step": 1823
    },
    {
      "epoch": 0.3563892145369285,
      "grad_norm": 9.599242210388184,
      "learning_rate": 1.4965247813485571e-05,
      "loss": 9.8869,
      "step": 1824
    },
    {
      "epoch": 0.35658460336068776,
      "grad_norm": 11.355889320373535,
      "learning_rate": 1.4959753319774968e-05,
      "loss": 9.2784,
      "step": 1825
    },
    {
      "epoch": 0.35677999218444706,
      "grad_norm": 11.106788635253906,
      "learning_rate": 1.495425683952918e-05,
      "loss": 10.4052,
      "step": 1826
    },
    {
      "epoch": 0.3569753810082063,
      "grad_norm": 8.836578369140625,
      "learning_rate": 1.4948758374949715e-05,
      "loss": 8.5516,
      "step": 1827
    },
    {
      "epoch": 0.3571707698319656,
      "grad_norm": 11.409369468688965,
      "learning_rate": 1.4943257928238879e-05,
      "loss": 9.7967,
      "step": 1828
    },
    {
      "epoch": 0.3573661586557249,
      "grad_norm": 7.801117420196533,
      "learning_rate": 1.4937755501599774e-05,
      "loss": 8.9674,
      "step": 1829
    },
    {
      "epoch": 0.35756154747948415,
      "grad_norm": 8.499917984008789,
      "learning_rate": 1.493225109723629e-05,
      "loss": 9.0892,
      "step": 1830
    },
    {
      "epoch": 0.35775693630324346,
      "grad_norm": 10.434982299804688,
      "learning_rate": 1.4926744717353112e-05,
      "loss": 9.8643,
      "step": 1831
    },
    {
      "epoch": 0.35795232512700276,
      "grad_norm": 9.691671371459961,
      "learning_rate": 1.4921236364155717e-05,
      "loss": 9.6132,
      "step": 1832
    },
    {
      "epoch": 0.358147713950762,
      "grad_norm": 7.935885429382324,
      "learning_rate": 1.4915726039850376e-05,
      "loss": 8.4978,
      "step": 1833
    },
    {
      "epoch": 0.3583431027745213,
      "grad_norm": 51.44169998168945,
      "learning_rate": 1.4910213746644136e-05,
      "loss": 9.921,
      "step": 1834
    },
    {
      "epoch": 0.35853849159828055,
      "grad_norm": 10.459129333496094,
      "learning_rate": 1.4904699486744848e-05,
      "loss": 9.5983,
      "step": 1835
    },
    {
      "epoch": 0.35873388042203985,
      "grad_norm": 9.238507270812988,
      "learning_rate": 1.489918326236114e-05,
      "loss": 8.4888,
      "step": 1836
    },
    {
      "epoch": 0.35892926924579915,
      "grad_norm": 17.272464752197266,
      "learning_rate": 1.4893665075702435e-05,
      "loss": 9.608,
      "step": 1837
    },
    {
      "epoch": 0.3591246580695584,
      "grad_norm": 8.849822044372559,
      "learning_rate": 1.4888144928978936e-05,
      "loss": 9.3512,
      "step": 1838
    },
    {
      "epoch": 0.3593200468933177,
      "grad_norm": 8.948539733886719,
      "learning_rate": 1.4882622824401632e-05,
      "loss": 8.9959,
      "step": 1839
    },
    {
      "epoch": 0.359515435717077,
      "grad_norm": 8.089892387390137,
      "learning_rate": 1.4877098764182301e-05,
      "loss": 9.1757,
      "step": 1840
    },
    {
      "epoch": 0.35971082454083625,
      "grad_norm": 9.682193756103516,
      "learning_rate": 1.4871572750533499e-05,
      "loss": 10.1555,
      "step": 1841
    },
    {
      "epoch": 0.35990621336459555,
      "grad_norm": 10.889745712280273,
      "learning_rate": 1.4866044785668563e-05,
      "loss": 8.9293,
      "step": 1842
    },
    {
      "epoch": 0.36010160218835485,
      "grad_norm": 8.232142448425293,
      "learning_rate": 1.486051487180162e-05,
      "loss": 9.5754,
      "step": 1843
    },
    {
      "epoch": 0.3602969910121141,
      "grad_norm": 8.604398727416992,
      "learning_rate": 1.4854983011147568e-05,
      "loss": 10.1241,
      "step": 1844
    },
    {
      "epoch": 0.3604923798358734,
      "grad_norm": 11.409536361694336,
      "learning_rate": 1.4849449205922088e-05,
      "loss": 10.0654,
      "step": 1845
    },
    {
      "epoch": 0.36068776865963265,
      "grad_norm": 12.24126148223877,
      "learning_rate": 1.4843913458341646e-05,
      "loss": 10.1095,
      "step": 1846
    },
    {
      "epoch": 0.36088315748339195,
      "grad_norm": 8.643593788146973,
      "learning_rate": 1.483837577062348e-05,
      "loss": 9.2632,
      "step": 1847
    },
    {
      "epoch": 0.36107854630715125,
      "grad_norm": 8.110797882080078,
      "learning_rate": 1.4832836144985601e-05,
      "loss": 8.6348,
      "step": 1848
    },
    {
      "epoch": 0.3612739351309105,
      "grad_norm": 10.314324378967285,
      "learning_rate": 1.4827294583646804e-05,
      "loss": 8.5782,
      "step": 1849
    },
    {
      "epoch": 0.3614693239546698,
      "grad_norm": 9.651543617248535,
      "learning_rate": 1.4821751088826655e-05,
      "loss": 9.8624,
      "step": 1850
    },
    {
      "epoch": 0.3616647127784291,
      "grad_norm": 9.259075164794922,
      "learning_rate": 1.4816205662745498e-05,
      "loss": 9.2482,
      "step": 1851
    },
    {
      "epoch": 0.36186010160218834,
      "grad_norm": 11.75296688079834,
      "learning_rate": 1.4810658307624445e-05,
      "loss": 8.6998,
      "step": 1852
    },
    {
      "epoch": 0.36205549042594765,
      "grad_norm": 12.283703804016113,
      "learning_rate": 1.4805109025685391e-05,
      "loss": 10.5427,
      "step": 1853
    },
    {
      "epoch": 0.3622508792497069,
      "grad_norm": 8.814277648925781,
      "learning_rate": 1.479955781915099e-05,
      "loss": 10.1233,
      "step": 1854
    },
    {
      "epoch": 0.3624462680734662,
      "grad_norm": 8.639643669128418,
      "learning_rate": 1.479400469024467e-05,
      "loss": 9.5759,
      "step": 1855
    },
    {
      "epoch": 0.3626416568972255,
      "grad_norm": 10.140213012695312,
      "learning_rate": 1.4788449641190637e-05,
      "loss": 10.0116,
      "step": 1856
    },
    {
      "epoch": 0.36283704572098474,
      "grad_norm": 9.60897445678711,
      "learning_rate": 1.4782892674213855e-05,
      "loss": 9.7824,
      "step": 1857
    },
    {
      "epoch": 0.36303243454474404,
      "grad_norm": 11.40339469909668,
      "learning_rate": 1.477733379154007e-05,
      "loss": 9.0601,
      "step": 1858
    },
    {
      "epoch": 0.36322782336850334,
      "grad_norm": 10.77801513671875,
      "learning_rate": 1.4771772995395779e-05,
      "loss": 8.0438,
      "step": 1859
    },
    {
      "epoch": 0.3634232121922626,
      "grad_norm": 10.345446586608887,
      "learning_rate": 1.476621028800826e-05,
      "loss": 9.349,
      "step": 1860
    },
    {
      "epoch": 0.3636186010160219,
      "grad_norm": 10.751293182373047,
      "learning_rate": 1.4760645671605545e-05,
      "loss": 9.2973,
      "step": 1861
    },
    {
      "epoch": 0.36381398983978114,
      "grad_norm": 12.415898323059082,
      "learning_rate": 1.475507914841644e-05,
      "loss": 9.0775,
      "step": 1862
    },
    {
      "epoch": 0.36400937866354044,
      "grad_norm": 9.695807456970215,
      "learning_rate": 1.4749510720670506e-05,
      "loss": 9.8121,
      "step": 1863
    },
    {
      "epoch": 0.36420476748729974,
      "grad_norm": 14.2394437789917,
      "learning_rate": 1.4743940390598072e-05,
      "loss": 10.325,
      "step": 1864
    },
    {
      "epoch": 0.364400156311059,
      "grad_norm": 10.003837585449219,
      "learning_rate": 1.4738368160430229e-05,
      "loss": 8.9347,
      "step": 1865
    },
    {
      "epoch": 0.3645955451348183,
      "grad_norm": 9.896443367004395,
      "learning_rate": 1.4732794032398832e-05,
      "loss": 9.6103,
      "step": 1866
    },
    {
      "epoch": 0.3647909339585776,
      "grad_norm": 8.426640510559082,
      "learning_rate": 1.4727218008736485e-05,
      "loss": 8.767,
      "step": 1867
    },
    {
      "epoch": 0.36498632278233684,
      "grad_norm": 8.99334716796875,
      "learning_rate": 1.4721640091676566e-05,
      "loss": 9.5023,
      "step": 1868
    },
    {
      "epoch": 0.36518171160609614,
      "grad_norm": 10.854475021362305,
      "learning_rate": 1.4716060283453198e-05,
      "loss": 9.2807,
      "step": 1869
    },
    {
      "epoch": 0.36537710042985544,
      "grad_norm": 10.936485290527344,
      "learning_rate": 1.4710478586301268e-05,
      "loss": 8.7673,
      "step": 1870
    },
    {
      "epoch": 0.3655724892536147,
      "grad_norm": 10.797525405883789,
      "learning_rate": 1.4704895002456424e-05,
      "loss": 9.6278,
      "step": 1871
    },
    {
      "epoch": 0.365767878077374,
      "grad_norm": 10.396178245544434,
      "learning_rate": 1.469930953415506e-05,
      "loss": 9.7169,
      "step": 1872
    },
    {
      "epoch": 0.36596326690113323,
      "grad_norm": 35.18453598022461,
      "learning_rate": 1.4693722183634332e-05,
      "loss": 9.0131,
      "step": 1873
    },
    {
      "epoch": 0.36615865572489253,
      "grad_norm": 11.25657844543457,
      "learning_rate": 1.4688132953132145e-05,
      "loss": 10.948,
      "step": 1874
    },
    {
      "epoch": 0.36635404454865184,
      "grad_norm": 11.835681915283203,
      "learning_rate": 1.4682541844887163e-05,
      "loss": 9.5515,
      "step": 1875
    },
    {
      "epoch": 0.3665494333724111,
      "grad_norm": 8.967255592346191,
      "learning_rate": 1.4676948861138793e-05,
      "loss": 9.7112,
      "step": 1876
    },
    {
      "epoch": 0.3667448221961704,
      "grad_norm": 12.371103286743164,
      "learning_rate": 1.4671354004127203e-05,
      "loss": 9.8506,
      "step": 1877
    },
    {
      "epoch": 0.3669402110199297,
      "grad_norm": 10.877632141113281,
      "learning_rate": 1.4665757276093306e-05,
      "loss": 10.1536,
      "step": 1878
    },
    {
      "epoch": 0.36713559984368893,
      "grad_norm": 12.124642372131348,
      "learning_rate": 1.4660158679278764e-05,
      "loss": 10.031,
      "step": 1879
    },
    {
      "epoch": 0.36733098866744823,
      "grad_norm": 9.705178260803223,
      "learning_rate": 1.465455821592599e-05,
      "loss": 9.7276,
      "step": 1880
    },
    {
      "epoch": 0.3675263774912075,
      "grad_norm": 9.293903350830078,
      "learning_rate": 1.4648955888278144e-05,
      "loss": 9.5486,
      "step": 1881
    },
    {
      "epoch": 0.3677217663149668,
      "grad_norm": 8.219917297363281,
      "learning_rate": 1.464335169857913e-05,
      "loss": 9.5627,
      "step": 1882
    },
    {
      "epoch": 0.3679171551387261,
      "grad_norm": 9.594404220581055,
      "learning_rate": 1.4637745649073602e-05,
      "loss": 9.9368,
      "step": 1883
    },
    {
      "epoch": 0.36811254396248533,
      "grad_norm": 18.886594772338867,
      "learning_rate": 1.4632137742006955e-05,
      "loss": 9.9665,
      "step": 1884
    },
    {
      "epoch": 0.36830793278624463,
      "grad_norm": 8.912412643432617,
      "learning_rate": 1.4626527979625334e-05,
      "loss": 9.4875,
      "step": 1885
    },
    {
      "epoch": 0.36850332161000393,
      "grad_norm": 8.886992454528809,
      "learning_rate": 1.4620916364175617e-05,
      "loss": 9.886,
      "step": 1886
    },
    {
      "epoch": 0.3686987104337632,
      "grad_norm": 9.128472328186035,
      "learning_rate": 1.4615302897905435e-05,
      "loss": 8.9634,
      "step": 1887
    },
    {
      "epoch": 0.3688940992575225,
      "grad_norm": 8.628862380981445,
      "learning_rate": 1.4609687583063155e-05,
      "loss": 9.6416,
      "step": 1888
    },
    {
      "epoch": 0.3690894880812817,
      "grad_norm": 10.561972618103027,
      "learning_rate": 1.4604070421897883e-05,
      "loss": 10.5866,
      "step": 1889
    },
    {
      "epoch": 0.369284876905041,
      "grad_norm": 7.685659885406494,
      "learning_rate": 1.4598451416659468e-05,
      "loss": 8.9428,
      "step": 1890
    },
    {
      "epoch": 0.36948026572880033,
      "grad_norm": 13.756418228149414,
      "learning_rate": 1.459283056959849e-05,
      "loss": 9.4792,
      "step": 1891
    },
    {
      "epoch": 0.3696756545525596,
      "grad_norm": 9.282796859741211,
      "learning_rate": 1.4587207882966283e-05,
      "loss": 9.8491,
      "step": 1892
    },
    {
      "epoch": 0.3698710433763189,
      "grad_norm": 9.124377250671387,
      "learning_rate": 1.45815833590149e-05,
      "loss": 9.6051,
      "step": 1893
    },
    {
      "epoch": 0.3700664322000782,
      "grad_norm": 9.004120826721191,
      "learning_rate": 1.4575956999997144e-05,
      "loss": 8.636,
      "step": 1894
    },
    {
      "epoch": 0.3702618210238374,
      "grad_norm": 8.152937889099121,
      "learning_rate": 1.4570328808166539e-05,
      "loss": 8.6843,
      "step": 1895
    },
    {
      "epoch": 0.3704572098475967,
      "grad_norm": 9.83988094329834,
      "learning_rate": 1.4564698785777353e-05,
      "loss": 9.4453,
      "step": 1896
    },
    {
      "epoch": 0.370652598671356,
      "grad_norm": 9.798221588134766,
      "learning_rate": 1.455906693508459e-05,
      "loss": 9.9185,
      "step": 1897
    },
    {
      "epoch": 0.37084798749511527,
      "grad_norm": 8.114480018615723,
      "learning_rate": 1.4553433258343974e-05,
      "loss": 9.3029,
      "step": 1898
    },
    {
      "epoch": 0.3710433763188746,
      "grad_norm": 10.6480712890625,
      "learning_rate": 1.454779775781197e-05,
      "loss": 10.0314,
      "step": 1899
    },
    {
      "epoch": 0.3712387651426338,
      "grad_norm": 9.375577926635742,
      "learning_rate": 1.4542160435745771e-05,
      "loss": 9.2519,
      "step": 1900
    },
    {
      "epoch": 0.3714341539663931,
      "grad_norm": 8.865910530090332,
      "learning_rate": 1.45365212944033e-05,
      "loss": 9.4088,
      "step": 1901
    },
    {
      "epoch": 0.3716295427901524,
      "grad_norm": 8.14194107055664,
      "learning_rate": 1.4530880336043212e-05,
      "loss": 8.474,
      "step": 1902
    },
    {
      "epoch": 0.37182493161391167,
      "grad_norm": 10.09233570098877,
      "learning_rate": 1.4525237562924879e-05,
      "loss": 9.8282,
      "step": 1903
    },
    {
      "epoch": 0.37202032043767097,
      "grad_norm": 7.718667507171631,
      "learning_rate": 1.4519592977308409e-05,
      "loss": 8.3733,
      "step": 1904
    },
    {
      "epoch": 0.37221570926143027,
      "grad_norm": 13.187790870666504,
      "learning_rate": 1.4513946581454634e-05,
      "loss": 10.4114,
      "step": 1905
    },
    {
      "epoch": 0.3724110980851895,
      "grad_norm": 7.69195556640625,
      "learning_rate": 1.4508298377625112e-05,
      "loss": 8.8665,
      "step": 1906
    },
    {
      "epoch": 0.3726064869089488,
      "grad_norm": 9.133148193359375,
      "learning_rate": 1.4502648368082123e-05,
      "loss": 8.5269,
      "step": 1907
    },
    {
      "epoch": 0.37280187573270807,
      "grad_norm": 14.639871597290039,
      "learning_rate": 1.4496996555088672e-05,
      "loss": 10.605,
      "step": 1908
    },
    {
      "epoch": 0.37299726455646737,
      "grad_norm": 10.028745651245117,
      "learning_rate": 1.4491342940908487e-05,
      "loss": 10.1821,
      "step": 1909
    },
    {
      "epoch": 0.37319265338022667,
      "grad_norm": 14.973143577575684,
      "learning_rate": 1.448568752780601e-05,
      "loss": 10.1627,
      "step": 1910
    },
    {
      "epoch": 0.3733880422039859,
      "grad_norm": 8.591376304626465,
      "learning_rate": 1.4480030318046417e-05,
      "loss": 8.6355,
      "step": 1911
    },
    {
      "epoch": 0.3735834310277452,
      "grad_norm": 9.579111099243164,
      "learning_rate": 1.447437131389559e-05,
      "loss": 9.7195,
      "step": 1912
    },
    {
      "epoch": 0.3737788198515045,
      "grad_norm": 8.835308074951172,
      "learning_rate": 1.446871051762014e-05,
      "loss": 8.7154,
      "step": 1913
    },
    {
      "epoch": 0.37397420867526376,
      "grad_norm": 10.81924057006836,
      "learning_rate": 1.4463047931487391e-05,
      "loss": 10.5608,
      "step": 1914
    },
    {
      "epoch": 0.37416959749902307,
      "grad_norm": 9.480253219604492,
      "learning_rate": 1.4457383557765385e-05,
      "loss": 9.7263,
      "step": 1915
    },
    {
      "epoch": 0.3743649863227823,
      "grad_norm": 16.763931274414062,
      "learning_rate": 1.4451717398722875e-05,
      "loss": 9.2355,
      "step": 1916
    },
    {
      "epoch": 0.3745603751465416,
      "grad_norm": 13.234987258911133,
      "learning_rate": 1.4446049456629339e-05,
      "loss": 10.6605,
      "step": 1917
    },
    {
      "epoch": 0.3747557639703009,
      "grad_norm": 8.773869514465332,
      "learning_rate": 1.4440379733754961e-05,
      "loss": 9.5099,
      "step": 1918
    },
    {
      "epoch": 0.37495115279406016,
      "grad_norm": 27.581687927246094,
      "learning_rate": 1.4434708232370642e-05,
      "loss": 10.4511,
      "step": 1919
    },
    {
      "epoch": 0.37514654161781946,
      "grad_norm": 10.067249298095703,
      "learning_rate": 1.4429034954747996e-05,
      "loss": 10.0346,
      "step": 1920
    },
    {
      "epoch": 0.37534193044157876,
      "grad_norm": 10.091471672058105,
      "learning_rate": 1.4423359903159342e-05,
      "loss": 9.173,
      "step": 1921
    },
    {
      "epoch": 0.375537319265338,
      "grad_norm": 7.579031467437744,
      "learning_rate": 1.441768307987772e-05,
      "loss": 9.0195,
      "step": 1922
    },
    {
      "epoch": 0.3757327080890973,
      "grad_norm": 8.659554481506348,
      "learning_rate": 1.4412004487176872e-05,
      "loss": 9.3103,
      "step": 1923
    },
    {
      "epoch": 0.37592809691285656,
      "grad_norm": 14.094969749450684,
      "learning_rate": 1.4406324127331246e-05,
      "loss": 10.1186,
      "step": 1924
    },
    {
      "epoch": 0.37612348573661586,
      "grad_norm": 9.010276794433594,
      "learning_rate": 1.440064200261601e-05,
      "loss": 10.1677,
      "step": 1925
    },
    {
      "epoch": 0.37631887456037516,
      "grad_norm": 7.850002288818359,
      "learning_rate": 1.4394958115307027e-05,
      "loss": 9.3536,
      "step": 1926
    },
    {
      "epoch": 0.3765142633841344,
      "grad_norm": 11.411676406860352,
      "learning_rate": 1.4389272467680871e-05,
      "loss": 10.2704,
      "step": 1927
    },
    {
      "epoch": 0.3767096522078937,
      "grad_norm": 8.324441909790039,
      "learning_rate": 1.4383585062014817e-05,
      "loss": 8.2264,
      "step": 1928
    },
    {
      "epoch": 0.376905041031653,
      "grad_norm": 17.55536460876465,
      "learning_rate": 1.4377895900586856e-05,
      "loss": 8.7914,
      "step": 1929
    },
    {
      "epoch": 0.37710042985541226,
      "grad_norm": 8.358119010925293,
      "learning_rate": 1.4372204985675666e-05,
      "loss": 8.6207,
      "step": 1930
    },
    {
      "epoch": 0.37729581867917156,
      "grad_norm": 8.256921768188477,
      "learning_rate": 1.4366512319560642e-05,
      "loss": 8.901,
      "step": 1931
    },
    {
      "epoch": 0.37749120750293086,
      "grad_norm": 9.609137535095215,
      "learning_rate": 1.4360817904521866e-05,
      "loss": 9.44,
      "step": 1932
    },
    {
      "epoch": 0.3776865963266901,
      "grad_norm": 8.766924858093262,
      "learning_rate": 1.4355121742840132e-05,
      "loss": 9.1647,
      "step": 1933
    },
    {
      "epoch": 0.3778819851504494,
      "grad_norm": 8.230937004089355,
      "learning_rate": 1.434942383679693e-05,
      "loss": 8.404,
      "step": 1934
    },
    {
      "epoch": 0.37807737397420865,
      "grad_norm": 7.7953643798828125,
      "learning_rate": 1.4343724188674446e-05,
      "loss": 10.0905,
      "step": 1935
    },
    {
      "epoch": 0.37827276279796795,
      "grad_norm": 9.239731788635254,
      "learning_rate": 1.4338022800755567e-05,
      "loss": 9.3502,
      "step": 1936
    },
    {
      "epoch": 0.37846815162172726,
      "grad_norm": 8.503083229064941,
      "learning_rate": 1.4332319675323879e-05,
      "loss": 9.2081,
      "step": 1937
    },
    {
      "epoch": 0.3786635404454865,
      "grad_norm": 9.946684837341309,
      "learning_rate": 1.4326614814663655e-05,
      "loss": 9.744,
      "step": 1938
    },
    {
      "epoch": 0.3788589292692458,
      "grad_norm": 10.687923431396484,
      "learning_rate": 1.4320908221059876e-05,
      "loss": 9.8484,
      "step": 1939
    },
    {
      "epoch": 0.3790543180930051,
      "grad_norm": 9.50954532623291,
      "learning_rate": 1.43151998967982e-05,
      "loss": 9.7537,
      "step": 1940
    },
    {
      "epoch": 0.37924970691676435,
      "grad_norm": 8.836617469787598,
      "learning_rate": 1.4309489844165e-05,
      "loss": 8.8867,
      "step": 1941
    },
    {
      "epoch": 0.37944509574052365,
      "grad_norm": 11.293417930603027,
      "learning_rate": 1.4303778065447319e-05,
      "loss": 9.4723,
      "step": 1942
    },
    {
      "epoch": 0.3796404845642829,
      "grad_norm": 9.453107833862305,
      "learning_rate": 1.4298064562932906e-05,
      "loss": 9.4513,
      "step": 1943
    },
    {
      "epoch": 0.3798358733880422,
      "grad_norm": 12.666526794433594,
      "learning_rate": 1.4292349338910199e-05,
      "loss": 10.6713,
      "step": 1944
    },
    {
      "epoch": 0.3800312622118015,
      "grad_norm": 7.939197540283203,
      "learning_rate": 1.4286632395668318e-05,
      "loss": 9.7284,
      "step": 1945
    },
    {
      "epoch": 0.38022665103556075,
      "grad_norm": 8.85693359375,
      "learning_rate": 1.4280913735497077e-05,
      "loss": 9.59,
      "step": 1946
    },
    {
      "epoch": 0.38042203985932005,
      "grad_norm": 7.0706658363342285,
      "learning_rate": 1.427519336068698e-05,
      "loss": 9.5534,
      "step": 1947
    },
    {
      "epoch": 0.38061742868307935,
      "grad_norm": 8.621312141418457,
      "learning_rate": 1.4269471273529212e-05,
      "loss": 8.332,
      "step": 1948
    },
    {
      "epoch": 0.3808128175068386,
      "grad_norm": 11.882975578308105,
      "learning_rate": 1.4263747476315646e-05,
      "loss": 10.2817,
      "step": 1949
    },
    {
      "epoch": 0.3810082063305979,
      "grad_norm": 10.135407447814941,
      "learning_rate": 1.4258021971338843e-05,
      "loss": 10.2998,
      "step": 1950
    },
    {
      "epoch": 0.38120359515435714,
      "grad_norm": 14.021071434020996,
      "learning_rate": 1.4252294760892042e-05,
      "loss": 9.3759,
      "step": 1951
    },
    {
      "epoch": 0.38139898397811645,
      "grad_norm": 161.7584991455078,
      "learning_rate": 1.4246565847269174e-05,
      "loss": 10.2743,
      "step": 1952
    },
    {
      "epoch": 0.38159437280187575,
      "grad_norm": 8.837624549865723,
      "learning_rate": 1.424083523276484e-05,
      "loss": 8.2313,
      "step": 1953
    },
    {
      "epoch": 0.381789761625635,
      "grad_norm": 16.58618927001953,
      "learning_rate": 1.4235102919674338e-05,
      "loss": 10.6641,
      "step": 1954
    },
    {
      "epoch": 0.3819851504493943,
      "grad_norm": 11.371005058288574,
      "learning_rate": 1.4229368910293625e-05,
      "loss": 8.9532,
      "step": 1955
    },
    {
      "epoch": 0.3821805392731536,
      "grad_norm": 10.293523788452148,
      "learning_rate": 1.422363320691936e-05,
      "loss": 8.8939,
      "step": 1956
    },
    {
      "epoch": 0.38237592809691284,
      "grad_norm": 9.805085182189941,
      "learning_rate": 1.4217895811848872e-05,
      "loss": 10.1148,
      "step": 1957
    },
    {
      "epoch": 0.38257131692067214,
      "grad_norm": 17.808725357055664,
      "learning_rate": 1.4212156727380155e-05,
      "loss": 9.7582,
      "step": 1958
    },
    {
      "epoch": 0.38276670574443145,
      "grad_norm": 11.809649467468262,
      "learning_rate": 1.42064159558119e-05,
      "loss": 9.4327,
      "step": 1959
    },
    {
      "epoch": 0.3829620945681907,
      "grad_norm": 8.403098106384277,
      "learning_rate": 1.4200673499443456e-05,
      "loss": 9.1216,
      "step": 1960
    },
    {
      "epoch": 0.38315748339195,
      "grad_norm": 22.475610733032227,
      "learning_rate": 1.4194929360574863e-05,
      "loss": 10.0235,
      "step": 1961
    },
    {
      "epoch": 0.38335287221570924,
      "grad_norm": 20.79180335998535,
      "learning_rate": 1.4189183541506821e-05,
      "loss": 9.8155,
      "step": 1962
    },
    {
      "epoch": 0.38354826103946854,
      "grad_norm": 9.602819442749023,
      "learning_rate": 1.4183436044540713e-05,
      "loss": 9.6023,
      "step": 1963
    },
    {
      "epoch": 0.38374364986322784,
      "grad_norm": 7.785224914550781,
      "learning_rate": 1.417768687197859e-05,
      "loss": 9.0554,
      "step": 1964
    },
    {
      "epoch": 0.3839390386869871,
      "grad_norm": 9.529844284057617,
      "learning_rate": 1.417193602612317e-05,
      "loss": 9.3794,
      "step": 1965
    },
    {
      "epoch": 0.3841344275107464,
      "grad_norm": 15.077787399291992,
      "learning_rate": 1.4166183509277848e-05,
      "loss": 9.3298,
      "step": 1966
    },
    {
      "epoch": 0.3843298163345057,
      "grad_norm": 11.620304107666016,
      "learning_rate": 1.4160429323746687e-05,
      "loss": 11.1838,
      "step": 1967
    },
    {
      "epoch": 0.38452520515826494,
      "grad_norm": 8.911892890930176,
      "learning_rate": 1.4154673471834414e-05,
      "loss": 9.7006,
      "step": 1968
    },
    {
      "epoch": 0.38472059398202424,
      "grad_norm": 7.756126403808594,
      "learning_rate": 1.4148915955846429e-05,
      "loss": 9.1773,
      "step": 1969
    },
    {
      "epoch": 0.3849159828057835,
      "grad_norm": 11.736945152282715,
      "learning_rate": 1.4143156778088795e-05,
      "loss": 9.1653,
      "step": 1970
    },
    {
      "epoch": 0.3851113716295428,
      "grad_norm": 10.644880294799805,
      "learning_rate": 1.4137395940868244e-05,
      "loss": 10.2337,
      "step": 1971
    },
    {
      "epoch": 0.3853067604533021,
      "grad_norm": 9.106091499328613,
      "learning_rate": 1.4131633446492163e-05,
      "loss": 8.9762,
      "step": 1972
    },
    {
      "epoch": 0.38550214927706133,
      "grad_norm": 10.716482162475586,
      "learning_rate": 1.4125869297268615e-05,
      "loss": 10.7699,
      "step": 1973
    },
    {
      "epoch": 0.38569753810082064,
      "grad_norm": 10.083828926086426,
      "learning_rate": 1.4120103495506321e-05,
      "loss": 9.2582,
      "step": 1974
    },
    {
      "epoch": 0.38589292692457994,
      "grad_norm": 9.934428215026855,
      "learning_rate": 1.4114336043514662e-05,
      "loss": 10.5586,
      "step": 1975
    },
    {
      "epoch": 0.3860883157483392,
      "grad_norm": 9.987054824829102,
      "learning_rate": 1.4108566943603683e-05,
      "loss": 9.2885,
      "step": 1976
    },
    {
      "epoch": 0.3862837045720985,
      "grad_norm": 18.581371307373047,
      "learning_rate": 1.4102796198084085e-05,
      "loss": 8.4393,
      "step": 1977
    },
    {
      "epoch": 0.38647909339585773,
      "grad_norm": 9.174590110778809,
      "learning_rate": 1.4097023809267235e-05,
      "loss": 8.9276,
      "step": 1978
    },
    {
      "epoch": 0.38667448221961703,
      "grad_norm": 9.374712944030762,
      "learning_rate": 1.409124977946515e-05,
      "loss": 9.4317,
      "step": 1979
    },
    {
      "epoch": 0.38686987104337633,
      "grad_norm": 10.881209373474121,
      "learning_rate": 1.408547411099051e-05,
      "loss": 9.9214,
      "step": 1980
    },
    {
      "epoch": 0.3870652598671356,
      "grad_norm": 8.33880615234375,
      "learning_rate": 1.4079696806156649e-05,
      "loss": 9.9988,
      "step": 1981
    },
    {
      "epoch": 0.3872606486908949,
      "grad_norm": 10.133203506469727,
      "learning_rate": 1.4073917867277557e-05,
      "loss": 10.4433,
      "step": 1982
    },
    {
      "epoch": 0.3874560375146542,
      "grad_norm": 9.079756736755371,
      "learning_rate": 1.4068137296667884e-05,
      "loss": 9.9795,
      "step": 1983
    },
    {
      "epoch": 0.38765142633841343,
      "grad_norm": 9.376028060913086,
      "learning_rate": 1.4062355096642917e-05,
      "loss": 9.1507,
      "step": 1984
    },
    {
      "epoch": 0.38784681516217273,
      "grad_norm": 8.1322660446167,
      "learning_rate": 1.4056571269518612e-05,
      "loss": 9.5236,
      "step": 1985
    },
    {
      "epoch": 0.388042203985932,
      "grad_norm": 11.051843643188477,
      "learning_rate": 1.4050785817611571e-05,
      "loss": 8.7444,
      "step": 1986
    },
    {
      "epoch": 0.3882375928096913,
      "grad_norm": 8.349790573120117,
      "learning_rate": 1.4044998743239051e-05,
      "loss": 10.1119,
      "step": 1987
    },
    {
      "epoch": 0.3884329816334506,
      "grad_norm": 75.24749755859375,
      "learning_rate": 1.403921004871895e-05,
      "loss": 9.6965,
      "step": 1988
    },
    {
      "epoch": 0.3886283704572098,
      "grad_norm": 19.189090728759766,
      "learning_rate": 1.4033419736369823e-05,
      "loss": 9.2351,
      "step": 1989
    },
    {
      "epoch": 0.38882375928096913,
      "grad_norm": 7.177303314208984,
      "learning_rate": 1.4027627808510866e-05,
      "loss": 9.4034,
      "step": 1990
    },
    {
      "epoch": 0.38901914810472843,
      "grad_norm": 7.710775375366211,
      "learning_rate": 1.4021834267461932e-05,
      "loss": 9.4265,
      "step": 1991
    },
    {
      "epoch": 0.3892145369284877,
      "grad_norm": 9.059228897094727,
      "learning_rate": 1.4016039115543508e-05,
      "loss": 9.1341,
      "step": 1992
    },
    {
      "epoch": 0.389409925752247,
      "grad_norm": 41.08426284790039,
      "learning_rate": 1.4010242355076733e-05,
      "loss": 9.6277,
      "step": 1993
    },
    {
      "epoch": 0.3896053145760063,
      "grad_norm": 8.896717071533203,
      "learning_rate": 1.4004443988383393e-05,
      "loss": 9.0697,
      "step": 1994
    },
    {
      "epoch": 0.3898007033997655,
      "grad_norm": 24.87018585205078,
      "learning_rate": 1.3998644017785912e-05,
      "loss": 9.1842,
      "step": 1995
    },
    {
      "epoch": 0.3899960922235248,
      "grad_norm": 10.012320518493652,
      "learning_rate": 1.3992842445607358e-05,
      "loss": 8.4979,
      "step": 1996
    },
    {
      "epoch": 0.39019148104728407,
      "grad_norm": 9.255597114562988,
      "learning_rate": 1.3987039274171441e-05,
      "loss": 9.1657,
      "step": 1997
    },
    {
      "epoch": 0.3903868698710434,
      "grad_norm": 10.156935691833496,
      "learning_rate": 1.3981234505802513e-05,
      "loss": 9.3369,
      "step": 1998
    },
    {
      "epoch": 0.3905822586948027,
      "grad_norm": 14.97067928314209,
      "learning_rate": 1.3975428142825562e-05,
      "loss": 10.339,
      "step": 1999
    },
    {
      "epoch": 0.3907776475185619,
      "grad_norm": 8.015941619873047,
      "learning_rate": 1.3969620187566216e-05,
      "loss": 10.3577,
      "step": 2000
    },
    {
      "epoch": 0.3909730363423212,
      "grad_norm": 8.700507164001465,
      "learning_rate": 1.3963810642350741e-05,
      "loss": 9.7226,
      "step": 2001
    },
    {
      "epoch": 0.3911684251660805,
      "grad_norm": 7.423156261444092,
      "learning_rate": 1.3957999509506044e-05,
      "loss": 8.3885,
      "step": 2002
    },
    {
      "epoch": 0.39136381398983977,
      "grad_norm": 10.251492500305176,
      "learning_rate": 1.395218679135966e-05,
      "loss": 10.1336,
      "step": 2003
    },
    {
      "epoch": 0.39155920281359907,
      "grad_norm": 8.170392990112305,
      "learning_rate": 1.3946372490239769e-05,
      "loss": 7.9025,
      "step": 2004
    },
    {
      "epoch": 0.3917545916373583,
      "grad_norm": 16.20621109008789,
      "learning_rate": 1.394055660847517e-05,
      "loss": 9.9719,
      "step": 2005
    },
    {
      "epoch": 0.3919499804611176,
      "grad_norm": 22.03818702697754,
      "learning_rate": 1.3934739148395311e-05,
      "loss": 8.8452,
      "step": 2006
    },
    {
      "epoch": 0.3921453692848769,
      "grad_norm": 24.292627334594727,
      "learning_rate": 1.3928920112330265e-05,
      "loss": 9.796,
      "step": 2007
    },
    {
      "epoch": 0.39234075810863617,
      "grad_norm": 9.747627258300781,
      "learning_rate": 1.3923099502610733e-05,
      "loss": 8.3834,
      "step": 2008
    },
    {
      "epoch": 0.39253614693239547,
      "grad_norm": 8.528168678283691,
      "learning_rate": 1.3917277321568052e-05,
      "loss": 9.0326,
      "step": 2009
    },
    {
      "epoch": 0.39273153575615477,
      "grad_norm": 10.370828628540039,
      "learning_rate": 1.3911453571534189e-05,
      "loss": 10.0083,
      "step": 2010
    },
    {
      "epoch": 0.392926924579914,
      "grad_norm": 9.663119316101074,
      "learning_rate": 1.390562825484173e-05,
      "loss": 8.2309,
      "step": 2011
    },
    {
      "epoch": 0.3931223134036733,
      "grad_norm": 9.895528793334961,
      "learning_rate": 1.3899801373823905e-05,
      "loss": 10.2185,
      "step": 2012
    },
    {
      "epoch": 0.39331770222743256,
      "grad_norm": 12.388712882995605,
      "learning_rate": 1.389397293081455e-05,
      "loss": 9.9484,
      "step": 2013
    },
    {
      "epoch": 0.39351309105119187,
      "grad_norm": 11.636601448059082,
      "learning_rate": 1.3888142928148143e-05,
      "loss": 8.8707,
      "step": 2014
    },
    {
      "epoch": 0.39370847987495117,
      "grad_norm": 11.408031463623047,
      "learning_rate": 1.3882311368159782e-05,
      "loss": 9.5439,
      "step": 2015
    },
    {
      "epoch": 0.3939038686987104,
      "grad_norm": 8.283649444580078,
      "learning_rate": 1.3876478253185183e-05,
      "loss": 8.3897,
      "step": 2016
    },
    {
      "epoch": 0.3940992575224697,
      "grad_norm": 11.562610626220703,
      "learning_rate": 1.3870643585560697e-05,
      "loss": 9.9744,
      "step": 2017
    },
    {
      "epoch": 0.394294646346229,
      "grad_norm": 9.294994354248047,
      "learning_rate": 1.3864807367623282e-05,
      "loss": 9.6295,
      "step": 2018
    },
    {
      "epoch": 0.39449003516998826,
      "grad_norm": 7.365604877471924,
      "learning_rate": 1.3858969601710533e-05,
      "loss": 9.4103,
      "step": 2019
    },
    {
      "epoch": 0.39468542399374756,
      "grad_norm": 10.17297077178955,
      "learning_rate": 1.3853130290160647e-05,
      "loss": 9.4888,
      "step": 2020
    },
    {
      "epoch": 0.39488081281750687,
      "grad_norm": 9.276739120483398,
      "learning_rate": 1.3847289435312455e-05,
      "loss": 9.889,
      "step": 2021
    },
    {
      "epoch": 0.3950762016412661,
      "grad_norm": 8.32766342163086,
      "learning_rate": 1.3841447039505398e-05,
      "loss": 8.3295,
      "step": 2022
    },
    {
      "epoch": 0.3952715904650254,
      "grad_norm": 9.893261909484863,
      "learning_rate": 1.383560310507954e-05,
      "loss": 9.938,
      "step": 2023
    },
    {
      "epoch": 0.39546697928878466,
      "grad_norm": 15.601113319396973,
      "learning_rate": 1.3829757634375556e-05,
      "loss": 9.2815,
      "step": 2024
    },
    {
      "epoch": 0.39566236811254396,
      "grad_norm": 13.371784210205078,
      "learning_rate": 1.3823910629734743e-05,
      "loss": 9.9588,
      "step": 2025
    },
    {
      "epoch": 0.39585775693630326,
      "grad_norm": 29.384428024291992,
      "learning_rate": 1.3818062093499e-05,
      "loss": 9.993,
      "step": 2026
    },
    {
      "epoch": 0.3960531457600625,
      "grad_norm": 9.52156925201416,
      "learning_rate": 1.3812212028010854e-05,
      "loss": 9.4732,
      "step": 2027
    },
    {
      "epoch": 0.3962485345838218,
      "grad_norm": 9.161752700805664,
      "learning_rate": 1.3806360435613433e-05,
      "loss": 9.4128,
      "step": 2028
    },
    {
      "epoch": 0.3964439234075811,
      "grad_norm": 13.252968788146973,
      "learning_rate": 1.3800507318650485e-05,
      "loss": 10.0995,
      "step": 2029
    },
    {
      "epoch": 0.39663931223134036,
      "grad_norm": 10.196971893310547,
      "learning_rate": 1.3794652679466363e-05,
      "loss": 9.456,
      "step": 2030
    },
    {
      "epoch": 0.39683470105509966,
      "grad_norm": 11.998358726501465,
      "learning_rate": 1.3788796520406033e-05,
      "loss": 9.2391,
      "step": 2031
    },
    {
      "epoch": 0.3970300898788589,
      "grad_norm": 8.397653579711914,
      "learning_rate": 1.3782938843815068e-05,
      "loss": 9.5083,
      "step": 2032
    },
    {
      "epoch": 0.3972254787026182,
      "grad_norm": 9.211454391479492,
      "learning_rate": 1.3777079652039649e-05,
      "loss": 8.3023,
      "step": 2033
    },
    {
      "epoch": 0.3974208675263775,
      "grad_norm": 17.611833572387695,
      "learning_rate": 1.3771218947426566e-05,
      "loss": 9.8341,
      "step": 2034
    },
    {
      "epoch": 0.39761625635013675,
      "grad_norm": 7.4314656257629395,
      "learning_rate": 1.3765356732323209e-05,
      "loss": 8.9265,
      "step": 2035
    },
    {
      "epoch": 0.39781164517389606,
      "grad_norm": 18.303077697753906,
      "learning_rate": 1.3759493009077583e-05,
      "loss": 9.1009,
      "step": 2036
    },
    {
      "epoch": 0.39800703399765536,
      "grad_norm": 15.656684875488281,
      "learning_rate": 1.3753627780038286e-05,
      "loss": 8.9401,
      "step": 2037
    },
    {
      "epoch": 0.3982024228214146,
      "grad_norm": 46.9091682434082,
      "learning_rate": 1.3747761047554531e-05,
      "loss": 10.1341,
      "step": 2038
    },
    {
      "epoch": 0.3983978116451739,
      "grad_norm": 8.730496406555176,
      "learning_rate": 1.3741892813976117e-05,
      "loss": 8.5978,
      "step": 2039
    },
    {
      "epoch": 0.39859320046893315,
      "grad_norm": 7.332820892333984,
      "learning_rate": 1.373602308165346e-05,
      "loss": 8.8639,
      "step": 2040
    },
    {
      "epoch": 0.39878858929269245,
      "grad_norm": 8.697312355041504,
      "learning_rate": 1.3730151852937567e-05,
      "loss": 8.9305,
      "step": 2041
    },
    {
      "epoch": 0.39898397811645175,
      "grad_norm": 8.44970703125,
      "learning_rate": 1.372427913018005e-05,
      "loss": 8.6963,
      "step": 2042
    },
    {
      "epoch": 0.399179366940211,
      "grad_norm": 9.921573638916016,
      "learning_rate": 1.371840491573312e-05,
      "loss": 9.7979,
      "step": 2043
    },
    {
      "epoch": 0.3993747557639703,
      "grad_norm": 20.099802017211914,
      "learning_rate": 1.3712529211949575e-05,
      "loss": 8.9746,
      "step": 2044
    },
    {
      "epoch": 0.3995701445877296,
      "grad_norm": 8.397631645202637,
      "learning_rate": 1.3706652021182818e-05,
      "loss": 9.6269,
      "step": 2045
    },
    {
      "epoch": 0.39976553341148885,
      "grad_norm": 14.477246284484863,
      "learning_rate": 1.370077334578685e-05,
      "loss": 9.4552,
      "step": 2046
    },
    {
      "epoch": 0.39996092223524815,
      "grad_norm": 18.402416229248047,
      "learning_rate": 1.369489318811626e-05,
      "loss": 8.1664,
      "step": 2047
    },
    {
      "epoch": 0.40015631105900745,
      "grad_norm": 56.74452590942383,
      "learning_rate": 1.3689011550526237e-05,
      "loss": 10.9544,
      "step": 2048
    },
    {
      "epoch": 0.4003516998827667,
      "grad_norm": 10.717354774475098,
      "learning_rate": 1.3683128435372558e-05,
      "loss": 9.1311,
      "step": 2049
    },
    {
      "epoch": 0.400547088706526,
      "grad_norm": 7.775894641876221,
      "learning_rate": 1.3677243845011589e-05,
      "loss": 9.6921,
      "step": 2050
    },
    {
      "epoch": 0.40074247753028525,
      "grad_norm": 8.910210609436035,
      "learning_rate": 1.36713577818003e-05,
      "loss": 9.7363,
      "step": 2051
    },
    {
      "epoch": 0.40093786635404455,
      "grad_norm": 7.493995189666748,
      "learning_rate": 1.3665470248096236e-05,
      "loss": 8.3686,
      "step": 2052
    },
    {
      "epoch": 0.40113325517780385,
      "grad_norm": 7.656559467315674,
      "learning_rate": 1.3659581246257533e-05,
      "loss": 8.1144,
      "step": 2053
    },
    {
      "epoch": 0.4013286440015631,
      "grad_norm": 7.61382532119751,
      "learning_rate": 1.365369077864293e-05,
      "loss": 8.3759,
      "step": 2054
    },
    {
      "epoch": 0.4015240328253224,
      "grad_norm": 7.527225971221924,
      "learning_rate": 1.3647798847611733e-05,
      "loss": 9.0688,
      "step": 2055
    },
    {
      "epoch": 0.4017194216490817,
      "grad_norm": 11.350974082946777,
      "learning_rate": 1.3641905455523846e-05,
      "loss": 10.2505,
      "step": 2056
    },
    {
      "epoch": 0.40191481047284094,
      "grad_norm": 9.301215171813965,
      "learning_rate": 1.3636010604739757e-05,
      "loss": 9.9992,
      "step": 2057
    },
    {
      "epoch": 0.40211019929660025,
      "grad_norm": 8.098173141479492,
      "learning_rate": 1.3630114297620535e-05,
      "loss": 9.7471,
      "step": 2058
    },
    {
      "epoch": 0.4023055881203595,
      "grad_norm": 7.174412727355957,
      "learning_rate": 1.362421653652784e-05,
      "loss": 9.3435,
      "step": 2059
    },
    {
      "epoch": 0.4025009769441188,
      "grad_norm": 11.994549751281738,
      "learning_rate": 1.3618317323823897e-05,
      "loss": 8.4048,
      "step": 2060
    },
    {
      "epoch": 0.4026963657678781,
      "grad_norm": 9.520535469055176,
      "learning_rate": 1.3612416661871532e-05,
      "loss": 8.5934,
      "step": 2061
    },
    {
      "epoch": 0.40289175459163734,
      "grad_norm": 11.950779914855957,
      "learning_rate": 1.360651455303414e-05,
      "loss": 9.5535,
      "step": 2062
    },
    {
      "epoch": 0.40308714341539664,
      "grad_norm": 8.245354652404785,
      "learning_rate": 1.3600610999675703e-05,
      "loss": 9.6193,
      "step": 2063
    },
    {
      "epoch": 0.40328253223915594,
      "grad_norm": 7.312138080596924,
      "learning_rate": 1.3594706004160772e-05,
      "loss": 9.3063,
      "step": 2064
    },
    {
      "epoch": 0.4034779210629152,
      "grad_norm": 8.881611824035645,
      "learning_rate": 1.3588799568854485e-05,
      "loss": 10.3356,
      "step": 2065
    },
    {
      "epoch": 0.4036733098866745,
      "grad_norm": 10.751225471496582,
      "learning_rate": 1.358289169612255e-05,
      "loss": 10.3858,
      "step": 2066
    },
    {
      "epoch": 0.40386869871043374,
      "grad_norm": 8.918109893798828,
      "learning_rate": 1.3576982388331258e-05,
      "loss": 9.4066,
      "step": 2067
    },
    {
      "epoch": 0.40406408753419304,
      "grad_norm": 8.959725379943848,
      "learning_rate": 1.3571071647847465e-05,
      "loss": 9.8884,
      "step": 2068
    },
    {
      "epoch": 0.40425947635795234,
      "grad_norm": 9.951619148254395,
      "learning_rate": 1.3565159477038608e-05,
      "loss": 8.6897,
      "step": 2069
    },
    {
      "epoch": 0.4044548651817116,
      "grad_norm": 16.672077178955078,
      "learning_rate": 1.3559245878272695e-05,
      "loss": 9.3297,
      "step": 2070
    },
    {
      "epoch": 0.4046502540054709,
      "grad_norm": 9.029973983764648,
      "learning_rate": 1.3553330853918305e-05,
      "loss": 10.0507,
      "step": 2071
    },
    {
      "epoch": 0.4048456428292302,
      "grad_norm": 9.713790893554688,
      "learning_rate": 1.354741440634459e-05,
      "loss": 10.0247,
      "step": 2072
    },
    {
      "epoch": 0.40504103165298944,
      "grad_norm": 7.559501647949219,
      "learning_rate": 1.3541496537921273e-05,
      "loss": 9.4727,
      "step": 2073
    },
    {
      "epoch": 0.40523642047674874,
      "grad_norm": 21.059707641601562,
      "learning_rate": 1.353557725101864e-05,
      "loss": 9.0248,
      "step": 2074
    },
    {
      "epoch": 0.405431809300508,
      "grad_norm": 7.639827728271484,
      "learning_rate": 1.352965654800755e-05,
      "loss": 9.2796,
      "step": 2075
    },
    {
      "epoch": 0.4056271981242673,
      "grad_norm": 8.415217399597168,
      "learning_rate": 1.3523734431259432e-05,
      "loss": 9.8698,
      "step": 2076
    },
    {
      "epoch": 0.4058225869480266,
      "grad_norm": 8.58553695678711,
      "learning_rate": 1.3517810903146273e-05,
      "loss": 8.7702,
      "step": 2077
    },
    {
      "epoch": 0.40601797577178583,
      "grad_norm": 7.218175888061523,
      "learning_rate": 1.3511885966040632e-05,
      "loss": 9.2946,
      "step": 2078
    },
    {
      "epoch": 0.40621336459554513,
      "grad_norm": 8.998862266540527,
      "learning_rate": 1.3505959622315632e-05,
      "loss": 9.6173,
      "step": 2079
    },
    {
      "epoch": 0.40640875341930444,
      "grad_norm": 27.732744216918945,
      "learning_rate": 1.3500031874344956e-05,
      "loss": 10.2917,
      "step": 2080
    },
    {
      "epoch": 0.4066041422430637,
      "grad_norm": 8.803112983703613,
      "learning_rate": 1.3494102724502849e-05,
      "loss": 10.2354,
      "step": 2081
    },
    {
      "epoch": 0.406799531066823,
      "grad_norm": 10.404772758483887,
      "learning_rate": 1.3488172175164122e-05,
      "loss": 9.5116,
      "step": 2082
    },
    {
      "epoch": 0.4069949198905823,
      "grad_norm": 8.896175384521484,
      "learning_rate": 1.3482240228704143e-05,
      "loss": 9.2856,
      "step": 2083
    },
    {
      "epoch": 0.40719030871434153,
      "grad_norm": 7.903563022613525,
      "learning_rate": 1.3476306887498842e-05,
      "loss": 8.9899,
      "step": 2084
    },
    {
      "epoch": 0.40738569753810083,
      "grad_norm": 9.580747604370117,
      "learning_rate": 1.3470372153924707e-05,
      "loss": 9.3801,
      "step": 2085
    },
    {
      "epoch": 0.4075810863618601,
      "grad_norm": 8.003578186035156,
      "learning_rate": 1.346443603035878e-05,
      "loss": 8.8302,
      "step": 2086
    },
    {
      "epoch": 0.4077764751856194,
      "grad_norm": 9.539640426635742,
      "learning_rate": 1.3458498519178665e-05,
      "loss": 9.8276,
      "step": 2087
    },
    {
      "epoch": 0.4079718640093787,
      "grad_norm": 9.275850296020508,
      "learning_rate": 1.3452559622762518e-05,
      "loss": 8.6797,
      "step": 2088
    },
    {
      "epoch": 0.40816725283313793,
      "grad_norm": 6.939610481262207,
      "learning_rate": 1.3446619343489053e-05,
      "loss": 8.6884,
      "step": 2089
    },
    {
      "epoch": 0.40836264165689723,
      "grad_norm": 12.585261344909668,
      "learning_rate": 1.3440677683737538e-05,
      "loss": 10.2965,
      "step": 2090
    },
    {
      "epoch": 0.40855803048065653,
      "grad_norm": 7.093527793884277,
      "learning_rate": 1.3434734645887788e-05,
      "loss": 8.9571,
      "step": 2091
    },
    {
      "epoch": 0.4087534193044158,
      "grad_norm": 7.72236967086792,
      "learning_rate": 1.3428790232320177e-05,
      "loss": 8.7031,
      "step": 2092
    },
    {
      "epoch": 0.4089488081281751,
      "grad_norm": 7.893250465393066,
      "learning_rate": 1.3422844445415628e-05,
      "loss": 8.4756,
      "step": 2093
    },
    {
      "epoch": 0.4091441969519343,
      "grad_norm": 10.6410551071167,
      "learning_rate": 1.341689728755561e-05,
      "loss": 9.0713,
      "step": 2094
    },
    {
      "epoch": 0.4093395857756936,
      "grad_norm": 9.671658515930176,
      "learning_rate": 1.3410948761122145e-05,
      "loss": 10.2595,
      "step": 2095
    },
    {
      "epoch": 0.40953497459945293,
      "grad_norm": 13.780243873596191,
      "learning_rate": 1.3404998868497803e-05,
      "loss": 9.4971,
      "step": 2096
    },
    {
      "epoch": 0.4097303634232122,
      "grad_norm": 22.337764739990234,
      "learning_rate": 1.3399047612065703e-05,
      "loss": 8.2947,
      "step": 2097
    },
    {
      "epoch": 0.4099257522469715,
      "grad_norm": 8.596807479858398,
      "learning_rate": 1.3393094994209505e-05,
      "loss": 8.9651,
      "step": 2098
    },
    {
      "epoch": 0.4101211410707308,
      "grad_norm": 8.460862159729004,
      "learning_rate": 1.338714101731342e-05,
      "loss": 8.8179,
      "step": 2099
    },
    {
      "epoch": 0.41031652989449,
      "grad_norm": 8.91547679901123,
      "learning_rate": 1.3381185683762197e-05,
      "loss": 9.2624,
      "step": 2100
    },
    {
      "epoch": 0.4105119187182493,
      "grad_norm": 8.833627700805664,
      "learning_rate": 1.3375228995941135e-05,
      "loss": 9.1694,
      "step": 2101
    },
    {
      "epoch": 0.41070730754200857,
      "grad_norm": 8.893487930297852,
      "learning_rate": 1.336927095623607e-05,
      "loss": 9.4226,
      "step": 2102
    },
    {
      "epoch": 0.41090269636576787,
      "grad_norm": 25.64516830444336,
      "learning_rate": 1.3363311567033384e-05,
      "loss": 9.3881,
      "step": 2103
    },
    {
      "epoch": 0.4110980851895272,
      "grad_norm": 9.39368724822998,
      "learning_rate": 1.3357350830719997e-05,
      "loss": 9.8792,
      "step": 2104
    },
    {
      "epoch": 0.4112934740132864,
      "grad_norm": 7.454868316650391,
      "learning_rate": 1.3351388749683369e-05,
      "loss": 9.0325,
      "step": 2105
    },
    {
      "epoch": 0.4114888628370457,
      "grad_norm": 13.555609703063965,
      "learning_rate": 1.33454253263115e-05,
      "loss": 9.6081,
      "step": 2106
    },
    {
      "epoch": 0.411684251660805,
      "grad_norm": 8.101183891296387,
      "learning_rate": 1.3339460562992922e-05,
      "loss": 10.0975,
      "step": 2107
    },
    {
      "epoch": 0.41187964048456427,
      "grad_norm": 11.35245418548584,
      "learning_rate": 1.333349446211671e-05,
      "loss": 10.0821,
      "step": 2108
    },
    {
      "epoch": 0.41207502930832357,
      "grad_norm": 10.807251930236816,
      "learning_rate": 1.3327527026072473e-05,
      "loss": 10.1509,
      "step": 2109
    },
    {
      "epoch": 0.41227041813208287,
      "grad_norm": 6.899622440338135,
      "learning_rate": 1.3321558257250355e-05,
      "loss": 8.7787,
      "step": 2110
    },
    {
      "epoch": 0.4124658069558421,
      "grad_norm": 9.099929809570312,
      "learning_rate": 1.3315588158041032e-05,
      "loss": 9.8039,
      "step": 2111
    },
    {
      "epoch": 0.4126611957796014,
      "grad_norm": 7.551453590393066,
      "learning_rate": 1.3309616730835715e-05,
      "loss": 9.2101,
      "step": 2112
    },
    {
      "epoch": 0.41285658460336067,
      "grad_norm": 11.950295448303223,
      "learning_rate": 1.3303643978026145e-05,
      "loss": 10.7997,
      "step": 2113
    },
    {
      "epoch": 0.41305197342711997,
      "grad_norm": 8.216484069824219,
      "learning_rate": 1.3297669902004598e-05,
      "loss": 9.7015,
      "step": 2114
    },
    {
      "epoch": 0.41324736225087927,
      "grad_norm": 7.783416748046875,
      "learning_rate": 1.3291694505163872e-05,
      "loss": 8.1667,
      "step": 2115
    },
    {
      "epoch": 0.4134427510746385,
      "grad_norm": 7.72036600112915,
      "learning_rate": 1.32857177898973e-05,
      "loss": 9.1759,
      "step": 2116
    },
    {
      "epoch": 0.4136381398983978,
      "grad_norm": 9.248005867004395,
      "learning_rate": 1.3279739758598746e-05,
      "loss": 9.5641,
      "step": 2117
    },
    {
      "epoch": 0.4138335287221571,
      "grad_norm": 8.15285587310791,
      "learning_rate": 1.3273760413662596e-05,
      "loss": 8.9547,
      "step": 2118
    },
    {
      "epoch": 0.41402891754591636,
      "grad_norm": 9.635090827941895,
      "learning_rate": 1.3267779757483761e-05,
      "loss": 9.9121,
      "step": 2119
    },
    {
      "epoch": 0.41422430636967567,
      "grad_norm": 8.915787696838379,
      "learning_rate": 1.3261797792457678e-05,
      "loss": 9.6434,
      "step": 2120
    },
    {
      "epoch": 0.4144196951934349,
      "grad_norm": 7.4869303703308105,
      "learning_rate": 1.3255814520980312e-05,
      "loss": 9.0176,
      "step": 2121
    },
    {
      "epoch": 0.4146150840171942,
      "grad_norm": 13.5423002243042,
      "learning_rate": 1.3249829945448151e-05,
      "loss": 9.7015,
      "step": 2122
    },
    {
      "epoch": 0.4148104728409535,
      "grad_norm": 11.67705249786377,
      "learning_rate": 1.32438440682582e-05,
      "loss": 10.2476,
      "step": 2123
    },
    {
      "epoch": 0.41500586166471276,
      "grad_norm": 9.512430191040039,
      "learning_rate": 1.3237856891807992e-05,
      "loss": 9.7233,
      "step": 2124
    },
    {
      "epoch": 0.41520125048847206,
      "grad_norm": 88.60699462890625,
      "learning_rate": 1.3231868418495573e-05,
      "loss": 11.1191,
      "step": 2125
    },
    {
      "epoch": 0.41539663931223136,
      "grad_norm": 7.701420307159424,
      "learning_rate": 1.3225878650719514e-05,
      "loss": 9.1693,
      "step": 2126
    },
    {
      "epoch": 0.4155920281359906,
      "grad_norm": 9.494232177734375,
      "learning_rate": 1.3219887590878903e-05,
      "loss": 9.4935,
      "step": 2127
    },
    {
      "epoch": 0.4157874169597499,
      "grad_norm": 8.114104270935059,
      "learning_rate": 1.3213895241373348e-05,
      "loss": 9.3104,
      "step": 2128
    },
    {
      "epoch": 0.41598280578350916,
      "grad_norm": 8.193770408630371,
      "learning_rate": 1.320790160460297e-05,
      "loss": 10.0441,
      "step": 2129
    },
    {
      "epoch": 0.41617819460726846,
      "grad_norm": 9.657182693481445,
      "learning_rate": 1.3201906682968404e-05,
      "loss": 9.3909,
      "step": 2130
    },
    {
      "epoch": 0.41637358343102776,
      "grad_norm": 9.406392097473145,
      "learning_rate": 1.3195910478870804e-05,
      "loss": 9.7109,
      "step": 2131
    },
    {
      "epoch": 0.416568972254787,
      "grad_norm": 7.698124885559082,
      "learning_rate": 1.3189912994711836e-05,
      "loss": 8.3727,
      "step": 2132
    },
    {
      "epoch": 0.4167643610785463,
      "grad_norm": 10.945155143737793,
      "learning_rate": 1.3183914232893675e-05,
      "loss": 9.561,
      "step": 2133
    },
    {
      "epoch": 0.4169597499023056,
      "grad_norm": 8.722308158874512,
      "learning_rate": 1.3177914195819018e-05,
      "loss": 10.1848,
      "step": 2134
    },
    {
      "epoch": 0.41715513872606486,
      "grad_norm": 10.87885856628418,
      "learning_rate": 1.3171912885891063e-05,
      "loss": 9.3673,
      "step": 2135
    },
    {
      "epoch": 0.41735052754982416,
      "grad_norm": 7.66211462020874,
      "learning_rate": 1.3165910305513521e-05,
      "loss": 9.4162,
      "step": 2136
    },
    {
      "epoch": 0.4175459163735834,
      "grad_norm": 11.434732437133789,
      "learning_rate": 1.3159906457090607e-05,
      "loss": 9.6539,
      "step": 2137
    },
    {
      "epoch": 0.4177413051973427,
      "grad_norm": 38.60456085205078,
      "learning_rate": 1.3153901343027058e-05,
      "loss": 8.932,
      "step": 2138
    },
    {
      "epoch": 0.417936694021102,
      "grad_norm": 9.288236618041992,
      "learning_rate": 1.3147894965728103e-05,
      "loss": 9.7608,
      "step": 2139
    },
    {
      "epoch": 0.41813208284486125,
      "grad_norm": 9.36833667755127,
      "learning_rate": 1.314188732759948e-05,
      "loss": 9.2929,
      "step": 2140
    },
    {
      "epoch": 0.41832747166862055,
      "grad_norm": 9.080362319946289,
      "learning_rate": 1.3135878431047442e-05,
      "loss": 9.847,
      "step": 2141
    },
    {
      "epoch": 0.41852286049237986,
      "grad_norm": 8.36670970916748,
      "learning_rate": 1.3129868278478733e-05,
      "loss": 9.0836,
      "step": 2142
    },
    {
      "epoch": 0.4187182493161391,
      "grad_norm": 8.57102108001709,
      "learning_rate": 1.3123856872300607e-05,
      "loss": 9.3054,
      "step": 2143
    },
    {
      "epoch": 0.4189136381398984,
      "grad_norm": 7.616203308105469,
      "learning_rate": 1.3117844214920818e-05,
      "loss": 8.6007,
      "step": 2144
    },
    {
      "epoch": 0.4191090269636577,
      "grad_norm": 6.933886528015137,
      "learning_rate": 1.3111830308747627e-05,
      "loss": 7.9322,
      "step": 2145
    },
    {
      "epoch": 0.41930441578741695,
      "grad_norm": 10.009140968322754,
      "learning_rate": 1.3105815156189782e-05,
      "loss": 8.7556,
      "step": 2146
    },
    {
      "epoch": 0.41949980461117625,
      "grad_norm": 9.128080368041992,
      "learning_rate": 1.3099798759656546e-05,
      "loss": 9.5527,
      "step": 2147
    },
    {
      "epoch": 0.4196951934349355,
      "grad_norm": 10.222684860229492,
      "learning_rate": 1.3093781121557667e-05,
      "loss": 9.1902,
      "step": 2148
    },
    {
      "epoch": 0.4198905822586948,
      "grad_norm": 10.032594680786133,
      "learning_rate": 1.3087762244303399e-05,
      "loss": 8.6057,
      "step": 2149
    },
    {
      "epoch": 0.4200859710824541,
      "grad_norm": 8.646735191345215,
      "learning_rate": 1.308174213030449e-05,
      "loss": 9.2424,
      "step": 2150
    },
    {
      "epoch": 0.42028135990621335,
      "grad_norm": 8.629133224487305,
      "learning_rate": 1.3075720781972176e-05,
      "loss": 10.0302,
      "step": 2151
    },
    {
      "epoch": 0.42047674872997265,
      "grad_norm": 10.1091890335083,
      "learning_rate": 1.3069698201718202e-05,
      "loss": 9.4035,
      "step": 2152
    },
    {
      "epoch": 0.42067213755373195,
      "grad_norm": 9.671299934387207,
      "learning_rate": 1.3063674391954795e-05,
      "loss": 10.2995,
      "step": 2153
    },
    {
      "epoch": 0.4208675263774912,
      "grad_norm": 7.490504264831543,
      "learning_rate": 1.3057649355094678e-05,
      "loss": 9.9019,
      "step": 2154
    },
    {
      "epoch": 0.4210629152012505,
      "grad_norm": 8.988183975219727,
      "learning_rate": 1.3051623093551062e-05,
      "loss": 8.8773,
      "step": 2155
    },
    {
      "epoch": 0.42125830402500974,
      "grad_norm": 11.667671203613281,
      "learning_rate": 1.3045595609737659e-05,
      "loss": 10.0266,
      "step": 2156
    },
    {
      "epoch": 0.42145369284876905,
      "grad_norm": 15.264968872070312,
      "learning_rate": 1.3039566906068657e-05,
      "loss": 9.9159,
      "step": 2157
    },
    {
      "epoch": 0.42164908167252835,
      "grad_norm": 11.067122459411621,
      "learning_rate": 1.3033536984958744e-05,
      "loss": 10.0485,
      "step": 2158
    },
    {
      "epoch": 0.4218444704962876,
      "grad_norm": 7.868844032287598,
      "learning_rate": 1.3027505848823086e-05,
      "loss": 9.4467,
      "step": 2159
    },
    {
      "epoch": 0.4220398593200469,
      "grad_norm": 7.70686149597168,
      "learning_rate": 1.3021473500077342e-05,
      "loss": 8.182,
      "step": 2160
    },
    {
      "epoch": 0.4222352481438062,
      "grad_norm": 7.336475849151611,
      "learning_rate": 1.301543994113766e-05,
      "loss": 8.5334,
      "step": 2161
    },
    {
      "epoch": 0.42243063696756544,
      "grad_norm": 9.529787063598633,
      "learning_rate": 1.300940517442066e-05,
      "loss": 10.0935,
      "step": 2162
    },
    {
      "epoch": 0.42262602579132474,
      "grad_norm": 9.209541320800781,
      "learning_rate": 1.3003369202343454e-05,
      "loss": 9.6814,
      "step": 2163
    },
    {
      "epoch": 0.422821414615084,
      "grad_norm": 8.053021430969238,
      "learning_rate": 1.299733202732364e-05,
      "loss": 8.7216,
      "step": 2164
    },
    {
      "epoch": 0.4230168034388433,
      "grad_norm": 8.257932662963867,
      "learning_rate": 1.2991293651779296e-05,
      "loss": 9.8704,
      "step": 2165
    },
    {
      "epoch": 0.4232121922626026,
      "grad_norm": 10.342300415039062,
      "learning_rate": 1.298525407812897e-05,
      "loss": 9.5254,
      "step": 2166
    },
    {
      "epoch": 0.42340758108636184,
      "grad_norm": 10.803786277770996,
      "learning_rate": 1.2979213308791709e-05,
      "loss": 9.3382,
      "step": 2167
    },
    {
      "epoch": 0.42360296991012114,
      "grad_norm": 7.010225296020508,
      "learning_rate": 1.2973171346187021e-05,
      "loss": 8.1361,
      "step": 2168
    },
    {
      "epoch": 0.42379835873388044,
      "grad_norm": 8.592170715332031,
      "learning_rate": 1.2967128192734903e-05,
      "loss": 9.1814,
      "step": 2169
    },
    {
      "epoch": 0.4239937475576397,
      "grad_norm": 11.579129219055176,
      "learning_rate": 1.2961083850855826e-05,
      "loss": 9.909,
      "step": 2170
    },
    {
      "epoch": 0.424189136381399,
      "grad_norm": 7.7738118171691895,
      "learning_rate": 1.2955038322970737e-05,
      "loss": 8.9426,
      "step": 2171
    },
    {
      "epoch": 0.4243845252051583,
      "grad_norm": 9.89653205871582,
      "learning_rate": 1.2948991611501053e-05,
      "loss": 10.6861,
      "step": 2172
    },
    {
      "epoch": 0.42457991402891754,
      "grad_norm": 7.241786479949951,
      "learning_rate": 1.2942943718868675e-05,
      "loss": 8.2366,
      "step": 2173
    },
    {
      "epoch": 0.42477530285267684,
      "grad_norm": 9.199349403381348,
      "learning_rate": 1.293689464749597e-05,
      "loss": 8.7557,
      "step": 2174
    },
    {
      "epoch": 0.4249706916764361,
      "grad_norm": 6.656001567840576,
      "learning_rate": 1.293084439980578e-05,
      "loss": 8.5495,
      "step": 2175
    },
    {
      "epoch": 0.4251660805001954,
      "grad_norm": 9.568596839904785,
      "learning_rate": 1.2924792978221414e-05,
      "loss": 10.432,
      "step": 2176
    },
    {
      "epoch": 0.4253614693239547,
      "grad_norm": 7.979373931884766,
      "learning_rate": 1.2918740385166655e-05,
      "loss": 8.3202,
      "step": 2177
    },
    {
      "epoch": 0.42555685814771393,
      "grad_norm": 8.833368301391602,
      "learning_rate": 1.2912686623065755e-05,
      "loss": 9.2741,
      "step": 2178
    },
    {
      "epoch": 0.42575224697147324,
      "grad_norm": 6.991685390472412,
      "learning_rate": 1.2906631694343436e-05,
      "loss": 8.5693,
      "step": 2179
    },
    {
      "epoch": 0.42594763579523254,
      "grad_norm": 10.28195571899414,
      "learning_rate": 1.2900575601424883e-05,
      "loss": 10.0393,
      "step": 2180
    },
    {
      "epoch": 0.4261430246189918,
      "grad_norm": 8.920546531677246,
      "learning_rate": 1.289451834673575e-05,
      "loss": 9.9295,
      "step": 2181
    },
    {
      "epoch": 0.4263384134427511,
      "grad_norm": 8.196661949157715,
      "learning_rate": 1.288845993270215e-05,
      "loss": 8.2215,
      "step": 2182
    },
    {
      "epoch": 0.42653380226651033,
      "grad_norm": 9.465472221374512,
      "learning_rate": 1.2882400361750676e-05,
      "loss": 8.8123,
      "step": 2183
    },
    {
      "epoch": 0.42672919109026963,
      "grad_norm": 8.361612319946289,
      "learning_rate": 1.2876339636308366e-05,
      "loss": 8.811,
      "step": 2184
    },
    {
      "epoch": 0.42692457991402893,
      "grad_norm": 8.781848907470703,
      "learning_rate": 1.287027775880273e-05,
      "loss": 9.1459,
      "step": 2185
    },
    {
      "epoch": 0.4271199687377882,
      "grad_norm": 10.067546844482422,
      "learning_rate": 1.2864214731661743e-05,
      "loss": 9.4827,
      "step": 2186
    },
    {
      "epoch": 0.4273153575615475,
      "grad_norm": 8.98030948638916,
      "learning_rate": 1.2858150557313832e-05,
      "loss": 9.0296,
      "step": 2187
    },
    {
      "epoch": 0.4275107463853068,
      "grad_norm": 8.424452781677246,
      "learning_rate": 1.285208523818789e-05,
      "loss": 8.3527,
      "step": 2188
    },
    {
      "epoch": 0.42770613520906603,
      "grad_norm": 12.284327507019043,
      "learning_rate": 1.2846018776713257e-05,
      "loss": 9.3219,
      "step": 2189
    },
    {
      "epoch": 0.42790152403282533,
      "grad_norm": 7.31266975402832,
      "learning_rate": 1.2839951175319752e-05,
      "loss": 8.6924,
      "step": 2190
    },
    {
      "epoch": 0.4280969128565846,
      "grad_norm": 8.193886756896973,
      "learning_rate": 1.2833882436437626e-05,
      "loss": 8.9165,
      "step": 2191
    },
    {
      "epoch": 0.4282923016803439,
      "grad_norm": 9.161237716674805,
      "learning_rate": 1.2827812562497603e-05,
      "loss": 9.3125,
      "step": 2192
    },
    {
      "epoch": 0.4284876905041032,
      "grad_norm": 8.897019386291504,
      "learning_rate": 1.2821741555930856e-05,
      "loss": 9.0099,
      "step": 2193
    },
    {
      "epoch": 0.4286830793278624,
      "grad_norm": 8.073427200317383,
      "learning_rate": 1.2815669419169013e-05,
      "loss": 9.546,
      "step": 2194
    },
    {
      "epoch": 0.42887846815162173,
      "grad_norm": 26.498016357421875,
      "learning_rate": 1.2809596154644148e-05,
      "loss": 9.2152,
      "step": 2195
    },
    {
      "epoch": 0.42907385697538103,
      "grad_norm": 9.393672943115234,
      "learning_rate": 1.2803521764788795e-05,
      "loss": 8.4541,
      "step": 2196
    },
    {
      "epoch": 0.4292692457991403,
      "grad_norm": 8.790255546569824,
      "learning_rate": 1.279744625203594e-05,
      "loss": 10.2608,
      "step": 2197
    },
    {
      "epoch": 0.4294646346228996,
      "grad_norm": 8.156511306762695,
      "learning_rate": 1.2791369618819008e-05,
      "loss": 7.7263,
      "step": 2198
    },
    {
      "epoch": 0.4296600234466589,
      "grad_norm": 6.234302997589111,
      "learning_rate": 1.2785291867571883e-05,
      "loss": 8.707,
      "step": 2199
    },
    {
      "epoch": 0.4298554122704181,
      "grad_norm": 10.79444408416748,
      "learning_rate": 1.2779213000728889e-05,
      "loss": 10.0896,
      "step": 2200
    },
    {
      "epoch": 0.4300508010941774,
      "grad_norm": 9.629950523376465,
      "learning_rate": 1.2773133020724804e-05,
      "loss": 10.357,
      "step": 2201
    },
    {
      "epoch": 0.43024618991793667,
      "grad_norm": 9.38863754272461,
      "learning_rate": 1.2767051929994848e-05,
      "loss": 9.3469,
      "step": 2202
    },
    {
      "epoch": 0.430441578741696,
      "grad_norm": 7.55531120300293,
      "learning_rate": 1.2760969730974692e-05,
      "loss": 9.691,
      "step": 2203
    },
    {
      "epoch": 0.4306369675654553,
      "grad_norm": 7.783170223236084,
      "learning_rate": 1.2754886426100436e-05,
      "loss": 9.4499,
      "step": 2204
    },
    {
      "epoch": 0.4308323563892145,
      "grad_norm": 7.9393086433410645,
      "learning_rate": 1.2748802017808637e-05,
      "loss": 10.4184,
      "step": 2205
    },
    {
      "epoch": 0.4310277452129738,
      "grad_norm": 8.78271770477295,
      "learning_rate": 1.274271650853629e-05,
      "loss": 8.9384,
      "step": 2206
    },
    {
      "epoch": 0.4312231340367331,
      "grad_norm": 38.12883377075195,
      "learning_rate": 1.2736629900720832e-05,
      "loss": 9.033,
      "step": 2207
    },
    {
      "epoch": 0.43141852286049237,
      "grad_norm": 9.394990921020508,
      "learning_rate": 1.2730542196800132e-05,
      "loss": 9.2351,
      "step": 2208
    },
    {
      "epoch": 0.43161391168425167,
      "grad_norm": 9.552518844604492,
      "learning_rate": 1.2724453399212513e-05,
      "loss": 8.9445,
      "step": 2209
    },
    {
      "epoch": 0.4318093005080109,
      "grad_norm": 8.833136558532715,
      "learning_rate": 1.2718363510396721e-05,
      "loss": 9.0274,
      "step": 2210
    },
    {
      "epoch": 0.4320046893317702,
      "grad_norm": 8.848443984985352,
      "learning_rate": 1.2712272532791947e-05,
      "loss": 10.6383,
      "step": 2211
    },
    {
      "epoch": 0.4322000781555295,
      "grad_norm": 8.959117889404297,
      "learning_rate": 1.270618046883782e-05,
      "loss": 9.0019,
      "step": 2212
    },
    {
      "epoch": 0.43239546697928877,
      "grad_norm": 8.169061660766602,
      "learning_rate": 1.2700087320974394e-05,
      "loss": 8.4105,
      "step": 2213
    },
    {
      "epoch": 0.43259085580304807,
      "grad_norm": 7.978032112121582,
      "learning_rate": 1.269399309164217e-05,
      "loss": 8.4926,
      "step": 2214
    },
    {
      "epoch": 0.43278624462680737,
      "grad_norm": 8.165763854980469,
      "learning_rate": 1.2687897783282072e-05,
      "loss": 8.1933,
      "step": 2215
    },
    {
      "epoch": 0.4329816334505666,
      "grad_norm": 8.540172576904297,
      "learning_rate": 1.2681801398335467e-05,
      "loss": 9.6549,
      "step": 2216
    },
    {
      "epoch": 0.4331770222743259,
      "grad_norm": 20.1662654876709,
      "learning_rate": 1.267570393924414e-05,
      "loss": 9.2204,
      "step": 2217
    },
    {
      "epoch": 0.43337241109808516,
      "grad_norm": 20.048967361450195,
      "learning_rate": 1.2669605408450314e-05,
      "loss": 9.9814,
      "step": 2218
    },
    {
      "epoch": 0.43356779992184447,
      "grad_norm": 9.27750015258789,
      "learning_rate": 1.2663505808396639e-05,
      "loss": 9.7194,
      "step": 2219
    },
    {
      "epoch": 0.43376318874560377,
      "grad_norm": 8.13779354095459,
      "learning_rate": 1.2657405141526196e-05,
      "loss": 9.7392,
      "step": 2220
    },
    {
      "epoch": 0.433958577569363,
      "grad_norm": 12.067617416381836,
      "learning_rate": 1.2651303410282489e-05,
      "loss": 9.1701,
      "step": 2221
    },
    {
      "epoch": 0.4341539663931223,
      "grad_norm": 5.860738277435303,
      "learning_rate": 1.2645200617109456e-05,
      "loss": 8.9809,
      "step": 2222
    },
    {
      "epoch": 0.4343493552168816,
      "grad_norm": 19.50615119934082,
      "learning_rate": 1.2639096764451451e-05,
      "loss": 10.4136,
      "step": 2223
    },
    {
      "epoch": 0.43454474404064086,
      "grad_norm": 9.074341773986816,
      "learning_rate": 1.2632991854753257e-05,
      "loss": 9.2914,
      "step": 2224
    },
    {
      "epoch": 0.43474013286440016,
      "grad_norm": 10.70407485961914,
      "learning_rate": 1.2626885890460079e-05,
      "loss": 10.488,
      "step": 2225
    },
    {
      "epoch": 0.4349355216881594,
      "grad_norm": 40.406402587890625,
      "learning_rate": 1.262077887401755e-05,
      "loss": 9.7503,
      "step": 2226
    },
    {
      "epoch": 0.4351309105119187,
      "grad_norm": 10.195911407470703,
      "learning_rate": 1.2614670807871711e-05,
      "loss": 9.8839,
      "step": 2227
    },
    {
      "epoch": 0.435326299335678,
      "grad_norm": 7.769820690155029,
      "learning_rate": 1.2608561694469042e-05,
      "loss": 8.9693,
      "step": 2228
    },
    {
      "epoch": 0.43552168815943726,
      "grad_norm": 10.964240074157715,
      "learning_rate": 1.2602451536256426e-05,
      "loss": 9.7091,
      "step": 2229
    },
    {
      "epoch": 0.43571707698319656,
      "grad_norm": 10.297403335571289,
      "learning_rate": 1.2596340335681174e-05,
      "loss": 9.5607,
      "step": 2230
    },
    {
      "epoch": 0.43591246580695586,
      "grad_norm": 8.6085205078125,
      "learning_rate": 1.2590228095191011e-05,
      "loss": 9.6563,
      "step": 2231
    },
    {
      "epoch": 0.4361078546307151,
      "grad_norm": 7.507328987121582,
      "learning_rate": 1.2584114817234076e-05,
      "loss": 8.9352,
      "step": 2232
    },
    {
      "epoch": 0.4363032434544744,
      "grad_norm": 10.990617752075195,
      "learning_rate": 1.2578000504258931e-05,
      "loss": 9.4201,
      "step": 2233
    },
    {
      "epoch": 0.4364986322782337,
      "grad_norm": 8.079955101013184,
      "learning_rate": 1.2571885158714545e-05,
      "loss": 10.1628,
      "step": 2234
    },
    {
      "epoch": 0.43669402110199296,
      "grad_norm": 10.56502628326416,
      "learning_rate": 1.2565768783050304e-05,
      "loss": 9.422,
      "step": 2235
    },
    {
      "epoch": 0.43688940992575226,
      "grad_norm": 8.347862243652344,
      "learning_rate": 1.255965137971601e-05,
      "loss": 10.2135,
      "step": 2236
    },
    {
      "epoch": 0.4370847987495115,
      "grad_norm": 13.142292022705078,
      "learning_rate": 1.255353295116187e-05,
      "loss": 9.6481,
      "step": 2237
    },
    {
      "epoch": 0.4372801875732708,
      "grad_norm": 7.691909313201904,
      "learning_rate": 1.2547413499838506e-05,
      "loss": 9.0773,
      "step": 2238
    },
    {
      "epoch": 0.4374755763970301,
      "grad_norm": 14.73984146118164,
      "learning_rate": 1.2541293028196946e-05,
      "loss": 9.5216,
      "step": 2239
    },
    {
      "epoch": 0.43767096522078935,
      "grad_norm": 9.408361434936523,
      "learning_rate": 1.2535171538688632e-05,
      "loss": 9.2028,
      "step": 2240
    },
    {
      "epoch": 0.43786635404454866,
      "grad_norm": 10.333508491516113,
      "learning_rate": 1.2529049033765405e-05,
      "loss": 10.0852,
      "step": 2241
    },
    {
      "epoch": 0.43806174286830796,
      "grad_norm": 9.904654502868652,
      "learning_rate": 1.2522925515879525e-05,
      "loss": 10.235,
      "step": 2242
    },
    {
      "epoch": 0.4382571316920672,
      "grad_norm": 10.235962867736816,
      "learning_rate": 1.2516800987483651e-05,
      "loss": 10.2423,
      "step": 2243
    },
    {
      "epoch": 0.4384525205158265,
      "grad_norm": 9.246352195739746,
      "learning_rate": 1.251067545103084e-05,
      "loss": 9.4063,
      "step": 2244
    },
    {
      "epoch": 0.43864790933958575,
      "grad_norm": 28.940309524536133,
      "learning_rate": 1.2504548908974562e-05,
      "loss": 10.0929,
      "step": 2245
    },
    {
      "epoch": 0.43884329816334505,
      "grad_norm": 8.908502578735352,
      "learning_rate": 1.2498421363768692e-05,
      "loss": 9.6554,
      "step": 2246
    },
    {
      "epoch": 0.43903868698710435,
      "grad_norm": 8.206972122192383,
      "learning_rate": 1.2492292817867498e-05,
      "loss": 8.6696,
      "step": 2247
    },
    {
      "epoch": 0.4392340758108636,
      "grad_norm": 10.232550621032715,
      "learning_rate": 1.248616327372565e-05,
      "loss": 9.515,
      "step": 2248
    },
    {
      "epoch": 0.4394294646346229,
      "grad_norm": 8.533775329589844,
      "learning_rate": 1.2480032733798229e-05,
      "loss": 9.4387,
      "step": 2249
    },
    {
      "epoch": 0.4396248534583822,
      "grad_norm": 10.445351600646973,
      "learning_rate": 1.2473901200540699e-05,
      "loss": 10.2788,
      "step": 2250
    },
    {
      "epoch": 0.43982024228214145,
      "grad_norm": 11.097278594970703,
      "learning_rate": 1.2467768676408936e-05,
      "loss": 9.391,
      "step": 2251
    },
    {
      "epoch": 0.44001563110590075,
      "grad_norm": 10.348945617675781,
      "learning_rate": 1.24616351638592e-05,
      "loss": 9.4205,
      "step": 2252
    },
    {
      "epoch": 0.44021101992966,
      "grad_norm": 12.105810165405273,
      "learning_rate": 1.2455500665348154e-05,
      "loss": 10.2328,
      "step": 2253
    },
    {
      "epoch": 0.4404064087534193,
      "grad_norm": 14.303231239318848,
      "learning_rate": 1.2449365183332862e-05,
      "loss": 9.1009,
      "step": 2254
    },
    {
      "epoch": 0.4406017975771786,
      "grad_norm": 8.684207916259766,
      "learning_rate": 1.2443228720270768e-05,
      "loss": 8.7528,
      "step": 2255
    },
    {
      "epoch": 0.44079718640093785,
      "grad_norm": 10.567802429199219,
      "learning_rate": 1.243709127861972e-05,
      "loss": 9.976,
      "step": 2256
    },
    {
      "epoch": 0.44099257522469715,
      "grad_norm": 7.749636173248291,
      "learning_rate": 1.2430952860837948e-05,
      "loss": 9.5294,
      "step": 2257
    },
    {
      "epoch": 0.44118796404845645,
      "grad_norm": 13.847437858581543,
      "learning_rate": 1.2424813469384084e-05,
      "loss": 11.1775,
      "step": 2258
    },
    {
      "epoch": 0.4413833528722157,
      "grad_norm": 7.527254104614258,
      "learning_rate": 1.2418673106717145e-05,
      "loss": 9.2885,
      "step": 2259
    },
    {
      "epoch": 0.441578741695975,
      "grad_norm": 8.540417671203613,
      "learning_rate": 1.2412531775296534e-05,
      "loss": 8.9638,
      "step": 2260
    },
    {
      "epoch": 0.4417741305197343,
      "grad_norm": 8.876973152160645,
      "learning_rate": 1.2406389477582048e-05,
      "loss": 10.072,
      "step": 2261
    },
    {
      "epoch": 0.44196951934349354,
      "grad_norm": 8.543230056762695,
      "learning_rate": 1.2400246216033867e-05,
      "loss": 8.6821,
      "step": 2262
    },
    {
      "epoch": 0.44216490816725285,
      "grad_norm": 12.249687194824219,
      "learning_rate": 1.2394101993112558e-05,
      "loss": 9.7186,
      "step": 2263
    },
    {
      "epoch": 0.4423602969910121,
      "grad_norm": 7.321566581726074,
      "learning_rate": 1.2387956811279069e-05,
      "loss": 9.8925,
      "step": 2264
    },
    {
      "epoch": 0.4425556858147714,
      "grad_norm": 6.195285797119141,
      "learning_rate": 1.2381810672994742e-05,
      "loss": 9.0145,
      "step": 2265
    },
    {
      "epoch": 0.4427510746385307,
      "grad_norm": 9.123198509216309,
      "learning_rate": 1.2375663580721296e-05,
      "loss": 9.6302,
      "step": 2266
    },
    {
      "epoch": 0.44294646346228994,
      "grad_norm": 10.450803756713867,
      "learning_rate": 1.2369515536920826e-05,
      "loss": 10.1972,
      "step": 2267
    },
    {
      "epoch": 0.44314185228604924,
      "grad_norm": 8.73688793182373,
      "learning_rate": 1.2363366544055821e-05,
      "loss": 9.3621,
      "step": 2268
    },
    {
      "epoch": 0.44333724110980854,
      "grad_norm": 8.812823295593262,
      "learning_rate": 1.2357216604589139e-05,
      "loss": 9.6422,
      "step": 2269
    },
    {
      "epoch": 0.4435326299335678,
      "grad_norm": 7.929076194763184,
      "learning_rate": 1.2351065720984027e-05,
      "loss": 9.6526,
      "step": 2270
    },
    {
      "epoch": 0.4437280187573271,
      "grad_norm": 7.984181880950928,
      "learning_rate": 1.2344913895704099e-05,
      "loss": 9.0684,
      "step": 2271
    },
    {
      "epoch": 0.44392340758108634,
      "grad_norm": 9.382901191711426,
      "learning_rate": 1.2338761131213357e-05,
      "loss": 9.6309,
      "step": 2272
    },
    {
      "epoch": 0.44411879640484564,
      "grad_norm": 8.028267860412598,
      "learning_rate": 1.2332607429976169e-05,
      "loss": 8.8354,
      "step": 2273
    },
    {
      "epoch": 0.44431418522860494,
      "grad_norm": 8.45240306854248,
      "learning_rate": 1.2326452794457289e-05,
      "loss": 10.0709,
      "step": 2274
    },
    {
      "epoch": 0.4445095740523642,
      "grad_norm": 7.99829626083374,
      "learning_rate": 1.2320297227121834e-05,
      "loss": 9.2649,
      "step": 2275
    },
    {
      "epoch": 0.4447049628761235,
      "grad_norm": 7.859771728515625,
      "learning_rate": 1.2314140730435305e-05,
      "loss": 8.5114,
      "step": 2276
    },
    {
      "epoch": 0.4449003516998828,
      "grad_norm": 11.960403442382812,
      "learning_rate": 1.2307983306863565e-05,
      "loss": 10.3548,
      "step": 2277
    },
    {
      "epoch": 0.44509574052364204,
      "grad_norm": 7.42790412902832,
      "learning_rate": 1.2301824958872857e-05,
      "loss": 9.0629,
      "step": 2278
    },
    {
      "epoch": 0.44529112934740134,
      "grad_norm": 7.955683708190918,
      "learning_rate": 1.229566568892979e-05,
      "loss": 8.9718,
      "step": 2279
    },
    {
      "epoch": 0.4454865181711606,
      "grad_norm": 14.33918571472168,
      "learning_rate": 1.2289505499501341e-05,
      "loss": 9.9159,
      "step": 2280
    },
    {
      "epoch": 0.4456819069949199,
      "grad_norm": 9.731377601623535,
      "learning_rate": 1.2283344393054862e-05,
      "loss": 9.4328,
      "step": 2281
    },
    {
      "epoch": 0.4458772958186792,
      "grad_norm": 10.249810218811035,
      "learning_rate": 1.227718237205806e-05,
      "loss": 9.5776,
      "step": 2282
    },
    {
      "epoch": 0.44607268464243843,
      "grad_norm": 7.93415641784668,
      "learning_rate": 1.2271019438979023e-05,
      "loss": 9.3234,
      "step": 2283
    },
    {
      "epoch": 0.44626807346619773,
      "grad_norm": 10.255139350891113,
      "learning_rate": 1.226485559628619e-05,
      "loss": 9.2531,
      "step": 2284
    },
    {
      "epoch": 0.44646346228995704,
      "grad_norm": 10.489916801452637,
      "learning_rate": 1.2258690846448382e-05,
      "loss": 9.2626,
      "step": 2285
    },
    {
      "epoch": 0.4466588511137163,
      "grad_norm": 8.567852973937988,
      "learning_rate": 1.225252519193476e-05,
      "loss": 9.6984,
      "step": 2286
    },
    {
      "epoch": 0.4468542399374756,
      "grad_norm": 9.522313117980957,
      "learning_rate": 1.2246358635214868e-05,
      "loss": 9.6132,
      "step": 2287
    },
    {
      "epoch": 0.44704962876123483,
      "grad_norm": 8.825810432434082,
      "learning_rate": 1.2240191178758598e-05,
      "loss": 9.7715,
      "step": 2288
    },
    {
      "epoch": 0.44724501758499413,
      "grad_norm": 9.867698669433594,
      "learning_rate": 1.2234022825036213e-05,
      "loss": 10.4578,
      "step": 2289
    },
    {
      "epoch": 0.44744040640875343,
      "grad_norm": 7.061227798461914,
      "learning_rate": 1.2227853576518329e-05,
      "loss": 9.2461,
      "step": 2290
    },
    {
      "epoch": 0.4476357952325127,
      "grad_norm": 9.363693237304688,
      "learning_rate": 1.222168343567592e-05,
      "loss": 10.1934,
      "step": 2291
    },
    {
      "epoch": 0.447831184056272,
      "grad_norm": 7.705371856689453,
      "learning_rate": 1.2215512404980321e-05,
      "loss": 9.0384,
      "step": 2292
    },
    {
      "epoch": 0.4480265728800313,
      "grad_norm": 10.034061431884766,
      "learning_rate": 1.2209340486903218e-05,
      "loss": 9.4011,
      "step": 2293
    },
    {
      "epoch": 0.44822196170379053,
      "grad_norm": 9.613932609558105,
      "learning_rate": 1.2203167683916658e-05,
      "loss": 9.2143,
      "step": 2294
    },
    {
      "epoch": 0.44841735052754983,
      "grad_norm": 10.576947212219238,
      "learning_rate": 1.2196993998493043e-05,
      "loss": 9.119,
      "step": 2295
    },
    {
      "epoch": 0.44861273935130913,
      "grad_norm": 7.123161315917969,
      "learning_rate": 1.219081943310512e-05,
      "loss": 9.2717,
      "step": 2296
    },
    {
      "epoch": 0.4488081281750684,
      "grad_norm": 10.57288932800293,
      "learning_rate": 1.2184643990225998e-05,
      "loss": 9.8432,
      "step": 2297
    },
    {
      "epoch": 0.4490035169988277,
      "grad_norm": 9.966277122497559,
      "learning_rate": 1.2178467672329136e-05,
      "loss": 9.3059,
      "step": 2298
    },
    {
      "epoch": 0.4491989058225869,
      "grad_norm": 8.62827205657959,
      "learning_rate": 1.2172290481888331e-05,
      "loss": 10.1742,
      "step": 2299
    },
    {
      "epoch": 0.4493942946463462,
      "grad_norm": 7.116940975189209,
      "learning_rate": 1.2166112421377749e-05,
      "loss": 8.0496,
      "step": 2300
    },
    {
      "epoch": 0.44958968347010553,
      "grad_norm": 8.18814754486084,
      "learning_rate": 1.2159933493271894e-05,
      "loss": 8.8701,
      "step": 2301
    },
    {
      "epoch": 0.4497850722938648,
      "grad_norm": 13.047163009643555,
      "learning_rate": 1.2153753700045614e-05,
      "loss": 9.4084,
      "step": 2302
    },
    {
      "epoch": 0.4499804611176241,
      "grad_norm": 6.840201377868652,
      "learning_rate": 1.2147573044174113e-05,
      "loss": 8.7816,
      "step": 2303
    },
    {
      "epoch": 0.4501758499413834,
      "grad_norm": 9.08859920501709,
      "learning_rate": 1.2141391528132931e-05,
      "loss": 9.1049,
      "step": 2304
    },
    {
      "epoch": 0.4503712387651426,
      "grad_norm": 8.56104850769043,
      "learning_rate": 1.2135209154397962e-05,
      "loss": 9.6122,
      "step": 2305
    },
    {
      "epoch": 0.4505666275889019,
      "grad_norm": 9.848429679870605,
      "learning_rate": 1.2129025925445435e-05,
      "loss": 9.074,
      "step": 2306
    },
    {
      "epoch": 0.45076201641266117,
      "grad_norm": 9.95374584197998,
      "learning_rate": 1.2122841843751925e-05,
      "loss": 10.2276,
      "step": 2307
    },
    {
      "epoch": 0.45095740523642047,
      "grad_norm": 7.756079196929932,
      "learning_rate": 1.211665691179435e-05,
      "loss": 8.8907,
      "step": 2308
    },
    {
      "epoch": 0.4511527940601798,
      "grad_norm": 8.182064056396484,
      "learning_rate": 1.2110471132049971e-05,
      "loss": 8.6392,
      "step": 2309
    },
    {
      "epoch": 0.451348182883939,
      "grad_norm": 8.239133834838867,
      "learning_rate": 1.2104284506996378e-05,
      "loss": 8.9845,
      "step": 2310
    },
    {
      "epoch": 0.4515435717076983,
      "grad_norm": 9.457928657531738,
      "learning_rate": 1.2098097039111511e-05,
      "loss": 9.528,
      "step": 2311
    },
    {
      "epoch": 0.4517389605314576,
      "grad_norm": 9.49638843536377,
      "learning_rate": 1.2091908730873641e-05,
      "loss": 10.0566,
      "step": 2312
    },
    {
      "epoch": 0.45193434935521687,
      "grad_norm": 22.126733779907227,
      "learning_rate": 1.2085719584761378e-05,
      "loss": 8.9613,
      "step": 2313
    },
    {
      "epoch": 0.45212973817897617,
      "grad_norm": 30.99988555908203,
      "learning_rate": 1.2079529603253666e-05,
      "loss": 10.345,
      "step": 2314
    },
    {
      "epoch": 0.4523251270027354,
      "grad_norm": 7.838778972625732,
      "learning_rate": 1.2073338788829787e-05,
      "loss": 9.2702,
      "step": 2315
    },
    {
      "epoch": 0.4525205158264947,
      "grad_norm": 7.920363903045654,
      "learning_rate": 1.2067147143969353e-05,
      "loss": 9.3191,
      "step": 2316
    },
    {
      "epoch": 0.452715904650254,
      "grad_norm": 9.7103271484375,
      "learning_rate": 1.206095467115231e-05,
      "loss": 8.8647,
      "step": 2317
    },
    {
      "epoch": 0.45291129347401327,
      "grad_norm": 7.919499397277832,
      "learning_rate": 1.2054761372858937e-05,
      "loss": 9.1471,
      "step": 2318
    },
    {
      "epoch": 0.45310668229777257,
      "grad_norm": 6.757690906524658,
      "learning_rate": 1.204856725156984e-05,
      "loss": 9.449,
      "step": 2319
    },
    {
      "epoch": 0.45330207112153187,
      "grad_norm": 6.938695430755615,
      "learning_rate": 1.204237230976596e-05,
      "loss": 8.7082,
      "step": 2320
    },
    {
      "epoch": 0.4534974599452911,
      "grad_norm": 6.6420369148254395,
      "learning_rate": 1.203617654992856e-05,
      "loss": 8.5528,
      "step": 2321
    },
    {
      "epoch": 0.4536928487690504,
      "grad_norm": 6.399874687194824,
      "learning_rate": 1.2029979974539233e-05,
      "loss": 8.5661,
      "step": 2322
    },
    {
      "epoch": 0.4538882375928097,
      "grad_norm": 7.898861885070801,
      "learning_rate": 1.2023782586079908e-05,
      "loss": 9.201,
      "step": 2323
    },
    {
      "epoch": 0.45408362641656896,
      "grad_norm": 9.081944465637207,
      "learning_rate": 1.2017584387032826e-05,
      "loss": 9.1618,
      "step": 2324
    },
    {
      "epoch": 0.45427901524032827,
      "grad_norm": 9.277242660522461,
      "learning_rate": 1.2011385379880555e-05,
      "loss": 9.4893,
      "step": 2325
    },
    {
      "epoch": 0.4544744040640875,
      "grad_norm": 8.17261791229248,
      "learning_rate": 1.2005185567105994e-05,
      "loss": 9.4795,
      "step": 2326
    },
    {
      "epoch": 0.4546697928878468,
      "grad_norm": 9.705378532409668,
      "learning_rate": 1.1998984951192361e-05,
      "loss": 8.5081,
      "step": 2327
    },
    {
      "epoch": 0.4548651817116061,
      "grad_norm": 8.096948623657227,
      "learning_rate": 1.1992783534623191e-05,
      "loss": 9.8414,
      "step": 2328
    },
    {
      "epoch": 0.45506057053536536,
      "grad_norm": 9.253615379333496,
      "learning_rate": 1.198658131988235e-05,
      "loss": 8.4891,
      "step": 2329
    },
    {
      "epoch": 0.45525595935912466,
      "grad_norm": 9.506009101867676,
      "learning_rate": 1.198037830945401e-05,
      "loss": 8.114,
      "step": 2330
    },
    {
      "epoch": 0.45545134818288396,
      "grad_norm": 8.674689292907715,
      "learning_rate": 1.1974174505822677e-05,
      "loss": 8.6011,
      "step": 2331
    },
    {
      "epoch": 0.4556467370066432,
      "grad_norm": 8.619670867919922,
      "learning_rate": 1.1967969911473162e-05,
      "loss": 9.3079,
      "step": 2332
    },
    {
      "epoch": 0.4558421258304025,
      "grad_norm": 9.665189743041992,
      "learning_rate": 1.1961764528890599e-05,
      "loss": 9.5076,
      "step": 2333
    },
    {
      "epoch": 0.45603751465416176,
      "grad_norm": 16.26235008239746,
      "learning_rate": 1.1955558360560438e-05,
      "loss": 9.5821,
      "step": 2334
    },
    {
      "epoch": 0.45623290347792106,
      "grad_norm": 9.290892601013184,
      "learning_rate": 1.194935140896844e-05,
      "loss": 9.7788,
      "step": 2335
    },
    {
      "epoch": 0.45642829230168036,
      "grad_norm": 8.773561477661133,
      "learning_rate": 1.194314367660068e-05,
      "loss": 9.1339,
      "step": 2336
    },
    {
      "epoch": 0.4566236811254396,
      "grad_norm": 10.138145446777344,
      "learning_rate": 1.1936935165943553e-05,
      "loss": 9.6766,
      "step": 2337
    },
    {
      "epoch": 0.4568190699491989,
      "grad_norm": 10.880075454711914,
      "learning_rate": 1.1930725879483756e-05,
      "loss": 9.9402,
      "step": 2338
    },
    {
      "epoch": 0.4570144587729582,
      "grad_norm": 9.441678047180176,
      "learning_rate": 1.19245158197083e-05,
      "loss": 9.9798,
      "step": 2339
    },
    {
      "epoch": 0.45720984759671746,
      "grad_norm": 9.227924346923828,
      "learning_rate": 1.1918304989104515e-05,
      "loss": 9.4905,
      "step": 2340
    },
    {
      "epoch": 0.45740523642047676,
      "grad_norm": 11.096552848815918,
      "learning_rate": 1.191209339016002e-05,
      "loss": 8.7374,
      "step": 2341
    },
    {
      "epoch": 0.457600625244236,
      "grad_norm": 8.83130931854248,
      "learning_rate": 1.190588102536276e-05,
      "loss": 9.031,
      "step": 2342
    },
    {
      "epoch": 0.4577960140679953,
      "grad_norm": 9.083894729614258,
      "learning_rate": 1.1899667897200978e-05,
      "loss": 9.2363,
      "step": 2343
    },
    {
      "epoch": 0.4579914028917546,
      "grad_norm": 10.509827613830566,
      "learning_rate": 1.1893454008163226e-05,
      "loss": 9.105,
      "step": 2344
    },
    {
      "epoch": 0.45818679171551385,
      "grad_norm": 8.675061225891113,
      "learning_rate": 1.1887239360738356e-05,
      "loss": 9.1982,
      "step": 2345
    },
    {
      "epoch": 0.45838218053927315,
      "grad_norm": 9.243605613708496,
      "learning_rate": 1.1881023957415535e-05,
      "loss": 9.5825,
      "step": 2346
    },
    {
      "epoch": 0.45857756936303246,
      "grad_norm": 17.001510620117188,
      "learning_rate": 1.1874807800684217e-05,
      "loss": 10.4223,
      "step": 2347
    },
    {
      "epoch": 0.4587729581867917,
      "grad_norm": 8.395930290222168,
      "learning_rate": 1.186859089303417e-05,
      "loss": 9.7094,
      "step": 2348
    },
    {
      "epoch": 0.458968347010551,
      "grad_norm": 9.87113094329834,
      "learning_rate": 1.1862373236955454e-05,
      "loss": 9.786,
      "step": 2349
    },
    {
      "epoch": 0.4591637358343103,
      "grad_norm": 10.49590015411377,
      "learning_rate": 1.1856154834938437e-05,
      "loss": 9.9623,
      "step": 2350
    },
    {
      "epoch": 0.45935912465806955,
      "grad_norm": 10.947039604187012,
      "learning_rate": 1.1849935689473786e-05,
      "loss": 10.264,
      "step": 2351
    },
    {
      "epoch": 0.45955451348182885,
      "grad_norm": 8.210731506347656,
      "learning_rate": 1.1843715803052453e-05,
      "loss": 9.5774,
      "step": 2352
    },
    {
      "epoch": 0.4597499023055881,
      "grad_norm": 6.829330921173096,
      "learning_rate": 1.1837495178165706e-05,
      "loss": 9.1155,
      "step": 2353
    },
    {
      "epoch": 0.4599452911293474,
      "grad_norm": 7.158986568450928,
      "learning_rate": 1.1831273817305088e-05,
      "loss": 9.4125,
      "step": 2354
    },
    {
      "epoch": 0.4601406799531067,
      "grad_norm": 11.706311225891113,
      "learning_rate": 1.1825051722962454e-05,
      "loss": 9.8573,
      "step": 2355
    },
    {
      "epoch": 0.46033606877686595,
      "grad_norm": 40.86763000488281,
      "learning_rate": 1.1818828897629941e-05,
      "loss": 9.155,
      "step": 2356
    },
    {
      "epoch": 0.46053145760062525,
      "grad_norm": 10.87190055847168,
      "learning_rate": 1.1812605343799989e-05,
      "loss": 9.4604,
      "step": 2357
    },
    {
      "epoch": 0.46072684642438455,
      "grad_norm": 9.73469352722168,
      "learning_rate": 1.180638106396532e-05,
      "loss": 9.801,
      "step": 2358
    },
    {
      "epoch": 0.4609222352481438,
      "grad_norm": 7.995573997497559,
      "learning_rate": 1.1800156060618955e-05,
      "loss": 10.1415,
      "step": 2359
    },
    {
      "epoch": 0.4611176240719031,
      "grad_norm": 8.469133377075195,
      "learning_rate": 1.1793930336254195e-05,
      "loss": 8.9077,
      "step": 2360
    },
    {
      "epoch": 0.46131301289566234,
      "grad_norm": 18.08845329284668,
      "learning_rate": 1.1787703893364645e-05,
      "loss": 9.7153,
      "step": 2361
    },
    {
      "epoch": 0.46150840171942165,
      "grad_norm": 9.333691596984863,
      "learning_rate": 1.1781476734444181e-05,
      "loss": 9.3963,
      "step": 2362
    },
    {
      "epoch": 0.46170379054318095,
      "grad_norm": 9.010549545288086,
      "learning_rate": 1.1775248861986978e-05,
      "loss": 9.5966,
      "step": 2363
    },
    {
      "epoch": 0.4618991793669402,
      "grad_norm": 11.042677879333496,
      "learning_rate": 1.176902027848749e-05,
      "loss": 9.724,
      "step": 2364
    },
    {
      "epoch": 0.4620945681906995,
      "grad_norm": 10.08812427520752,
      "learning_rate": 1.176279098644046e-05,
      "loss": 9.7276,
      "step": 2365
    },
    {
      "epoch": 0.4622899570144588,
      "grad_norm": 13.08978271484375,
      "learning_rate": 1.1756560988340912e-05,
      "loss": 9.7234,
      "step": 2366
    },
    {
      "epoch": 0.46248534583821804,
      "grad_norm": 10.918002128601074,
      "learning_rate": 1.1750330286684153e-05,
      "loss": 8.2166,
      "step": 2367
    },
    {
      "epoch": 0.46268073466197734,
      "grad_norm": 8.465288162231445,
      "learning_rate": 1.174409888396577e-05,
      "loss": 8.8377,
      "step": 2368
    },
    {
      "epoch": 0.4628761234857366,
      "grad_norm": 11.169697761535645,
      "learning_rate": 1.173786678268164e-05,
      "loss": 8.7151,
      "step": 2369
    },
    {
      "epoch": 0.4630715123094959,
      "grad_norm": 9.257155418395996,
      "learning_rate": 1.1731633985327906e-05,
      "loss": 9.4025,
      "step": 2370
    },
    {
      "epoch": 0.4632669011332552,
      "grad_norm": 8.561311721801758,
      "learning_rate": 1.1725400494401e-05,
      "loss": 9.0156,
      "step": 2371
    },
    {
      "epoch": 0.46346228995701444,
      "grad_norm": 8.800094604492188,
      "learning_rate": 1.171916631239763e-05,
      "loss": 9.1216,
      "step": 2372
    },
    {
      "epoch": 0.46365767878077374,
      "grad_norm": 11.316170692443848,
      "learning_rate": 1.1712931441814776e-05,
      "loss": 9.9333,
      "step": 2373
    },
    {
      "epoch": 0.46385306760453304,
      "grad_norm": 9.657631874084473,
      "learning_rate": 1.1706695885149702e-05,
      "loss": 10.1079,
      "step": 2374
    },
    {
      "epoch": 0.4640484564282923,
      "grad_norm": 8.509729385375977,
      "learning_rate": 1.1700459644899938e-05,
      "loss": 9.6815,
      "step": 2375
    },
    {
      "epoch": 0.4642438452520516,
      "grad_norm": 9.730488777160645,
      "learning_rate": 1.1694222723563291e-05,
      "loss": 9.8981,
      "step": 2376
    },
    {
      "epoch": 0.46443923407581084,
      "grad_norm": 6.831174373626709,
      "learning_rate": 1.1687985123637846e-05,
      "loss": 8.5715,
      "step": 2377
    },
    {
      "epoch": 0.46463462289957014,
      "grad_norm": 36.317893981933594,
      "learning_rate": 1.1681746847621951e-05,
      "loss": 10.3437,
      "step": 2378
    },
    {
      "epoch": 0.46483001172332944,
      "grad_norm": 9.85550308227539,
      "learning_rate": 1.1675507898014235e-05,
      "loss": 9.3706,
      "step": 2379
    },
    {
      "epoch": 0.4650254005470887,
      "grad_norm": 10.740527153015137,
      "learning_rate": 1.1669268277313585e-05,
      "loss": 9.1447,
      "step": 2380
    },
    {
      "epoch": 0.465220789370848,
      "grad_norm": 6.926548957824707,
      "learning_rate": 1.1663027988019166e-05,
      "loss": 9.047,
      "step": 2381
    },
    {
      "epoch": 0.4654161781946073,
      "grad_norm": 8.198529243469238,
      "learning_rate": 1.1656787032630405e-05,
      "loss": 8.1967,
      "step": 2382
    },
    {
      "epoch": 0.46561156701836653,
      "grad_norm": 9.073921203613281,
      "learning_rate": 1.1650545413647001e-05,
      "loss": 9.9589,
      "step": 2383
    },
    {
      "epoch": 0.46580695584212584,
      "grad_norm": 13.952261924743652,
      "learning_rate": 1.1644303133568915e-05,
      "loss": 9.2896,
      "step": 2384
    },
    {
      "epoch": 0.46600234466588514,
      "grad_norm": 8.263925552368164,
      "learning_rate": 1.1638060194896378e-05,
      "loss": 9.2413,
      "step": 2385
    },
    {
      "epoch": 0.4661977334896444,
      "grad_norm": 22.786531448364258,
      "learning_rate": 1.163181660012987e-05,
      "loss": 9.0289,
      "step": 2386
    },
    {
      "epoch": 0.4663931223134037,
      "grad_norm": 7.734263896942139,
      "learning_rate": 1.1625572351770157e-05,
      "loss": 8.9236,
      "step": 2387
    },
    {
      "epoch": 0.46658851113716293,
      "grad_norm": 17.39560890197754,
      "learning_rate": 1.1619327452318248e-05,
      "loss": 9.8094,
      "step": 2388
    },
    {
      "epoch": 0.46678389996092223,
      "grad_norm": 7.780709743499756,
      "learning_rate": 1.1613081904275419e-05,
      "loss": 8.9536,
      "step": 2389
    },
    {
      "epoch": 0.46697928878468153,
      "grad_norm": 8.669890403747559,
      "learning_rate": 1.1606835710143207e-05,
      "loss": 8.8066,
      "step": 2390
    },
    {
      "epoch": 0.4671746776084408,
      "grad_norm": 25.645999908447266,
      "learning_rate": 1.1600588872423406e-05,
      "loss": 9.4202,
      "step": 2391
    },
    {
      "epoch": 0.4673700664322001,
      "grad_norm": 9.888745307922363,
      "learning_rate": 1.1594341393618071e-05,
      "loss": 10.3101,
      "step": 2392
    },
    {
      "epoch": 0.4675654552559594,
      "grad_norm": 10.650516510009766,
      "learning_rate": 1.158809327622951e-05,
      "loss": 9.7502,
      "step": 2393
    },
    {
      "epoch": 0.46776084407971863,
      "grad_norm": 9.753227233886719,
      "learning_rate": 1.1581844522760287e-05,
      "loss": 8.6243,
      "step": 2394
    },
    {
      "epoch": 0.46795623290347793,
      "grad_norm": 8.801817893981934,
      "learning_rate": 1.157559513571322e-05,
      "loss": 8.637,
      "step": 2395
    },
    {
      "epoch": 0.4681516217272372,
      "grad_norm": 11.485738754272461,
      "learning_rate": 1.1569345117591384e-05,
      "loss": 10.5581,
      "step": 2396
    },
    {
      "epoch": 0.4683470105509965,
      "grad_norm": 43.89450454711914,
      "learning_rate": 1.1563094470898106e-05,
      "loss": 9.4397,
      "step": 2397
    },
    {
      "epoch": 0.4685423993747558,
      "grad_norm": 11.00967788696289,
      "learning_rate": 1.1556843198136966e-05,
      "loss": 10.098,
      "step": 2398
    },
    {
      "epoch": 0.468737788198515,
      "grad_norm": 8.109318733215332,
      "learning_rate": 1.155059130181179e-05,
      "loss": 9.6306,
      "step": 2399
    },
    {
      "epoch": 0.46893317702227433,
      "grad_norm": 7.781679630279541,
      "learning_rate": 1.1544338784426656e-05,
      "loss": 8.9134,
      "step": 2400
    },
    {
      "epoch": 0.46912856584603363,
      "grad_norm": 15.912310600280762,
      "learning_rate": 1.1538085648485895e-05,
      "loss": 8.9027,
      "step": 2401
    },
    {
      "epoch": 0.4693239546697929,
      "grad_norm": 9.630857467651367,
      "learning_rate": 1.153183189649408e-05,
      "loss": 9.6047,
      "step": 2402
    },
    {
      "epoch": 0.4695193434935522,
      "grad_norm": 7.696481704711914,
      "learning_rate": 1.1525577530956029e-05,
      "loss": 9.589,
      "step": 2403
    },
    {
      "epoch": 0.4697147323173114,
      "grad_norm": 11.7659273147583,
      "learning_rate": 1.1519322554376816e-05,
      "loss": 10.0371,
      "step": 2404
    },
    {
      "epoch": 0.4699101211410707,
      "grad_norm": 8.038793563842773,
      "learning_rate": 1.151306696926175e-05,
      "loss": 9.0711,
      "step": 2405
    },
    {
      "epoch": 0.47010550996483,
      "grad_norm": 8.523825645446777,
      "learning_rate": 1.1506810778116389e-05,
      "loss": 8.9531,
      "step": 2406
    },
    {
      "epoch": 0.47030089878858927,
      "grad_norm": 8.917826652526855,
      "learning_rate": 1.1500553983446527e-05,
      "loss": 9.3239,
      "step": 2407
    },
    {
      "epoch": 0.4704962876123486,
      "grad_norm": 12.260668754577637,
      "learning_rate": 1.1494296587758212e-05,
      "loss": 10.3139,
      "step": 2408
    },
    {
      "epoch": 0.4706916764361079,
      "grad_norm": 9.119510650634766,
      "learning_rate": 1.148803859355772e-05,
      "loss": 9.6787,
      "step": 2409
    },
    {
      "epoch": 0.4708870652598671,
      "grad_norm": 8.856493949890137,
      "learning_rate": 1.148178000335157e-05,
      "loss": 10.5187,
      "step": 2410
    },
    {
      "epoch": 0.4710824540836264,
      "grad_norm": 11.283249855041504,
      "learning_rate": 1.1475520819646526e-05,
      "loss": 10.0481,
      "step": 2411
    },
    {
      "epoch": 0.4712778429073857,
      "grad_norm": 6.657747745513916,
      "learning_rate": 1.1469261044949584e-05,
      "loss": 7.8504,
      "step": 2412
    },
    {
      "epoch": 0.47147323173114497,
      "grad_norm": 10.082439422607422,
      "learning_rate": 1.146300068176798e-05,
      "loss": 10.1751,
      "step": 2413
    },
    {
      "epoch": 0.47166862055490427,
      "grad_norm": 7.962443828582764,
      "learning_rate": 1.1456739732609179e-05,
      "loss": 9.62,
      "step": 2414
    },
    {
      "epoch": 0.4718640093786635,
      "grad_norm": 6.457741737365723,
      "learning_rate": 1.145047819998089e-05,
      "loss": 8.9817,
      "step": 2415
    },
    {
      "epoch": 0.4720593982024228,
      "grad_norm": 9.077668190002441,
      "learning_rate": 1.1444216086391052e-05,
      "loss": 9.7593,
      "step": 2416
    },
    {
      "epoch": 0.4722547870261821,
      "grad_norm": 9.026140213012695,
      "learning_rate": 1.143795339434783e-05,
      "loss": 9.2714,
      "step": 2417
    },
    {
      "epoch": 0.47245017584994137,
      "grad_norm": 6.757838249206543,
      "learning_rate": 1.1431690126359632e-05,
      "loss": 8.3541,
      "step": 2418
    },
    {
      "epoch": 0.47264556467370067,
      "grad_norm": 10.788949966430664,
      "learning_rate": 1.142542628493509e-05,
      "loss": 8.5847,
      "step": 2419
    },
    {
      "epoch": 0.47284095349745997,
      "grad_norm": 8.040674209594727,
      "learning_rate": 1.1419161872583065e-05,
      "loss": 8.9276,
      "step": 2420
    },
    {
      "epoch": 0.4730363423212192,
      "grad_norm": 7.822730541229248,
      "learning_rate": 1.1412896891812655e-05,
      "loss": 8.8715,
      "step": 2421
    },
    {
      "epoch": 0.4732317311449785,
      "grad_norm": 9.462562561035156,
      "learning_rate": 1.140663134513317e-05,
      "loss": 8.9585,
      "step": 2422
    },
    {
      "epoch": 0.47342711996873776,
      "grad_norm": 10.111047744750977,
      "learning_rate": 1.140036523505416e-05,
      "loss": 9.907,
      "step": 2423
    },
    {
      "epoch": 0.47362250879249707,
      "grad_norm": 9.930951118469238,
      "learning_rate": 1.1394098564085399e-05,
      "loss": 8.905,
      "step": 2424
    },
    {
      "epoch": 0.47381789761625637,
      "grad_norm": 12.305794715881348,
      "learning_rate": 1.1387831334736878e-05,
      "loss": 10.4031,
      "step": 2425
    },
    {
      "epoch": 0.4740132864400156,
      "grad_norm": 9.197370529174805,
      "learning_rate": 1.1381563549518823e-05,
      "loss": 9.1929,
      "step": 2426
    },
    {
      "epoch": 0.4742086752637749,
      "grad_norm": 10.884814262390137,
      "learning_rate": 1.1375295210941674e-05,
      "loss": 8.6907,
      "step": 2427
    },
    {
      "epoch": 0.4744040640875342,
      "grad_norm": 7.242812156677246,
      "learning_rate": 1.1369026321516094e-05,
      "loss": 9.2762,
      "step": 2428
    },
    {
      "epoch": 0.47459945291129346,
      "grad_norm": 13.42102336883545,
      "learning_rate": 1.136275688375297e-05,
      "loss": 9.0727,
      "step": 2429
    },
    {
      "epoch": 0.47479484173505276,
      "grad_norm": 8.243850708007812,
      "learning_rate": 1.1356486900163404e-05,
      "loss": 9.6639,
      "step": 2430
    },
    {
      "epoch": 0.474990230558812,
      "grad_norm": 9.975531578063965,
      "learning_rate": 1.1350216373258723e-05,
      "loss": 9.575,
      "step": 2431
    },
    {
      "epoch": 0.4751856193825713,
      "grad_norm": 9.604329109191895,
      "learning_rate": 1.1343945305550464e-05,
      "loss": 9.822,
      "step": 2432
    },
    {
      "epoch": 0.4753810082063306,
      "grad_norm": 9.020915985107422,
      "learning_rate": 1.1337673699550382e-05,
      "loss": 8.9249,
      "step": 2433
    },
    {
      "epoch": 0.47557639703008986,
      "grad_norm": 7.667820453643799,
      "learning_rate": 1.1331401557770461e-05,
      "loss": 9.5894,
      "step": 2434
    },
    {
      "epoch": 0.47577178585384916,
      "grad_norm": 7.572841644287109,
      "learning_rate": 1.1325128882722876e-05,
      "loss": 8.8098,
      "step": 2435
    },
    {
      "epoch": 0.47596717467760846,
      "grad_norm": 11.557333946228027,
      "learning_rate": 1.131885567692003e-05,
      "loss": 10.3338,
      "step": 2436
    },
    {
      "epoch": 0.4761625635013677,
      "grad_norm": 7.1445841789245605,
      "learning_rate": 1.1312581942874546e-05,
      "loss": 8.5131,
      "step": 2437
    },
    {
      "epoch": 0.476357952325127,
      "grad_norm": 7.750548362731934,
      "learning_rate": 1.1306307683099237e-05,
      "loss": 9.4567,
      "step": 2438
    },
    {
      "epoch": 0.4765533411488863,
      "grad_norm": 9.577622413635254,
      "learning_rate": 1.1300032900107147e-05,
      "loss": 9.1148,
      "step": 2439
    },
    {
      "epoch": 0.47674872997264556,
      "grad_norm": 10.267053604125977,
      "learning_rate": 1.1293757596411522e-05,
      "loss": 9.7057,
      "step": 2440
    },
    {
      "epoch": 0.47694411879640486,
      "grad_norm": 9.346061706542969,
      "learning_rate": 1.128748177452581e-05,
      "loss": 9.019,
      "step": 2441
    },
    {
      "epoch": 0.4771395076201641,
      "grad_norm": 7.570725917816162,
      "learning_rate": 1.1281205436963676e-05,
      "loss": 8.3706,
      "step": 2442
    },
    {
      "epoch": 0.4773348964439234,
      "grad_norm": 7.138862133026123,
      "learning_rate": 1.1274928586238987e-05,
      "loss": 8.349,
      "step": 2443
    },
    {
      "epoch": 0.4775302852676827,
      "grad_norm": 10.648791313171387,
      "learning_rate": 1.1268651224865818e-05,
      "loss": 9.6219,
      "step": 2444
    },
    {
      "epoch": 0.47772567409144195,
      "grad_norm": 9.657983779907227,
      "learning_rate": 1.1262373355358447e-05,
      "loss": 8.6772,
      "step": 2445
    },
    {
      "epoch": 0.47792106291520126,
      "grad_norm": 7.355937480926514,
      "learning_rate": 1.1256094980231356e-05,
      "loss": 8.7624,
      "step": 2446
    },
    {
      "epoch": 0.47811645173896056,
      "grad_norm": 9.921847343444824,
      "learning_rate": 1.1249816101999227e-05,
      "loss": 8.6981,
      "step": 2447
    },
    {
      "epoch": 0.4783118405627198,
      "grad_norm": 10.855193138122559,
      "learning_rate": 1.1243536723176947e-05,
      "loss": 9.7601,
      "step": 2448
    },
    {
      "epoch": 0.4785072293864791,
      "grad_norm": 8.516083717346191,
      "learning_rate": 1.1237256846279603e-05,
      "loss": 8.7526,
      "step": 2449
    },
    {
      "epoch": 0.47870261821023835,
      "grad_norm": 9.826043128967285,
      "learning_rate": 1.1230976473822478e-05,
      "loss": 9.6618,
      "step": 2450
    },
    {
      "epoch": 0.47889800703399765,
      "grad_norm": 8.2230224609375,
      "learning_rate": 1.1224695608321057e-05,
      "loss": 9.1489,
      "step": 2451
    },
    {
      "epoch": 0.47909339585775695,
      "grad_norm": 53.27069091796875,
      "learning_rate": 1.1218414252291024e-05,
      "loss": 10.0209,
      "step": 2452
    },
    {
      "epoch": 0.4792887846815162,
      "grad_norm": 9.501374244689941,
      "learning_rate": 1.1212132408248256e-05,
      "loss": 9.2953,
      "step": 2453
    },
    {
      "epoch": 0.4794841735052755,
      "grad_norm": 8.565271377563477,
      "learning_rate": 1.1205850078708824e-05,
      "loss": 9.3541,
      "step": 2454
    },
    {
      "epoch": 0.4796795623290348,
      "grad_norm": 7.600032806396484,
      "learning_rate": 1.1199567266189e-05,
      "loss": 9.3391,
      "step": 2455
    },
    {
      "epoch": 0.47987495115279405,
      "grad_norm": 7.484233856201172,
      "learning_rate": 1.1193283973205242e-05,
      "loss": 9.673,
      "step": 2456
    },
    {
      "epoch": 0.48007033997655335,
      "grad_norm": 9.007365226745605,
      "learning_rate": 1.1187000202274207e-05,
      "loss": 10.5114,
      "step": 2457
    },
    {
      "epoch": 0.4802657288003126,
      "grad_norm": 10.141722679138184,
      "learning_rate": 1.118071595591274e-05,
      "loss": 10.2439,
      "step": 2458
    },
    {
      "epoch": 0.4804611176240719,
      "grad_norm": 26.891250610351562,
      "learning_rate": 1.1174431236637875e-05,
      "loss": 9.4861,
      "step": 2459
    },
    {
      "epoch": 0.4806565064478312,
      "grad_norm": 9.505500793457031,
      "learning_rate": 1.1168146046966838e-05,
      "loss": 9.4145,
      "step": 2460
    },
    {
      "epoch": 0.48085189527159045,
      "grad_norm": 8.005602836608887,
      "learning_rate": 1.1161860389417041e-05,
      "loss": 8.5351,
      "step": 2461
    },
    {
      "epoch": 0.48104728409534975,
      "grad_norm": 7.418105602264404,
      "learning_rate": 1.1155574266506089e-05,
      "loss": 9.4643,
      "step": 2462
    },
    {
      "epoch": 0.48124267291910905,
      "grad_norm": 9.839845657348633,
      "learning_rate": 1.1149287680751763e-05,
      "loss": 10.0309,
      "step": 2463
    },
    {
      "epoch": 0.4814380617428683,
      "grad_norm": 8.617295265197754,
      "learning_rate": 1.1143000634672043e-05,
      "loss": 8.7564,
      "step": 2464
    },
    {
      "epoch": 0.4816334505666276,
      "grad_norm": 10.28796100616455,
      "learning_rate": 1.1136713130785079e-05,
      "loss": 9.6796,
      "step": 2465
    },
    {
      "epoch": 0.48182883939038684,
      "grad_norm": 12.032356262207031,
      "learning_rate": 1.1130425171609218e-05,
      "loss": 9.0274,
      "step": 2466
    },
    {
      "epoch": 0.48202422821414614,
      "grad_norm": 15.842183113098145,
      "learning_rate": 1.112413675966298e-05,
      "loss": 10.0809,
      "step": 2467
    },
    {
      "epoch": 0.48221961703790545,
      "grad_norm": 8.806893348693848,
      "learning_rate": 1.1117847897465071e-05,
      "loss": 9.4578,
      "step": 2468
    },
    {
      "epoch": 0.4824150058616647,
      "grad_norm": 24.973302841186523,
      "learning_rate": 1.111155858753437e-05,
      "loss": 9.3003,
      "step": 2469
    },
    {
      "epoch": 0.482610394685424,
      "grad_norm": 63.59320831298828,
      "learning_rate": 1.1105268832389946e-05,
      "loss": 9.683,
      "step": 2470
    },
    {
      "epoch": 0.4828057835091833,
      "grad_norm": 8.895898818969727,
      "learning_rate": 1.109897863455104e-05,
      "loss": 9.9377,
      "step": 2471
    },
    {
      "epoch": 0.48300117233294254,
      "grad_norm": 8.31209659576416,
      "learning_rate": 1.1092687996537075e-05,
      "loss": 9.3796,
      "step": 2472
    },
    {
      "epoch": 0.48319656115670184,
      "grad_norm": 8.310145378112793,
      "learning_rate": 1.1086396920867642e-05,
      "loss": 9.5811,
      "step": 2473
    },
    {
      "epoch": 0.48339194998046114,
      "grad_norm": 9.453574180603027,
      "learning_rate": 1.1080105410062512e-05,
      "loss": 9.3392,
      "step": 2474
    },
    {
      "epoch": 0.4835873388042204,
      "grad_norm": 9.653833389282227,
      "learning_rate": 1.1073813466641633e-05,
      "loss": 8.8944,
      "step": 2475
    },
    {
      "epoch": 0.4837827276279797,
      "grad_norm": 9.032988548278809,
      "learning_rate": 1.1067521093125123e-05,
      "loss": 9.4389,
      "step": 2476
    },
    {
      "epoch": 0.48397811645173894,
      "grad_norm": 9.604551315307617,
      "learning_rate": 1.106122829203327e-05,
      "loss": 9.1107,
      "step": 2477
    },
    {
      "epoch": 0.48417350527549824,
      "grad_norm": 10.697450637817383,
      "learning_rate": 1.1054935065886541e-05,
      "loss": 9.6396,
      "step": 2478
    },
    {
      "epoch": 0.48436889409925754,
      "grad_norm": 8.172533988952637,
      "learning_rate": 1.1048641417205563e-05,
      "loss": 8.8635,
      "step": 2479
    },
    {
      "epoch": 0.4845642829230168,
      "grad_norm": 9.067036628723145,
      "learning_rate": 1.1042347348511145e-05,
      "loss": 10.2947,
      "step": 2480
    },
    {
      "epoch": 0.4847596717467761,
      "grad_norm": 12.501672744750977,
      "learning_rate": 1.103605286232425e-05,
      "loss": 9.8823,
      "step": 2481
    },
    {
      "epoch": 0.4849550605705354,
      "grad_norm": 13.687049865722656,
      "learning_rate": 1.1029757961166019e-05,
      "loss": 9.2825,
      "step": 2482
    },
    {
      "epoch": 0.48515044939429464,
      "grad_norm": 9.006345748901367,
      "learning_rate": 1.1023462647557752e-05,
      "loss": 9.474,
      "step": 2483
    },
    {
      "epoch": 0.48534583821805394,
      "grad_norm": 8.601507186889648,
      "learning_rate": 1.1017166924020921e-05,
      "loss": 9.5004,
      "step": 2484
    },
    {
      "epoch": 0.4855412270418132,
      "grad_norm": 7.638338565826416,
      "learning_rate": 1.101087079307716e-05,
      "loss": 8.4473,
      "step": 2485
    },
    {
      "epoch": 0.4857366158655725,
      "grad_norm": 10.555083274841309,
      "learning_rate": 1.100457425724826e-05,
      "loss": 8.6913,
      "step": 2486
    },
    {
      "epoch": 0.4859320046893318,
      "grad_norm": 8.18532943725586,
      "learning_rate": 1.0998277319056181e-05,
      "loss": 9.7484,
      "step": 2487
    },
    {
      "epoch": 0.48612739351309103,
      "grad_norm": 9.937145233154297,
      "learning_rate": 1.0991979981023044e-05,
      "loss": 8.4777,
      "step": 2488
    },
    {
      "epoch": 0.48632278233685033,
      "grad_norm": 9.026718139648438,
      "learning_rate": 1.098568224567113e-05,
      "loss": 8.9193,
      "step": 2489
    },
    {
      "epoch": 0.48651817116060964,
      "grad_norm": 9.864764213562012,
      "learning_rate": 1.0979384115522871e-05,
      "loss": 10.205,
      "step": 2490
    },
    {
      "epoch": 0.4867135599843689,
      "grad_norm": 7.276735305786133,
      "learning_rate": 1.097308559310087e-05,
      "loss": 8.0698,
      "step": 2491
    },
    {
      "epoch": 0.4869089488081282,
      "grad_norm": 7.894010543823242,
      "learning_rate": 1.0966786680927875e-05,
      "loss": 10.2766,
      "step": 2492
    },
    {
      "epoch": 0.48710433763188743,
      "grad_norm": 8.936746597290039,
      "learning_rate": 1.0960487381526801e-05,
      "loss": 9.2279,
      "step": 2493
    },
    {
      "epoch": 0.48729972645564673,
      "grad_norm": 8.326064109802246,
      "learning_rate": 1.095418769742071e-05,
      "loss": 8.7967,
      "step": 2494
    },
    {
      "epoch": 0.48749511527940603,
      "grad_norm": 10.585051536560059,
      "learning_rate": 1.0947887631132823e-05,
      "loss": 9.2622,
      "step": 2495
    },
    {
      "epoch": 0.4876905041031653,
      "grad_norm": 7.402519702911377,
      "learning_rate": 1.0941587185186512e-05,
      "loss": 8.8416,
      "step": 2496
    },
    {
      "epoch": 0.4878858929269246,
      "grad_norm": 8.184988975524902,
      "learning_rate": 1.0935286362105298e-05,
      "loss": 9.509,
      "step": 2497
    },
    {
      "epoch": 0.4880812817506839,
      "grad_norm": 10.121121406555176,
      "learning_rate": 1.092898516441286e-05,
      "loss": 9.354,
      "step": 2498
    },
    {
      "epoch": 0.48827667057444313,
      "grad_norm": 9.630965232849121,
      "learning_rate": 1.092268359463302e-05,
      "loss": 8.4544,
      "step": 2499
    },
    {
      "epoch": 0.48847205939820243,
      "grad_norm": 7.4092183113098145,
      "learning_rate": 1.0916381655289756e-05,
      "loss": 8.9632,
      "step": 2500
    },
    {
      "epoch": 0.48866744822196173,
      "grad_norm": 9.727022171020508,
      "learning_rate": 1.091007934890719e-05,
      "loss": 8.9544,
      "step": 2501
    },
    {
      "epoch": 0.488862837045721,
      "grad_norm": 8.538817405700684,
      "learning_rate": 1.090377667800959e-05,
      "loss": 8.3954,
      "step": 2502
    },
    {
      "epoch": 0.4890582258694803,
      "grad_norm": 9.476492881774902,
      "learning_rate": 1.0897473645121373e-05,
      "loss": 9.1442,
      "step": 2503
    },
    {
      "epoch": 0.4892536146932395,
      "grad_norm": 18.45824432373047,
      "learning_rate": 1.0891170252767094e-05,
      "loss": 8.5243,
      "step": 2504
    },
    {
      "epoch": 0.4894490035169988,
      "grad_norm": 9.022347450256348,
      "learning_rate": 1.0884866503471463e-05,
      "loss": 9.0409,
      "step": 2505
    },
    {
      "epoch": 0.48964439234075813,
      "grad_norm": 11.24704647064209,
      "learning_rate": 1.0878562399759324e-05,
      "loss": 9.3097,
      "step": 2506
    },
    {
      "epoch": 0.4898397811645174,
      "grad_norm": 14.451510429382324,
      "learning_rate": 1.087225794415567e-05,
      "loss": 9.3767,
      "step": 2507
    },
    {
      "epoch": 0.4900351699882767,
      "grad_norm": 17.859617233276367,
      "learning_rate": 1.0865953139185625e-05,
      "loss": 10.09,
      "step": 2508
    },
    {
      "epoch": 0.490230558812036,
      "grad_norm": 6.762642860412598,
      "learning_rate": 1.0859647987374467e-05,
      "loss": 8.9209,
      "step": 2509
    },
    {
      "epoch": 0.4904259476357952,
      "grad_norm": 10.67780876159668,
      "learning_rate": 1.0853342491247598e-05,
      "loss": 9.9544,
      "step": 2510
    },
    {
      "epoch": 0.4906213364595545,
      "grad_norm": 10.100765228271484,
      "learning_rate": 1.084703665333057e-05,
      "loss": 9.2581,
      "step": 2511
    },
    {
      "epoch": 0.49081672528331377,
      "grad_norm": 9.13901138305664,
      "learning_rate": 1.0840730476149064e-05,
      "loss": 9.1497,
      "step": 2512
    },
    {
      "epoch": 0.49101211410707307,
      "grad_norm": 10.920374870300293,
      "learning_rate": 1.0834423962228902e-05,
      "loss": 9.2326,
      "step": 2513
    },
    {
      "epoch": 0.4912075029308324,
      "grad_norm": 10.368356704711914,
      "learning_rate": 1.0828117114096035e-05,
      "loss": 9.8312,
      "step": 2514
    },
    {
      "epoch": 0.4914028917545916,
      "grad_norm": 6.879678249359131,
      "learning_rate": 1.0821809934276555e-05,
      "loss": 9.4539,
      "step": 2515
    },
    {
      "epoch": 0.4915982805783509,
      "grad_norm": 8.457839012145996,
      "learning_rate": 1.0815502425296685e-05,
      "loss": 9.1,
      "step": 2516
    },
    {
      "epoch": 0.4917936694021102,
      "grad_norm": 8.530153274536133,
      "learning_rate": 1.0809194589682772e-05,
      "loss": 10.0003,
      "step": 2517
    },
    {
      "epoch": 0.49198905822586947,
      "grad_norm": 8.18465518951416,
      "learning_rate": 1.0802886429961303e-05,
      "loss": 9.053,
      "step": 2518
    },
    {
      "epoch": 0.49218444704962877,
      "grad_norm": 9.027351379394531,
      "learning_rate": 1.0796577948658893e-05,
      "loss": 9.2156,
      "step": 2519
    },
    {
      "epoch": 0.492379835873388,
      "grad_norm": 10.641546249389648,
      "learning_rate": 1.0790269148302283e-05,
      "loss": 9.5959,
      "step": 2520
    },
    {
      "epoch": 0.4925752246971473,
      "grad_norm": 10.388471603393555,
      "learning_rate": 1.0783960031418345e-05,
      "loss": 10.1863,
      "step": 2521
    },
    {
      "epoch": 0.4927706135209066,
      "grad_norm": 10.69629192352295,
      "learning_rate": 1.0777650600534076e-05,
      "loss": 9.5259,
      "step": 2522
    },
    {
      "epoch": 0.49296600234466587,
      "grad_norm": 8.71397876739502,
      "learning_rate": 1.0771340858176595e-05,
      "loss": 9.2118,
      "step": 2523
    },
    {
      "epoch": 0.49316139116842517,
      "grad_norm": 8.098600387573242,
      "learning_rate": 1.0765030806873155e-05,
      "loss": 9.6289,
      "step": 2524
    },
    {
      "epoch": 0.49335677999218447,
      "grad_norm": 10.088919639587402,
      "learning_rate": 1.0758720449151125e-05,
      "loss": 8.5031,
      "step": 2525
    },
    {
      "epoch": 0.4935521688159437,
      "grad_norm": 7.400728225708008,
      "learning_rate": 1.0752409787538e-05,
      "loss": 9.5247,
      "step": 2526
    },
    {
      "epoch": 0.493747557639703,
      "grad_norm": 8.490511894226074,
      "learning_rate": 1.0746098824561397e-05,
      "loss": 9.7487,
      "step": 2527
    },
    {
      "epoch": 0.49394294646346226,
      "grad_norm": 10.189058303833008,
      "learning_rate": 1.0739787562749049e-05,
      "loss": 9.1151,
      "step": 2528
    },
    {
      "epoch": 0.49413833528722156,
      "grad_norm": 9.887974739074707,
      "learning_rate": 1.0733476004628817e-05,
      "loss": 10.338,
      "step": 2529
    },
    {
      "epoch": 0.49433372411098087,
      "grad_norm": 27.863605499267578,
      "learning_rate": 1.0727164152728671e-05,
      "loss": 9.5593,
      "step": 2530
    },
    {
      "epoch": 0.4945291129347401,
      "grad_norm": 7.720983028411865,
      "learning_rate": 1.0720852009576705e-05,
      "loss": 9.0009,
      "step": 2531
    },
    {
      "epoch": 0.4947245017584994,
      "grad_norm": 9.274893760681152,
      "learning_rate": 1.0714539577701131e-05,
      "loss": 8.6465,
      "step": 2532
    },
    {
      "epoch": 0.4949198905822587,
      "grad_norm": 8.545907974243164,
      "learning_rate": 1.0708226859630273e-05,
      "loss": 9.3616,
      "step": 2533
    },
    {
      "epoch": 0.49511527940601796,
      "grad_norm": 9.02899169921875,
      "learning_rate": 1.0701913857892567e-05,
      "loss": 8.4511,
      "step": 2534
    },
    {
      "epoch": 0.49531066822977726,
      "grad_norm": 7.919510841369629,
      "learning_rate": 1.0695600575016571e-05,
      "loss": 8.8204,
      "step": 2535
    },
    {
      "epoch": 0.49550605705353656,
      "grad_norm": 9.303332328796387,
      "learning_rate": 1.0689287013530949e-05,
      "loss": 10.0516,
      "step": 2536
    },
    {
      "epoch": 0.4957014458772958,
      "grad_norm": 10.700124740600586,
      "learning_rate": 1.0682973175964476e-05,
      "loss": 9.2763,
      "step": 2537
    },
    {
      "epoch": 0.4958968347010551,
      "grad_norm": 8.245739936828613,
      "learning_rate": 1.0676659064846044e-05,
      "loss": 8.3476,
      "step": 2538
    },
    {
      "epoch": 0.49609222352481436,
      "grad_norm": 10.72991943359375,
      "learning_rate": 1.067034468270465e-05,
      "loss": 9.4349,
      "step": 2539
    },
    {
      "epoch": 0.49628761234857366,
      "grad_norm": 9.243351936340332,
      "learning_rate": 1.0664030032069396e-05,
      "loss": 7.9045,
      "step": 2540
    },
    {
      "epoch": 0.49648300117233296,
      "grad_norm": 9.342486381530762,
      "learning_rate": 1.06577151154695e-05,
      "loss": 9.3108,
      "step": 2541
    },
    {
      "epoch": 0.4966783899960922,
      "grad_norm": 7.926912307739258,
      "learning_rate": 1.0651399935434281e-05,
      "loss": 9.807,
      "step": 2542
    },
    {
      "epoch": 0.4968737788198515,
      "grad_norm": 9.867131233215332,
      "learning_rate": 1.0645084494493166e-05,
      "loss": 10.9449,
      "step": 2543
    },
    {
      "epoch": 0.4970691676436108,
      "grad_norm": 7.8673834800720215,
      "learning_rate": 1.063876879517568e-05,
      "loss": 9.6844,
      "step": 2544
    },
    {
      "epoch": 0.49726455646737006,
      "grad_norm": 9.949667930603027,
      "learning_rate": 1.0632452840011463e-05,
      "loss": 9.6345,
      "step": 2545
    },
    {
      "epoch": 0.49745994529112936,
      "grad_norm": 9.161530494689941,
      "learning_rate": 1.0626136631530245e-05,
      "loss": 9.8698,
      "step": 2546
    },
    {
      "epoch": 0.4976553341148886,
      "grad_norm": 8.347764015197754,
      "learning_rate": 1.0619820172261867e-05,
      "loss": 9.2639,
      "step": 2547
    },
    {
      "epoch": 0.4978507229386479,
      "grad_norm": 7.256474494934082,
      "learning_rate": 1.0613503464736267e-05,
      "loss": 8.2839,
      "step": 2548
    },
    {
      "epoch": 0.4980461117624072,
      "grad_norm": 10.699699401855469,
      "learning_rate": 1.0607186511483485e-05,
      "loss": 9.0246,
      "step": 2549
    },
    {
      "epoch": 0.49824150058616645,
      "grad_norm": 8.31127643585205,
      "learning_rate": 1.0600869315033651e-05,
      "loss": 8.6217,
      "step": 2550
    },
    {
      "epoch": 0.49843688940992575,
      "grad_norm": 8.519987106323242,
      "learning_rate": 1.0594551877917003e-05,
      "loss": 8.2195,
      "step": 2551
    },
    {
      "epoch": 0.49863227823368506,
      "grad_norm": 10.063212394714355,
      "learning_rate": 1.058823420266387e-05,
      "loss": 10.2044,
      "step": 2552
    },
    {
      "epoch": 0.4988276670574443,
      "grad_norm": 12.953171730041504,
      "learning_rate": 1.0581916291804675e-05,
      "loss": 8.8876,
      "step": 2553
    },
    {
      "epoch": 0.4990230558812036,
      "grad_norm": 11.009001731872559,
      "learning_rate": 1.0575598147869939e-05,
      "loss": 9.1471,
      "step": 2554
    },
    {
      "epoch": 0.49921844470496285,
      "grad_norm": 11.144474983215332,
      "learning_rate": 1.0569279773390273e-05,
      "loss": 9.1009,
      "step": 2555
    },
    {
      "epoch": 0.49941383352872215,
      "grad_norm": 8.041287422180176,
      "learning_rate": 1.0562961170896384e-05,
      "loss": 8.7397,
      "step": 2556
    },
    {
      "epoch": 0.49960922235248145,
      "grad_norm": 9.339242935180664,
      "learning_rate": 1.0556642342919071e-05,
      "loss": 10.0537,
      "step": 2557
    },
    {
      "epoch": 0.4998046111762407,
      "grad_norm": 9.279107093811035,
      "learning_rate": 1.0550323291989217e-05,
      "loss": 8.9358,
      "step": 2558
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.196199417114258,
      "learning_rate": 1.0544004020637799e-05,
      "loss": 8.1397,
      "step": 2559
    },
    {
      "epoch": 0.5001953888237592,
      "grad_norm": 7.969322681427002,
      "learning_rate": 1.053768453139588e-05,
      "loss": 8.3753,
      "step": 2560
    },
    {
      "epoch": 0.5003907776475186,
      "grad_norm": 9.300406455993652,
      "learning_rate": 1.0531364826794612e-05,
      "loss": 9.3357,
      "step": 2561
    },
    {
      "epoch": 0.5005861664712778,
      "grad_norm": 12.185202598571777,
      "learning_rate": 1.0525044909365237e-05,
      "loss": 8.2229,
      "step": 2562
    },
    {
      "epoch": 0.5007815552950371,
      "grad_norm": 17.077205657958984,
      "learning_rate": 1.0518724781639073e-05,
      "loss": 9.7043,
      "step": 2563
    },
    {
      "epoch": 0.5009769441187965,
      "grad_norm": 11.250316619873047,
      "learning_rate": 1.0512404446147531e-05,
      "loss": 9.4062,
      "step": 2564
    },
    {
      "epoch": 0.5011723329425557,
      "grad_norm": 7.564905166625977,
      "learning_rate": 1.0506083905422097e-05,
      "loss": 9.1755,
      "step": 2565
    },
    {
      "epoch": 0.5013677217663149,
      "grad_norm": 15.252607345581055,
      "learning_rate": 1.049976316199435e-05,
      "loss": 9.1116,
      "step": 2566
    },
    {
      "epoch": 0.5015631105900743,
      "grad_norm": 9.757753372192383,
      "learning_rate": 1.0493442218395938e-05,
      "loss": 8.4352,
      "step": 2567
    },
    {
      "epoch": 0.5017584994138335,
      "grad_norm": 11.019709587097168,
      "learning_rate": 1.04871210771586e-05,
      "loss": 9.4381,
      "step": 2568
    },
    {
      "epoch": 0.5019538882375928,
      "grad_norm": 8.746038436889648,
      "learning_rate": 1.0480799740814145e-05,
      "loss": 8.9848,
      "step": 2569
    },
    {
      "epoch": 0.502149277061352,
      "grad_norm": 8.94467544555664,
      "learning_rate": 1.047447821189447e-05,
      "loss": 9.3281,
      "step": 2570
    },
    {
      "epoch": 0.5023446658851114,
      "grad_norm": 13.826160430908203,
      "learning_rate": 1.046815649293154e-05,
      "loss": 10.3383,
      "step": 2571
    },
    {
      "epoch": 0.5025400547088706,
      "grad_norm": 8.326482772827148,
      "learning_rate": 1.0461834586457398e-05,
      "loss": 9.0154,
      "step": 2572
    },
    {
      "epoch": 0.5027354435326299,
      "grad_norm": 10.193694114685059,
      "learning_rate": 1.0455512495004167e-05,
      "loss": 10.2128,
      "step": 2573
    },
    {
      "epoch": 0.5029308323563892,
      "grad_norm": 8.055763244628906,
      "learning_rate": 1.044919022110404e-05,
      "loss": 9.2993,
      "step": 2574
    },
    {
      "epoch": 0.5031262211801485,
      "grad_norm": 7.485759735107422,
      "learning_rate": 1.0442867767289282e-05,
      "loss": 8.1856,
      "step": 2575
    },
    {
      "epoch": 0.5033216100039077,
      "grad_norm": 11.85636043548584,
      "learning_rate": 1.0436545136092234e-05,
      "loss": 10.6178,
      "step": 2576
    },
    {
      "epoch": 0.5035169988276671,
      "grad_norm": 12.691350936889648,
      "learning_rate": 1.0430222330045306e-05,
      "loss": 9.6323,
      "step": 2577
    },
    {
      "epoch": 0.5037123876514263,
      "grad_norm": 7.107717514038086,
      "learning_rate": 1.0423899351680979e-05,
      "loss": 9.3238,
      "step": 2578
    },
    {
      "epoch": 0.5039077764751856,
      "grad_norm": 8.05132007598877,
      "learning_rate": 1.0417576203531801e-05,
      "loss": 9.8049,
      "step": 2579
    },
    {
      "epoch": 0.5041031652989449,
      "grad_norm": 8.098458290100098,
      "learning_rate": 1.0411252888130388e-05,
      "loss": 8.9939,
      "step": 2580
    },
    {
      "epoch": 0.5042985541227042,
      "grad_norm": 15.144835472106934,
      "learning_rate": 1.0404929408009425e-05,
      "loss": 9.4043,
      "step": 2581
    },
    {
      "epoch": 0.5044939429464634,
      "grad_norm": 8.488953590393066,
      "learning_rate": 1.0398605765701663e-05,
      "loss": 9.0053,
      "step": 2582
    },
    {
      "epoch": 0.5046893317702228,
      "grad_norm": 7.072780132293701,
      "learning_rate": 1.0392281963739918e-05,
      "loss": 8.9614,
      "step": 2583
    },
    {
      "epoch": 0.504884720593982,
      "grad_norm": 13.088480949401855,
      "learning_rate": 1.0385958004657069e-05,
      "loss": 9.5415,
      "step": 2584
    },
    {
      "epoch": 0.5050801094177413,
      "grad_norm": 10.07627010345459,
      "learning_rate": 1.0379633890986054e-05,
      "loss": 9.8575,
      "step": 2585
    },
    {
      "epoch": 0.5052754982415005,
      "grad_norm": 9.738160133361816,
      "learning_rate": 1.037330962525988e-05,
      "loss": 9.3236,
      "step": 2586
    },
    {
      "epoch": 0.5054708870652599,
      "grad_norm": 11.041217803955078,
      "learning_rate": 1.036698521001161e-05,
      "loss": 9.0937,
      "step": 2587
    },
    {
      "epoch": 0.5056662758890191,
      "grad_norm": 6.805891990661621,
      "learning_rate": 1.036066064777437e-05,
      "loss": 9.6693,
      "step": 2588
    },
    {
      "epoch": 0.5058616647127784,
      "grad_norm": 7.843379497528076,
      "learning_rate": 1.0354335941081344e-05,
      "loss": 8.8285,
      "step": 2589
    },
    {
      "epoch": 0.5060570535365377,
      "grad_norm": 9.05453872680664,
      "learning_rate": 1.0348011092465772e-05,
      "loss": 8.7001,
      "step": 2590
    },
    {
      "epoch": 0.506252442360297,
      "grad_norm": 7.629778861999512,
      "learning_rate": 1.0341686104460954e-05,
      "loss": 9.5514,
      "step": 2591
    },
    {
      "epoch": 0.5064478311840562,
      "grad_norm": 12.36606502532959,
      "learning_rate": 1.0335360979600244e-05,
      "loss": 9.2567,
      "step": 2592
    },
    {
      "epoch": 0.5066432200078156,
      "grad_norm": 9.524723052978516,
      "learning_rate": 1.0329035720417049e-05,
      "loss": 9.4839,
      "step": 2593
    },
    {
      "epoch": 0.5068386088315748,
      "grad_norm": 6.781744956970215,
      "learning_rate": 1.0322710329444831e-05,
      "loss": 8.4025,
      "step": 2594
    },
    {
      "epoch": 0.5070339976553341,
      "grad_norm": 7.558021068572998,
      "learning_rate": 1.0316384809217108e-05,
      "loss": 8.1913,
      "step": 2595
    },
    {
      "epoch": 0.5072293864790934,
      "grad_norm": 10.1206693649292,
      "learning_rate": 1.0310059162267448e-05,
      "loss": 8.9099,
      "step": 2596
    },
    {
      "epoch": 0.5074247753028527,
      "grad_norm": 10.372179985046387,
      "learning_rate": 1.0303733391129467e-05,
      "loss": 9.707,
      "step": 2597
    },
    {
      "epoch": 0.5076201641266119,
      "grad_norm": 8.356359481811523,
      "learning_rate": 1.0297407498336832e-05,
      "loss": 9.1372,
      "step": 2598
    },
    {
      "epoch": 0.5078155529503713,
      "grad_norm": 10.27808666229248,
      "learning_rate": 1.029108148642326e-05,
      "loss": 8.5818,
      "step": 2599
    },
    {
      "epoch": 0.5080109417741305,
      "grad_norm": 7.9865336418151855,
      "learning_rate": 1.0284755357922515e-05,
      "loss": 8.9857,
      "step": 2600
    },
    {
      "epoch": 0.5082063305978898,
      "grad_norm": 7.481165409088135,
      "learning_rate": 1.027842911536841e-05,
      "loss": 9.6992,
      "step": 2601
    },
    {
      "epoch": 0.5084017194216491,
      "grad_norm": 8.574689865112305,
      "learning_rate": 1.02721027612948e-05,
      "loss": 8.7809,
      "step": 2602
    },
    {
      "epoch": 0.5085971082454084,
      "grad_norm": 9.692758560180664,
      "learning_rate": 1.0265776298235586e-05,
      "loss": 9.4764,
      "step": 2603
    },
    {
      "epoch": 0.5087924970691676,
      "grad_norm": 9.18383502960205,
      "learning_rate": 1.0259449728724712e-05,
      "loss": 9.5221,
      "step": 2604
    },
    {
      "epoch": 0.5089878858929269,
      "grad_norm": 9.158015251159668,
      "learning_rate": 1.0253123055296167e-05,
      "loss": 8.7137,
      "step": 2605
    },
    {
      "epoch": 0.5091832747166862,
      "grad_norm": 9.627050399780273,
      "learning_rate": 1.0246796280483982e-05,
      "loss": 9.5247,
      "step": 2606
    },
    {
      "epoch": 0.5093786635404455,
      "grad_norm": 7.9211812019348145,
      "learning_rate": 1.0240469406822221e-05,
      "loss": 8.6351,
      "step": 2607
    },
    {
      "epoch": 0.5095740523642047,
      "grad_norm": 11.340261459350586,
      "learning_rate": 1.0234142436845e-05,
      "loss": 9.7535,
      "step": 2608
    },
    {
      "epoch": 0.5097694411879641,
      "grad_norm": 8.429910659790039,
      "learning_rate": 1.0227815373086463e-05,
      "loss": 8.8982,
      "step": 2609
    },
    {
      "epoch": 0.5099648300117233,
      "grad_norm": 14.658758163452148,
      "learning_rate": 1.0221488218080796e-05,
      "loss": 9.8136,
      "step": 2610
    },
    {
      "epoch": 0.5101602188354826,
      "grad_norm": 9.807266235351562,
      "learning_rate": 1.0215160974362224e-05,
      "loss": 9.6128,
      "step": 2611
    },
    {
      "epoch": 0.5103556076592419,
      "grad_norm": 8.956068992614746,
      "learning_rate": 1.0208833644464997e-05,
      "loss": 10.4443,
      "step": 2612
    },
    {
      "epoch": 0.5105509964830012,
      "grad_norm": 8.865155220031738,
      "learning_rate": 1.0202506230923417e-05,
      "loss": 8.4965,
      "step": 2613
    },
    {
      "epoch": 0.5107463853067604,
      "grad_norm": 9.319807052612305,
      "learning_rate": 1.0196178736271803e-05,
      "loss": 9.5889,
      "step": 2614
    },
    {
      "epoch": 0.5109417741305198,
      "grad_norm": 11.961516380310059,
      "learning_rate": 1.0189851163044511e-05,
      "loss": 8.8658,
      "step": 2615
    },
    {
      "epoch": 0.511137162954279,
      "grad_norm": 9.452198028564453,
      "learning_rate": 1.0183523513775935e-05,
      "loss": 8.5611,
      "step": 2616
    },
    {
      "epoch": 0.5113325517780383,
      "grad_norm": 7.758265972137451,
      "learning_rate": 1.0177195791000496e-05,
      "loss": 8.7061,
      "step": 2617
    },
    {
      "epoch": 0.5115279406017976,
      "grad_norm": 15.241104125976562,
      "learning_rate": 1.017086799725264e-05,
      "loss": 9.2781,
      "step": 2618
    },
    {
      "epoch": 0.5117233294255569,
      "grad_norm": 8.775419235229492,
      "learning_rate": 1.0164540135066846e-05,
      "loss": 8.6912,
      "step": 2619
    },
    {
      "epoch": 0.5119187182493161,
      "grad_norm": 10.642983436584473,
      "learning_rate": 1.015821220697762e-05,
      "loss": 9.1323,
      "step": 2620
    },
    {
      "epoch": 0.5121141070730755,
      "grad_norm": 8.464726448059082,
      "learning_rate": 1.0151884215519491e-05,
      "loss": 9.4481,
      "step": 2621
    },
    {
      "epoch": 0.5123094958968347,
      "grad_norm": 10.788630485534668,
      "learning_rate": 1.0145556163227021e-05,
      "loss": 8.8519,
      "step": 2622
    },
    {
      "epoch": 0.512504884720594,
      "grad_norm": 10.610641479492188,
      "learning_rate": 1.013922805263479e-05,
      "loss": 8.0585,
      "step": 2623
    },
    {
      "epoch": 0.5127002735443532,
      "grad_norm": 11.172237396240234,
      "learning_rate": 1.0132899886277398e-05,
      "loss": 10.3538,
      "step": 2624
    },
    {
      "epoch": 0.5128956623681126,
      "grad_norm": 7.80220890045166,
      "learning_rate": 1.0126571666689475e-05,
      "loss": 9.3621,
      "step": 2625
    },
    {
      "epoch": 0.5130910511918718,
      "grad_norm": 7.723788261413574,
      "learning_rate": 1.0120243396405676e-05,
      "loss": 8.9266,
      "step": 2626
    },
    {
      "epoch": 0.5132864400156311,
      "grad_norm": 14.545475959777832,
      "learning_rate": 1.011391507796066e-05,
      "loss": 10.129,
      "step": 2627
    },
    {
      "epoch": 0.5134818288393904,
      "grad_norm": 7.488824844360352,
      "learning_rate": 1.0107586713889117e-05,
      "loss": 9.7446,
      "step": 2628
    },
    {
      "epoch": 0.5136772176631497,
      "grad_norm": 8.845382690429688,
      "learning_rate": 1.0101258306725754e-05,
      "loss": 9.5032,
      "step": 2629
    },
    {
      "epoch": 0.5138726064869089,
      "grad_norm": 8.629199028015137,
      "learning_rate": 1.0094929859005299e-05,
      "loss": 9.31,
      "step": 2630
    },
    {
      "epoch": 0.5140679953106683,
      "grad_norm": 11.23619556427002,
      "learning_rate": 1.0088601373262486e-05,
      "loss": 9.6429,
      "step": 2631
    },
    {
      "epoch": 0.5142633841344275,
      "grad_norm": 11.032987594604492,
      "learning_rate": 1.0082272852032073e-05,
      "loss": 8.6772,
      "step": 2632
    },
    {
      "epoch": 0.5144587729581868,
      "grad_norm": 10.961579322814941,
      "learning_rate": 1.0075944297848827e-05,
      "loss": 8.5175,
      "step": 2633
    },
    {
      "epoch": 0.5146541617819461,
      "grad_norm": 7.2397894859313965,
      "learning_rate": 1.0069615713247534e-05,
      "loss": 8.9605,
      "step": 2634
    },
    {
      "epoch": 0.5148495506057054,
      "grad_norm": 7.577977180480957,
      "learning_rate": 1.0063287100762987e-05,
      "loss": 9.0421,
      "step": 2635
    },
    {
      "epoch": 0.5150449394294646,
      "grad_norm": 8.063026428222656,
      "learning_rate": 1.005695846292999e-05,
      "loss": 8.8904,
      "step": 2636
    },
    {
      "epoch": 0.515240328253224,
      "grad_norm": 10.646279335021973,
      "learning_rate": 1.0050629802283357e-05,
      "loss": 9.6499,
      "step": 2637
    },
    {
      "epoch": 0.5154357170769832,
      "grad_norm": 16.412921905517578,
      "learning_rate": 1.0044301121357923e-05,
      "loss": 9.8436,
      "step": 2638
    },
    {
      "epoch": 0.5156311059007425,
      "grad_norm": 10.519708633422852,
      "learning_rate": 1.003797242268851e-05,
      "loss": 7.6298,
      "step": 2639
    },
    {
      "epoch": 0.5158264947245017,
      "grad_norm": 9.18188762664795,
      "learning_rate": 1.0031643708809965e-05,
      "loss": 10.5935,
      "step": 2640
    },
    {
      "epoch": 0.5160218835482611,
      "grad_norm": 8.290685653686523,
      "learning_rate": 1.0025314982257131e-05,
      "loss": 8.5635,
      "step": 2641
    },
    {
      "epoch": 0.5162172723720203,
      "grad_norm": 11.62686538696289,
      "learning_rate": 1.0018986245564862e-05,
      "loss": 9.1133,
      "step": 2642
    },
    {
      "epoch": 0.5164126611957796,
      "grad_norm": 9.008712768554688,
      "learning_rate": 1.001265750126801e-05,
      "loss": 10.4709,
      "step": 2643
    },
    {
      "epoch": 0.5166080500195389,
      "grad_norm": 7.980377674102783,
      "learning_rate": 1.0006328751901438e-05,
      "loss": 9.1196,
      "step": 2644
    },
    {
      "epoch": 0.5168034388432982,
      "grad_norm": 12.153450012207031,
      "learning_rate": 1e-05,
      "loss": 9.0372,
      "step": 2645
    },
    {
      "epoch": 0.5169988276670574,
      "grad_norm": 9.671964645385742,
      "learning_rate": 9.993671248098567e-06,
      "loss": 9.165,
      "step": 2646
    },
    {
      "epoch": 0.5171942164908168,
      "grad_norm": 7.296621322631836,
      "learning_rate": 9.987342498731993e-06,
      "loss": 8.58,
      "step": 2647
    },
    {
      "epoch": 0.517389605314576,
      "grad_norm": 7.797286033630371,
      "learning_rate": 9.981013754435142e-06,
      "loss": 9.5087,
      "step": 2648
    },
    {
      "epoch": 0.5175849941383353,
      "grad_norm": 7.952873706817627,
      "learning_rate": 9.97468501774287e-06,
      "loss": 9.1935,
      "step": 2649
    },
    {
      "epoch": 0.5177803829620946,
      "grad_norm": 11.546721458435059,
      "learning_rate": 9.968356291190038e-06,
      "loss": 8.7435,
      "step": 2650
    },
    {
      "epoch": 0.5179757717858539,
      "grad_norm": 9.659433364868164,
      "learning_rate": 9.962027577311491e-06,
      "loss": 7.5912,
      "step": 2651
    },
    {
      "epoch": 0.5181711606096131,
      "grad_norm": 9.96871566772461,
      "learning_rate": 9.955698878642082e-06,
      "loss": 9.4127,
      "step": 2652
    },
    {
      "epoch": 0.5183665494333725,
      "grad_norm": 8.684602737426758,
      "learning_rate": 9.949370197716644e-06,
      "loss": 9.9006,
      "step": 2653
    },
    {
      "epoch": 0.5185619382571317,
      "grad_norm": 12.929977416992188,
      "learning_rate": 9.943041537070014e-06,
      "loss": 9.6474,
      "step": 2654
    },
    {
      "epoch": 0.518757327080891,
      "grad_norm": 10.658129692077637,
      "learning_rate": 9.936712899237018e-06,
      "loss": 9.7753,
      "step": 2655
    },
    {
      "epoch": 0.5189527159046503,
      "grad_norm": 33.21316146850586,
      "learning_rate": 9.930384286752469e-06,
      "loss": 7.8055,
      "step": 2656
    },
    {
      "epoch": 0.5191481047284096,
      "grad_norm": 7.257182598114014,
      "learning_rate": 9.924055702151176e-06,
      "loss": 9.2182,
      "step": 2657
    },
    {
      "epoch": 0.5193434935521688,
      "grad_norm": 8.233306884765625,
      "learning_rate": 9.91772714796793e-06,
      "loss": 9.055,
      "step": 2658
    },
    {
      "epoch": 0.519538882375928,
      "grad_norm": 8.045198440551758,
      "learning_rate": 9.911398626737517e-06,
      "loss": 8.596,
      "step": 2659
    },
    {
      "epoch": 0.5197342711996874,
      "grad_norm": 8.076264381408691,
      "learning_rate": 9.905070140994705e-06,
      "loss": 9.4418,
      "step": 2660
    },
    {
      "epoch": 0.5199296600234466,
      "grad_norm": 10.012219429016113,
      "learning_rate": 9.898741693274247e-06,
      "loss": 9.1936,
      "step": 2661
    },
    {
      "epoch": 0.5201250488472059,
      "grad_norm": 8.335098266601562,
      "learning_rate": 9.892413286110886e-06,
      "loss": 9.2575,
      "step": 2662
    },
    {
      "epoch": 0.5203204376709653,
      "grad_norm": 8.148777961730957,
      "learning_rate": 9.886084922039344e-06,
      "loss": 9.0273,
      "step": 2663
    },
    {
      "epoch": 0.5205158264947245,
      "grad_norm": 6.868932247161865,
      "learning_rate": 9.879756603594328e-06,
      "loss": 9.055,
      "step": 2664
    },
    {
      "epoch": 0.5207112153184837,
      "grad_norm": 8.413737297058105,
      "learning_rate": 9.873428333310527e-06,
      "loss": 9.4662,
      "step": 2665
    },
    {
      "epoch": 0.5209066041422431,
      "grad_norm": 8.817794799804688,
      "learning_rate": 9.867100113722605e-06,
      "loss": 9.1913,
      "step": 2666
    },
    {
      "epoch": 0.5211019929660023,
      "grad_norm": 10.635193824768066,
      "learning_rate": 9.860771947365214e-06,
      "loss": 9.1389,
      "step": 2667
    },
    {
      "epoch": 0.5212973817897616,
      "grad_norm": 7.9307026863098145,
      "learning_rate": 9.85444383677298e-06,
      "loss": 8.9327,
      "step": 2668
    },
    {
      "epoch": 0.521492770613521,
      "grad_norm": 8.434563636779785,
      "learning_rate": 9.84811578448051e-06,
      "loss": 9.2359,
      "step": 2669
    },
    {
      "epoch": 0.5216881594372802,
      "grad_norm": 7.498775482177734,
      "learning_rate": 9.841787793022383e-06,
      "loss": 9.6017,
      "step": 2670
    },
    {
      "epoch": 0.5218835482610394,
      "grad_norm": 10.593565940856934,
      "learning_rate": 9.835459864933156e-06,
      "loss": 10.0484,
      "step": 2671
    },
    {
      "epoch": 0.5220789370847988,
      "grad_norm": 7.587101459503174,
      "learning_rate": 9.829132002747363e-06,
      "loss": 8.3536,
      "step": 2672
    },
    {
      "epoch": 0.522274325908558,
      "grad_norm": 10.2027587890625,
      "learning_rate": 9.822804208999508e-06,
      "loss": 10.2716,
      "step": 2673
    },
    {
      "epoch": 0.5224697147323173,
      "grad_norm": 9.8612699508667,
      "learning_rate": 9.816476486224067e-06,
      "loss": 10.2911,
      "step": 2674
    },
    {
      "epoch": 0.5226651035560765,
      "grad_norm": 6.963353633880615,
      "learning_rate": 9.810148836955492e-06,
      "loss": 9.2523,
      "step": 2675
    },
    {
      "epoch": 0.5228604923798359,
      "grad_norm": 9.44482135772705,
      "learning_rate": 9.8038212637282e-06,
      "loss": 9.9173,
      "step": 2676
    },
    {
      "epoch": 0.5230558812035951,
      "grad_norm": 9.125898361206055,
      "learning_rate": 9.797493769076587e-06,
      "loss": 8.8245,
      "step": 2677
    },
    {
      "epoch": 0.5232512700273544,
      "grad_norm": 10.172958374023438,
      "learning_rate": 9.791166355535004e-06,
      "loss": 9.8728,
      "step": 2678
    },
    {
      "epoch": 0.5234466588511137,
      "grad_norm": 10.32078742980957,
      "learning_rate": 9.78483902563778e-06,
      "loss": 8.6425,
      "step": 2679
    },
    {
      "epoch": 0.523642047674873,
      "grad_norm": 11.205142974853516,
      "learning_rate": 9.778511781919205e-06,
      "loss": 10.0009,
      "step": 2680
    },
    {
      "epoch": 0.5238374364986322,
      "grad_norm": 7.736554145812988,
      "learning_rate": 9.772184626913537e-06,
      "loss": 8.3589,
      "step": 2681
    },
    {
      "epoch": 0.5240328253223916,
      "grad_norm": 8.641042709350586,
      "learning_rate": 9.765857563155e-06,
      "loss": 9.821,
      "step": 2682
    },
    {
      "epoch": 0.5242282141461508,
      "grad_norm": 8.605805397033691,
      "learning_rate": 9.759530593177777e-06,
      "loss": 9.7728,
      "step": 2683
    },
    {
      "epoch": 0.5244236029699101,
      "grad_norm": 8.84467887878418,
      "learning_rate": 9.75320371951602e-06,
      "loss": 8.7332,
      "step": 2684
    },
    {
      "epoch": 0.5246189917936694,
      "grad_norm": 10.527255058288574,
      "learning_rate": 9.746876944703833e-06,
      "loss": 9.2599,
      "step": 2685
    },
    {
      "epoch": 0.5248143806174287,
      "grad_norm": 10.367681503295898,
      "learning_rate": 9.740550271275287e-06,
      "loss": 8.7876,
      "step": 2686
    },
    {
      "epoch": 0.5250097694411879,
      "grad_norm": 8.358341217041016,
      "learning_rate": 9.734223701764415e-06,
      "loss": 8.8515,
      "step": 2687
    },
    {
      "epoch": 0.5252051582649473,
      "grad_norm": 9.358244895935059,
      "learning_rate": 9.7278972387052e-06,
      "loss": 9.2275,
      "step": 2688
    },
    {
      "epoch": 0.5254005470887065,
      "grad_norm": 9.502291679382324,
      "learning_rate": 9.721570884631591e-06,
      "loss": 9.2014,
      "step": 2689
    },
    {
      "epoch": 0.5255959359124658,
      "grad_norm": 9.14945125579834,
      "learning_rate": 9.715244642077485e-06,
      "loss": 9.0812,
      "step": 2690
    },
    {
      "epoch": 0.5257913247362251,
      "grad_norm": 10.896734237670898,
      "learning_rate": 9.708918513576742e-06,
      "loss": 9.2459,
      "step": 2691
    },
    {
      "epoch": 0.5259867135599844,
      "grad_norm": 8.33903980255127,
      "learning_rate": 9.702592501663172e-06,
      "loss": 9.0402,
      "step": 2692
    },
    {
      "epoch": 0.5261821023837436,
      "grad_norm": 8.217243194580078,
      "learning_rate": 9.696266608870536e-06,
      "loss": 8.9903,
      "step": 2693
    },
    {
      "epoch": 0.5263774912075029,
      "grad_norm": 9.530319213867188,
      "learning_rate": 9.689940837732554e-06,
      "loss": 8.7124,
      "step": 2694
    },
    {
      "epoch": 0.5265728800312622,
      "grad_norm": 7.9275360107421875,
      "learning_rate": 9.683615190782892e-06,
      "loss": 9.1352,
      "step": 2695
    },
    {
      "epoch": 0.5267682688550215,
      "grad_norm": 9.213961601257324,
      "learning_rate": 9.67728967055517e-06,
      "loss": 8.9705,
      "step": 2696
    },
    {
      "epoch": 0.5269636576787807,
      "grad_norm": 7.555400848388672,
      "learning_rate": 9.670964279582953e-06,
      "loss": 8.2625,
      "step": 2697
    },
    {
      "epoch": 0.5271590465025401,
      "grad_norm": 8.472102165222168,
      "learning_rate": 9.664639020399757e-06,
      "loss": 9.0743,
      "step": 2698
    },
    {
      "epoch": 0.5273544353262993,
      "grad_norm": 10.481719970703125,
      "learning_rate": 9.658313895539045e-06,
      "loss": 9.0627,
      "step": 2699
    },
    {
      "epoch": 0.5275498241500586,
      "grad_norm": 8.405017852783203,
      "learning_rate": 9.651988907534228e-06,
      "loss": 9.979,
      "step": 2700
    },
    {
      "epoch": 0.5277452129738179,
      "grad_norm": 7.347189903259277,
      "learning_rate": 9.645664058918656e-06,
      "loss": 9.4459,
      "step": 2701
    },
    {
      "epoch": 0.5279406017975772,
      "grad_norm": 6.290065765380859,
      "learning_rate": 9.63933935222563e-06,
      "loss": 8.7453,
      "step": 2702
    },
    {
      "epoch": 0.5281359906213364,
      "grad_norm": 7.323775291442871,
      "learning_rate": 9.633014789988391e-06,
      "loss": 8.7035,
      "step": 2703
    },
    {
      "epoch": 0.5283313794450958,
      "grad_norm": 11.120918273925781,
      "learning_rate": 9.626690374740124e-06,
      "loss": 9.9083,
      "step": 2704
    },
    {
      "epoch": 0.528526768268855,
      "grad_norm": 8.071630477905273,
      "learning_rate": 9.620366109013949e-06,
      "loss": 9.4713,
      "step": 2705
    },
    {
      "epoch": 0.5287221570926143,
      "grad_norm": 7.903960227966309,
      "learning_rate": 9.614041995342935e-06,
      "loss": 8.9713,
      "step": 2706
    },
    {
      "epoch": 0.5289175459163736,
      "grad_norm": 9.752288818359375,
      "learning_rate": 9.607718036260083e-06,
      "loss": 8.2165,
      "step": 2707
    },
    {
      "epoch": 0.5291129347401329,
      "grad_norm": 7.624960899353027,
      "learning_rate": 9.601394234298337e-06,
      "loss": 8.3963,
      "step": 2708
    },
    {
      "epoch": 0.5293083235638921,
      "grad_norm": 9.899323463439941,
      "learning_rate": 9.595070591990575e-06,
      "loss": 9.0401,
      "step": 2709
    },
    {
      "epoch": 0.5295037123876515,
      "grad_norm": 7.964188098907471,
      "learning_rate": 9.588747111869614e-06,
      "loss": 9.2519,
      "step": 2710
    },
    {
      "epoch": 0.5296991012114107,
      "grad_norm": 6.899359703063965,
      "learning_rate": 9.5824237964682e-06,
      "loss": 8.4557,
      "step": 2711
    },
    {
      "epoch": 0.52989449003517,
      "grad_norm": 6.853302955627441,
      "learning_rate": 9.576100648319023e-06,
      "loss": 8.5504,
      "step": 2712
    },
    {
      "epoch": 0.5300898788589292,
      "grad_norm": 10.143784523010254,
      "learning_rate": 9.569777669954694e-06,
      "loss": 9.248,
      "step": 2713
    },
    {
      "epoch": 0.5302852676826886,
      "grad_norm": 10.853752136230469,
      "learning_rate": 9.563454863907771e-06,
      "loss": 9.5435,
      "step": 2714
    },
    {
      "epoch": 0.5304806565064478,
      "grad_norm": 9.94910717010498,
      "learning_rate": 9.557132232710725e-06,
      "loss": 9.2622,
      "step": 2715
    },
    {
      "epoch": 0.5306760453302071,
      "grad_norm": 8.490160942077637,
      "learning_rate": 9.550809778895967e-06,
      "loss": 8.1029,
      "step": 2716
    },
    {
      "epoch": 0.5308714341539664,
      "grad_norm": 8.913311958312988,
      "learning_rate": 9.544487504995838e-06,
      "loss": 9.7678,
      "step": 2717
    },
    {
      "epoch": 0.5310668229777257,
      "grad_norm": 7.795229911804199,
      "learning_rate": 9.538165413542607e-06,
      "loss": 10.2682,
      "step": 2718
    },
    {
      "epoch": 0.5312622118014849,
      "grad_norm": 9.823403358459473,
      "learning_rate": 9.531843507068466e-06,
      "loss": 9.1957,
      "step": 2719
    },
    {
      "epoch": 0.5314576006252443,
      "grad_norm": 8.27929401397705,
      "learning_rate": 9.525521788105534e-06,
      "loss": 9.0988,
      "step": 2720
    },
    {
      "epoch": 0.5316529894490035,
      "grad_norm": 12.38388729095459,
      "learning_rate": 9.519200259185856e-06,
      "loss": 9.1459,
      "step": 2721
    },
    {
      "epoch": 0.5318483782727628,
      "grad_norm": 7.560203552246094,
      "learning_rate": 9.512878922841403e-06,
      "loss": 8.4563,
      "step": 2722
    },
    {
      "epoch": 0.5320437670965221,
      "grad_norm": 6.650539398193359,
      "learning_rate": 9.506557781604067e-06,
      "loss": 9.0784,
      "step": 2723
    },
    {
      "epoch": 0.5322391559202814,
      "grad_norm": 8.1847505569458,
      "learning_rate": 9.500236838005656e-06,
      "loss": 9.5058,
      "step": 2724
    },
    {
      "epoch": 0.5324345447440406,
      "grad_norm": 6.476805686950684,
      "learning_rate": 9.493916094577908e-06,
      "loss": 8.3882,
      "step": 2725
    },
    {
      "epoch": 0.5326299335678,
      "grad_norm": 14.108317375183105,
      "learning_rate": 9.487595553852476e-06,
      "loss": 9.6569,
      "step": 2726
    },
    {
      "epoch": 0.5328253223915592,
      "grad_norm": 8.654928207397461,
      "learning_rate": 9.481275218360933e-06,
      "loss": 9.746,
      "step": 2727
    },
    {
      "epoch": 0.5330207112153185,
      "grad_norm": 7.9840521812438965,
      "learning_rate": 9.474955090634768e-06,
      "loss": 9.4968,
      "step": 2728
    },
    {
      "epoch": 0.5332161000390777,
      "grad_norm": 9.199403762817383,
      "learning_rate": 9.468635173205392e-06,
      "loss": 9.8293,
      "step": 2729
    },
    {
      "epoch": 0.5334114888628371,
      "grad_norm": 7.720449447631836,
      "learning_rate": 9.462315468604126e-06,
      "loss": 8.8497,
      "step": 2730
    },
    {
      "epoch": 0.5336068776865963,
      "grad_norm": 8.197383880615234,
      "learning_rate": 9.455995979362206e-06,
      "loss": 9.5901,
      "step": 2731
    },
    {
      "epoch": 0.5338022665103556,
      "grad_norm": 7.391048908233643,
      "learning_rate": 9.449676708010788e-06,
      "loss": 8.9208,
      "step": 2732
    },
    {
      "epoch": 0.5339976553341149,
      "grad_norm": 7.571177005767822,
      "learning_rate": 9.443357657080932e-06,
      "loss": 8.9337,
      "step": 2733
    },
    {
      "epoch": 0.5341930441578742,
      "grad_norm": 6.633815288543701,
      "learning_rate": 9.437038829103618e-06,
      "loss": 9.1504,
      "step": 2734
    },
    {
      "epoch": 0.5343884329816334,
      "grad_norm": 15.273905754089355,
      "learning_rate": 9.43072022660973e-06,
      "loss": 10.3509,
      "step": 2735
    },
    {
      "epoch": 0.5345838218053928,
      "grad_norm": 20.139923095703125,
      "learning_rate": 9.424401852130068e-06,
      "loss": 10.224,
      "step": 2736
    },
    {
      "epoch": 0.534779210629152,
      "grad_norm": 9.288772583007812,
      "learning_rate": 9.418083708195332e-06,
      "loss": 9.6518,
      "step": 2737
    },
    {
      "epoch": 0.5349745994529113,
      "grad_norm": 9.217354774475098,
      "learning_rate": 9.411765797336136e-06,
      "loss": 8.8565,
      "step": 2738
    },
    {
      "epoch": 0.5351699882766706,
      "grad_norm": 9.57825756072998,
      "learning_rate": 9.405448122083002e-06,
      "loss": 9.4921,
      "step": 2739
    },
    {
      "epoch": 0.5353653771004299,
      "grad_norm": 8.773568153381348,
      "learning_rate": 9.399130684966354e-06,
      "loss": 9.2346,
      "step": 2740
    },
    {
      "epoch": 0.5355607659241891,
      "grad_norm": 7.933095932006836,
      "learning_rate": 9.392813488516521e-06,
      "loss": 10.1161,
      "step": 2741
    },
    {
      "epoch": 0.5357561547479485,
      "grad_norm": 8.663396835327148,
      "learning_rate": 9.386496535263736e-06,
      "loss": 8.5159,
      "step": 2742
    },
    {
      "epoch": 0.5359515435717077,
      "grad_norm": 9.957633972167969,
      "learning_rate": 9.380179827738138e-06,
      "loss": 9.9456,
      "step": 2743
    },
    {
      "epoch": 0.536146932395467,
      "grad_norm": 9.319618225097656,
      "learning_rate": 9.37386336846976e-06,
      "loss": 9.237,
      "step": 2744
    },
    {
      "epoch": 0.5363423212192263,
      "grad_norm": 8.81218433380127,
      "learning_rate": 9.367547159988542e-06,
      "loss": 8.0087,
      "step": 2745
    },
    {
      "epoch": 0.5365377100429856,
      "grad_norm": 9.204264640808105,
      "learning_rate": 9.361231204824324e-06,
      "loss": 9.5934,
      "step": 2746
    },
    {
      "epoch": 0.5367330988667448,
      "grad_norm": 6.535369396209717,
      "learning_rate": 9.354915505506839e-06,
      "loss": 8.9945,
      "step": 2747
    },
    {
      "epoch": 0.536928487690504,
      "grad_norm": 9.098174095153809,
      "learning_rate": 9.34860006456572e-06,
      "loss": 9.3909,
      "step": 2748
    },
    {
      "epoch": 0.5371238765142634,
      "grad_norm": 8.996417045593262,
      "learning_rate": 9.342284884530504e-06,
      "loss": 8.5684,
      "step": 2749
    },
    {
      "epoch": 0.5373192653380227,
      "grad_norm": 14.172801971435547,
      "learning_rate": 9.335969967930607e-06,
      "loss": 9.9654,
      "step": 2750
    },
    {
      "epoch": 0.5375146541617819,
      "grad_norm": 11.021417617797852,
      "learning_rate": 9.329655317295354e-06,
      "loss": 9.4326,
      "step": 2751
    },
    {
      "epoch": 0.5377100429855413,
      "grad_norm": 12.336243629455566,
      "learning_rate": 9.323340935153958e-06,
      "loss": 9.0942,
      "step": 2752
    },
    {
      "epoch": 0.5379054318093005,
      "grad_norm": 8.203696250915527,
      "learning_rate": 9.317026824035525e-06,
      "loss": 8.7552,
      "step": 2753
    },
    {
      "epoch": 0.5381008206330598,
      "grad_norm": 8.273969650268555,
      "learning_rate": 9.310712986469055e-06,
      "loss": 9.7479,
      "step": 2754
    },
    {
      "epoch": 0.5382962094568191,
      "grad_norm": 7.83505392074585,
      "learning_rate": 9.30439942498343e-06,
      "loss": 9.5113,
      "step": 2755
    },
    {
      "epoch": 0.5384915982805784,
      "grad_norm": 10.40118408203125,
      "learning_rate": 9.298086142107436e-06,
      "loss": 9.2196,
      "step": 2756
    },
    {
      "epoch": 0.5386869871043376,
      "grad_norm": 9.551358222961426,
      "learning_rate": 9.291773140369732e-06,
      "loss": 8.7518,
      "step": 2757
    },
    {
      "epoch": 0.538882375928097,
      "grad_norm": 8.112876892089844,
      "learning_rate": 9.28546042229887e-06,
      "loss": 7.146,
      "step": 2758
    },
    {
      "epoch": 0.5390777647518562,
      "grad_norm": 10.979324340820312,
      "learning_rate": 9.279147990423297e-06,
      "loss": 9.6383,
      "step": 2759
    },
    {
      "epoch": 0.5392731535756154,
      "grad_norm": 9.588387489318848,
      "learning_rate": 9.272835847271332e-06,
      "loss": 10.5762,
      "step": 2760
    },
    {
      "epoch": 0.5394685423993748,
      "grad_norm": 8.50763988494873,
      "learning_rate": 9.266523995371186e-06,
      "loss": 8.8917,
      "step": 2761
    },
    {
      "epoch": 0.539663931223134,
      "grad_norm": 9.856245040893555,
      "learning_rate": 9.260212437250955e-06,
      "loss": 9.8669,
      "step": 2762
    },
    {
      "epoch": 0.5398593200468933,
      "grad_norm": 8.32020092010498,
      "learning_rate": 9.253901175438607e-06,
      "loss": 8.4172,
      "step": 2763
    },
    {
      "epoch": 0.5400547088706525,
      "grad_norm": 8.922029495239258,
      "learning_rate": 9.247590212462001e-06,
      "loss": 9.3984,
      "step": 2764
    },
    {
      "epoch": 0.5402500976944119,
      "grad_norm": 6.328679084777832,
      "learning_rate": 9.241279550848877e-06,
      "loss": 8.091,
      "step": 2765
    },
    {
      "epoch": 0.5404454865181711,
      "grad_norm": 7.473681449890137,
      "learning_rate": 9.234969193126847e-06,
      "loss": 8.7435,
      "step": 2766
    },
    {
      "epoch": 0.5406408753419304,
      "grad_norm": 7.918021202087402,
      "learning_rate": 9.228659141823407e-06,
      "loss": 8.6152,
      "step": 2767
    },
    {
      "epoch": 0.5408362641656898,
      "grad_norm": 7.37488317489624,
      "learning_rate": 9.222349399465927e-06,
      "loss": 8.4447,
      "step": 2768
    },
    {
      "epoch": 0.541031652989449,
      "grad_norm": 10.067622184753418,
      "learning_rate": 9.216039968581656e-06,
      "loss": 9.1785,
      "step": 2769
    },
    {
      "epoch": 0.5412270418132082,
      "grad_norm": 10.376371383666992,
      "learning_rate": 9.209730851697718e-06,
      "loss": 10.3289,
      "step": 2770
    },
    {
      "epoch": 0.5414224306369676,
      "grad_norm": 7.040582180023193,
      "learning_rate": 9.203422051341109e-06,
      "loss": 8.9531,
      "step": 2771
    },
    {
      "epoch": 0.5416178194607268,
      "grad_norm": 7.681628704071045,
      "learning_rate": 9.197113570038699e-06,
      "loss": 9.2074,
      "step": 2772
    },
    {
      "epoch": 0.5418132082844861,
      "grad_norm": 8.371284484863281,
      "learning_rate": 9.190805410317231e-06,
      "loss": 9.2613,
      "step": 2773
    },
    {
      "epoch": 0.5420085971082454,
      "grad_norm": 8.331684112548828,
      "learning_rate": 9.184497574703318e-06,
      "loss": 9.1944,
      "step": 2774
    },
    {
      "epoch": 0.5422039859320047,
      "grad_norm": 10.406688690185547,
      "learning_rate": 9.178190065723447e-06,
      "loss": 9.0204,
      "step": 2775
    },
    {
      "epoch": 0.5423993747557639,
      "grad_norm": 10.491332054138184,
      "learning_rate": 9.171882885903967e-06,
      "loss": 9.5959,
      "step": 2776
    },
    {
      "epoch": 0.5425947635795233,
      "grad_norm": 7.148529529571533,
      "learning_rate": 9.165576037771102e-06,
      "loss": 8.2787,
      "step": 2777
    },
    {
      "epoch": 0.5427901524032825,
      "grad_norm": 6.761819362640381,
      "learning_rate": 9.159269523850939e-06,
      "loss": 9.3,
      "step": 2778
    },
    {
      "epoch": 0.5429855412270418,
      "grad_norm": 8.8622407913208,
      "learning_rate": 9.152963346669432e-06,
      "loss": 9.5788,
      "step": 2779
    },
    {
      "epoch": 0.5431809300508011,
      "grad_norm": 9.01783561706543,
      "learning_rate": 9.146657508752404e-06,
      "loss": 9.5756,
      "step": 2780
    },
    {
      "epoch": 0.5433763188745604,
      "grad_norm": 7.692517280578613,
      "learning_rate": 9.140352012625538e-06,
      "loss": 7.3885,
      "step": 2781
    },
    {
      "epoch": 0.5435717076983196,
      "grad_norm": 8.276142120361328,
      "learning_rate": 9.134046860814377e-06,
      "loss": 8.3795,
      "step": 2782
    },
    {
      "epoch": 0.5437670965220789,
      "grad_norm": 7.867055416107178,
      "learning_rate": 9.127742055844334e-06,
      "loss": 9.4628,
      "step": 2783
    },
    {
      "epoch": 0.5439624853458382,
      "grad_norm": 8.413628578186035,
      "learning_rate": 9.121437600240679e-06,
      "loss": 8.7052,
      "step": 2784
    },
    {
      "epoch": 0.5441578741695975,
      "grad_norm": 8.16491413116455,
      "learning_rate": 9.11513349652854e-06,
      "loss": 10.1515,
      "step": 2785
    },
    {
      "epoch": 0.5443532629933567,
      "grad_norm": 7.399144649505615,
      "learning_rate": 9.108829747232909e-06,
      "loss": 8.2059,
      "step": 2786
    },
    {
      "epoch": 0.5445486518171161,
      "grad_norm": 8.344947814941406,
      "learning_rate": 9.102526354878632e-06,
      "loss": 9.4465,
      "step": 2787
    },
    {
      "epoch": 0.5447440406408753,
      "grad_norm": 7.7732157707214355,
      "learning_rate": 9.096223321990414e-06,
      "loss": 9.7633,
      "step": 2788
    },
    {
      "epoch": 0.5449394294646346,
      "grad_norm": 9.581931114196777,
      "learning_rate": 9.089920651092812e-06,
      "loss": 8.6713,
      "step": 2789
    },
    {
      "epoch": 0.5451348182883939,
      "grad_norm": 10.937223434448242,
      "learning_rate": 9.083618344710246e-06,
      "loss": 9.332,
      "step": 2790
    },
    {
      "epoch": 0.5453302071121532,
      "grad_norm": 8.541001319885254,
      "learning_rate": 9.07731640536698e-06,
      "loss": 9.2417,
      "step": 2791
    },
    {
      "epoch": 0.5455255959359124,
      "grad_norm": 8.981481552124023,
      "learning_rate": 9.071014835587144e-06,
      "loss": 9.0391,
      "step": 2792
    },
    {
      "epoch": 0.5457209847596718,
      "grad_norm": 7.546681880950928,
      "learning_rate": 9.064713637894706e-06,
      "loss": 8.8833,
      "step": 2793
    },
    {
      "epoch": 0.545916373583431,
      "grad_norm": 10.02629280090332,
      "learning_rate": 9.058412814813493e-06,
      "loss": 10.0756,
      "step": 2794
    },
    {
      "epoch": 0.5461117624071903,
      "grad_norm": 8.090208053588867,
      "learning_rate": 9.05211236886718e-06,
      "loss": 9.1745,
      "step": 2795
    },
    {
      "epoch": 0.5463071512309496,
      "grad_norm": 10.446197509765625,
      "learning_rate": 9.045812302579293e-06,
      "loss": 9.5297,
      "step": 2796
    },
    {
      "epoch": 0.5465025400547089,
      "grad_norm": 7.993022441864014,
      "learning_rate": 9.039512618473202e-06,
      "loss": 9.2405,
      "step": 2797
    },
    {
      "epoch": 0.5466979288784681,
      "grad_norm": 7.007857799530029,
      "learning_rate": 9.033213319072126e-06,
      "loss": 8.2401,
      "step": 2798
    },
    {
      "epoch": 0.5468933177022274,
      "grad_norm": 7.0248260498046875,
      "learning_rate": 9.026914406899134e-06,
      "loss": 8.3184,
      "step": 2799
    },
    {
      "epoch": 0.5470887065259867,
      "grad_norm": 7.492818355560303,
      "learning_rate": 9.02061588447713e-06,
      "loss": 8.7988,
      "step": 2800
    },
    {
      "epoch": 0.547284095349746,
      "grad_norm": 8.054252624511719,
      "learning_rate": 9.014317754328873e-06,
      "loss": 9.4373,
      "step": 2801
    },
    {
      "epoch": 0.5474794841735052,
      "grad_norm": 8.825780868530273,
      "learning_rate": 9.008020018976959e-06,
      "loss": 8.9278,
      "step": 2802
    },
    {
      "epoch": 0.5476748729972646,
      "grad_norm": 7.805006980895996,
      "learning_rate": 9.00172268094382e-06,
      "loss": 9.6183,
      "step": 2803
    },
    {
      "epoch": 0.5478702618210238,
      "grad_norm": 8.974769592285156,
      "learning_rate": 8.995425742751743e-06,
      "loss": 9.6281,
      "step": 2804
    },
    {
      "epoch": 0.5480656506447831,
      "grad_norm": 9.293949127197266,
      "learning_rate": 8.989129206922844e-06,
      "loss": 7.8442,
      "step": 2805
    },
    {
      "epoch": 0.5482610394685424,
      "grad_norm": 8.424934387207031,
      "learning_rate": 8.98283307597908e-06,
      "loss": 8.1946,
      "step": 2806
    },
    {
      "epoch": 0.5484564282923017,
      "grad_norm": 8.770522117614746,
      "learning_rate": 8.97653735244225e-06,
      "loss": 9.0785,
      "step": 2807
    },
    {
      "epoch": 0.5486518171160609,
      "grad_norm": 8.615560531616211,
      "learning_rate": 8.970242038833985e-06,
      "loss": 9.4919,
      "step": 2808
    },
    {
      "epoch": 0.5488472059398203,
      "grad_norm": 10.273186683654785,
      "learning_rate": 8.963947137675754e-06,
      "loss": 9.844,
      "step": 2809
    },
    {
      "epoch": 0.5490425947635795,
      "grad_norm": 9.942178726196289,
      "learning_rate": 8.95765265148886e-06,
      "loss": 9.7945,
      "step": 2810
    },
    {
      "epoch": 0.5492379835873388,
      "grad_norm": 9.615091323852539,
      "learning_rate": 8.951358582794438e-06,
      "loss": 9.2248,
      "step": 2811
    },
    {
      "epoch": 0.5494333724110981,
      "grad_norm": 8.68079948425293,
      "learning_rate": 8.945064934113464e-06,
      "loss": 9.2415,
      "step": 2812
    },
    {
      "epoch": 0.5496287612348574,
      "grad_norm": 10.706026077270508,
      "learning_rate": 8.938771707966733e-06,
      "loss": 9.3499,
      "step": 2813
    },
    {
      "epoch": 0.5498241500586166,
      "grad_norm": 7.729515075683594,
      "learning_rate": 8.93247890687488e-06,
      "loss": 8.0236,
      "step": 2814
    },
    {
      "epoch": 0.550019538882376,
      "grad_norm": 10.87379264831543,
      "learning_rate": 8.92618653335837e-06,
      "loss": 10.2506,
      "step": 2815
    },
    {
      "epoch": 0.5502149277061352,
      "grad_norm": 7.861601829528809,
      "learning_rate": 8.919894589937492e-06,
      "loss": 8.2541,
      "step": 2816
    },
    {
      "epoch": 0.5504103165298945,
      "grad_norm": 9.677508354187012,
      "learning_rate": 8.913603079132361e-06,
      "loss": 9.235,
      "step": 2817
    },
    {
      "epoch": 0.5506057053536537,
      "grad_norm": 8.08492660522461,
      "learning_rate": 8.907312003462926e-06,
      "loss": 9.1055,
      "step": 2818
    },
    {
      "epoch": 0.5508010941774131,
      "grad_norm": 7.463259696960449,
      "learning_rate": 8.901021365448959e-06,
      "loss": 8.8301,
      "step": 2819
    },
    {
      "epoch": 0.5509964830011723,
      "grad_norm": 8.580756187438965,
      "learning_rate": 8.894731167610054e-06,
      "loss": 9.5622,
      "step": 2820
    },
    {
      "epoch": 0.5511918718249316,
      "grad_norm": 9.969331741333008,
      "learning_rate": 8.888441412465631e-06,
      "loss": 9.2752,
      "step": 2821
    },
    {
      "epoch": 0.5513872606486909,
      "grad_norm": 6.705500602722168,
      "learning_rate": 8.88215210253493e-06,
      "loss": 9.0879,
      "step": 2822
    },
    {
      "epoch": 0.5515826494724502,
      "grad_norm": 8.789013862609863,
      "learning_rate": 8.87586324033702e-06,
      "loss": 10.0995,
      "step": 2823
    },
    {
      "epoch": 0.5517780382962094,
      "grad_norm": 8.352804183959961,
      "learning_rate": 8.869574828390782e-06,
      "loss": 9.2406,
      "step": 2824
    },
    {
      "epoch": 0.5519734271199688,
      "grad_norm": 9.722862243652344,
      "learning_rate": 8.86328686921492e-06,
      "loss": 8.4509,
      "step": 2825
    },
    {
      "epoch": 0.552168815943728,
      "grad_norm": 45.539649963378906,
      "learning_rate": 8.856999365327959e-06,
      "loss": 9.6746,
      "step": 2826
    },
    {
      "epoch": 0.5523642047674873,
      "grad_norm": 7.403289794921875,
      "learning_rate": 8.850712319248237e-06,
      "loss": 9.0779,
      "step": 2827
    },
    {
      "epoch": 0.5525595935912466,
      "grad_norm": 9.596833229064941,
      "learning_rate": 8.844425733493915e-06,
      "loss": 10.3165,
      "step": 2828
    },
    {
      "epoch": 0.5527549824150059,
      "grad_norm": 7.67540168762207,
      "learning_rate": 8.83813961058296e-06,
      "loss": 9.3652,
      "step": 2829
    },
    {
      "epoch": 0.5529503712387651,
      "grad_norm": 7.21089506149292,
      "learning_rate": 8.831853953033165e-06,
      "loss": 8.4772,
      "step": 2830
    },
    {
      "epoch": 0.5531457600625245,
      "grad_norm": 7.019738674163818,
      "learning_rate": 8.825568763362126e-06,
      "loss": 9.5753,
      "step": 2831
    },
    {
      "epoch": 0.5533411488862837,
      "grad_norm": 7.952762126922607,
      "learning_rate": 8.81928404408726e-06,
      "loss": 9.3136,
      "step": 2832
    },
    {
      "epoch": 0.553536537710043,
      "grad_norm": 9.55434513092041,
      "learning_rate": 8.812999797725793e-06,
      "loss": 9.0351,
      "step": 2833
    },
    {
      "epoch": 0.5537319265338023,
      "grad_norm": 7.781782627105713,
      "learning_rate": 8.806716026794758e-06,
      "loss": 8.1702,
      "step": 2834
    },
    {
      "epoch": 0.5539273153575616,
      "grad_norm": 7.829460144042969,
      "learning_rate": 8.800432733811e-06,
      "loss": 8.661,
      "step": 2835
    },
    {
      "epoch": 0.5541227041813208,
      "grad_norm": 8.869157791137695,
      "learning_rate": 8.794149921291176e-06,
      "loss": 9.5003,
      "step": 2836
    },
    {
      "epoch": 0.5543180930050801,
      "grad_norm": 7.07666540145874,
      "learning_rate": 8.787867591751746e-06,
      "loss": 8.7779,
      "step": 2837
    },
    {
      "epoch": 0.5545134818288394,
      "grad_norm": 7.235918045043945,
      "learning_rate": 8.781585747708976e-06,
      "loss": 9.5523,
      "step": 2838
    },
    {
      "epoch": 0.5547088706525987,
      "grad_norm": 8.273935317993164,
      "learning_rate": 8.775304391678943e-06,
      "loss": 10.2941,
      "step": 2839
    },
    {
      "epoch": 0.5549042594763579,
      "grad_norm": 12.045279502868652,
      "learning_rate": 8.769023526177524e-06,
      "loss": 9.8395,
      "step": 2840
    },
    {
      "epoch": 0.5550996483001173,
      "grad_norm": 10.213648796081543,
      "learning_rate": 8.7627431537204e-06,
      "loss": 9.3265,
      "step": 2841
    },
    {
      "epoch": 0.5552950371238765,
      "grad_norm": 9.113309860229492,
      "learning_rate": 8.756463276823054e-06,
      "loss": 9.5764,
      "step": 2842
    },
    {
      "epoch": 0.5554904259476358,
      "grad_norm": 9.318734169006348,
      "learning_rate": 8.750183898000775e-06,
      "loss": 9.885,
      "step": 2843
    },
    {
      "epoch": 0.5556858147713951,
      "grad_norm": 9.162623405456543,
      "learning_rate": 8.743905019768646e-06,
      "loss": 10.3492,
      "step": 2844
    },
    {
      "epoch": 0.5558812035951544,
      "grad_norm": 7.756526470184326,
      "learning_rate": 8.737626644641553e-06,
      "loss": 8.9615,
      "step": 2845
    },
    {
      "epoch": 0.5560765924189136,
      "grad_norm": 18.73146629333496,
      "learning_rate": 8.731348775134182e-06,
      "loss": 8.7326,
      "step": 2846
    },
    {
      "epoch": 0.556271981242673,
      "grad_norm": 18.97466468811035,
      "learning_rate": 8.725071413761013e-06,
      "loss": 9.347,
      "step": 2847
    },
    {
      "epoch": 0.5564673700664322,
      "grad_norm": 7.583729267120361,
      "learning_rate": 8.718794563036326e-06,
      "loss": 8.5347,
      "step": 2848
    },
    {
      "epoch": 0.5566627588901915,
      "grad_norm": 9.999347686767578,
      "learning_rate": 8.712518225474191e-06,
      "loss": 8.9671,
      "step": 2849
    },
    {
      "epoch": 0.5568581477139508,
      "grad_norm": 8.69705867767334,
      "learning_rate": 8.70624240358848e-06,
      "loss": 9.2673,
      "step": 2850
    },
    {
      "epoch": 0.5570535365377101,
      "grad_norm": 7.3665618896484375,
      "learning_rate": 8.699967099892851e-06,
      "loss": 9.4397,
      "step": 2851
    },
    {
      "epoch": 0.5572489253614693,
      "grad_norm": 8.915785789489746,
      "learning_rate": 8.693692316900768e-06,
      "loss": 9.6176,
      "step": 2852
    },
    {
      "epoch": 0.5574443141852286,
      "grad_norm": 8.13288402557373,
      "learning_rate": 8.68741805712546e-06,
      "loss": 9.3996,
      "step": 2853
    },
    {
      "epoch": 0.5576397030089879,
      "grad_norm": 6.944223403930664,
      "learning_rate": 8.681144323079973e-06,
      "loss": 9.267,
      "step": 2854
    },
    {
      "epoch": 0.5578350918327472,
      "grad_norm": 9.68083667755127,
      "learning_rate": 8.67487111727713e-06,
      "loss": 8.8771,
      "step": 2855
    },
    {
      "epoch": 0.5580304806565064,
      "grad_norm": 9.229497909545898,
      "learning_rate": 8.668598442229545e-06,
      "loss": 9.5669,
      "step": 2856
    },
    {
      "epoch": 0.5582258694802658,
      "grad_norm": 8.748711585998535,
      "learning_rate": 8.66232630044962e-06,
      "loss": 9.3087,
      "step": 2857
    },
    {
      "epoch": 0.558421258304025,
      "grad_norm": 8.638025283813477,
      "learning_rate": 8.656054694449541e-06,
      "loss": 8.0628,
      "step": 2858
    },
    {
      "epoch": 0.5586166471277842,
      "grad_norm": 11.083617210388184,
      "learning_rate": 8.64978362674128e-06,
      "loss": 9.7794,
      "step": 2859
    },
    {
      "epoch": 0.5588120359515436,
      "grad_norm": 12.017905235290527,
      "learning_rate": 8.6435130998366e-06,
      "loss": 8.9731,
      "step": 2860
    },
    {
      "epoch": 0.5590074247753029,
      "grad_norm": 8.4308500289917,
      "learning_rate": 8.637243116247036e-06,
      "loss": 7.6834,
      "step": 2861
    },
    {
      "epoch": 0.5592028135990621,
      "grad_norm": 8.182673454284668,
      "learning_rate": 8.630973678483911e-06,
      "loss": 8.5101,
      "step": 2862
    },
    {
      "epoch": 0.5593982024228215,
      "grad_norm": 8.208181381225586,
      "learning_rate": 8.624704789058331e-06,
      "loss": 9.0657,
      "step": 2863
    },
    {
      "epoch": 0.5595935912465807,
      "grad_norm": 10.765786170959473,
      "learning_rate": 8.618436450481182e-06,
      "loss": 9.9012,
      "step": 2864
    },
    {
      "epoch": 0.55978898007034,
      "grad_norm": 10.154743194580078,
      "learning_rate": 8.612168665263125e-06,
      "loss": 7.8456,
      "step": 2865
    },
    {
      "epoch": 0.5599843688940993,
      "grad_norm": 13.1641263961792,
      "learning_rate": 8.605901435914608e-06,
      "loss": 9.3338,
      "step": 2866
    },
    {
      "epoch": 0.5601797577178586,
      "grad_norm": 7.61334228515625,
      "learning_rate": 8.599634764945845e-06,
      "loss": 7.9552,
      "step": 2867
    },
    {
      "epoch": 0.5603751465416178,
      "grad_norm": 9.793645858764648,
      "learning_rate": 8.593368654866835e-06,
      "loss": 8.9891,
      "step": 2868
    },
    {
      "epoch": 0.5605705353653772,
      "grad_norm": 8.298504829406738,
      "learning_rate": 8.587103108187352e-06,
      "loss": 8.6191,
      "step": 2869
    },
    {
      "epoch": 0.5607659241891364,
      "grad_norm": 10.942178726196289,
      "learning_rate": 8.580838127416937e-06,
      "loss": 9.8686,
      "step": 2870
    },
    {
      "epoch": 0.5609613130128956,
      "grad_norm": 8.616482734680176,
      "learning_rate": 8.574573715064913e-06,
      "loss": 9.1715,
      "step": 2871
    },
    {
      "epoch": 0.5611567018366549,
      "grad_norm": 16.707321166992188,
      "learning_rate": 8.56830987364037e-06,
      "loss": 9.2851,
      "step": 2872
    },
    {
      "epoch": 0.5613520906604142,
      "grad_norm": 8.741340637207031,
      "learning_rate": 8.562046605652174e-06,
      "loss": 10.2423,
      "step": 2873
    },
    {
      "epoch": 0.5615474794841735,
      "grad_norm": 9.816158294677734,
      "learning_rate": 8.555783913608955e-06,
      "loss": 9.1263,
      "step": 2874
    },
    {
      "epoch": 0.5617428683079327,
      "grad_norm": 12.084359169006348,
      "learning_rate": 8.549521800019115e-06,
      "loss": 9.1611,
      "step": 2875
    },
    {
      "epoch": 0.5619382571316921,
      "grad_norm": 7.766233921051025,
      "learning_rate": 8.543260267390824e-06,
      "loss": 9.5226,
      "step": 2876
    },
    {
      "epoch": 0.5621336459554513,
      "grad_norm": 9.25195026397705,
      "learning_rate": 8.536999318232025e-06,
      "loss": 9.8658,
      "step": 2877
    },
    {
      "epoch": 0.5623290347792106,
      "grad_norm": 7.865414619445801,
      "learning_rate": 8.53073895505042e-06,
      "loss": 7.745,
      "step": 2878
    },
    {
      "epoch": 0.56252442360297,
      "grad_norm": 9.51463508605957,
      "learning_rate": 8.52447918035348e-06,
      "loss": 9.9384,
      "step": 2879
    },
    {
      "epoch": 0.5627198124267292,
      "grad_norm": 11.417374610900879,
      "learning_rate": 8.518219996648435e-06,
      "loss": 9.4388,
      "step": 2880
    },
    {
      "epoch": 0.5629152012504884,
      "grad_norm": 10.57020092010498,
      "learning_rate": 8.511961406442286e-06,
      "loss": 8.6496,
      "step": 2881
    },
    {
      "epoch": 0.5631105900742478,
      "grad_norm": 20.026182174682617,
      "learning_rate": 8.505703412241793e-06,
      "loss": 10.1787,
      "step": 2882
    },
    {
      "epoch": 0.563305978898007,
      "grad_norm": 10.828873634338379,
      "learning_rate": 8.499446016553475e-06,
      "loss": 9.5856,
      "step": 2883
    },
    {
      "epoch": 0.5635013677217663,
      "grad_norm": 9.27635383605957,
      "learning_rate": 8.493189221883615e-06,
      "loss": 7.8634,
      "step": 2884
    },
    {
      "epoch": 0.5636967565455256,
      "grad_norm": 15.371089935302734,
      "learning_rate": 8.486933030738252e-06,
      "loss": 9.4457,
      "step": 2885
    },
    {
      "epoch": 0.5638921453692849,
      "grad_norm": 8.049420356750488,
      "learning_rate": 8.480677445623187e-06,
      "loss": 9.6421,
      "step": 2886
    },
    {
      "epoch": 0.5640875341930441,
      "grad_norm": 11.85643482208252,
      "learning_rate": 8.474422469043974e-06,
      "loss": 9.5314,
      "step": 2887
    },
    {
      "epoch": 0.5642829230168034,
      "grad_norm": 9.827743530273438,
      "learning_rate": 8.468168103505926e-06,
      "loss": 9.3676,
      "step": 2888
    },
    {
      "epoch": 0.5644783118405627,
      "grad_norm": 7.995309352874756,
      "learning_rate": 8.461914351514109e-06,
      "loss": 8.5075,
      "step": 2889
    },
    {
      "epoch": 0.564673700664322,
      "grad_norm": 9.013062477111816,
      "learning_rate": 8.455661215573345e-06,
      "loss": 8.5168,
      "step": 2890
    },
    {
      "epoch": 0.5648690894880812,
      "grad_norm": 12.336776733398438,
      "learning_rate": 8.449408698188212e-06,
      "loss": 10.6045,
      "step": 2891
    },
    {
      "epoch": 0.5650644783118406,
      "grad_norm": 8.412564277648926,
      "learning_rate": 8.443156801863037e-06,
      "loss": 8.6074,
      "step": 2892
    },
    {
      "epoch": 0.5652598671355998,
      "grad_norm": 6.834690093994141,
      "learning_rate": 8.436905529101896e-06,
      "loss": 8.5053,
      "step": 2893
    },
    {
      "epoch": 0.5654552559593591,
      "grad_norm": 7.084744930267334,
      "learning_rate": 8.430654882408619e-06,
      "loss": 8.0449,
      "step": 2894
    },
    {
      "epoch": 0.5656506447831184,
      "grad_norm": 8.889432907104492,
      "learning_rate": 8.424404864286784e-06,
      "loss": 8.8036,
      "step": 2895
    },
    {
      "epoch": 0.5658460336068777,
      "grad_norm": 8.221407890319824,
      "learning_rate": 8.418155477239718e-06,
      "loss": 9.1227,
      "step": 2896
    },
    {
      "epoch": 0.5660414224306369,
      "grad_norm": 8.30092716217041,
      "learning_rate": 8.411906723770493e-06,
      "loss": 8.8562,
      "step": 2897
    },
    {
      "epoch": 0.5662368112543963,
      "grad_norm": 8.753860473632812,
      "learning_rate": 8.40565860638193e-06,
      "loss": 8.9539,
      "step": 2898
    },
    {
      "epoch": 0.5664322000781555,
      "grad_norm": 9.127002716064453,
      "learning_rate": 8.399411127576597e-06,
      "loss": 9.6091,
      "step": 2899
    },
    {
      "epoch": 0.5666275889019148,
      "grad_norm": 9.005349159240723,
      "learning_rate": 8.393164289856797e-06,
      "loss": 9.0778,
      "step": 2900
    },
    {
      "epoch": 0.5668229777256741,
      "grad_norm": 8.229484558105469,
      "learning_rate": 8.386918095724584e-06,
      "loss": 9.9055,
      "step": 2901
    },
    {
      "epoch": 0.5670183665494334,
      "grad_norm": 6.597065448760986,
      "learning_rate": 8.380672547681757e-06,
      "loss": 8.0167,
      "step": 2902
    },
    {
      "epoch": 0.5672137553731926,
      "grad_norm": 9.146075248718262,
      "learning_rate": 8.374427648229845e-06,
      "loss": 8.7352,
      "step": 2903
    },
    {
      "epoch": 0.567409144196952,
      "grad_norm": 9.293710708618164,
      "learning_rate": 8.368183399870132e-06,
      "loss": 9.692,
      "step": 2904
    },
    {
      "epoch": 0.5676045330207112,
      "grad_norm": 7.453103542327881,
      "learning_rate": 8.361939805103628e-06,
      "loss": 8.6005,
      "step": 2905
    },
    {
      "epoch": 0.5677999218444705,
      "grad_norm": 6.154191970825195,
      "learning_rate": 8.355696866431087e-06,
      "loss": 9.0064,
      "step": 2906
    },
    {
      "epoch": 0.5679953106682297,
      "grad_norm": 10.142219543457031,
      "learning_rate": 8.349454586353e-06,
      "loss": 10.9779,
      "step": 2907
    },
    {
      "epoch": 0.5681906994919891,
      "grad_norm": 6.828899383544922,
      "learning_rate": 8.343212967369598e-06,
      "loss": 9.5795,
      "step": 2908
    },
    {
      "epoch": 0.5683860883157483,
      "grad_norm": 8.226543426513672,
      "learning_rate": 8.336972011980837e-06,
      "loss": 8.374,
      "step": 2909
    },
    {
      "epoch": 0.5685814771395076,
      "grad_norm": 7.978690147399902,
      "learning_rate": 8.330731722686416e-06,
      "loss": 8.69,
      "step": 2910
    },
    {
      "epoch": 0.5687768659632669,
      "grad_norm": 7.589662075042725,
      "learning_rate": 8.324492101985766e-06,
      "loss": 9.587,
      "step": 2911
    },
    {
      "epoch": 0.5689722547870262,
      "grad_norm": 7.409912109375,
      "learning_rate": 8.31825315237805e-06,
      "loss": 8.7215,
      "step": 2912
    },
    {
      "epoch": 0.5691676436107854,
      "grad_norm": 8.238944053649902,
      "learning_rate": 8.312014876362157e-06,
      "loss": 8.2848,
      "step": 2913
    },
    {
      "epoch": 0.5693630324345448,
      "grad_norm": 8.096856117248535,
      "learning_rate": 8.305777276436712e-06,
      "loss": 8.7827,
      "step": 2914
    },
    {
      "epoch": 0.569558421258304,
      "grad_norm": 7.199119567871094,
      "learning_rate": 8.299540355100066e-06,
      "loss": 8.9888,
      "step": 2915
    },
    {
      "epoch": 0.5697538100820633,
      "grad_norm": 9.162145614624023,
      "learning_rate": 8.293304114850302e-06,
      "loss": 9.1039,
      "step": 2916
    },
    {
      "epoch": 0.5699491989058226,
      "grad_norm": 8.96921157836914,
      "learning_rate": 8.287068558185225e-06,
      "loss": 9.3564,
      "step": 2917
    },
    {
      "epoch": 0.5701445877295819,
      "grad_norm": 8.391603469848633,
      "learning_rate": 8.280833687602372e-06,
      "loss": 9.2859,
      "step": 2918
    },
    {
      "epoch": 0.5703399765533411,
      "grad_norm": 8.860815048217773,
      "learning_rate": 8.274599505599002e-06,
      "loss": 8.9089,
      "step": 2919
    },
    {
      "epoch": 0.5705353653771005,
      "grad_norm": 10.831894874572754,
      "learning_rate": 8.268366014672095e-06,
      "loss": 8.3156,
      "step": 2920
    },
    {
      "epoch": 0.5707307542008597,
      "grad_norm": 8.517152786254883,
      "learning_rate": 8.262133217318363e-06,
      "loss": 7.8869,
      "step": 2921
    },
    {
      "epoch": 0.570926143024619,
      "grad_norm": 8.543846130371094,
      "learning_rate": 8.255901116034231e-06,
      "loss": 8.3345,
      "step": 2922
    },
    {
      "epoch": 0.5711215318483783,
      "grad_norm": 32.423622131347656,
      "learning_rate": 8.249669713315849e-06,
      "loss": 10.3243,
      "step": 2923
    },
    {
      "epoch": 0.5713169206721376,
      "grad_norm": 9.708186149597168,
      "learning_rate": 8.24343901165909e-06,
      "loss": 9.1622,
      "step": 2924
    },
    {
      "epoch": 0.5715123094958968,
      "grad_norm": 10.76029109954834,
      "learning_rate": 8.237209013559542e-06,
      "loss": 9.2075,
      "step": 2925
    },
    {
      "epoch": 0.5717076983196561,
      "grad_norm": 9.259683609008789,
      "learning_rate": 8.230979721512511e-06,
      "loss": 9.4542,
      "step": 2926
    },
    {
      "epoch": 0.5719030871434154,
      "grad_norm": 7.613196849822998,
      "learning_rate": 8.224751138013025e-06,
      "loss": 9.2146,
      "step": 2927
    },
    {
      "epoch": 0.5720984759671747,
      "grad_norm": 8.023900985717773,
      "learning_rate": 8.21852326555582e-06,
      "loss": 8.7284,
      "step": 2928
    },
    {
      "epoch": 0.5722938647909339,
      "grad_norm": 7.808635711669922,
      "learning_rate": 8.212296106635358e-06,
      "loss": 9.2191,
      "step": 2929
    },
    {
      "epoch": 0.5724892536146933,
      "grad_norm": 10.434918403625488,
      "learning_rate": 8.206069663745806e-06,
      "loss": 9.3047,
      "step": 2930
    },
    {
      "epoch": 0.5726846424384525,
      "grad_norm": 9.0089111328125,
      "learning_rate": 8.19984393938105e-06,
      "loss": 9.5568,
      "step": 2931
    },
    {
      "epoch": 0.5728800312622118,
      "grad_norm": 9.408156394958496,
      "learning_rate": 8.193618936034683e-06,
      "loss": 9.1575,
      "step": 2932
    },
    {
      "epoch": 0.5730754200859711,
      "grad_norm": 8.474098205566406,
      "learning_rate": 8.187394656200013e-06,
      "loss": 10.0839,
      "step": 2933
    },
    {
      "epoch": 0.5732708089097304,
      "grad_norm": 7.609988689422607,
      "learning_rate": 8.181171102370062e-06,
      "loss": 8.9113,
      "step": 2934
    },
    {
      "epoch": 0.5734661977334896,
      "grad_norm": 9.248665809631348,
      "learning_rate": 8.174948277037548e-06,
      "loss": 8.7736,
      "step": 2935
    },
    {
      "epoch": 0.573661586557249,
      "grad_norm": 8.241129875183105,
      "learning_rate": 8.168726182694913e-06,
      "loss": 9.8128,
      "step": 2936
    },
    {
      "epoch": 0.5738569753810082,
      "grad_norm": 9.227700233459473,
      "learning_rate": 8.162504821834296e-06,
      "loss": 9.1306,
      "step": 2937
    },
    {
      "epoch": 0.5740523642047675,
      "grad_norm": 7.878133773803711,
      "learning_rate": 8.156284196947549e-06,
      "loss": 8.7135,
      "step": 2938
    },
    {
      "epoch": 0.5742477530285268,
      "grad_norm": 8.619709014892578,
      "learning_rate": 8.150064310526217e-06,
      "loss": 8.6783,
      "step": 2939
    },
    {
      "epoch": 0.5744431418522861,
      "grad_norm": 9.555558204650879,
      "learning_rate": 8.143845165061564e-06,
      "loss": 9.4556,
      "step": 2940
    },
    {
      "epoch": 0.5746385306760453,
      "grad_norm": 7.761295795440674,
      "learning_rate": 8.13762676304455e-06,
      "loss": 9.2814,
      "step": 2941
    },
    {
      "epoch": 0.5748339194998046,
      "grad_norm": 8.48897647857666,
      "learning_rate": 8.131409106965836e-06,
      "loss": 8.5359,
      "step": 2942
    },
    {
      "epoch": 0.5750293083235639,
      "grad_norm": 6.357192039489746,
      "learning_rate": 8.125192199315788e-06,
      "loss": 9.0464,
      "step": 2943
    },
    {
      "epoch": 0.5752246971473232,
      "grad_norm": 8.602022171020508,
      "learning_rate": 8.11897604258447e-06,
      "loss": 8.2716,
      "step": 2944
    },
    {
      "epoch": 0.5754200859710824,
      "grad_norm": 12.048428535461426,
      "learning_rate": 8.112760639261646e-06,
      "loss": 9.6382,
      "step": 2945
    },
    {
      "epoch": 0.5756154747948418,
      "grad_norm": 8.437899589538574,
      "learning_rate": 8.106545991836778e-06,
      "loss": 10.1358,
      "step": 2946
    },
    {
      "epoch": 0.575810863618601,
      "grad_norm": 8.29260540008545,
      "learning_rate": 8.100332102799026e-06,
      "loss": 9.3783,
      "step": 2947
    },
    {
      "epoch": 0.5760062524423603,
      "grad_norm": 10.251049995422363,
      "learning_rate": 8.094118974637245e-06,
      "loss": 9.6561,
      "step": 2948
    },
    {
      "epoch": 0.5762016412661196,
      "grad_norm": 7.310251235961914,
      "learning_rate": 8.087906609839984e-06,
      "loss": 9.7492,
      "step": 2949
    },
    {
      "epoch": 0.5763970300898789,
      "grad_norm": 9.511523246765137,
      "learning_rate": 8.081695010895489e-06,
      "loss": 9.3339,
      "step": 2950
    },
    {
      "epoch": 0.5765924189136381,
      "grad_norm": 9.167460441589355,
      "learning_rate": 8.075484180291702e-06,
      "loss": 9.8695,
      "step": 2951
    },
    {
      "epoch": 0.5767878077373975,
      "grad_norm": 9.307744026184082,
      "learning_rate": 8.069274120516249e-06,
      "loss": 9.5672,
      "step": 2952
    },
    {
      "epoch": 0.5769831965611567,
      "grad_norm": 6.490602493286133,
      "learning_rate": 8.06306483405645e-06,
      "loss": 9.5354,
      "step": 2953
    },
    {
      "epoch": 0.577178585384916,
      "grad_norm": 8.028519630432129,
      "learning_rate": 8.056856323399321e-06,
      "loss": 8.4964,
      "step": 2954
    },
    {
      "epoch": 0.5773739742086753,
      "grad_norm": 9.199743270874023,
      "learning_rate": 8.050648591031562e-06,
      "loss": 9.4259,
      "step": 2955
    },
    {
      "epoch": 0.5775693630324346,
      "grad_norm": 8.184178352355957,
      "learning_rate": 8.044441639439564e-06,
      "loss": 10.0353,
      "step": 2956
    },
    {
      "epoch": 0.5777647518561938,
      "grad_norm": 7.361752510070801,
      "learning_rate": 8.038235471109401e-06,
      "loss": 8.7964,
      "step": 2957
    },
    {
      "epoch": 0.5779601406799532,
      "grad_norm": 8.456011772155762,
      "learning_rate": 8.03203008852684e-06,
      "loss": 9.1222,
      "step": 2958
    },
    {
      "epoch": 0.5781555295037124,
      "grad_norm": 34.3315315246582,
      "learning_rate": 8.025825494177324e-06,
      "loss": 8.6756,
      "step": 2959
    },
    {
      "epoch": 0.5783509183274717,
      "grad_norm": 7.625585556030273,
      "learning_rate": 8.01962169054599e-06,
      "loss": 9.5105,
      "step": 2960
    },
    {
      "epoch": 0.5785463071512309,
      "grad_norm": 10.44150447845459,
      "learning_rate": 8.013418680117652e-06,
      "loss": 9.686,
      "step": 2961
    },
    {
      "epoch": 0.5787416959749903,
      "grad_norm": 7.208309650421143,
      "learning_rate": 8.007216465376809e-06,
      "loss": 8.249,
      "step": 2962
    },
    {
      "epoch": 0.5789370847987495,
      "grad_norm": 10.5735502243042,
      "learning_rate": 8.00101504880764e-06,
      "loss": 8.962,
      "step": 2963
    },
    {
      "epoch": 0.5791324736225087,
      "grad_norm": 7.800056457519531,
      "learning_rate": 7.994814432894008e-06,
      "loss": 8.4317,
      "step": 2964
    },
    {
      "epoch": 0.5793278624462681,
      "grad_norm": 14.920567512512207,
      "learning_rate": 7.988614620119446e-06,
      "loss": 8.8176,
      "step": 2965
    },
    {
      "epoch": 0.5795232512700274,
      "grad_norm": 6.216339588165283,
      "learning_rate": 7.982415612967177e-06,
      "loss": 8.0652,
      "step": 2966
    },
    {
      "epoch": 0.5797186400937866,
      "grad_norm": 11.770721435546875,
      "learning_rate": 7.976217413920093e-06,
      "loss": 9.7224,
      "step": 2967
    },
    {
      "epoch": 0.579914028917546,
      "grad_norm": 14.045543670654297,
      "learning_rate": 7.970020025460765e-06,
      "loss": 8.9005,
      "step": 2968
    },
    {
      "epoch": 0.5801094177413052,
      "grad_norm": 7.276342868804932,
      "learning_rate": 7.963823450071442e-06,
      "loss": 9.1735,
      "step": 2969
    },
    {
      "epoch": 0.5803048065650644,
      "grad_norm": 10.889382362365723,
      "learning_rate": 7.957627690234042e-06,
      "loss": 9.7824,
      "step": 2970
    },
    {
      "epoch": 0.5805001953888238,
      "grad_norm": 8.489128112792969,
      "learning_rate": 7.95143274843016e-06,
      "loss": 9.0536,
      "step": 2971
    },
    {
      "epoch": 0.580695584212583,
      "grad_norm": 7.522155284881592,
      "learning_rate": 7.945238627141063e-06,
      "loss": 9.0415,
      "step": 2972
    },
    {
      "epoch": 0.5808909730363423,
      "grad_norm": 9.565140724182129,
      "learning_rate": 7.93904532884769e-06,
      "loss": 9.0799,
      "step": 2973
    },
    {
      "epoch": 0.5810863618601017,
      "grad_norm": 10.56672477722168,
      "learning_rate": 7.932852856030646e-06,
      "loss": 9.5377,
      "step": 2974
    },
    {
      "epoch": 0.5812817506838609,
      "grad_norm": 7.589129447937012,
      "learning_rate": 7.926661211170213e-06,
      "loss": 8.8639,
      "step": 2975
    },
    {
      "epoch": 0.5814771395076201,
      "grad_norm": 9.25190544128418,
      "learning_rate": 7.920470396746334e-06,
      "loss": 9.2866,
      "step": 2976
    },
    {
      "epoch": 0.5816725283313794,
      "grad_norm": 9.842716217041016,
      "learning_rate": 7.914280415238624e-06,
      "loss": 9.493,
      "step": 2977
    },
    {
      "epoch": 0.5818679171551387,
      "grad_norm": 8.670297622680664,
      "learning_rate": 7.90809126912636e-06,
      "loss": 8.3575,
      "step": 2978
    },
    {
      "epoch": 0.582063305978898,
      "grad_norm": 115.61127471923828,
      "learning_rate": 7.90190296088849e-06,
      "loss": 10.4128,
      "step": 2979
    },
    {
      "epoch": 0.5822586948026572,
      "grad_norm": 8.339439392089844,
      "learning_rate": 7.895715493003623e-06,
      "loss": 8.7615,
      "step": 2980
    },
    {
      "epoch": 0.5824540836264166,
      "grad_norm": 7.4925537109375,
      "learning_rate": 7.88952886795003e-06,
      "loss": 9.2757,
      "step": 2981
    },
    {
      "epoch": 0.5826494724501758,
      "grad_norm": 9.57435131072998,
      "learning_rate": 7.883343088205648e-06,
      "loss": 8.7239,
      "step": 2982
    },
    {
      "epoch": 0.5828448612739351,
      "grad_norm": 9.265599250793457,
      "learning_rate": 7.877158156248075e-06,
      "loss": 9.0901,
      "step": 2983
    },
    {
      "epoch": 0.5830402500976944,
      "grad_norm": 8.441311836242676,
      "learning_rate": 7.870974074554566e-06,
      "loss": 8.9896,
      "step": 2984
    },
    {
      "epoch": 0.5832356389214537,
      "grad_norm": 9.167561531066895,
      "learning_rate": 7.86479084560204e-06,
      "loss": 9.6028,
      "step": 2985
    },
    {
      "epoch": 0.5834310277452129,
      "grad_norm": 8.28317642211914,
      "learning_rate": 7.858608471867069e-06,
      "loss": 8.3372,
      "step": 2986
    },
    {
      "epoch": 0.5836264165689723,
      "grad_norm": 9.223546981811523,
      "learning_rate": 7.852426955825889e-06,
      "loss": 9.7081,
      "step": 2987
    },
    {
      "epoch": 0.5838218053927315,
      "grad_norm": 8.843958854675293,
      "learning_rate": 7.846246299954386e-06,
      "loss": 9.6832,
      "step": 2988
    },
    {
      "epoch": 0.5840171942164908,
      "grad_norm": 7.481940269470215,
      "learning_rate": 7.840066506728108e-06,
      "loss": 8.7187,
      "step": 2989
    },
    {
      "epoch": 0.5842125830402501,
      "grad_norm": 8.704890251159668,
      "learning_rate": 7.833887578622254e-06,
      "loss": 9.2202,
      "step": 2990
    },
    {
      "epoch": 0.5844079718640094,
      "grad_norm": 70.9208755493164,
      "learning_rate": 7.827709518111674e-06,
      "loss": 10.2329,
      "step": 2991
    },
    {
      "epoch": 0.5846033606877686,
      "grad_norm": 7.739467144012451,
      "learning_rate": 7.821532327670871e-06,
      "loss": 9.1888,
      "step": 2992
    },
    {
      "epoch": 0.584798749511528,
      "grad_norm": 11.358851432800293,
      "learning_rate": 7.815356009774006e-06,
      "loss": 9.6033,
      "step": 2993
    },
    {
      "epoch": 0.5849941383352872,
      "grad_norm": 10.573582649230957,
      "learning_rate": 7.809180566894882e-06,
      "loss": 8.9731,
      "step": 2994
    },
    {
      "epoch": 0.5851895271590465,
      "grad_norm": 8.335413932800293,
      "learning_rate": 7.80300600150696e-06,
      "loss": 8.6475,
      "step": 2995
    },
    {
      "epoch": 0.5853849159828057,
      "grad_norm": 8.019357681274414,
      "learning_rate": 7.796832316083347e-06,
      "loss": 8.9185,
      "step": 2996
    },
    {
      "epoch": 0.5855803048065651,
      "grad_norm": 8.108970642089844,
      "learning_rate": 7.790659513096787e-06,
      "loss": 9.8417,
      "step": 2997
    },
    {
      "epoch": 0.5857756936303243,
      "grad_norm": 6.562138557434082,
      "learning_rate": 7.784487595019686e-06,
      "loss": 9.0271,
      "step": 2998
    },
    {
      "epoch": 0.5859710824540836,
      "grad_norm": 6.990952014923096,
      "learning_rate": 7.778316564324085e-06,
      "loss": 9.229,
      "step": 2999
    },
    {
      "epoch": 0.5861664712778429,
      "grad_norm": 9.679683685302734,
      "learning_rate": 7.772146423481676e-06,
      "loss": 9.4981,
      "step": 3000
    },
    {
      "epoch": 0.5863618601016022,
      "grad_norm": 8.7763090133667,
      "learning_rate": 7.76597717496379e-06,
      "loss": 8.3711,
      "step": 3001
    },
    {
      "epoch": 0.5865572489253614,
      "grad_norm": 9.195281028747559,
      "learning_rate": 7.759808821241406e-06,
      "loss": 9.6284,
      "step": 3002
    },
    {
      "epoch": 0.5867526377491208,
      "grad_norm": 6.703279972076416,
      "learning_rate": 7.753641364785139e-06,
      "loss": 8.2272,
      "step": 3003
    },
    {
      "epoch": 0.58694802657288,
      "grad_norm": 7.0960001945495605,
      "learning_rate": 7.747474808065244e-06,
      "loss": 7.9143,
      "step": 3004
    },
    {
      "epoch": 0.5871434153966393,
      "grad_norm": 7.982921123504639,
      "learning_rate": 7.741309153551625e-06,
      "loss": 8.9272,
      "step": 3005
    },
    {
      "epoch": 0.5873388042203986,
      "grad_norm": 49.128475189208984,
      "learning_rate": 7.735144403713811e-06,
      "loss": 10.1243,
      "step": 3006
    },
    {
      "epoch": 0.5875341930441579,
      "grad_norm": 7.292541027069092,
      "learning_rate": 7.72898056102098e-06,
      "loss": 8.2249,
      "step": 3007
    },
    {
      "epoch": 0.5877295818679171,
      "grad_norm": 7.324792861938477,
      "learning_rate": 7.722817627941943e-06,
      "loss": 9.1744,
      "step": 3008
    },
    {
      "epoch": 0.5879249706916765,
      "grad_norm": 7.497281551361084,
      "learning_rate": 7.716655606945143e-06,
      "loss": 10.0566,
      "step": 3009
    },
    {
      "epoch": 0.5881203595154357,
      "grad_norm": 6.438875675201416,
      "learning_rate": 7.710494500498662e-06,
      "loss": 8.9169,
      "step": 3010
    },
    {
      "epoch": 0.588315748339195,
      "grad_norm": 8.44289779663086,
      "learning_rate": 7.704334311070215e-06,
      "loss": 9.1258,
      "step": 3011
    },
    {
      "epoch": 0.5885111371629543,
      "grad_norm": 9.694618225097656,
      "learning_rate": 7.698175041127148e-06,
      "loss": 10.1942,
      "step": 3012
    },
    {
      "epoch": 0.5887065259867136,
      "grad_norm": 10.61161994934082,
      "learning_rate": 7.69201669313644e-06,
      "loss": 8.6437,
      "step": 3013
    },
    {
      "epoch": 0.5889019148104728,
      "grad_norm": 8.284984588623047,
      "learning_rate": 7.685859269564702e-06,
      "loss": 9.9219,
      "step": 3014
    },
    {
      "epoch": 0.5890973036342321,
      "grad_norm": 8.922584533691406,
      "learning_rate": 7.679702772878171e-06,
      "loss": 9.5434,
      "step": 3015
    },
    {
      "epoch": 0.5892926924579914,
      "grad_norm": 7.449310302734375,
      "learning_rate": 7.673547205542718e-06,
      "loss": 8.126,
      "step": 3016
    },
    {
      "epoch": 0.5894880812817507,
      "grad_norm": 8.603376388549805,
      "learning_rate": 7.667392570023834e-06,
      "loss": 9.302,
      "step": 3017
    },
    {
      "epoch": 0.5896834701055099,
      "grad_norm": 53.00111389160156,
      "learning_rate": 7.661238868786648e-06,
      "loss": 9.0841,
      "step": 3018
    },
    {
      "epoch": 0.5898788589292693,
      "grad_norm": 8.923293113708496,
      "learning_rate": 7.655086104295904e-06,
      "loss": 8.7745,
      "step": 3019
    },
    {
      "epoch": 0.5900742477530285,
      "grad_norm": 8.259012222290039,
      "learning_rate": 7.648934279015977e-06,
      "loss": 9.5202,
      "step": 3020
    },
    {
      "epoch": 0.5902696365767878,
      "grad_norm": 30.650358200073242,
      "learning_rate": 7.642783395410863e-06,
      "loss": 8.6799,
      "step": 3021
    },
    {
      "epoch": 0.5904650254005471,
      "grad_norm": 8.582069396972656,
      "learning_rate": 7.636633455944184e-06,
      "loss": 9.9083,
      "step": 3022
    },
    {
      "epoch": 0.5906604142243064,
      "grad_norm": 7.850262641906738,
      "learning_rate": 7.630484463079177e-06,
      "loss": 8.974,
      "step": 3023
    },
    {
      "epoch": 0.5908558030480656,
      "grad_norm": 9.134495735168457,
      "learning_rate": 7.6243364192787095e-06,
      "loss": 8.9006,
      "step": 3024
    },
    {
      "epoch": 0.591051191871825,
      "grad_norm": 7.085219383239746,
      "learning_rate": 7.618189327005261e-06,
      "loss": 8.6565,
      "step": 3025
    },
    {
      "epoch": 0.5912465806955842,
      "grad_norm": 7.791258335113525,
      "learning_rate": 7.612043188720933e-06,
      "loss": 9.6032,
      "step": 3026
    },
    {
      "epoch": 0.5914419695193435,
      "grad_norm": 10.506172180175781,
      "learning_rate": 7.605898006887447e-06,
      "loss": 9.6001,
      "step": 3027
    },
    {
      "epoch": 0.5916373583431028,
      "grad_norm": 9.15788459777832,
      "learning_rate": 7.599753783966136e-06,
      "loss": 9.9401,
      "step": 3028
    },
    {
      "epoch": 0.5918327471668621,
      "grad_norm": 10.521397590637207,
      "learning_rate": 7.5936105224179534e-06,
      "loss": 9.1462,
      "step": 3029
    },
    {
      "epoch": 0.5920281359906213,
      "grad_norm": 11.96612548828125,
      "learning_rate": 7.587468224703467e-06,
      "loss": 9.2608,
      "step": 3030
    },
    {
      "epoch": 0.5922235248143806,
      "grad_norm": 8.097817420959473,
      "learning_rate": 7.581326893282858e-06,
      "loss": 8.5244,
      "step": 3031
    },
    {
      "epoch": 0.5924189136381399,
      "grad_norm": 7.517138481140137,
      "learning_rate": 7.5751865306159174e-06,
      "loss": 8.6608,
      "step": 3032
    },
    {
      "epoch": 0.5926143024618992,
      "grad_norm": 8.84758472442627,
      "learning_rate": 7.569047139162054e-06,
      "loss": 9.0976,
      "step": 3033
    },
    {
      "epoch": 0.5928096912856584,
      "grad_norm": 12.096185684204102,
      "learning_rate": 7.5629087213802846e-06,
      "loss": 8.6873,
      "step": 3034
    },
    {
      "epoch": 0.5930050801094178,
      "grad_norm": 9.214247703552246,
      "learning_rate": 7.556771279729236e-06,
      "loss": 9.2499,
      "step": 3035
    },
    {
      "epoch": 0.593200468933177,
      "grad_norm": 7.316731929779053,
      "learning_rate": 7.550634816667142e-06,
      "loss": 9.0062,
      "step": 3036
    },
    {
      "epoch": 0.5933958577569363,
      "grad_norm": 9.038198471069336,
      "learning_rate": 7.544499334651847e-06,
      "loss": 8.8277,
      "step": 3037
    },
    {
      "epoch": 0.5935912465806956,
      "grad_norm": 7.669400691986084,
      "learning_rate": 7.538364836140804e-06,
      "loss": 9.1077,
      "step": 3038
    },
    {
      "epoch": 0.5937866354044549,
      "grad_norm": 11.094863891601562,
      "learning_rate": 7.532231323591068e-06,
      "loss": 7.842,
      "step": 3039
    },
    {
      "epoch": 0.5939820242282141,
      "grad_norm": 10.717040061950684,
      "learning_rate": 7.526098799459303e-06,
      "loss": 10.1978,
      "step": 3040
    },
    {
      "epoch": 0.5941774130519735,
      "grad_norm": 7.92678165435791,
      "learning_rate": 7.519967266201774e-06,
      "loss": 8.9518,
      "step": 3041
    },
    {
      "epoch": 0.5943728018757327,
      "grad_norm": 8.335107803344727,
      "learning_rate": 7.513836726274352e-06,
      "loss": 9.5342,
      "step": 3042
    },
    {
      "epoch": 0.594568190699492,
      "grad_norm": 8.594441413879395,
      "learning_rate": 7.507707182132507e-06,
      "loss": 9.543,
      "step": 3043
    },
    {
      "epoch": 0.5947635795232513,
      "grad_norm": 8.061004638671875,
      "learning_rate": 7.5015786362313115e-06,
      "loss": 9.0036,
      "step": 3044
    },
    {
      "epoch": 0.5949589683470106,
      "grad_norm": 8.662178039550781,
      "learning_rate": 7.495451091025441e-06,
      "loss": 8.7278,
      "step": 3045
    },
    {
      "epoch": 0.5951543571707698,
      "grad_norm": 8.901294708251953,
      "learning_rate": 7.489324548969163e-06,
      "loss": 10.3345,
      "step": 3046
    },
    {
      "epoch": 0.5953497459945292,
      "grad_norm": 10.195141792297363,
      "learning_rate": 7.483199012516353e-06,
      "loss": 8.9794,
      "step": 3047
    },
    {
      "epoch": 0.5955451348182884,
      "grad_norm": 10.048354148864746,
      "learning_rate": 7.477074484120477e-06,
      "loss": 8.6309,
      "step": 3048
    },
    {
      "epoch": 0.5957405236420477,
      "grad_norm": 10.495071411132812,
      "learning_rate": 7.470950966234597e-06,
      "loss": 9.6058,
      "step": 3049
    },
    {
      "epoch": 0.5959359124658069,
      "grad_norm": 7.866878032684326,
      "learning_rate": 7.464828461311372e-06,
      "loss": 9.4768,
      "step": 3050
    },
    {
      "epoch": 0.5961313012895663,
      "grad_norm": 10.282525062561035,
      "learning_rate": 7.458706971803056e-06,
      "loss": 9.3288,
      "step": 3051
    },
    {
      "epoch": 0.5963266901133255,
      "grad_norm": 6.528744220733643,
      "learning_rate": 7.452586500161496e-06,
      "loss": 9.0563,
      "step": 3052
    },
    {
      "epoch": 0.5965220789370848,
      "grad_norm": 8.274612426757812,
      "learning_rate": 7.446467048838131e-06,
      "loss": 8.4749,
      "step": 3053
    },
    {
      "epoch": 0.5967174677608441,
      "grad_norm": 8.588279724121094,
      "learning_rate": 7.44034862028399e-06,
      "loss": 8.847,
      "step": 3054
    },
    {
      "epoch": 0.5969128565846034,
      "grad_norm": 7.771967887878418,
      "learning_rate": 7.434231216949696e-06,
      "loss": 9.1657,
      "step": 3055
    },
    {
      "epoch": 0.5971082454083626,
      "grad_norm": 9.351832389831543,
      "learning_rate": 7.428114841285458e-06,
      "loss": 10.4129,
      "step": 3056
    },
    {
      "epoch": 0.597303634232122,
      "grad_norm": 13.710846900939941,
      "learning_rate": 7.421999495741072e-06,
      "loss": 9.5978,
      "step": 3057
    },
    {
      "epoch": 0.5974990230558812,
      "grad_norm": 7.8330464363098145,
      "learning_rate": 7.415885182765927e-06,
      "loss": 9.15,
      "step": 3058
    },
    {
      "epoch": 0.5976944118796405,
      "grad_norm": 9.916016578674316,
      "learning_rate": 7.409771904808993e-06,
      "loss": 8.9455,
      "step": 3059
    },
    {
      "epoch": 0.5978898007033998,
      "grad_norm": 14.84218692779541,
      "learning_rate": 7.403659664318828e-06,
      "loss": 9.5175,
      "step": 3060
    },
    {
      "epoch": 0.5980851895271591,
      "grad_norm": 7.279349327087402,
      "learning_rate": 7.397548463743578e-06,
      "loss": 9.2482,
      "step": 3061
    },
    {
      "epoch": 0.5982805783509183,
      "grad_norm": 10.927651405334473,
      "learning_rate": 7.391438305530961e-06,
      "loss": 8.7706,
      "step": 3062
    },
    {
      "epoch": 0.5984759671746777,
      "grad_norm": 8.275943756103516,
      "learning_rate": 7.38532919212829e-06,
      "loss": 8.6868,
      "step": 3063
    },
    {
      "epoch": 0.5986713559984369,
      "grad_norm": 7.2512712478637695,
      "learning_rate": 7.379221125982454e-06,
      "loss": 9.0557,
      "step": 3064
    },
    {
      "epoch": 0.5988667448221962,
      "grad_norm": 9.323055267333984,
      "learning_rate": 7.3731141095399225e-06,
      "loss": 9.4869,
      "step": 3065
    },
    {
      "epoch": 0.5990621336459554,
      "grad_norm": 9.154170036315918,
      "learning_rate": 7.367008145246746e-06,
      "loss": 9.1243,
      "step": 3066
    },
    {
      "epoch": 0.5992575224697148,
      "grad_norm": 7.6077775955200195,
      "learning_rate": 7.360903235548552e-06,
      "loss": 8.9403,
      "step": 3067
    },
    {
      "epoch": 0.599452911293474,
      "grad_norm": 13.280308723449707,
      "learning_rate": 7.354799382890546e-06,
      "loss": 10.0786,
      "step": 3068
    },
    {
      "epoch": 0.5996483001172332,
      "grad_norm": 20.785051345825195,
      "learning_rate": 7.348696589717512e-06,
      "loss": 9.6481,
      "step": 3069
    },
    {
      "epoch": 0.5998436889409926,
      "grad_norm": 7.255066394805908,
      "learning_rate": 7.342594858473807e-06,
      "loss": 9.4078,
      "step": 3070
    },
    {
      "epoch": 0.6000390777647518,
      "grad_norm": 8.109085083007812,
      "learning_rate": 7.336494191603364e-06,
      "loss": 8.3056,
      "step": 3071
    },
    {
      "epoch": 0.6002344665885111,
      "grad_norm": 6.681050777435303,
      "learning_rate": 7.33039459154969e-06,
      "loss": 9.0156,
      "step": 3072
    },
    {
      "epoch": 0.6004298554122705,
      "grad_norm": 7.881631851196289,
      "learning_rate": 7.3242960607558626e-06,
      "loss": 8.4381,
      "step": 3073
    },
    {
      "epoch": 0.6006252442360297,
      "grad_norm": 12.575008392333984,
      "learning_rate": 7.318198601664537e-06,
      "loss": 10.4451,
      "step": 3074
    },
    {
      "epoch": 0.6008206330597889,
      "grad_norm": 10.129982948303223,
      "learning_rate": 7.312102216717929e-06,
      "loss": 9.4406,
      "step": 3075
    },
    {
      "epoch": 0.6010160218835483,
      "grad_norm": 6.688687801361084,
      "learning_rate": 7.3060069083578325e-06,
      "loss": 8.0307,
      "step": 3076
    },
    {
      "epoch": 0.6012114107073075,
      "grad_norm": 8.346129417419434,
      "learning_rate": 7.299912679025609e-06,
      "loss": 8.1342,
      "step": 3077
    },
    {
      "epoch": 0.6014067995310668,
      "grad_norm": 8.57900619506836,
      "learning_rate": 7.293819531162185e-06,
      "loss": 9.8277,
      "step": 3078
    },
    {
      "epoch": 0.6016021883548262,
      "grad_norm": 8.378705978393555,
      "learning_rate": 7.287727467208055e-06,
      "loss": 7.9267,
      "step": 3079
    },
    {
      "epoch": 0.6017975771785854,
      "grad_norm": 9.134478569030762,
      "learning_rate": 7.2816364896032835e-06,
      "loss": 9.6305,
      "step": 3080
    },
    {
      "epoch": 0.6019929660023446,
      "grad_norm": 10.45907974243164,
      "learning_rate": 7.2755466007874905e-06,
      "loss": 9.1696,
      "step": 3081
    },
    {
      "epoch": 0.602188354826104,
      "grad_norm": 10.994975090026855,
      "learning_rate": 7.269457803199868e-06,
      "loss": 9.1492,
      "step": 3082
    },
    {
      "epoch": 0.6023837436498632,
      "grad_norm": 10.184895515441895,
      "learning_rate": 7.263370099279173e-06,
      "loss": 9.3663,
      "step": 3083
    },
    {
      "epoch": 0.6025791324736225,
      "grad_norm": 8.31871509552002,
      "learning_rate": 7.257283491463711e-06,
      "loss": 9.104,
      "step": 3084
    },
    {
      "epoch": 0.6027745212973817,
      "grad_norm": 6.585756778717041,
      "learning_rate": 7.251197982191365e-06,
      "loss": 8.4756,
      "step": 3085
    },
    {
      "epoch": 0.6029699101211411,
      "grad_norm": 8.996659278869629,
      "learning_rate": 7.245113573899567e-06,
      "loss": 9.7396,
      "step": 3086
    },
    {
      "epoch": 0.6031652989449003,
      "grad_norm": 9.247672080993652,
      "learning_rate": 7.239030269025311e-06,
      "loss": 10.0921,
      "step": 3087
    },
    {
      "epoch": 0.6033606877686596,
      "grad_norm": 9.721524238586426,
      "learning_rate": 7.232948070005153e-06,
      "loss": 8.7793,
      "step": 3088
    },
    {
      "epoch": 0.6035560765924189,
      "grad_norm": 9.967495918273926,
      "learning_rate": 7.226866979275198e-06,
      "loss": 8.1611,
      "step": 3089
    },
    {
      "epoch": 0.6037514654161782,
      "grad_norm": 8.950152397155762,
      "learning_rate": 7.220786999271114e-06,
      "loss": 9.3363,
      "step": 3090
    },
    {
      "epoch": 0.6039468542399374,
      "grad_norm": 7.841548442840576,
      "learning_rate": 7.214708132428121e-06,
      "loss": 8.4895,
      "step": 3091
    },
    {
      "epoch": 0.6041422430636968,
      "grad_norm": 6.933669567108154,
      "learning_rate": 7.208630381180995e-06,
      "loss": 8.7605,
      "step": 3092
    },
    {
      "epoch": 0.604337631887456,
      "grad_norm": 9.261148452758789,
      "learning_rate": 7.202553747964063e-06,
      "loss": 9.047,
      "step": 3093
    },
    {
      "epoch": 0.6045330207112153,
      "grad_norm": 7.2570672035217285,
      "learning_rate": 7.196478235211204e-06,
      "loss": 8.7401,
      "step": 3094
    },
    {
      "epoch": 0.6047284095349746,
      "grad_norm": 7.566317558288574,
      "learning_rate": 7.190403845355853e-06,
      "loss": 9.5184,
      "step": 3095
    },
    {
      "epoch": 0.6049237983587339,
      "grad_norm": 9.053235054016113,
      "learning_rate": 7.18433058083099e-06,
      "loss": 9.1205,
      "step": 3096
    },
    {
      "epoch": 0.6051191871824931,
      "grad_norm": 8.755714416503906,
      "learning_rate": 7.178258444069144e-06,
      "loss": 10.2415,
      "step": 3097
    },
    {
      "epoch": 0.6053145760062525,
      "grad_norm": 9.840388298034668,
      "learning_rate": 7.172187437502397e-06,
      "loss": 8.8317,
      "step": 3098
    },
    {
      "epoch": 0.6055099648300117,
      "grad_norm": 8.49913501739502,
      "learning_rate": 7.166117563562375e-06,
      "loss": 9.3803,
      "step": 3099
    },
    {
      "epoch": 0.605705353653771,
      "grad_norm": 8.731346130371094,
      "learning_rate": 7.160048824680251e-06,
      "loss": 9.8032,
      "step": 3100
    },
    {
      "epoch": 0.6059007424775303,
      "grad_norm": 7.560633182525635,
      "learning_rate": 7.153981223286743e-06,
      "loss": 9.2163,
      "step": 3101
    },
    {
      "epoch": 0.6060961313012896,
      "grad_norm": 10.798296928405762,
      "learning_rate": 7.1479147618121135e-06,
      "loss": 9.709,
      "step": 3102
    },
    {
      "epoch": 0.6062915201250488,
      "grad_norm": 8.901002883911133,
      "learning_rate": 7.141849442686168e-06,
      "loss": 9.0237,
      "step": 3103
    },
    {
      "epoch": 0.6064869089488081,
      "grad_norm": 9.370978355407715,
      "learning_rate": 7.1357852683382565e-06,
      "loss": 9.0017,
      "step": 3104
    },
    {
      "epoch": 0.6066822977725674,
      "grad_norm": 9.1856050491333,
      "learning_rate": 7.129722241197269e-06,
      "loss": 9.2504,
      "step": 3105
    },
    {
      "epoch": 0.6068776865963267,
      "grad_norm": 8.396125793457031,
      "learning_rate": 7.123660363691636e-06,
      "loss": 9.5331,
      "step": 3106
    },
    {
      "epoch": 0.6070730754200859,
      "grad_norm": 8.387005805969238,
      "learning_rate": 7.117599638249326e-06,
      "loss": 8.3431,
      "step": 3107
    },
    {
      "epoch": 0.6072684642438453,
      "grad_norm": 13.175034523010254,
      "learning_rate": 7.11154006729785e-06,
      "loss": 9.4098,
      "step": 3108
    },
    {
      "epoch": 0.6074638530676045,
      "grad_norm": 9.21987533569336,
      "learning_rate": 7.1054816532642525e-06,
      "loss": 9.9374,
      "step": 3109
    },
    {
      "epoch": 0.6076592418913638,
      "grad_norm": 6.446108341217041,
      "learning_rate": 7.0994243985751175e-06,
      "loss": 7.8664,
      "step": 3110
    },
    {
      "epoch": 0.6078546307151231,
      "grad_norm": 10.859238624572754,
      "learning_rate": 7.093368305656564e-06,
      "loss": 8.7809,
      "step": 3111
    },
    {
      "epoch": 0.6080500195388824,
      "grad_norm": 9.779617309570312,
      "learning_rate": 7.087313376934244e-06,
      "loss": 9.1884,
      "step": 3112
    },
    {
      "epoch": 0.6082454083626416,
      "grad_norm": 9.77717399597168,
      "learning_rate": 7.081259614833345e-06,
      "loss": 9.1779,
      "step": 3113
    },
    {
      "epoch": 0.608440797186401,
      "grad_norm": 9.52149486541748,
      "learning_rate": 7.07520702177859e-06,
      "loss": 9.1229,
      "step": 3114
    },
    {
      "epoch": 0.6086361860101602,
      "grad_norm": 8.325057983398438,
      "learning_rate": 7.069155600194223e-06,
      "loss": 9.2613,
      "step": 3115
    },
    {
      "epoch": 0.6088315748339195,
      "grad_norm": 10.121953010559082,
      "learning_rate": 7.063105352504031e-06,
      "loss": 9.1125,
      "step": 3116
    },
    {
      "epoch": 0.6090269636576788,
      "grad_norm": 10.042227745056152,
      "learning_rate": 7.057056281131325e-06,
      "loss": 9.4723,
      "step": 3117
    },
    {
      "epoch": 0.6092223524814381,
      "grad_norm": 10.98670768737793,
      "learning_rate": 7.0510083884989476e-06,
      "loss": 9.4069,
      "step": 3118
    },
    {
      "epoch": 0.6094177413051973,
      "grad_norm": 10.467682838439941,
      "learning_rate": 7.044961677029265e-06,
      "loss": 9.74,
      "step": 3119
    },
    {
      "epoch": 0.6096131301289566,
      "grad_norm": 7.917074203491211,
      "learning_rate": 7.038916149144175e-06,
      "loss": 8.6953,
      "step": 3120
    },
    {
      "epoch": 0.6098085189527159,
      "grad_norm": 7.649165153503418,
      "learning_rate": 7.032871807265097e-06,
      "loss": 9.1816,
      "step": 3121
    },
    {
      "epoch": 0.6100039077764752,
      "grad_norm": 7.314424991607666,
      "learning_rate": 7.02682865381298e-06,
      "loss": 8.7005,
      "step": 3122
    },
    {
      "epoch": 0.6101992966002344,
      "grad_norm": 8.90090560913086,
      "learning_rate": 7.020786691208292e-06,
      "loss": 9.215,
      "step": 3123
    },
    {
      "epoch": 0.6103946854239938,
      "grad_norm": 7.860108375549316,
      "learning_rate": 7.014745921871029e-06,
      "loss": 9.1524,
      "step": 3124
    },
    {
      "epoch": 0.610590074247753,
      "grad_norm": 7.705721855163574,
      "learning_rate": 7.008706348220706e-06,
      "loss": 9.3513,
      "step": 3125
    },
    {
      "epoch": 0.6107854630715123,
      "grad_norm": 8.295076370239258,
      "learning_rate": 7.00266797267636e-06,
      "loss": 9.2205,
      "step": 3126
    },
    {
      "epoch": 0.6109808518952716,
      "grad_norm": 9.50456714630127,
      "learning_rate": 6.996630797656547e-06,
      "loss": 9.4783,
      "step": 3127
    },
    {
      "epoch": 0.6111762407190309,
      "grad_norm": 7.332135200500488,
      "learning_rate": 6.990594825579347e-06,
      "loss": 8.861,
      "step": 3128
    },
    {
      "epoch": 0.6113716295427901,
      "grad_norm": 7.384237766265869,
      "learning_rate": 6.984560058862345e-06,
      "loss": 8.6675,
      "step": 3129
    },
    {
      "epoch": 0.6115670183665495,
      "grad_norm": 8.145926475524902,
      "learning_rate": 6.97852649992266e-06,
      "loss": 8.709,
      "step": 3130
    },
    {
      "epoch": 0.6117624071903087,
      "grad_norm": 9.77009391784668,
      "learning_rate": 6.972494151176917e-06,
      "loss": 8.2443,
      "step": 3131
    },
    {
      "epoch": 0.611957796014068,
      "grad_norm": 9.653775215148926,
      "learning_rate": 6.966463015041262e-06,
      "loss": 8.9126,
      "step": 3132
    },
    {
      "epoch": 0.6121531848378273,
      "grad_norm": 7.483040809631348,
      "learning_rate": 6.9604330939313466e-06,
      "loss": 8.1147,
      "step": 3133
    },
    {
      "epoch": 0.6123485736615866,
      "grad_norm": 7.499786376953125,
      "learning_rate": 6.954404390262346e-06,
      "loss": 8.5026,
      "step": 3134
    },
    {
      "epoch": 0.6125439624853458,
      "grad_norm": 8.2280912399292,
      "learning_rate": 6.948376906448941e-06,
      "loss": 8.6048,
      "step": 3135
    },
    {
      "epoch": 0.6127393513091052,
      "grad_norm": 9.438916206359863,
      "learning_rate": 6.942350644905329e-06,
      "loss": 9.9455,
      "step": 3136
    },
    {
      "epoch": 0.6129347401328644,
      "grad_norm": 7.607224941253662,
      "learning_rate": 6.936325608045211e-06,
      "loss": 8.7415,
      "step": 3137
    },
    {
      "epoch": 0.6131301289566237,
      "grad_norm": 8.274602890014648,
      "learning_rate": 6.930301798281803e-06,
      "loss": 8.9917,
      "step": 3138
    },
    {
      "epoch": 0.6133255177803829,
      "grad_norm": 18.421751022338867,
      "learning_rate": 6.924279218027828e-06,
      "loss": 9.6311,
      "step": 3139
    },
    {
      "epoch": 0.6135209066041423,
      "grad_norm": 8.55823040008545,
      "learning_rate": 6.918257869695517e-06,
      "loss": 8.6787,
      "step": 3140
    },
    {
      "epoch": 0.6137162954279015,
      "grad_norm": 8.536458969116211,
      "learning_rate": 6.9122377556966045e-06,
      "loss": 9.1508,
      "step": 3141
    },
    {
      "epoch": 0.6139116842516608,
      "grad_norm": 9.035019874572754,
      "learning_rate": 6.906218878442337e-06,
      "loss": 9.7352,
      "step": 3142
    },
    {
      "epoch": 0.6141070730754201,
      "grad_norm": 12.555957794189453,
      "learning_rate": 6.900201240343458e-06,
      "loss": 9.1511,
      "step": 3143
    },
    {
      "epoch": 0.6143024618991794,
      "grad_norm": 7.259243011474609,
      "learning_rate": 6.894184843810221e-06,
      "loss": 9.4217,
      "step": 3144
    },
    {
      "epoch": 0.6144978507229386,
      "grad_norm": 9.768951416015625,
      "learning_rate": 6.88816969125238e-06,
      "loss": 8.5827,
      "step": 3145
    },
    {
      "epoch": 0.614693239546698,
      "grad_norm": 10.937029838562012,
      "learning_rate": 6.882155785079186e-06,
      "loss": 9.6322,
      "step": 3146
    },
    {
      "epoch": 0.6148886283704572,
      "grad_norm": 8.791969299316406,
      "learning_rate": 6.876143127699397e-06,
      "loss": 8.9697,
      "step": 3147
    },
    {
      "epoch": 0.6150840171942165,
      "grad_norm": 26.33353614807129,
      "learning_rate": 6.8701317215212716e-06,
      "loss": 9.2435,
      "step": 3148
    },
    {
      "epoch": 0.6152794060179758,
      "grad_norm": 8.440177917480469,
      "learning_rate": 6.8641215689525635e-06,
      "loss": 8.714,
      "step": 3149
    },
    {
      "epoch": 0.6154747948417351,
      "grad_norm": 8.96982479095459,
      "learning_rate": 6.858112672400523e-06,
      "loss": 9.534,
      "step": 3150
    },
    {
      "epoch": 0.6156701836654943,
      "grad_norm": 10.628748893737793,
      "learning_rate": 6.852105034271903e-06,
      "loss": 8.9206,
      "step": 3151
    },
    {
      "epoch": 0.6158655724892537,
      "grad_norm": 10.883689880371094,
      "learning_rate": 6.846098656972947e-06,
      "loss": 9.4001,
      "step": 3152
    },
    {
      "epoch": 0.6160609613130129,
      "grad_norm": 63.99466323852539,
      "learning_rate": 6.840093542909396e-06,
      "loss": 8.4326,
      "step": 3153
    },
    {
      "epoch": 0.6162563501367722,
      "grad_norm": 8.935728073120117,
      "learning_rate": 6.834089694486485e-06,
      "loss": 8.2268,
      "step": 3154
    },
    {
      "epoch": 0.6164517389605314,
      "grad_norm": 9.16843032836914,
      "learning_rate": 6.8280871141089415e-06,
      "loss": 9.8084,
      "step": 3155
    },
    {
      "epoch": 0.6166471277842908,
      "grad_norm": 33.58384323120117,
      "learning_rate": 6.822085804180985e-06,
      "loss": 9.3495,
      "step": 3156
    },
    {
      "epoch": 0.61684251660805,
      "grad_norm": 34.05607223510742,
      "learning_rate": 6.816085767106328e-06,
      "loss": 8.9961,
      "step": 3157
    },
    {
      "epoch": 0.6170379054318093,
      "grad_norm": 9.352027893066406,
      "learning_rate": 6.810087005288169e-06,
      "loss": 9.0201,
      "step": 3158
    },
    {
      "epoch": 0.6172332942555686,
      "grad_norm": 9.532946586608887,
      "learning_rate": 6.804089521129202e-06,
      "loss": 10.2474,
      "step": 3159
    },
    {
      "epoch": 0.6174286830793279,
      "grad_norm": 11.039137840270996,
      "learning_rate": 6.798093317031601e-06,
      "loss": 9.4527,
      "step": 3160
    },
    {
      "epoch": 0.6176240719030871,
      "grad_norm": 7.1295576095581055,
      "learning_rate": 6.792098395397036e-06,
      "loss": 9.0233,
      "step": 3161
    },
    {
      "epoch": 0.6178194607268465,
      "grad_norm": 6.612754821777344,
      "learning_rate": 6.786104758626655e-06,
      "loss": 8.1259,
      "step": 3162
    },
    {
      "epoch": 0.6180148495506057,
      "grad_norm": 9.287490844726562,
      "learning_rate": 6.780112409121099e-06,
      "loss": 9.7588,
      "step": 3163
    },
    {
      "epoch": 0.618210238374365,
      "grad_norm": 7.553799629211426,
      "learning_rate": 6.774121349280489e-06,
      "loss": 8.5617,
      "step": 3164
    },
    {
      "epoch": 0.6184056271981243,
      "grad_norm": 10.90973949432373,
      "learning_rate": 6.7681315815044316e-06,
      "loss": 9.4657,
      "step": 3165
    },
    {
      "epoch": 0.6186010160218836,
      "grad_norm": 10.601751327514648,
      "learning_rate": 6.762143108192012e-06,
      "loss": 8.1576,
      "step": 3166
    },
    {
      "epoch": 0.6187964048456428,
      "grad_norm": 6.629915237426758,
      "learning_rate": 6.756155931741803e-06,
      "loss": 8.4618,
      "step": 3167
    },
    {
      "epoch": 0.6189917936694022,
      "grad_norm": 10.512444496154785,
      "learning_rate": 6.750170054551852e-06,
      "loss": 8.9375,
      "step": 3168
    },
    {
      "epoch": 0.6191871824931614,
      "grad_norm": 13.67101001739502,
      "learning_rate": 6.744185479019689e-06,
      "loss": 9.5742,
      "step": 3169
    },
    {
      "epoch": 0.6193825713169206,
      "grad_norm": 8.136218070983887,
      "learning_rate": 6.738202207542325e-06,
      "loss": 9.0165,
      "step": 3170
    },
    {
      "epoch": 0.61957796014068,
      "grad_norm": 9.810447692871094,
      "learning_rate": 6.732220242516243e-06,
      "loss": 8.9211,
      "step": 3171
    },
    {
      "epoch": 0.6197733489644393,
      "grad_norm": 9.089319229125977,
      "learning_rate": 6.726239586337408e-06,
      "loss": 8.4533,
      "step": 3172
    },
    {
      "epoch": 0.6199687377881985,
      "grad_norm": 7.8400559425354,
      "learning_rate": 6.7202602414012555e-06,
      "loss": 9.4795,
      "step": 3173
    },
    {
      "epoch": 0.6201641266119577,
      "grad_norm": 11.087610244750977,
      "learning_rate": 6.714282210102701e-06,
      "loss": 8.8104,
      "step": 3174
    },
    {
      "epoch": 0.6203595154357171,
      "grad_norm": 7.8216352462768555,
      "learning_rate": 6.708305494836131e-06,
      "loss": 9.7963,
      "step": 3175
    },
    {
      "epoch": 0.6205549042594763,
      "grad_norm": 7.982954502105713,
      "learning_rate": 6.702330097995406e-06,
      "loss": 8.6174,
      "step": 3176
    },
    {
      "epoch": 0.6207502930832356,
      "grad_norm": 9.812914848327637,
      "learning_rate": 6.696356021973856e-06,
      "loss": 9.1891,
      "step": 3177
    },
    {
      "epoch": 0.620945681906995,
      "grad_norm": 8.950902938842773,
      "learning_rate": 6.690383269164287e-06,
      "loss": 8.8326,
      "step": 3178
    },
    {
      "epoch": 0.6211410707307542,
      "grad_norm": 8.162497520446777,
      "learning_rate": 6.68441184195897e-06,
      "loss": 8.6706,
      "step": 3179
    },
    {
      "epoch": 0.6213364595545134,
      "grad_norm": 11.237481117248535,
      "learning_rate": 6.6784417427496465e-06,
      "loss": 8.7699,
      "step": 3180
    },
    {
      "epoch": 0.6215318483782728,
      "grad_norm": 12.55990982055664,
      "learning_rate": 6.672472973927528e-06,
      "loss": 10.0447,
      "step": 3181
    },
    {
      "epoch": 0.621727237202032,
      "grad_norm": 10.461010932922363,
      "learning_rate": 6.666505537883292e-06,
      "loss": 8.6269,
      "step": 3182
    },
    {
      "epoch": 0.6219226260257913,
      "grad_norm": 7.173439979553223,
      "learning_rate": 6.660539437007081e-06,
      "loss": 8.3415,
      "step": 3183
    },
    {
      "epoch": 0.6221180148495506,
      "grad_norm": 20.17717933654785,
      "learning_rate": 6.654574673688503e-06,
      "loss": 9.4778,
      "step": 3184
    },
    {
      "epoch": 0.6223134036733099,
      "grad_norm": 7.767277717590332,
      "learning_rate": 6.648611250316635e-06,
      "loss": 9.8497,
      "step": 3185
    },
    {
      "epoch": 0.6225087924970691,
      "grad_norm": 16.488759994506836,
      "learning_rate": 6.642649169280005e-06,
      "loss": 9.7972,
      "step": 3186
    },
    {
      "epoch": 0.6227041813208285,
      "grad_norm": 8.96546745300293,
      "learning_rate": 6.636688432966618e-06,
      "loss": 9.7068,
      "step": 3187
    },
    {
      "epoch": 0.6228995701445877,
      "grad_norm": 8.427123069763184,
      "learning_rate": 6.6307290437639326e-06,
      "loss": 9.4458,
      "step": 3188
    },
    {
      "epoch": 0.623094958968347,
      "grad_norm": 7.777038097381592,
      "learning_rate": 6.624771004058869e-06,
      "loss": 8.7883,
      "step": 3189
    },
    {
      "epoch": 0.6232903477921062,
      "grad_norm": 9.135398864746094,
      "learning_rate": 6.618814316237807e-06,
      "loss": 9.5184,
      "step": 3190
    },
    {
      "epoch": 0.6234857366158656,
      "grad_norm": 11.99392032623291,
      "learning_rate": 6.612858982686584e-06,
      "loss": 9.166,
      "step": 3191
    },
    {
      "epoch": 0.6236811254396248,
      "grad_norm": 8.496952056884766,
      "learning_rate": 6.606905005790498e-06,
      "loss": 9.4242,
      "step": 3192
    },
    {
      "epoch": 0.6238765142633841,
      "grad_norm": 8.410484313964844,
      "learning_rate": 6.6009523879343e-06,
      "loss": 9.6058,
      "step": 3193
    },
    {
      "epoch": 0.6240719030871434,
      "grad_norm": 7.824978351593018,
      "learning_rate": 6.5950011315021986e-06,
      "loss": 9.4955,
      "step": 3194
    },
    {
      "epoch": 0.6242672919109027,
      "grad_norm": 6.903197288513184,
      "learning_rate": 6.589051238877858e-06,
      "loss": 8.9792,
      "step": 3195
    },
    {
      "epoch": 0.6244626807346619,
      "grad_norm": 7.065006256103516,
      "learning_rate": 6.5831027124443935e-06,
      "loss": 9.3928,
      "step": 3196
    },
    {
      "epoch": 0.6246580695584213,
      "grad_norm": 7.14594841003418,
      "learning_rate": 6.5771555545843755e-06,
      "loss": 8.5202,
      "step": 3197
    },
    {
      "epoch": 0.6248534583821805,
      "grad_norm": 11.351675033569336,
      "learning_rate": 6.571209767679827e-06,
      "loss": 9.4747,
      "step": 3198
    },
    {
      "epoch": 0.6250488472059398,
      "grad_norm": 8.51473617553711,
      "learning_rate": 6.565265354112215e-06,
      "loss": 8.8936,
      "step": 3199
    },
    {
      "epoch": 0.6252442360296991,
      "grad_norm": 8.77932071685791,
      "learning_rate": 6.559322316262465e-06,
      "loss": 9.1062,
      "step": 3200
    },
    {
      "epoch": 0.6254396248534584,
      "grad_norm": 7.614633560180664,
      "learning_rate": 6.553380656510948e-06,
      "loss": 8.5036,
      "step": 3201
    },
    {
      "epoch": 0.6256350136772176,
      "grad_norm": 9.768440246582031,
      "learning_rate": 6.547440377237484e-06,
      "loss": 9.7145,
      "step": 3202
    },
    {
      "epoch": 0.625830402500977,
      "grad_norm": 10.376811027526855,
      "learning_rate": 6.5415014808213385e-06,
      "loss": 8.7293,
      "step": 3203
    },
    {
      "epoch": 0.6260257913247362,
      "grad_norm": 8.296785354614258,
      "learning_rate": 6.535563969641223e-06,
      "loss": 8.2191,
      "step": 3204
    },
    {
      "epoch": 0.6262211801484955,
      "grad_norm": 10.103215217590332,
      "learning_rate": 6.529627846075297e-06,
      "loss": 9.3273,
      "step": 3205
    },
    {
      "epoch": 0.6264165689722548,
      "grad_norm": 8.04591178894043,
      "learning_rate": 6.52369311250116e-06,
      "loss": 9.2947,
      "step": 3206
    },
    {
      "epoch": 0.6266119577960141,
      "grad_norm": 90.45257568359375,
      "learning_rate": 6.51775977129586e-06,
      "loss": 9.6114,
      "step": 3207
    },
    {
      "epoch": 0.6268073466197733,
      "grad_norm": 7.731884956359863,
      "learning_rate": 6.511827824835881e-06,
      "loss": 9.0109,
      "step": 3208
    },
    {
      "epoch": 0.6270027354435326,
      "grad_norm": 8.872273445129395,
      "learning_rate": 6.5058972754971535e-06,
      "loss": 9.6837,
      "step": 3209
    },
    {
      "epoch": 0.6271981242672919,
      "grad_norm": 13.767102241516113,
      "learning_rate": 6.499968125655047e-06,
      "loss": 9.097,
      "step": 3210
    },
    {
      "epoch": 0.6273935130910512,
      "grad_norm": 6.631365776062012,
      "learning_rate": 6.4940403776843715e-06,
      "loss": 8.7749,
      "step": 3211
    },
    {
      "epoch": 0.6275889019148104,
      "grad_norm": 9.937586784362793,
      "learning_rate": 6.488114033959371e-06,
      "loss": 9.6075,
      "step": 3212
    },
    {
      "epoch": 0.6277842907385698,
      "grad_norm": 7.5149617195129395,
      "learning_rate": 6.48218909685373e-06,
      "loss": 9.0256,
      "step": 3213
    },
    {
      "epoch": 0.627979679562329,
      "grad_norm": 9.131972312927246,
      "learning_rate": 6.476265568740572e-06,
      "loss": 8.9044,
      "step": 3214
    },
    {
      "epoch": 0.6281750683860883,
      "grad_norm": 15.428956031799316,
      "learning_rate": 6.470343451992452e-06,
      "loss": 9.7636,
      "step": 3215
    },
    {
      "epoch": 0.6283704572098476,
      "grad_norm": 8.987717628479004,
      "learning_rate": 6.464422748981364e-06,
      "loss": 8.9946,
      "step": 3216
    },
    {
      "epoch": 0.6285658460336069,
      "grad_norm": 10.1642484664917,
      "learning_rate": 6.458503462078731e-06,
      "loss": 10.0214,
      "step": 3217
    },
    {
      "epoch": 0.6287612348573661,
      "grad_norm": 7.6031413078308105,
      "learning_rate": 6.452585593655412e-06,
      "loss": 9.4287,
      "step": 3218
    },
    {
      "epoch": 0.6289566236811255,
      "grad_norm": 8.994832038879395,
      "learning_rate": 6.446669146081697e-06,
      "loss": 9.3486,
      "step": 3219
    },
    {
      "epoch": 0.6291520125048847,
      "grad_norm": 15.768937110900879,
      "learning_rate": 6.440754121727308e-06,
      "loss": 9.4399,
      "step": 3220
    },
    {
      "epoch": 0.629347401328644,
      "grad_norm": 8.522614479064941,
      "learning_rate": 6.434840522961395e-06,
      "loss": 8.7106,
      "step": 3221
    },
    {
      "epoch": 0.6295427901524033,
      "grad_norm": 9.83200740814209,
      "learning_rate": 6.428928352152538e-06,
      "loss": 9.5416,
      "step": 3222
    },
    {
      "epoch": 0.6297381789761626,
      "grad_norm": 9.205872535705566,
      "learning_rate": 6.423017611668745e-06,
      "loss": 8.5827,
      "step": 3223
    },
    {
      "epoch": 0.6299335677999218,
      "grad_norm": 9.592184066772461,
      "learning_rate": 6.4171083038774525e-06,
      "loss": 10.8158,
      "step": 3224
    },
    {
      "epoch": 0.6301289566236812,
      "grad_norm": 12.21749210357666,
      "learning_rate": 6.411200431145519e-06,
      "loss": 10.0817,
      "step": 3225
    },
    {
      "epoch": 0.6303243454474404,
      "grad_norm": 11.177057266235352,
      "learning_rate": 6.40529399583923e-06,
      "loss": 9.248,
      "step": 3226
    },
    {
      "epoch": 0.6305197342711997,
      "grad_norm": 8.112342834472656,
      "learning_rate": 6.399389000324301e-06,
      "loss": 9.1992,
      "step": 3227
    },
    {
      "epoch": 0.6307151230949589,
      "grad_norm": 14.266334533691406,
      "learning_rate": 6.393485446965861e-06,
      "loss": 8.742,
      "step": 3228
    },
    {
      "epoch": 0.6309105119187183,
      "grad_norm": 8.99083423614502,
      "learning_rate": 6.387583338128471e-06,
      "loss": 9.5529,
      "step": 3229
    },
    {
      "epoch": 0.6311059007424775,
      "grad_norm": 7.230621337890625,
      "learning_rate": 6.381682676176106e-06,
      "loss": 8.9044,
      "step": 3230
    },
    {
      "epoch": 0.6313012895662368,
      "grad_norm": 10.015031814575195,
      "learning_rate": 6.375783463472165e-06,
      "loss": 8.8948,
      "step": 3231
    },
    {
      "epoch": 0.6314966783899961,
      "grad_norm": 6.90847110748291,
      "learning_rate": 6.369885702379465e-06,
      "loss": 9.1416,
      "step": 3232
    },
    {
      "epoch": 0.6316920672137554,
      "grad_norm": 10.970864295959473,
      "learning_rate": 6.363989395260244e-06,
      "loss": 10.2324,
      "step": 3233
    },
    {
      "epoch": 0.6318874560375146,
      "grad_norm": 8.515042304992676,
      "learning_rate": 6.358094544476154e-06,
      "loss": 8.9114,
      "step": 3234
    },
    {
      "epoch": 0.632082844861274,
      "grad_norm": 8.391950607299805,
      "learning_rate": 6.352201152388269e-06,
      "loss": 9.033,
      "step": 3235
    },
    {
      "epoch": 0.6322782336850332,
      "grad_norm": 7.665505886077881,
      "learning_rate": 6.346309221357072e-06,
      "loss": 10.047,
      "step": 3236
    },
    {
      "epoch": 0.6324736225087925,
      "grad_norm": 7.0385212898254395,
      "learning_rate": 6.340418753742468e-06,
      "loss": 9.8185,
      "step": 3237
    },
    {
      "epoch": 0.6326690113325518,
      "grad_norm": 8.163466453552246,
      "learning_rate": 6.334529751903768e-06,
      "loss": 9.4894,
      "step": 3238
    },
    {
      "epoch": 0.6328644001563111,
      "grad_norm": 8.51423454284668,
      "learning_rate": 6.328642218199702e-06,
      "loss": 9.8436,
      "step": 3239
    },
    {
      "epoch": 0.6330597889800703,
      "grad_norm": 7.965399742126465,
      "learning_rate": 6.32275615498841e-06,
      "loss": 9.2198,
      "step": 3240
    },
    {
      "epoch": 0.6332551778038297,
      "grad_norm": 9.111120223999023,
      "learning_rate": 6.316871564627443e-06,
      "loss": 9.3589,
      "step": 3241
    },
    {
      "epoch": 0.6334505666275889,
      "grad_norm": 7.943230152130127,
      "learning_rate": 6.310988449473763e-06,
      "loss": 8.1335,
      "step": 3242
    },
    {
      "epoch": 0.6336459554513482,
      "grad_norm": 8.767099380493164,
      "learning_rate": 6.30510681188374e-06,
      "loss": 8.2788,
      "step": 3243
    },
    {
      "epoch": 0.6338413442751074,
      "grad_norm": 9.133307456970215,
      "learning_rate": 6.299226654213151e-06,
      "loss": 9.5274,
      "step": 3244
    },
    {
      "epoch": 0.6340367330988668,
      "grad_norm": 12.576370239257812,
      "learning_rate": 6.293347978817182e-06,
      "loss": 8.6891,
      "step": 3245
    },
    {
      "epoch": 0.634232121922626,
      "grad_norm": 7.583223342895508,
      "learning_rate": 6.287470788050427e-06,
      "loss": 9.4539,
      "step": 3246
    },
    {
      "epoch": 0.6344275107463853,
      "grad_norm": 9.225921630859375,
      "learning_rate": 6.281595084266883e-06,
      "loss": 9.8242,
      "step": 3247
    },
    {
      "epoch": 0.6346228995701446,
      "grad_norm": 9.983611106872559,
      "learning_rate": 6.2757208698199475e-06,
      "loss": 9.5421,
      "step": 3248
    },
    {
      "epoch": 0.6348182883939039,
      "grad_norm": 13.373017311096191,
      "learning_rate": 6.269848147062431e-06,
      "loss": 9.2219,
      "step": 3249
    },
    {
      "epoch": 0.6350136772176631,
      "grad_norm": 9.064508438110352,
      "learning_rate": 6.263976918346542e-06,
      "loss": 9.4761,
      "step": 3250
    },
    {
      "epoch": 0.6352090660414225,
      "grad_norm": 10.210753440856934,
      "learning_rate": 6.258107186023885e-06,
      "loss": 8.3957,
      "step": 3251
    },
    {
      "epoch": 0.6354044548651817,
      "grad_norm": 7.373657703399658,
      "learning_rate": 6.252238952445473e-06,
      "loss": 8.2487,
      "step": 3252
    },
    {
      "epoch": 0.635599843688941,
      "grad_norm": 7.989492416381836,
      "learning_rate": 6.246372219961714e-06,
      "loss": 8.9867,
      "step": 3253
    },
    {
      "epoch": 0.6357952325127003,
      "grad_norm": 7.448361396789551,
      "learning_rate": 6.240506990922418e-06,
      "loss": 9.0057,
      "step": 3254
    },
    {
      "epoch": 0.6359906213364596,
      "grad_norm": 8.415063858032227,
      "learning_rate": 6.23464326767679e-06,
      "loss": 9.4472,
      "step": 3255
    },
    {
      "epoch": 0.6361860101602188,
      "grad_norm": 10.542451858520508,
      "learning_rate": 6.228781052573436e-06,
      "loss": 9.0757,
      "step": 3256
    },
    {
      "epoch": 0.6363813989839782,
      "grad_norm": 9.972257614135742,
      "learning_rate": 6.22292034796035e-06,
      "loss": 9.8713,
      "step": 3257
    },
    {
      "epoch": 0.6365767878077374,
      "grad_norm": 9.886518478393555,
      "learning_rate": 6.217061156184932e-06,
      "loss": 8.1271,
      "step": 3258
    },
    {
      "epoch": 0.6367721766314967,
      "grad_norm": 9.851444244384766,
      "learning_rate": 6.211203479593968e-06,
      "loss": 9.6882,
      "step": 3259
    },
    {
      "epoch": 0.636967565455256,
      "grad_norm": 9.61854362487793,
      "learning_rate": 6.205347320533637e-06,
      "loss": 9.0446,
      "step": 3260
    },
    {
      "epoch": 0.6371629542790153,
      "grad_norm": 10.928366661071777,
      "learning_rate": 6.199492681349516e-06,
      "loss": 10.1104,
      "step": 3261
    },
    {
      "epoch": 0.6373583431027745,
      "grad_norm": 7.869286060333252,
      "learning_rate": 6.193639564386567e-06,
      "loss": 8.8156,
      "step": 3262
    },
    {
      "epoch": 0.6375537319265338,
      "grad_norm": 10.400713920593262,
      "learning_rate": 6.18778797198915e-06,
      "loss": 8.5672,
      "step": 3263
    },
    {
      "epoch": 0.6377491207502931,
      "grad_norm": 10.945712089538574,
      "learning_rate": 6.181937906501001e-06,
      "loss": 10.0861,
      "step": 3264
    },
    {
      "epoch": 0.6379445095740524,
      "grad_norm": 9.357305526733398,
      "learning_rate": 6.176089370265259e-06,
      "loss": 8.6621,
      "step": 3265
    },
    {
      "epoch": 0.6381398983978116,
      "grad_norm": 11.457706451416016,
      "learning_rate": 6.170242365624446e-06,
      "loss": 9.3661,
      "step": 3266
    },
    {
      "epoch": 0.638335287221571,
      "grad_norm": 13.932377815246582,
      "learning_rate": 6.164396894920463e-06,
      "loss": 10.1762,
      "step": 3267
    },
    {
      "epoch": 0.6385306760453302,
      "grad_norm": 8.705086708068848,
      "learning_rate": 6.158552960494605e-06,
      "loss": 9.3154,
      "step": 3268
    },
    {
      "epoch": 0.6387260648690894,
      "grad_norm": 9.752387046813965,
      "learning_rate": 6.152710564687552e-06,
      "loss": 8.626,
      "step": 3269
    },
    {
      "epoch": 0.6389214536928488,
      "grad_norm": 10.422157287597656,
      "learning_rate": 6.1468697098393585e-06,
      "loss": 9.6922,
      "step": 3270
    },
    {
      "epoch": 0.639116842516608,
      "grad_norm": 8.750487327575684,
      "learning_rate": 6.141030398289474e-06,
      "loss": 8.7594,
      "step": 3271
    },
    {
      "epoch": 0.6393122313403673,
      "grad_norm": 10.304443359375,
      "learning_rate": 6.135192632376721e-06,
      "loss": 9.5961,
      "step": 3272
    },
    {
      "epoch": 0.6395076201641267,
      "grad_norm": 7.825802326202393,
      "learning_rate": 6.129356414439308e-06,
      "loss": 9.8164,
      "step": 3273
    },
    {
      "epoch": 0.6397030089878859,
      "grad_norm": 7.505426406860352,
      "learning_rate": 6.12352174681482e-06,
      "loss": 8.8713,
      "step": 3274
    },
    {
      "epoch": 0.6398983978116451,
      "grad_norm": 8.099054336547852,
      "learning_rate": 6.117688631840224e-06,
      "loss": 9.2513,
      "step": 3275
    },
    {
      "epoch": 0.6400937866354045,
      "grad_norm": 6.938951015472412,
      "learning_rate": 6.111857071851861e-06,
      "loss": 9.7799,
      "step": 3276
    },
    {
      "epoch": 0.6402891754591638,
      "grad_norm": 7.928799629211426,
      "learning_rate": 6.106027069185455e-06,
      "loss": 9.7725,
      "step": 3277
    },
    {
      "epoch": 0.640484564282923,
      "grad_norm": 14.206740379333496,
      "learning_rate": 6.100198626176101e-06,
      "loss": 9.5218,
      "step": 3278
    },
    {
      "epoch": 0.6406799531066822,
      "grad_norm": 8.585370063781738,
      "learning_rate": 6.094371745158273e-06,
      "loss": 9.4407,
      "step": 3279
    },
    {
      "epoch": 0.6408753419304416,
      "grad_norm": 10.159416198730469,
      "learning_rate": 6.088546428465816e-06,
      "loss": 10.2069,
      "step": 3280
    },
    {
      "epoch": 0.6410707307542008,
      "grad_norm": 8.027912139892578,
      "learning_rate": 6.082722678431952e-06,
      "loss": 10.0021,
      "step": 3281
    },
    {
      "epoch": 0.6412661195779601,
      "grad_norm": 7.191110134124756,
      "learning_rate": 6.076900497389273e-06,
      "loss": 9.1605,
      "step": 3282
    },
    {
      "epoch": 0.6414615084017194,
      "grad_norm": 9.171540260314941,
      "learning_rate": 6.071079887669742e-06,
      "loss": 9.4411,
      "step": 3283
    },
    {
      "epoch": 0.6416568972254787,
      "grad_norm": 9.93891716003418,
      "learning_rate": 6.065260851604694e-06,
      "loss": 10.1845,
      "step": 3284
    },
    {
      "epoch": 0.6418522860492379,
      "grad_norm": 7.612746238708496,
      "learning_rate": 6.0594433915248355e-06,
      "loss": 7.7668,
      "step": 3285
    },
    {
      "epoch": 0.6420476748729973,
      "grad_norm": 9.86822509765625,
      "learning_rate": 6.053627509760238e-06,
      "loss": 9.5375,
      "step": 3286
    },
    {
      "epoch": 0.6422430636967565,
      "grad_norm": 9.132811546325684,
      "learning_rate": 6.047813208640343e-06,
      "loss": 9.6339,
      "step": 3287
    },
    {
      "epoch": 0.6424384525205158,
      "grad_norm": 8.128195762634277,
      "learning_rate": 6.04200049049396e-06,
      "loss": 8.6282,
      "step": 3288
    },
    {
      "epoch": 0.6426338413442751,
      "grad_norm": 10.335332870483398,
      "learning_rate": 6.036189357649263e-06,
      "loss": 9.745,
      "step": 3289
    },
    {
      "epoch": 0.6428292301680344,
      "grad_norm": 8.72690486907959,
      "learning_rate": 6.030379812433788e-06,
      "loss": 8.6669,
      "step": 3290
    },
    {
      "epoch": 0.6430246189917936,
      "grad_norm": 7.441038131713867,
      "learning_rate": 6.024571857174443e-06,
      "loss": 8.4453,
      "step": 3291
    },
    {
      "epoch": 0.643220007815553,
      "grad_norm": 7.772334098815918,
      "learning_rate": 6.018765494197492e-06,
      "loss": 9.6616,
      "step": 3292
    },
    {
      "epoch": 0.6434153966393122,
      "grad_norm": 8.764260292053223,
      "learning_rate": 6.012960725828561e-06,
      "loss": 9.2197,
      "step": 3293
    },
    {
      "epoch": 0.6436107854630715,
      "grad_norm": 9.021172523498535,
      "learning_rate": 6.0071575543926445e-06,
      "loss": 9.8582,
      "step": 3294
    },
    {
      "epoch": 0.6438061742868308,
      "grad_norm": 7.492588043212891,
      "learning_rate": 6.001355982214092e-06,
      "loss": 9.0723,
      "step": 3295
    },
    {
      "epoch": 0.6440015631105901,
      "grad_norm": 8.09508228302002,
      "learning_rate": 5.995556011616611e-06,
      "loss": 8.796,
      "step": 3296
    },
    {
      "epoch": 0.6441969519343493,
      "grad_norm": 9.950896263122559,
      "learning_rate": 5.989757644923271e-06,
      "loss": 7.8733,
      "step": 3297
    },
    {
      "epoch": 0.6443923407581086,
      "grad_norm": 8.723230361938477,
      "learning_rate": 5.983960884456496e-06,
      "loss": 9.9915,
      "step": 3298
    },
    {
      "epoch": 0.6445877295818679,
      "grad_norm": 31.13210105895996,
      "learning_rate": 5.978165732538073e-06,
      "loss": 9.8638,
      "step": 3299
    },
    {
      "epoch": 0.6447831184056272,
      "grad_norm": 6.883072376251221,
      "learning_rate": 5.972372191489136e-06,
      "loss": 8.3624,
      "step": 3300
    },
    {
      "epoch": 0.6449785072293864,
      "grad_norm": 7.361454010009766,
      "learning_rate": 5.9665802636301805e-06,
      "loss": 8.5977,
      "step": 3301
    },
    {
      "epoch": 0.6451738960531458,
      "grad_norm": 7.5528459548950195,
      "learning_rate": 5.960789951281052e-06,
      "loss": 9.2596,
      "step": 3302
    },
    {
      "epoch": 0.645369284876905,
      "grad_norm": 9.67246150970459,
      "learning_rate": 5.955001256760951e-06,
      "loss": 9.8974,
      "step": 3303
    },
    {
      "epoch": 0.6455646737006643,
      "grad_norm": 8.841197967529297,
      "learning_rate": 5.9492141823884295e-06,
      "loss": 9.5671,
      "step": 3304
    },
    {
      "epoch": 0.6457600625244236,
      "grad_norm": 8.962235450744629,
      "learning_rate": 5.943428730481391e-06,
      "loss": 7.8073,
      "step": 3305
    },
    {
      "epoch": 0.6459554513481829,
      "grad_norm": 10.199047088623047,
      "learning_rate": 5.937644903357087e-06,
      "loss": 9.4195,
      "step": 3306
    },
    {
      "epoch": 0.6461508401719421,
      "grad_norm": 9.305266380310059,
      "learning_rate": 5.931862703332121e-06,
      "loss": 8.6202,
      "step": 3307
    },
    {
      "epoch": 0.6463462289957015,
      "grad_norm": 8.97860336303711,
      "learning_rate": 5.926082132722445e-06,
      "loss": 8.9567,
      "step": 3308
    },
    {
      "epoch": 0.6465416178194607,
      "grad_norm": 6.764257431030273,
      "learning_rate": 5.920303193843353e-06,
      "loss": 8.338,
      "step": 3309
    },
    {
      "epoch": 0.64673700664322,
      "grad_norm": 7.807435035705566,
      "learning_rate": 5.914525889009493e-06,
      "loss": 8.907,
      "step": 3310
    },
    {
      "epoch": 0.6469323954669793,
      "grad_norm": 10.246101379394531,
      "learning_rate": 5.908750220534853e-06,
      "loss": 8.8581,
      "step": 3311
    },
    {
      "epoch": 0.6471277842907386,
      "grad_norm": 8.756179809570312,
      "learning_rate": 5.902976190732768e-06,
      "loss": 8.8028,
      "step": 3312
    },
    {
      "epoch": 0.6473231731144978,
      "grad_norm": 8.108232498168945,
      "learning_rate": 5.897203801915918e-06,
      "loss": 9.0011,
      "step": 3313
    },
    {
      "epoch": 0.6475185619382572,
      "grad_norm": 7.685898303985596,
      "learning_rate": 5.8914330563963205e-06,
      "loss": 9.0275,
      "step": 3314
    },
    {
      "epoch": 0.6477139507620164,
      "grad_norm": 7.592754364013672,
      "learning_rate": 5.885663956485341e-06,
      "loss": 8.0847,
      "step": 3315
    },
    {
      "epoch": 0.6479093395857757,
      "grad_norm": 8.844161033630371,
      "learning_rate": 5.8798965044936816e-06,
      "loss": 9.006,
      "step": 3316
    },
    {
      "epoch": 0.6481047284095349,
      "grad_norm": 10.648222923278809,
      "learning_rate": 5.8741307027313875e-06,
      "loss": 9.1198,
      "step": 3317
    },
    {
      "epoch": 0.6483001172332943,
      "grad_norm": 9.618973731994629,
      "learning_rate": 5.868366553507841e-06,
      "loss": 9.2452,
      "step": 3318
    },
    {
      "epoch": 0.6484955060570535,
      "grad_norm": 9.56131649017334,
      "learning_rate": 5.86260405913176e-06,
      "loss": 9.0775,
      "step": 3319
    },
    {
      "epoch": 0.6486908948808128,
      "grad_norm": 8.947144508361816,
      "learning_rate": 5.856843221911206e-06,
      "loss": 8.8293,
      "step": 3320
    },
    {
      "epoch": 0.6488862837045721,
      "grad_norm": 7.5485382080078125,
      "learning_rate": 5.851084044153573e-06,
      "loss": 9.0039,
      "step": 3321
    },
    {
      "epoch": 0.6490816725283314,
      "grad_norm": 11.291420936584473,
      "learning_rate": 5.845326528165588e-06,
      "loss": 8.9254,
      "step": 3322
    },
    {
      "epoch": 0.6492770613520906,
      "grad_norm": 8.44905948638916,
      "learning_rate": 5.839570676253317e-06,
      "loss": 9.5899,
      "step": 3323
    },
    {
      "epoch": 0.64947245017585,
      "grad_norm": 8.68692684173584,
      "learning_rate": 5.833816490722153e-06,
      "loss": 9.3606,
      "step": 3324
    },
    {
      "epoch": 0.6496678389996092,
      "grad_norm": 8.023199081420898,
      "learning_rate": 5.828063973876834e-06,
      "loss": 8.6232,
      "step": 3325
    },
    {
      "epoch": 0.6498632278233685,
      "grad_norm": 6.996118068695068,
      "learning_rate": 5.822313128021413e-06,
      "loss": 7.3301,
      "step": 3326
    },
    {
      "epoch": 0.6500586166471278,
      "grad_norm": 6.996118068695068,
      "learning_rate": 5.822313128021413e-06,
      "loss": 9.9519,
      "step": 3327
    },
    {
      "epoch": 0.6502540054708871,
      "grad_norm": 7.3202056884765625,
      "learning_rate": 5.81656395545929e-06,
      "loss": 8.1281,
      "step": 3328
    },
    {
      "epoch": 0.6504493942946463,
      "grad_norm": 9.246292114257812,
      "learning_rate": 5.810816458493184e-06,
      "loss": 8.9741,
      "step": 3329
    },
    {
      "epoch": 0.6506447831184057,
      "grad_norm": 14.636777877807617,
      "learning_rate": 5.805070639425141e-06,
      "loss": 9.053,
      "step": 3330
    },
    {
      "epoch": 0.6508401719421649,
      "grad_norm": 10.433954238891602,
      "learning_rate": 5.7993265005565494e-06,
      "loss": 8.3707,
      "step": 3331
    },
    {
      "epoch": 0.6510355607659242,
      "grad_norm": 9.631535530090332,
      "learning_rate": 5.793584044188105e-06,
      "loss": 8.9179,
      "step": 3332
    },
    {
      "epoch": 0.6512309495896834,
      "grad_norm": 9.813020706176758,
      "learning_rate": 5.7878432726198495e-06,
      "loss": 8.7182,
      "step": 3333
    },
    {
      "epoch": 0.6514263384134428,
      "grad_norm": 11.817461967468262,
      "learning_rate": 5.782104188151132e-06,
      "loss": 9.2989,
      "step": 3334
    },
    {
      "epoch": 0.651621727237202,
      "grad_norm": 9.830137252807617,
      "learning_rate": 5.776366793080641e-06,
      "loss": 9.5555,
      "step": 3335
    },
    {
      "epoch": 0.6518171160609613,
      "grad_norm": 7.121776103973389,
      "learning_rate": 5.770631089706374e-06,
      "loss": 8.3879,
      "step": 3336
    },
    {
      "epoch": 0.6520125048847206,
      "grad_norm": 11.094462394714355,
      "learning_rate": 5.7648970803256664e-06,
      "loss": 10.1041,
      "step": 3337
    },
    {
      "epoch": 0.6522078937084799,
      "grad_norm": 7.530900955200195,
      "learning_rate": 5.759164767235159e-06,
      "loss": 9.1009,
      "step": 3338
    },
    {
      "epoch": 0.6524032825322391,
      "grad_norm": 10.451367378234863,
      "learning_rate": 5.753434152730829e-06,
      "loss": 8.9201,
      "step": 3339
    },
    {
      "epoch": 0.6525986713559985,
      "grad_norm": 9.436930656433105,
      "learning_rate": 5.747705239107958e-06,
      "loss": 8.7656,
      "step": 3340
    },
    {
      "epoch": 0.6527940601797577,
      "grad_norm": 10.836393356323242,
      "learning_rate": 5.741978028661161e-06,
      "loss": 9.5958,
      "step": 3341
    },
    {
      "epoch": 0.652989449003517,
      "grad_norm": 10.820073127746582,
      "learning_rate": 5.736252523684359e-06,
      "loss": 9.6317,
      "step": 3342
    },
    {
      "epoch": 0.6531848378272763,
      "grad_norm": 6.66307258605957,
      "learning_rate": 5.730528726470792e-06,
      "loss": 7.7654,
      "step": 3343
    },
    {
      "epoch": 0.6533802266510356,
      "grad_norm": 9.497946739196777,
      "learning_rate": 5.724806639313025e-06,
      "loss": 8.1978,
      "step": 3344
    },
    {
      "epoch": 0.6535756154747948,
      "grad_norm": 8.500927925109863,
      "learning_rate": 5.719086264502924e-06,
      "loss": 9.5653,
      "step": 3345
    },
    {
      "epoch": 0.6537710042985542,
      "grad_norm": 8.197712898254395,
      "learning_rate": 5.713367604331688e-06,
      "loss": 8.9477,
      "step": 3346
    },
    {
      "epoch": 0.6539663931223134,
      "grad_norm": 9.65852165222168,
      "learning_rate": 5.707650661089804e-06,
      "loss": 9.4163,
      "step": 3347
    },
    {
      "epoch": 0.6541617819460727,
      "grad_norm": 7.677917957305908,
      "learning_rate": 5.701935437067096e-06,
      "loss": 9.2371,
      "step": 3348
    },
    {
      "epoch": 0.654357170769832,
      "grad_norm": 8.639566421508789,
      "learning_rate": 5.696221934552684e-06,
      "loss": 8.1582,
      "step": 3349
    },
    {
      "epoch": 0.6545525595935913,
      "grad_norm": 7.323366165161133,
      "learning_rate": 5.690510155835005e-06,
      "loss": 9.698,
      "step": 3350
    },
    {
      "epoch": 0.6547479484173505,
      "grad_norm": 8.872257232666016,
      "learning_rate": 5.6848001032018e-06,
      "loss": 8.4604,
      "step": 3351
    },
    {
      "epoch": 0.6549433372411098,
      "grad_norm": 8.639331817626953,
      "learning_rate": 5.679091778940129e-06,
      "loss": 9.3642,
      "step": 3352
    },
    {
      "epoch": 0.6551387260648691,
      "grad_norm": 7.850310325622559,
      "learning_rate": 5.6733851853363455e-06,
      "loss": 9.2343,
      "step": 3353
    },
    {
      "epoch": 0.6553341148886284,
      "grad_norm": 8.028477668762207,
      "learning_rate": 5.667680324676125e-06,
      "loss": 9.2818,
      "step": 3354
    },
    {
      "epoch": 0.6555295037123876,
      "grad_norm": 8.717341423034668,
      "learning_rate": 5.661977199244437e-06,
      "loss": 9.2635,
      "step": 3355
    },
    {
      "epoch": 0.655724892536147,
      "grad_norm": 8.757511138916016,
      "learning_rate": 5.656275811325557e-06,
      "loss": 9.2249,
      "step": 3356
    },
    {
      "epoch": 0.6559202813599062,
      "grad_norm": 11.633004188537598,
      "learning_rate": 5.650576163203075e-06,
      "loss": 8.787,
      "step": 3357
    },
    {
      "epoch": 0.6561156701836655,
      "grad_norm": 13.906158447265625,
      "learning_rate": 5.64487825715987e-06,
      "loss": 9.0062,
      "step": 3358
    },
    {
      "epoch": 0.6563110590074248,
      "grad_norm": 8.314506530761719,
      "learning_rate": 5.639182095478138e-06,
      "loss": 8.6612,
      "step": 3359
    },
    {
      "epoch": 0.6565064478311841,
      "grad_norm": 6.556262016296387,
      "learning_rate": 5.633487680439362e-06,
      "loss": 9.3609,
      "step": 3360
    },
    {
      "epoch": 0.6567018366549433,
      "grad_norm": 9.146658897399902,
      "learning_rate": 5.627795014324335e-06,
      "loss": 9.2437,
      "step": 3361
    },
    {
      "epoch": 0.6568972254787027,
      "grad_norm": 8.117045402526855,
      "learning_rate": 5.622104099413146e-06,
      "loss": 9.0189,
      "step": 3362
    },
    {
      "epoch": 0.6570926143024619,
      "grad_norm": 7.30604887008667,
      "learning_rate": 5.616414937985184e-06,
      "loss": 8.4643,
      "step": 3363
    },
    {
      "epoch": 0.6572880031262212,
      "grad_norm": 7.005485534667969,
      "learning_rate": 5.610727532319131e-06,
      "loss": 7.646,
      "step": 3364
    },
    {
      "epoch": 0.6574833919499805,
      "grad_norm": 8.44398307800293,
      "learning_rate": 5.605041884692977e-06,
      "loss": 9.31,
      "step": 3365
    },
    {
      "epoch": 0.6576787807737398,
      "grad_norm": 10.372838973999023,
      "learning_rate": 5.5993579973839915e-06,
      "loss": 9.2662,
      "step": 3366
    },
    {
      "epoch": 0.657874169597499,
      "grad_norm": 7.844415664672852,
      "learning_rate": 5.5936758726687554e-06,
      "loss": 9.3544,
      "step": 3367
    },
    {
      "epoch": 0.6580695584212582,
      "grad_norm": 10.088156700134277,
      "learning_rate": 5.587995512823133e-06,
      "loss": 8.2324,
      "step": 3368
    },
    {
      "epoch": 0.6582649472450176,
      "grad_norm": 10.11949348449707,
      "learning_rate": 5.582316920122281e-06,
      "loss": 9.4767,
      "step": 3369
    },
    {
      "epoch": 0.6584603360687769,
      "grad_norm": 7.221989631652832,
      "learning_rate": 5.57664009684066e-06,
      "loss": 8.1311,
      "step": 3370
    },
    {
      "epoch": 0.6586557248925361,
      "grad_norm": 10.301944732666016,
      "learning_rate": 5.570965045252006e-06,
      "loss": 8.7277,
      "step": 3371
    },
    {
      "epoch": 0.6588511137162955,
      "grad_norm": 7.021796703338623,
      "learning_rate": 5.565291767629359e-06,
      "loss": 9.1348,
      "step": 3372
    },
    {
      "epoch": 0.6590465025400547,
      "grad_norm": 7.4562883377075195,
      "learning_rate": 5.559620266245038e-06,
      "loss": 9.2466,
      "step": 3373
    },
    {
      "epoch": 0.659241891363814,
      "grad_norm": 8.977425575256348,
      "learning_rate": 5.553950543370663e-06,
      "loss": 10.1065,
      "step": 3374
    },
    {
      "epoch": 0.6594372801875733,
      "grad_norm": 10.201164245605469,
      "learning_rate": 5.548282601277124e-06,
      "loss": 9.538,
      "step": 3375
    },
    {
      "epoch": 0.6596326690113326,
      "grad_norm": 7.474252223968506,
      "learning_rate": 5.542616442234618e-06,
      "loss": 8.566,
      "step": 3376
    },
    {
      "epoch": 0.6598280578350918,
      "grad_norm": 13.249435424804688,
      "learning_rate": 5.536952068512609e-06,
      "loss": 9.1197,
      "step": 3377
    },
    {
      "epoch": 0.6600234466588512,
      "grad_norm": 7.932127952575684,
      "learning_rate": 5.53128948237986e-06,
      "loss": 8.6702,
      "step": 3378
    },
    {
      "epoch": 0.6602188354826104,
      "grad_norm": 8.74057674407959,
      "learning_rate": 5.52562868610441e-06,
      "loss": 9.1433,
      "step": 3379
    },
    {
      "epoch": 0.6604142243063696,
      "grad_norm": 9.633611679077148,
      "learning_rate": 5.5199696819535856e-06,
      "loss": 9.1596,
      "step": 3380
    },
    {
      "epoch": 0.660609613130129,
      "grad_norm": 8.033287048339844,
      "learning_rate": 5.514312472193992e-06,
      "loss": 8.9639,
      "step": 3381
    },
    {
      "epoch": 0.6608050019538882,
      "grad_norm": 9.30175495147705,
      "learning_rate": 5.508657059091516e-06,
      "loss": 9.6563,
      "step": 3382
    },
    {
      "epoch": 0.6610003907776475,
      "grad_norm": 9.922159194946289,
      "learning_rate": 5.50300344491133e-06,
      "loss": 9.0399,
      "step": 3383
    },
    {
      "epoch": 0.6611957796014069,
      "grad_norm": 11.326530456542969,
      "learning_rate": 5.497351631917878e-06,
      "loss": 8.8985,
      "step": 3384
    },
    {
      "epoch": 0.6613911684251661,
      "grad_norm": 8.612028121948242,
      "learning_rate": 5.49170162237489e-06,
      "loss": 9.1703,
      "step": 3385
    },
    {
      "epoch": 0.6615865572489253,
      "grad_norm": 8.003447532653809,
      "learning_rate": 5.4860534185453665e-06,
      "loss": 8.3363,
      "step": 3386
    },
    {
      "epoch": 0.6617819460726846,
      "grad_norm": 8.059162139892578,
      "learning_rate": 5.480407022691594e-06,
      "loss": 8.5584,
      "step": 3387
    },
    {
      "epoch": 0.661977334896444,
      "grad_norm": 8.822245597839355,
      "learning_rate": 5.474762437075123e-06,
      "loss": 9.7109,
      "step": 3388
    },
    {
      "epoch": 0.6621727237202032,
      "grad_norm": 8.332511901855469,
      "learning_rate": 5.46911966395679e-06,
      "loss": 8.1816,
      "step": 3389
    },
    {
      "epoch": 0.6623681125439624,
      "grad_norm": 7.900548934936523,
      "learning_rate": 5.463478705596696e-06,
      "loss": 9.266,
      "step": 3390
    },
    {
      "epoch": 0.6625635013677218,
      "grad_norm": 8.51298713684082,
      "learning_rate": 5.457839564254228e-06,
      "loss": 9.6674,
      "step": 3391
    },
    {
      "epoch": 0.662758890191481,
      "grad_norm": 7.403846740722656,
      "learning_rate": 5.452202242188029e-06,
      "loss": 8.6096,
      "step": 3392
    },
    {
      "epoch": 0.6629542790152403,
      "grad_norm": 7.903921127319336,
      "learning_rate": 5.446566741656028e-06,
      "loss": 9.7027,
      "step": 3393
    },
    {
      "epoch": 0.6631496678389996,
      "grad_norm": 10.345064163208008,
      "learning_rate": 5.440933064915414e-06,
      "loss": 8.7708,
      "step": 3394
    },
    {
      "epoch": 0.6633450566627589,
      "grad_norm": 9.622152328491211,
      "learning_rate": 5.435301214222646e-06,
      "loss": 9.8728,
      "step": 3395
    },
    {
      "epoch": 0.6635404454865181,
      "grad_norm": 10.184322357177734,
      "learning_rate": 5.429671191833464e-06,
      "loss": 9.5646,
      "step": 3396
    },
    {
      "epoch": 0.6637358343102775,
      "grad_norm": 7.466825008392334,
      "learning_rate": 5.424043000002857e-06,
      "loss": 8.7748,
      "step": 3397
    },
    {
      "epoch": 0.6639312231340367,
      "grad_norm": 8.833131790161133,
      "learning_rate": 5.4184166409851e-06,
      "loss": 9.9261,
      "step": 3398
    },
    {
      "epoch": 0.664126611957796,
      "grad_norm": 8.251693725585938,
      "learning_rate": 5.412792117033717e-06,
      "loss": 7.8546,
      "step": 3399
    },
    {
      "epoch": 0.6643220007815553,
      "grad_norm": 10.826041221618652,
      "learning_rate": 5.407169430401511e-06,
      "loss": 10.1836,
      "step": 3400
    },
    {
      "epoch": 0.6645173896053146,
      "grad_norm": 11.188115119934082,
      "learning_rate": 5.401548583340535e-06,
      "loss": 10.2475,
      "step": 3401
    },
    {
      "epoch": 0.6647127784290738,
      "grad_norm": 12.538986206054688,
      "learning_rate": 5.395929578102119e-06,
      "loss": 10.5756,
      "step": 3402
    },
    {
      "epoch": 0.6649081672528332,
      "grad_norm": 10.929691314697266,
      "learning_rate": 5.390312416936845e-06,
      "loss": 9.0069,
      "step": 3403
    },
    {
      "epoch": 0.6651035560765924,
      "grad_norm": 7.685513496398926,
      "learning_rate": 5.384697102094565e-06,
      "loss": 8.5899,
      "step": 3404
    },
    {
      "epoch": 0.6652989449003517,
      "grad_norm": 7.4986090660095215,
      "learning_rate": 5.379083635824385e-06,
      "loss": 9.4647,
      "step": 3405
    },
    {
      "epoch": 0.6654943337241109,
      "grad_norm": 9.000301361083984,
      "learning_rate": 5.3734720203746715e-06,
      "loss": 9.9078,
      "step": 3406
    },
    {
      "epoch": 0.6656897225478703,
      "grad_norm": 7.087594032287598,
      "learning_rate": 5.367862257993046e-06,
      "loss": 8.7551,
      "step": 3407
    },
    {
      "epoch": 0.6658851113716295,
      "grad_norm": 9.473038673400879,
      "learning_rate": 5.362254350926403e-06,
      "loss": 10.2642,
      "step": 3408
    },
    {
      "epoch": 0.6660805001953888,
      "grad_norm": 8.565691947937012,
      "learning_rate": 5.356648301420872e-06,
      "loss": 9.0764,
      "step": 3409
    },
    {
      "epoch": 0.6662758890191481,
      "grad_norm": 6.862060070037842,
      "learning_rate": 5.3510441117218615e-06,
      "loss": 8.8252,
      "step": 3410
    },
    {
      "epoch": 0.6664712778429074,
      "grad_norm": 8.263054847717285,
      "learning_rate": 5.3454417840740125e-06,
      "loss": 9.004,
      "step": 3411
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 9.111275672912598,
      "learning_rate": 5.339841320721239e-06,
      "loss": 8.0287,
      "step": 3412
    },
    {
      "epoch": 0.666862055490426,
      "grad_norm": 8.513102531433105,
      "learning_rate": 5.334242723906701e-06,
      "loss": 9.5652,
      "step": 3413
    },
    {
      "epoch": 0.6670574443141852,
      "grad_norm": 8.109259605407715,
      "learning_rate": 5.328645995872801e-06,
      "loss": 9.0984,
      "step": 3414
    },
    {
      "epoch": 0.6672528331379445,
      "grad_norm": 7.018557071685791,
      "learning_rate": 5.323051138861213e-06,
      "loss": 9.2285,
      "step": 3415
    },
    {
      "epoch": 0.6674482219617038,
      "grad_norm": 7.353377819061279,
      "learning_rate": 5.317458155112842e-06,
      "loss": 9.5638,
      "step": 3416
    },
    {
      "epoch": 0.6676436107854631,
      "grad_norm": 9.536938667297363,
      "learning_rate": 5.31186704686786e-06,
      "loss": 10.1526,
      "step": 3417
    },
    {
      "epoch": 0.6678389996092223,
      "grad_norm": 7.111334800720215,
      "learning_rate": 5.30627781636567e-06,
      "loss": 9.0047,
      "step": 3418
    },
    {
      "epoch": 0.6680343884329817,
      "grad_norm": 9.799473762512207,
      "learning_rate": 5.3006904658449445e-06,
      "loss": 8.7924,
      "step": 3419
    },
    {
      "epoch": 0.6682297772567409,
      "grad_norm": 8.26110553741455,
      "learning_rate": 5.295104997543579e-06,
      "loss": 9.8125,
      "step": 3420
    },
    {
      "epoch": 0.6684251660805002,
      "grad_norm": 6.976203441619873,
      "learning_rate": 5.289521413698735e-06,
      "loss": 9.0219,
      "step": 3421
    },
    {
      "epoch": 0.6686205549042594,
      "grad_norm": 9.517335891723633,
      "learning_rate": 5.283939716546805e-06,
      "loss": 8.918,
      "step": 3422
    },
    {
      "epoch": 0.6688159437280188,
      "grad_norm": 12.369084358215332,
      "learning_rate": 5.2783599083234395e-06,
      "loss": 9.2844,
      "step": 3423
    },
    {
      "epoch": 0.669011332551778,
      "grad_norm": 8.553173065185547,
      "learning_rate": 5.2727819912635155e-06,
      "loss": 9.8545,
      "step": 3424
    },
    {
      "epoch": 0.6692067213755373,
      "grad_norm": 77.29253387451172,
      "learning_rate": 5.267205967601172e-06,
      "loss": 9.5447,
      "step": 3425
    },
    {
      "epoch": 0.6694021101992966,
      "grad_norm": 10.61068058013916,
      "learning_rate": 5.261631839569776e-06,
      "loss": 9.4725,
      "step": 3426
    },
    {
      "epoch": 0.6695974990230559,
      "grad_norm": 8.081039428710938,
      "learning_rate": 5.256059609401931e-06,
      "loss": 9.1839,
      "step": 3427
    },
    {
      "epoch": 0.6697928878468151,
      "grad_norm": 8.719537734985352,
      "learning_rate": 5.250489279329501e-06,
      "loss": 9.3304,
      "step": 3428
    },
    {
      "epoch": 0.6699882766705745,
      "grad_norm": 7.648680686950684,
      "learning_rate": 5.244920851583565e-06,
      "loss": 8.7841,
      "step": 3429
    },
    {
      "epoch": 0.6701836654943337,
      "grad_norm": 7.385828971862793,
      "learning_rate": 5.2393543283944595e-06,
      "loss": 8.0707,
      "step": 3430
    },
    {
      "epoch": 0.670379054318093,
      "grad_norm": 8.194306373596191,
      "learning_rate": 5.233789711991744e-06,
      "loss": 10.0857,
      "step": 3431
    },
    {
      "epoch": 0.6705744431418523,
      "grad_norm": 12.389236450195312,
      "learning_rate": 5.228227004604225e-06,
      "loss": 9.7494,
      "step": 3432
    },
    {
      "epoch": 0.6707698319656116,
      "grad_norm": 7.089478492736816,
      "learning_rate": 5.222666208459933e-06,
      "loss": 8.8768,
      "step": 3433
    },
    {
      "epoch": 0.6709652207893708,
      "grad_norm": 9.06132984161377,
      "learning_rate": 5.217107325786148e-06,
      "loss": 9.0011,
      "step": 3434
    },
    {
      "epoch": 0.6711606096131302,
      "grad_norm": 6.347553253173828,
      "learning_rate": 5.211550358809368e-06,
      "loss": 8.3051,
      "step": 3435
    },
    {
      "epoch": 0.6713559984368894,
      "grad_norm": 9.32907485961914,
      "learning_rate": 5.205995309755335e-06,
      "loss": 9.5297,
      "step": 3436
    },
    {
      "epoch": 0.6715513872606487,
      "grad_norm": 9.510924339294434,
      "learning_rate": 5.2004421808490145e-06,
      "loss": 8.7285,
      "step": 3437
    },
    {
      "epoch": 0.671746776084408,
      "grad_norm": 6.453519344329834,
      "learning_rate": 5.194890974314612e-06,
      "loss": 8.2006,
      "step": 3438
    },
    {
      "epoch": 0.6719421649081673,
      "grad_norm": 19.823518753051758,
      "learning_rate": 5.189341692375558e-06,
      "loss": 9.7615,
      "step": 3439
    },
    {
      "epoch": 0.6721375537319265,
      "grad_norm": 9.389676094055176,
      "learning_rate": 5.183794337254504e-06,
      "loss": 9.473,
      "step": 3440
    },
    {
      "epoch": 0.6723329425556858,
      "grad_norm": 7.221741199493408,
      "learning_rate": 5.1782489111733494e-06,
      "loss": 8.0322,
      "step": 3441
    },
    {
      "epoch": 0.6725283313794451,
      "grad_norm": 9.929072380065918,
      "learning_rate": 5.172705416353199e-06,
      "loss": 8.7213,
      "step": 3442
    },
    {
      "epoch": 0.6727237202032044,
      "grad_norm": 13.621519088745117,
      "learning_rate": 5.167163855014403e-06,
      "loss": 8.6147,
      "step": 3443
    },
    {
      "epoch": 0.6729191090269636,
      "grad_norm": 7.856151103973389,
      "learning_rate": 5.161624229376523e-06,
      "loss": 8.6556,
      "step": 3444
    },
    {
      "epoch": 0.673114497850723,
      "grad_norm": 9.297572135925293,
      "learning_rate": 5.156086541658356e-06,
      "loss": 9.1263,
      "step": 3445
    },
    {
      "epoch": 0.6733098866744822,
      "grad_norm": 9.4236421585083,
      "learning_rate": 5.150550794077911e-06,
      "loss": 9.3593,
      "step": 3446
    },
    {
      "epoch": 0.6735052754982415,
      "grad_norm": 8.085559844970703,
      "learning_rate": 5.145016988852436e-06,
      "loss": 8.7282,
      "step": 3447
    },
    {
      "epoch": 0.6737006643220008,
      "grad_norm": 7.528972148895264,
      "learning_rate": 5.139485128198382e-06,
      "loss": 8.7748,
      "step": 3448
    },
    {
      "epoch": 0.6738960531457601,
      "grad_norm": 13.162374496459961,
      "learning_rate": 5.133955214331439e-06,
      "loss": 8.553,
      "step": 3449
    },
    {
      "epoch": 0.6740914419695193,
      "grad_norm": 10.401016235351562,
      "learning_rate": 5.128427249466503e-06,
      "loss": 9.7187,
      "step": 3450
    },
    {
      "epoch": 0.6742868307932787,
      "grad_norm": 7.216558933258057,
      "learning_rate": 5.122901235817701e-06,
      "loss": 9.4551,
      "step": 3451
    },
    {
      "epoch": 0.6744822196170379,
      "grad_norm": 9.662524223327637,
      "learning_rate": 5.117377175598372e-06,
      "loss": 8.5878,
      "step": 3452
    },
    {
      "epoch": 0.6746776084407972,
      "grad_norm": 7.320957660675049,
      "learning_rate": 5.111855071021068e-06,
      "loss": 8.5201,
      "step": 3453
    },
    {
      "epoch": 0.6748729972645565,
      "grad_norm": 10.163082122802734,
      "learning_rate": 5.10633492429757e-06,
      "loss": 8.9651,
      "step": 3454
    },
    {
      "epoch": 0.6750683860883158,
      "grad_norm": 8.646751403808594,
      "learning_rate": 5.100816737638863e-06,
      "loss": 9.3415,
      "step": 3455
    },
    {
      "epoch": 0.675263774912075,
      "grad_norm": 7.870480537414551,
      "learning_rate": 5.095300513255157e-06,
      "loss": 9.3154,
      "step": 3456
    },
    {
      "epoch": 0.6754591637358343,
      "grad_norm": 9.3548583984375,
      "learning_rate": 5.0897862533558665e-06,
      "loss": 8.4811,
      "step": 3457
    },
    {
      "epoch": 0.6756545525595936,
      "grad_norm": 8.656578063964844,
      "learning_rate": 5.084273960149629e-06,
      "loss": 8.9353,
      "step": 3458
    },
    {
      "epoch": 0.6758499413833529,
      "grad_norm": 9.321224212646484,
      "learning_rate": 5.0787636358442825e-06,
      "loss": 8.4416,
      "step": 3459
    },
    {
      "epoch": 0.6760453302071121,
      "grad_norm": 9.321224212646484,
      "learning_rate": 5.0787636358442825e-06,
      "loss": 10.2964,
      "step": 3460
    },
    {
      "epoch": 0.6762407190308715,
      "grad_norm": 8.607054710388184,
      "learning_rate": 5.07325528264689e-06,
      "loss": 8.4016,
      "step": 3461
    },
    {
      "epoch": 0.6764361078546307,
      "grad_norm": 7.984342098236084,
      "learning_rate": 5.0677489027637115e-06,
      "loss": 8.8981,
      "step": 3462
    },
    {
      "epoch": 0.67663149667839,
      "grad_norm": 9.526960372924805,
      "learning_rate": 5.062244498400228e-06,
      "loss": 8.2054,
      "step": 3463
    },
    {
      "epoch": 0.6768268855021493,
      "grad_norm": 12.335521697998047,
      "learning_rate": 5.05674207176112e-06,
      "loss": 9.5845,
      "step": 3464
    },
    {
      "epoch": 0.6770222743259086,
      "grad_norm": 10.526650428771973,
      "learning_rate": 5.051241625050287e-06,
      "loss": 9.766,
      "step": 3465
    },
    {
      "epoch": 0.6772176631496678,
      "grad_norm": 9.057307243347168,
      "learning_rate": 5.045743160470824e-06,
      "loss": 8.5928,
      "step": 3466
    },
    {
      "epoch": 0.6774130519734272,
      "grad_norm": 11.242501258850098,
      "learning_rate": 5.040246680225034e-06,
      "loss": 10.1617,
      "step": 3467
    },
    {
      "epoch": 0.6776084407971864,
      "grad_norm": 8.937871932983398,
      "learning_rate": 5.034752186514433e-06,
      "loss": 8.895,
      "step": 3468
    },
    {
      "epoch": 0.6778038296209457,
      "grad_norm": 8.857319831848145,
      "learning_rate": 5.029259681539732e-06,
      "loss": 8.7988,
      "step": 3469
    },
    {
      "epoch": 0.677999218444705,
      "grad_norm": 8.807754516601562,
      "learning_rate": 5.023769167500856e-06,
      "loss": 8.2176,
      "step": 3470
    },
    {
      "epoch": 0.6781946072684643,
      "grad_norm": 9.361241340637207,
      "learning_rate": 5.018280646596918e-06,
      "loss": 9.7489,
      "step": 3471
    },
    {
      "epoch": 0.6783899960922235,
      "grad_norm": 9.284899711608887,
      "learning_rate": 5.01279412102625e-06,
      "loss": 8.8437,
      "step": 3472
    },
    {
      "epoch": 0.6785853849159829,
      "grad_norm": 11.50267505645752,
      "learning_rate": 5.007309592986367e-06,
      "loss": 9.6544,
      "step": 3473
    },
    {
      "epoch": 0.6787807737397421,
      "grad_norm": 8.50446605682373,
      "learning_rate": 5.001827064674001e-06,
      "loss": 9.1788,
      "step": 3474
    },
    {
      "epoch": 0.6789761625635014,
      "grad_norm": 7.195436477661133,
      "learning_rate": 4.996346538285066e-06,
      "loss": 9.0272,
      "step": 3475
    },
    {
      "epoch": 0.6791715513872606,
      "grad_norm": 7.552397727966309,
      "learning_rate": 4.9908680160146934e-06,
      "loss": 8.7314,
      "step": 3476
    },
    {
      "epoch": 0.67936694021102,
      "grad_norm": 8.463982582092285,
      "learning_rate": 4.9853915000571915e-06,
      "loss": 9.6868,
      "step": 3477
    },
    {
      "epoch": 0.6795623290347792,
      "grad_norm": 14.675905227661133,
      "learning_rate": 4.979916992606083e-06,
      "loss": 9.1982,
      "step": 3478
    },
    {
      "epoch": 0.6797577178585384,
      "grad_norm": 7.972330570220947,
      "learning_rate": 4.974444495854075e-06,
      "loss": 8.4736,
      "step": 3479
    },
    {
      "epoch": 0.6799531066822978,
      "grad_norm": 9.317290306091309,
      "learning_rate": 4.968974011993067e-06,
      "loss": 9.9627,
      "step": 3480
    },
    {
      "epoch": 0.680148495506057,
      "grad_norm": 8.15661907196045,
      "learning_rate": 4.9635055432141675e-06,
      "loss": 8.6467,
      "step": 3481
    },
    {
      "epoch": 0.6803438843298163,
      "grad_norm": 7.992409706115723,
      "learning_rate": 4.958039091707659e-06,
      "loss": 10.0247,
      "step": 3482
    },
    {
      "epoch": 0.6805392731535757,
      "grad_norm": 7.441638946533203,
      "learning_rate": 4.952574659663032e-06,
      "loss": 8.4088,
      "step": 3483
    },
    {
      "epoch": 0.6807346619773349,
      "grad_norm": 7.223865509033203,
      "learning_rate": 4.947112249268955e-06,
      "loss": 8.8867,
      "step": 3484
    },
    {
      "epoch": 0.6809300508010941,
      "grad_norm": 9.672779083251953,
      "learning_rate": 4.9416518627133006e-06,
      "loss": 8.8267,
      "step": 3485
    },
    {
      "epoch": 0.6811254396248535,
      "grad_norm": 8.218040466308594,
      "learning_rate": 4.936193502183114e-06,
      "loss": 9.6503,
      "step": 3486
    },
    {
      "epoch": 0.6813208284486127,
      "grad_norm": 9.12499713897705,
      "learning_rate": 4.930737169864647e-06,
      "loss": 9.4916,
      "step": 3487
    },
    {
      "epoch": 0.681516217272372,
      "grad_norm": 9.555933952331543,
      "learning_rate": 4.925282867943322e-06,
      "loss": 8.9187,
      "step": 3488
    },
    {
      "epoch": 0.6817116060961314,
      "grad_norm": 7.154976844787598,
      "learning_rate": 4.9198305986037635e-06,
      "loss": 8.8264,
      "step": 3489
    },
    {
      "epoch": 0.6819069949198906,
      "grad_norm": 6.26767635345459,
      "learning_rate": 4.9143803640297675e-06,
      "loss": 8.361,
      "step": 3490
    },
    {
      "epoch": 0.6821023837436498,
      "grad_norm": 9.955586433410645,
      "learning_rate": 4.90893216640433e-06,
      "loss": 9.5551,
      "step": 3491
    },
    {
      "epoch": 0.6822977725674091,
      "grad_norm": 7.493185997009277,
      "learning_rate": 4.903486007909619e-06,
      "loss": 8.9033,
      "step": 3492
    },
    {
      "epoch": 0.6824931613911684,
      "grad_norm": 12.523082733154297,
      "learning_rate": 4.898041890726987e-06,
      "loss": 9.4942,
      "step": 3493
    },
    {
      "epoch": 0.6826885502149277,
      "grad_norm": 8.432913780212402,
      "learning_rate": 4.892599817036978e-06,
      "loss": 8.9897,
      "step": 3494
    },
    {
      "epoch": 0.6828839390386869,
      "grad_norm": 11.34340763092041,
      "learning_rate": 4.887159789019306e-06,
      "loss": 8.7441,
      "step": 3495
    },
    {
      "epoch": 0.6830793278624463,
      "grad_norm": 9.538162231445312,
      "learning_rate": 4.881721808852877e-06,
      "loss": 9.3453,
      "step": 3496
    },
    {
      "epoch": 0.6832747166862055,
      "grad_norm": 6.3256120681762695,
      "learning_rate": 4.876285878715764e-06,
      "loss": 9.0006,
      "step": 3497
    },
    {
      "epoch": 0.6834701055099648,
      "grad_norm": 10.030577659606934,
      "learning_rate": 4.870852000785233e-06,
      "loss": 10.0886,
      "step": 3498
    },
    {
      "epoch": 0.6836654943337241,
      "grad_norm": 7.294377326965332,
      "learning_rate": 4.865420177237714e-06,
      "loss": 9.1273,
      "step": 3499
    },
    {
      "epoch": 0.6838608831574834,
      "grad_norm": 8.801251411437988,
      "learning_rate": 4.859990410248828e-06,
      "loss": 9.648,
      "step": 3500
    },
    {
      "epoch": 0.6840562719812426,
      "grad_norm": 8.2620849609375,
      "learning_rate": 4.854562701993358e-06,
      "loss": 9.1979,
      "step": 3501
    },
    {
      "epoch": 0.684251660805002,
      "grad_norm": 10.138997077941895,
      "learning_rate": 4.849137054645276e-06,
      "loss": 9.3575,
      "step": 3502
    },
    {
      "epoch": 0.6844470496287612,
      "grad_norm": 8.214333534240723,
      "learning_rate": 4.843713470377716e-06,
      "loss": 8.9957,
      "step": 3503
    },
    {
      "epoch": 0.6846424384525205,
      "grad_norm": 11.297102928161621,
      "learning_rate": 4.838291951363e-06,
      "loss": 8.6428,
      "step": 3504
    },
    {
      "epoch": 0.6848378272762798,
      "grad_norm": 8.354808807373047,
      "learning_rate": 4.8328724997726095e-06,
      "loss": 8.0891,
      "step": 3505
    },
    {
      "epoch": 0.6850332161000391,
      "grad_norm": 10.444690704345703,
      "learning_rate": 4.827455117777199e-06,
      "loss": 9.1011,
      "step": 3506
    },
    {
      "epoch": 0.6852286049237983,
      "grad_norm": 8.39549732208252,
      "learning_rate": 4.822039807546607e-06,
      "loss": 9.0153,
      "step": 3507
    },
    {
      "epoch": 0.6854239937475577,
      "grad_norm": 13.161046981811523,
      "learning_rate": 4.816626571249825e-06,
      "loss": 10.1814,
      "step": 3508
    },
    {
      "epoch": 0.6856193825713169,
      "grad_norm": 9.3190279006958,
      "learning_rate": 4.811215411055032e-06,
      "loss": 9.6779,
      "step": 3509
    },
    {
      "epoch": 0.6858147713950762,
      "grad_norm": 7.86978006362915,
      "learning_rate": 4.805806329129554e-06,
      "loss": 9.9839,
      "step": 3510
    },
    {
      "epoch": 0.6860101602188354,
      "grad_norm": 11.596525192260742,
      "learning_rate": 4.800399327639907e-06,
      "loss": 9.7852,
      "step": 3511
    },
    {
      "epoch": 0.6862055490425948,
      "grad_norm": 7.724665641784668,
      "learning_rate": 4.794994408751753e-06,
      "loss": 9.2349,
      "step": 3512
    },
    {
      "epoch": 0.686400937866354,
      "grad_norm": 8.651827812194824,
      "learning_rate": 4.78959157462994e-06,
      "loss": 9.4594,
      "step": 3513
    },
    {
      "epoch": 0.6865963266901133,
      "grad_norm": 7.398945331573486,
      "learning_rate": 4.784190827438462e-06,
      "loss": 9.0394,
      "step": 3514
    },
    {
      "epoch": 0.6867917155138726,
      "grad_norm": 8.461224555969238,
      "learning_rate": 4.7787921693404934e-06,
      "loss": 9.575,
      "step": 3515
    },
    {
      "epoch": 0.6869871043376319,
      "grad_norm": 9.653675079345703,
      "learning_rate": 4.773395602498358e-06,
      "loss": 8.9556,
      "step": 3516
    },
    {
      "epoch": 0.6871824931613911,
      "grad_norm": 14.610803604125977,
      "learning_rate": 4.768001129073553e-06,
      "loss": 8.9779,
      "step": 3517
    },
    {
      "epoch": 0.6873778819851505,
      "grad_norm": 7.912567138671875,
      "learning_rate": 4.762608751226731e-06,
      "loss": 8.9374,
      "step": 3518
    },
    {
      "epoch": 0.6875732708089097,
      "grad_norm": 7.511148452758789,
      "learning_rate": 4.757218471117704e-06,
      "loss": 9.0939,
      "step": 3519
    },
    {
      "epoch": 0.687768659632669,
      "grad_norm": 7.186136722564697,
      "learning_rate": 4.7518302909054515e-06,
      "loss": 8.2911,
      "step": 3520
    },
    {
      "epoch": 0.6879640484564283,
      "grad_norm": 8.460360527038574,
      "learning_rate": 4.7464442127481e-06,
      "loss": 8.9182,
      "step": 3521
    },
    {
      "epoch": 0.6881594372801876,
      "grad_norm": 8.387063026428223,
      "learning_rate": 4.74106023880295e-06,
      "loss": 9.7751,
      "step": 3522
    },
    {
      "epoch": 0.6883548261039468,
      "grad_norm": 8.238409042358398,
      "learning_rate": 4.7356783712264405e-06,
      "loss": 9.041,
      "step": 3523
    },
    {
      "epoch": 0.6885502149277062,
      "grad_norm": 6.9534592628479,
      "learning_rate": 4.730298612174187e-06,
      "loss": 8.1406,
      "step": 3524
    },
    {
      "epoch": 0.6887456037514654,
      "grad_norm": 11.205235481262207,
      "learning_rate": 4.724920963800939e-06,
      "loss": 9.901,
      "step": 3525
    },
    {
      "epoch": 0.6889409925752247,
      "grad_norm": 7.504866600036621,
      "learning_rate": 4.719545428260623e-06,
      "loss": 8.7138,
      "step": 3526
    },
    {
      "epoch": 0.689136381398984,
      "grad_norm": 8.635149002075195,
      "learning_rate": 4.714172007706298e-06,
      "loss": 9.1709,
      "step": 3527
    },
    {
      "epoch": 0.6893317702227433,
      "grad_norm": 6.568149566650391,
      "learning_rate": 4.708800704290193e-06,
      "loss": 8.9039,
      "step": 3528
    },
    {
      "epoch": 0.6895271590465025,
      "grad_norm": 6.23305082321167,
      "learning_rate": 4.703431520163675e-06,
      "loss": 8.5542,
      "step": 3529
    },
    {
      "epoch": 0.6897225478702618,
      "grad_norm": 207.06385803222656,
      "learning_rate": 4.698064457477276e-06,
      "loss": 10.1821,
      "step": 3530
    },
    {
      "epoch": 0.6899179366940211,
      "grad_norm": 6.3483076095581055,
      "learning_rate": 4.692699518380664e-06,
      "loss": 7.4388,
      "step": 3531
    },
    {
      "epoch": 0.6901133255177804,
      "grad_norm": 7.407440662384033,
      "learning_rate": 4.687336705022672e-06,
      "loss": 7.9367,
      "step": 3532
    },
    {
      "epoch": 0.6903087143415396,
      "grad_norm": 8.763710975646973,
      "learning_rate": 4.681976019551268e-06,
      "loss": 8.208,
      "step": 3533
    },
    {
      "epoch": 0.690504103165299,
      "grad_norm": 9.159621238708496,
      "learning_rate": 4.676617464113571e-06,
      "loss": 9.6811,
      "step": 3534
    },
    {
      "epoch": 0.6906994919890582,
      "grad_norm": 7.8151679039001465,
      "learning_rate": 4.6712610408558546e-06,
      "loss": 9.3893,
      "step": 3535
    },
    {
      "epoch": 0.6908948808128175,
      "grad_norm": 8.498544692993164,
      "learning_rate": 4.665906751923526e-06,
      "loss": 8.9263,
      "step": 3536
    },
    {
      "epoch": 0.6910902696365768,
      "grad_norm": 7.852433681488037,
      "learning_rate": 4.660554599461151e-06,
      "loss": 9.2672,
      "step": 3537
    },
    {
      "epoch": 0.6912856584603361,
      "grad_norm": 7.7206220626831055,
      "learning_rate": 4.655204585612426e-06,
      "loss": 8.5579,
      "step": 3538
    },
    {
      "epoch": 0.6914810472840953,
      "grad_norm": 8.858148574829102,
      "learning_rate": 4.649856712520205e-06,
      "loss": 10.4965,
      "step": 3539
    },
    {
      "epoch": 0.6916764361078547,
      "grad_norm": 9.706315994262695,
      "learning_rate": 4.64451098232647e-06,
      "loss": 9.7618,
      "step": 3540
    },
    {
      "epoch": 0.6918718249316139,
      "grad_norm": 9.114653587341309,
      "learning_rate": 4.639167397172358e-06,
      "loss": 8.7886,
      "step": 3541
    },
    {
      "epoch": 0.6920672137553732,
      "grad_norm": 7.406974792480469,
      "learning_rate": 4.633825959198135e-06,
      "loss": 9.5343,
      "step": 3542
    },
    {
      "epoch": 0.6922626025791325,
      "grad_norm": 8.273829460144043,
      "learning_rate": 4.628486670543218e-06,
      "loss": 9.231,
      "step": 3543
    },
    {
      "epoch": 0.6924579914028918,
      "grad_norm": 8.927740097045898,
      "learning_rate": 4.623149533346158e-06,
      "loss": 9.1105,
      "step": 3544
    },
    {
      "epoch": 0.692653380226651,
      "grad_norm": 8.607205390930176,
      "learning_rate": 4.6178145497446404e-06,
      "loss": 8.9878,
      "step": 3545
    },
    {
      "epoch": 0.6928487690504103,
      "grad_norm": 9.740939140319824,
      "learning_rate": 4.61248172187549e-06,
      "loss": 9.4668,
      "step": 3546
    },
    {
      "epoch": 0.6930441578741696,
      "grad_norm": 10.739304542541504,
      "learning_rate": 4.607151051874677e-06,
      "loss": 9.0491,
      "step": 3547
    },
    {
      "epoch": 0.6932395466979289,
      "grad_norm": 7.079265117645264,
      "learning_rate": 4.601822541877291e-06,
      "loss": 8.5638,
      "step": 3548
    },
    {
      "epoch": 0.6934349355216881,
      "grad_norm": 10.578570365905762,
      "learning_rate": 4.596496194017575e-06,
      "loss": 9.3137,
      "step": 3549
    },
    {
      "epoch": 0.6936303243454475,
      "grad_norm": 15.079474449157715,
      "learning_rate": 4.591172010428895e-06,
      "loss": 8.961,
      "step": 3550
    },
    {
      "epoch": 0.6938257131692067,
      "grad_norm": 9.929420471191406,
      "learning_rate": 4.585849993243744e-06,
      "loss": 9.1616,
      "step": 3551
    },
    {
      "epoch": 0.694021101992966,
      "grad_norm": 6.517254829406738,
      "learning_rate": 4.580530144593765e-06,
      "loss": 6.7743,
      "step": 3552
    },
    {
      "epoch": 0.6942164908167253,
      "grad_norm": 10.221206665039062,
      "learning_rate": 4.575212466609713e-06,
      "loss": 10.227,
      "step": 3553
    },
    {
      "epoch": 0.6944118796404846,
      "grad_norm": 8.188820838928223,
      "learning_rate": 4.5698969614214924e-06,
      "loss": 9.3501,
      "step": 3554
    },
    {
      "epoch": 0.6946072684642438,
      "grad_norm": 10.827705383300781,
      "learning_rate": 4.5645836311581215e-06,
      "loss": 10.4984,
      "step": 3555
    },
    {
      "epoch": 0.6948026572880032,
      "grad_norm": 8.928410530090332,
      "learning_rate": 4.559272477947759e-06,
      "loss": 8.8778,
      "step": 3556
    },
    {
      "epoch": 0.6949980461117624,
      "grad_norm": 8.749245643615723,
      "learning_rate": 4.553963503917681e-06,
      "loss": 9.4237,
      "step": 3557
    },
    {
      "epoch": 0.6951934349355217,
      "grad_norm": 10.222722053527832,
      "learning_rate": 4.548656711194302e-06,
      "loss": 9.4023,
      "step": 3558
    },
    {
      "epoch": 0.695388823759281,
      "grad_norm": 7.6055450439453125,
      "learning_rate": 4.543352101903153e-06,
      "loss": 9.0012,
      "step": 3559
    },
    {
      "epoch": 0.6955842125830403,
      "grad_norm": 11.432832717895508,
      "learning_rate": 4.538049678168899e-06,
      "loss": 9.6928,
      "step": 3560
    },
    {
      "epoch": 0.6957796014067995,
      "grad_norm": 6.709242343902588,
      "learning_rate": 4.5327494421153195e-06,
      "loss": 9.0779,
      "step": 3561
    },
    {
      "epoch": 0.6959749902305589,
      "grad_norm": 6.713991641998291,
      "learning_rate": 4.527451395865331e-06,
      "loss": 7.6969,
      "step": 3562
    },
    {
      "epoch": 0.6961703790543181,
      "grad_norm": 8.824820518493652,
      "learning_rate": 4.522155541540961e-06,
      "loss": 9.0058,
      "step": 3563
    },
    {
      "epoch": 0.6963657678780774,
      "grad_norm": 8.883805274963379,
      "learning_rate": 4.51686188126336e-06,
      "loss": 9.5905,
      "step": 3564
    },
    {
      "epoch": 0.6965611567018366,
      "grad_norm": 7.526402950286865,
      "learning_rate": 4.5115704171528105e-06,
      "loss": 8.956,
      "step": 3565
    },
    {
      "epoch": 0.696756545525596,
      "grad_norm": 6.879123687744141,
      "learning_rate": 4.506281151328703e-06,
      "loss": 7.6253,
      "step": 3566
    },
    {
      "epoch": 0.6969519343493552,
      "grad_norm": 8.952786445617676,
      "learning_rate": 4.500994085909557e-06,
      "loss": 9.9273,
      "step": 3567
    },
    {
      "epoch": 0.6971473231731145,
      "grad_norm": 7.670935153961182,
      "learning_rate": 4.495709223013e-06,
      "loss": 8.6843,
      "step": 3568
    },
    {
      "epoch": 0.6973427119968738,
      "grad_norm": 8.049132347106934,
      "learning_rate": 4.490426564755792e-06,
      "loss": 8.277,
      "step": 3569
    },
    {
      "epoch": 0.6975381008206331,
      "grad_norm": 7.849486351013184,
      "learning_rate": 4.485146113253794e-06,
      "loss": 8.815,
      "step": 3570
    },
    {
      "epoch": 0.6977334896443923,
      "grad_norm": 7.602990627288818,
      "learning_rate": 4.479867870621996e-06,
      "loss": 9.5663,
      "step": 3571
    },
    {
      "epoch": 0.6979288784681517,
      "grad_norm": 7.465991020202637,
      "learning_rate": 4.474591838974495e-06,
      "loss": 9.0858,
      "step": 3572
    },
    {
      "epoch": 0.6981242672919109,
      "grad_norm": 10.337199211120605,
      "learning_rate": 4.469318020424508e-06,
      "loss": 9.0271,
      "step": 3573
    },
    {
      "epoch": 0.6983196561156702,
      "grad_norm": 8.851112365722656,
      "learning_rate": 4.464046417084359e-06,
      "loss": 8.8061,
      "step": 3574
    },
    {
      "epoch": 0.6985150449394295,
      "grad_norm": 8.679439544677734,
      "learning_rate": 4.458777031065496e-06,
      "loss": 9.4217,
      "step": 3575
    },
    {
      "epoch": 0.6987104337631888,
      "grad_norm": 8.675277709960938,
      "learning_rate": 4.453509864478464e-06,
      "loss": 8.6926,
      "step": 3576
    },
    {
      "epoch": 0.698905822586948,
      "grad_norm": 11.892107963562012,
      "learning_rate": 4.448244919432929e-06,
      "loss": 9.2724,
      "step": 3577
    },
    {
      "epoch": 0.6991012114107074,
      "grad_norm": 10.948375701904297,
      "learning_rate": 4.442982198037669e-06,
      "loss": 9.3081,
      "step": 3578
    },
    {
      "epoch": 0.6992966002344666,
      "grad_norm": 12.927075386047363,
      "learning_rate": 4.437721702400559e-06,
      "loss": 8.6354,
      "step": 3579
    },
    {
      "epoch": 0.6994919890582258,
      "grad_norm": 9.045134544372559,
      "learning_rate": 4.432463434628601e-06,
      "loss": 9.3044,
      "step": 3580
    },
    {
      "epoch": 0.6996873778819851,
      "grad_norm": 9.391888618469238,
      "learning_rate": 4.4272073968278865e-06,
      "loss": 9.1682,
      "step": 3581
    },
    {
      "epoch": 0.6998827667057445,
      "grad_norm": 9.01781177520752,
      "learning_rate": 4.421953591103627e-06,
      "loss": 8.4321,
      "step": 3582
    },
    {
      "epoch": 0.7000781555295037,
      "grad_norm": 9.802136421203613,
      "learning_rate": 4.416702019560129e-06,
      "loss": 9.8514,
      "step": 3583
    },
    {
      "epoch": 0.7002735443532629,
      "grad_norm": 8.904387474060059,
      "learning_rate": 4.411452684300818e-06,
      "loss": 9.3018,
      "step": 3584
    },
    {
      "epoch": 0.7004689331770223,
      "grad_norm": 7.979209899902344,
      "learning_rate": 4.406205587428206e-06,
      "loss": 8.6123,
      "step": 3585
    },
    {
      "epoch": 0.7006643220007815,
      "grad_norm": 17.200040817260742,
      "learning_rate": 4.400960731043929e-06,
      "loss": 9.4106,
      "step": 3586
    },
    {
      "epoch": 0.7008597108245408,
      "grad_norm": 7.51259708404541,
      "learning_rate": 4.395718117248704e-06,
      "loss": 9.1072,
      "step": 3587
    },
    {
      "epoch": 0.7010550996483002,
      "grad_norm": 8.189982414245605,
      "learning_rate": 4.39047774814237e-06,
      "loss": 8.2669,
      "step": 3588
    },
    {
      "epoch": 0.7012504884720594,
      "grad_norm": 6.8721442222595215,
      "learning_rate": 4.385239625823854e-06,
      "loss": 8.0266,
      "step": 3589
    },
    {
      "epoch": 0.7014458772958186,
      "grad_norm": 9.310396194458008,
      "learning_rate": 4.380003752391181e-06,
      "loss": 9.1325,
      "step": 3590
    },
    {
      "epoch": 0.701641266119578,
      "grad_norm": 9.49889850616455,
      "learning_rate": 4.374770129941491e-06,
      "loss": 10.0293,
      "step": 3591
    },
    {
      "epoch": 0.7018366549433372,
      "grad_norm": 9.905617713928223,
      "learning_rate": 4.369538760571003e-06,
      "loss": 9.4679,
      "step": 3592
    },
    {
      "epoch": 0.7020320437670965,
      "grad_norm": 8.300040245056152,
      "learning_rate": 4.3643096463750504e-06,
      "loss": 8.839,
      "step": 3593
    },
    {
      "epoch": 0.7022274325908558,
      "grad_norm": 9.779138565063477,
      "learning_rate": 4.359082789448048e-06,
      "loss": 8.1039,
      "step": 3594
    },
    {
      "epoch": 0.7024228214146151,
      "grad_norm": 8.973949432373047,
      "learning_rate": 4.3538581918835236e-06,
      "loss": 10.0138,
      "step": 3595
    },
    {
      "epoch": 0.7026182102383743,
      "grad_norm": 12.79630184173584,
      "learning_rate": 4.348635855774082e-06,
      "loss": 10.153,
      "step": 3596
    },
    {
      "epoch": 0.7028135990621337,
      "grad_norm": 10.279449462890625,
      "learning_rate": 4.343415783211438e-06,
      "loss": 9.1011,
      "step": 3597
    },
    {
      "epoch": 0.7030089878858929,
      "grad_norm": 7.635469913482666,
      "learning_rate": 4.338197976286385e-06,
      "loss": 8.5996,
      "step": 3598
    },
    {
      "epoch": 0.7032043767096522,
      "grad_norm": 9.318404197692871,
      "learning_rate": 4.332982437088825e-06,
      "loss": 9.6105,
      "step": 3599
    },
    {
      "epoch": 0.7033997655334114,
      "grad_norm": 8.180233001708984,
      "learning_rate": 4.3277691677077375e-06,
      "loss": 9.5781,
      "step": 3600
    },
    {
      "epoch": 0.7035951543571708,
      "grad_norm": 7.368651390075684,
      "learning_rate": 4.322558170231202e-06,
      "loss": 9.1421,
      "step": 3601
    },
    {
      "epoch": 0.70379054318093,
      "grad_norm": 10.323442459106445,
      "learning_rate": 4.317349446746382e-06,
      "loss": 9.5176,
      "step": 3602
    },
    {
      "epoch": 0.7039859320046893,
      "grad_norm": 8.985882759094238,
      "learning_rate": 4.312142999339537e-06,
      "loss": 9.0242,
      "step": 3603
    },
    {
      "epoch": 0.7041813208284486,
      "grad_norm": 9.284278869628906,
      "learning_rate": 4.306938830096008e-06,
      "loss": 9.0943,
      "step": 3604
    },
    {
      "epoch": 0.7043767096522079,
      "grad_norm": 6.843111038208008,
      "learning_rate": 4.301736941100223e-06,
      "loss": 8.3563,
      "step": 3605
    },
    {
      "epoch": 0.7045720984759671,
      "grad_norm": 6.5979905128479,
      "learning_rate": 4.296537334435707e-06,
      "loss": 7.6602,
      "step": 3606
    },
    {
      "epoch": 0.7047674872997265,
      "grad_norm": 10.094749450683594,
      "learning_rate": 4.291340012185058e-06,
      "loss": 9.9754,
      "step": 3607
    },
    {
      "epoch": 0.7049628761234857,
      "grad_norm": 7.74041748046875,
      "learning_rate": 4.286144976429971e-06,
      "loss": 9.5605,
      "step": 3608
    },
    {
      "epoch": 0.705158264947245,
      "grad_norm": 7.466304302215576,
      "learning_rate": 4.280952229251212e-06,
      "loss": 8.9152,
      "step": 3609
    },
    {
      "epoch": 0.7053536537710043,
      "grad_norm": 18.040563583374023,
      "learning_rate": 4.275761772728644e-06,
      "loss": 7.9846,
      "step": 3610
    },
    {
      "epoch": 0.7055490425947636,
      "grad_norm": 9.60568904876709,
      "learning_rate": 4.2705736089411995e-06,
      "loss": 9.5358,
      "step": 3611
    },
    {
      "epoch": 0.7057444314185228,
      "grad_norm": 10.833187103271484,
      "learning_rate": 4.265387739966907e-06,
      "loss": 10.1627,
      "step": 3612
    },
    {
      "epoch": 0.7059398202422822,
      "grad_norm": 15.13516902923584,
      "learning_rate": 4.26020416788286e-06,
      "loss": 8.5006,
      "step": 3613
    },
    {
      "epoch": 0.7061352090660414,
      "grad_norm": 8.97628402709961,
      "learning_rate": 4.255022894765247e-06,
      "loss": 9.3216,
      "step": 3614
    },
    {
      "epoch": 0.7063305978898007,
      "grad_norm": 8.268674850463867,
      "learning_rate": 4.249843922689322e-06,
      "loss": 9.4552,
      "step": 3615
    },
    {
      "epoch": 0.70652598671356,
      "grad_norm": 6.960999488830566,
      "learning_rate": 4.244667253729431e-06,
      "loss": 8.6508,
      "step": 3616
    },
    {
      "epoch": 0.7067213755373193,
      "grad_norm": 7.6391987800598145,
      "learning_rate": 4.2394928899589884e-06,
      "loss": 8.8044,
      "step": 3617
    },
    {
      "epoch": 0.7069167643610785,
      "grad_norm": 9.734604835510254,
      "learning_rate": 4.234320833450482e-06,
      "loss": 9.2713,
      "step": 3618
    },
    {
      "epoch": 0.7071121531848378,
      "grad_norm": 8.456120491027832,
      "learning_rate": 4.229151086275488e-06,
      "loss": 9.1271,
      "step": 3619
    },
    {
      "epoch": 0.7073075420085971,
      "grad_norm": 9.645123481750488,
      "learning_rate": 4.2239836505046465e-06,
      "loss": 9.9589,
      "step": 3620
    },
    {
      "epoch": 0.7075029308323564,
      "grad_norm": 10.434496879577637,
      "learning_rate": 4.21881852820768e-06,
      "loss": 9.1162,
      "step": 3621
    },
    {
      "epoch": 0.7076983196561156,
      "grad_norm": 11.226633071899414,
      "learning_rate": 4.213655721453373e-06,
      "loss": 9.5855,
      "step": 3622
    },
    {
      "epoch": 0.707893708479875,
      "grad_norm": 8.523823738098145,
      "learning_rate": 4.208495232309601e-06,
      "loss": 8.8937,
      "step": 3623
    },
    {
      "epoch": 0.7080890973036342,
      "grad_norm": 10.059098243713379,
      "learning_rate": 4.203337062843289e-06,
      "loss": 9.5106,
      "step": 3624
    },
    {
      "epoch": 0.7082844861273935,
      "grad_norm": 6.950419902801514,
      "learning_rate": 4.1981812151204515e-06,
      "loss": 9.1168,
      "step": 3625
    },
    {
      "epoch": 0.7084798749511528,
      "grad_norm": 7.646895408630371,
      "learning_rate": 4.19302769120616e-06,
      "loss": 8.8619,
      "step": 3626
    },
    {
      "epoch": 0.7086752637749121,
      "grad_norm": 10.336943626403809,
      "learning_rate": 4.187876493164569e-06,
      "loss": 9.8278,
      "step": 3627
    },
    {
      "epoch": 0.7088706525986713,
      "grad_norm": 7.599386692047119,
      "learning_rate": 4.182727623058883e-06,
      "loss": 9.5769,
      "step": 3628
    },
    {
      "epoch": 0.7090660414224307,
      "grad_norm": 7.983441352844238,
      "learning_rate": 4.177581082951393e-06,
      "loss": 8.6082,
      "step": 3629
    },
    {
      "epoch": 0.7092614302461899,
      "grad_norm": 7.941792011260986,
      "learning_rate": 4.1724368749034434e-06,
      "loss": 9.0121,
      "step": 3630
    },
    {
      "epoch": 0.7094568190699492,
      "grad_norm": 8.246491432189941,
      "learning_rate": 4.167295000975449e-06,
      "loss": 9.4766,
      "step": 3631
    },
    {
      "epoch": 0.7096522078937085,
      "grad_norm": 13.80324649810791,
      "learning_rate": 4.1621554632268935e-06,
      "loss": 9.6819,
      "step": 3632
    },
    {
      "epoch": 0.7098475967174678,
      "grad_norm": 8.703511238098145,
      "learning_rate": 4.1570182637163155e-06,
      "loss": 8.7774,
      "step": 3633
    },
    {
      "epoch": 0.710042985541227,
      "grad_norm": 8.409005165100098,
      "learning_rate": 4.15188340450133e-06,
      "loss": 8.604,
      "step": 3634
    },
    {
      "epoch": 0.7102383743649863,
      "grad_norm": 8.394705772399902,
      "learning_rate": 4.146750887638601e-06,
      "loss": 8.5099,
      "step": 3635
    },
    {
      "epoch": 0.7104337631887456,
      "grad_norm": 6.875105381011963,
      "learning_rate": 4.141620715183867e-06,
      "loss": 8.7819,
      "step": 3636
    },
    {
      "epoch": 0.7106291520125049,
      "grad_norm": 9.762279510498047,
      "learning_rate": 4.136492889191914e-06,
      "loss": 9.6374,
      "step": 3637
    },
    {
      "epoch": 0.7108245408362641,
      "grad_norm": 8.945638656616211,
      "learning_rate": 4.131367411716605e-06,
      "loss": 9.8514,
      "step": 3638
    },
    {
      "epoch": 0.7110199296600235,
      "grad_norm": 8.55778980255127,
      "learning_rate": 4.126244284810842e-06,
      "loss": 8.1234,
      "step": 3639
    },
    {
      "epoch": 0.7112153184837827,
      "grad_norm": 5.584824085235596,
      "learning_rate": 4.1211235105266065e-06,
      "loss": 8.2136,
      "step": 3640
    },
    {
      "epoch": 0.711410707307542,
      "grad_norm": 9.62353801727295,
      "learning_rate": 4.11600509091492e-06,
      "loss": 9.8375,
      "step": 3641
    },
    {
      "epoch": 0.7116060961313013,
      "grad_norm": 8.715863227844238,
      "learning_rate": 4.110889028025874e-06,
      "loss": 9.1487,
      "step": 3642
    },
    {
      "epoch": 0.7118014849550606,
      "grad_norm": 7.7625651359558105,
      "learning_rate": 4.105775323908608e-06,
      "loss": 8.7921,
      "step": 3643
    },
    {
      "epoch": 0.7119968737788198,
      "grad_norm": 11.389391899108887,
      "learning_rate": 4.100663980611317e-06,
      "loss": 8.8859,
      "step": 3644
    },
    {
      "epoch": 0.7121922626025792,
      "grad_norm": 7.64915132522583,
      "learning_rate": 4.095555000181257e-06,
      "loss": 8.9723,
      "step": 3645
    },
    {
      "epoch": 0.7123876514263384,
      "grad_norm": 10.494763374328613,
      "learning_rate": 4.090448384664728e-06,
      "loss": 10.0071,
      "step": 3646
    },
    {
      "epoch": 0.7125830402500977,
      "grad_norm": 8.759225845336914,
      "learning_rate": 4.085344136107096e-06,
      "loss": 9.3408,
      "step": 3647
    },
    {
      "epoch": 0.712778429073857,
      "grad_norm": 11.41947078704834,
      "learning_rate": 4.080242256552761e-06,
      "loss": 9.6381,
      "step": 3648
    },
    {
      "epoch": 0.7129738178976163,
      "grad_norm": 6.726463794708252,
      "learning_rate": 4.075142748045194e-06,
      "loss": 8.6356,
      "step": 3649
    },
    {
      "epoch": 0.7131692067213755,
      "grad_norm": 9.784330368041992,
      "learning_rate": 4.070045612626898e-06,
      "loss": 9.9572,
      "step": 3650
    },
    {
      "epoch": 0.7133645955451349,
      "grad_norm": 9.007823944091797,
      "learning_rate": 4.064950852339442e-06,
      "loss": 9.215,
      "step": 3651
    },
    {
      "epoch": 0.7135599843688941,
      "grad_norm": 7.706821441650391,
      "learning_rate": 4.059858469223428e-06,
      "loss": 7.903,
      "step": 3652
    },
    {
      "epoch": 0.7137553731926534,
      "grad_norm": 7.795345783233643,
      "learning_rate": 4.054768465318521e-06,
      "loss": 8.621,
      "step": 3653
    },
    {
      "epoch": 0.7139507620164126,
      "grad_norm": 8.038520812988281,
      "learning_rate": 4.049680842663416e-06,
      "loss": 8.8847,
      "step": 3654
    },
    {
      "epoch": 0.714146150840172,
      "grad_norm": 6.677659511566162,
      "learning_rate": 4.044595603295876e-06,
      "loss": 9.6192,
      "step": 3655
    },
    {
      "epoch": 0.7143415396639312,
      "grad_norm": 8.80317211151123,
      "learning_rate": 4.039512749252689e-06,
      "loss": 9.7371,
      "step": 3656
    },
    {
      "epoch": 0.7145369284876905,
      "grad_norm": 6.420666694641113,
      "learning_rate": 4.034432282569694e-06,
      "loss": 8.5476,
      "step": 3657
    },
    {
      "epoch": 0.7147323173114498,
      "grad_norm": 9.233664512634277,
      "learning_rate": 4.029354205281784e-06,
      "loss": 10.0034,
      "step": 3658
    },
    {
      "epoch": 0.7149277061352091,
      "grad_norm": 9.827657699584961,
      "learning_rate": 4.024278519422877e-06,
      "loss": 8.514,
      "step": 3659
    },
    {
      "epoch": 0.7151230949589683,
      "grad_norm": 8.356546401977539,
      "learning_rate": 4.019205227025952e-06,
      "loss": 8.3579,
      "step": 3660
    },
    {
      "epoch": 0.7153184837827277,
      "grad_norm": 9.356789588928223,
      "learning_rate": 4.014134330123012e-06,
      "loss": 8.6536,
      "step": 3661
    },
    {
      "epoch": 0.7155138726064869,
      "grad_norm": 10.166316032409668,
      "learning_rate": 4.009065830745116e-06,
      "loss": 8.6797,
      "step": 3662
    },
    {
      "epoch": 0.7157092614302462,
      "grad_norm": 79.7872314453125,
      "learning_rate": 4.0039997309223475e-06,
      "loss": 10.0384,
      "step": 3663
    },
    {
      "epoch": 0.7159046502540055,
      "grad_norm": 8.668107986450195,
      "learning_rate": 3.9989360326838455e-06,
      "loss": 8.5372,
      "step": 3664
    },
    {
      "epoch": 0.7161000390777648,
      "grad_norm": 8.511775970458984,
      "learning_rate": 3.993874738057771e-06,
      "loss": 8.6834,
      "step": 3665
    },
    {
      "epoch": 0.716295427901524,
      "grad_norm": 8.757607460021973,
      "learning_rate": 3.988815849071335e-06,
      "loss": 8.415,
      "step": 3666
    },
    {
      "epoch": 0.7164908167252834,
      "grad_norm": 9.092312812805176,
      "learning_rate": 3.983759367750772e-06,
      "loss": 8.8122,
      "step": 3667
    },
    {
      "epoch": 0.7166862055490426,
      "grad_norm": 9.101534843444824,
      "learning_rate": 3.978705296121372e-06,
      "loss": 9.1954,
      "step": 3668
    },
    {
      "epoch": 0.7168815943728019,
      "grad_norm": 8.66796588897705,
      "learning_rate": 3.973653636207437e-06,
      "loss": 8.7924,
      "step": 3669
    },
    {
      "epoch": 0.7170769831965611,
      "grad_norm": 23.35134506225586,
      "learning_rate": 3.968604390032316e-06,
      "loss": 8.6947,
      "step": 3670
    },
    {
      "epoch": 0.7172723720203205,
      "grad_norm": 9.638257026672363,
      "learning_rate": 3.963557559618392e-06,
      "loss": 8.8963,
      "step": 3671
    },
    {
      "epoch": 0.7174677608440797,
      "grad_norm": 7.746241569519043,
      "learning_rate": 3.958513146987073e-06,
      "loss": 8.4087,
      "step": 3672
    },
    {
      "epoch": 0.717663149667839,
      "grad_norm": 8.321249008178711,
      "learning_rate": 3.953471154158808e-06,
      "loss": 9.321,
      "step": 3673
    },
    {
      "epoch": 0.7178585384915983,
      "grad_norm": 7.365278720855713,
      "learning_rate": 3.948431583153064e-06,
      "loss": 9.0058,
      "step": 3674
    },
    {
      "epoch": 0.7180539273153576,
      "grad_norm": 9.134242057800293,
      "learning_rate": 3.943394435988356e-06,
      "loss": 9.9397,
      "step": 3675
    },
    {
      "epoch": 0.7182493161391168,
      "grad_norm": 8.319193840026855,
      "learning_rate": 3.9383597146822076e-06,
      "loss": 8.5295,
      "step": 3676
    },
    {
      "epoch": 0.7184447049628762,
      "grad_norm": 11.505075454711914,
      "learning_rate": 3.933327421251189e-06,
      "loss": 9.552,
      "step": 3677
    },
    {
      "epoch": 0.7186400937866354,
      "grad_norm": 8.731011390686035,
      "learning_rate": 3.928297557710883e-06,
      "loss": 8.065,
      "step": 3678
    },
    {
      "epoch": 0.7188354826103946,
      "grad_norm": 7.6019110679626465,
      "learning_rate": 3.923270126075913e-06,
      "loss": 8.741,
      "step": 3679
    },
    {
      "epoch": 0.719030871434154,
      "grad_norm": 6.32687520980835,
      "learning_rate": 3.9182451283599156e-06,
      "loss": 7.7967,
      "step": 3680
    },
    {
      "epoch": 0.7192262602579133,
      "grad_norm": 18.116029739379883,
      "learning_rate": 3.913222566575564e-06,
      "loss": 9.2571,
      "step": 3681
    },
    {
      "epoch": 0.7194216490816725,
      "grad_norm": 8.875486373901367,
      "learning_rate": 3.908202442734546e-06,
      "loss": 8.6051,
      "step": 3682
    },
    {
      "epoch": 0.7196170379054319,
      "grad_norm": 7.803152561187744,
      "learning_rate": 3.90318475884758e-06,
      "loss": 8.6105,
      "step": 3683
    },
    {
      "epoch": 0.7198124267291911,
      "grad_norm": 7.7268171310424805,
      "learning_rate": 3.8981695169243986e-06,
      "loss": 8.451,
      "step": 3684
    },
    {
      "epoch": 0.7200078155529503,
      "grad_norm": 7.699978828430176,
      "learning_rate": 3.893156718973767e-06,
      "loss": 8.5678,
      "step": 3685
    },
    {
      "epoch": 0.7202032043767097,
      "grad_norm": 7.997290134429932,
      "learning_rate": 3.888146367003464e-06,
      "loss": 8.8491,
      "step": 3686
    },
    {
      "epoch": 0.720398593200469,
      "grad_norm": 7.056305408477783,
      "learning_rate": 3.883138463020294e-06,
      "loss": 9.4386,
      "step": 3687
    },
    {
      "epoch": 0.7205939820242282,
      "grad_norm": 7.773458957672119,
      "learning_rate": 3.878133009030076e-06,
      "loss": 8.9443,
      "step": 3688
    },
    {
      "epoch": 0.7207893708479874,
      "grad_norm": 6.357101917266846,
      "learning_rate": 3.873130007037647e-06,
      "loss": 7.9893,
      "step": 3689
    },
    {
      "epoch": 0.7209847596717468,
      "grad_norm": 14.02954387664795,
      "learning_rate": 3.868129459046871e-06,
      "loss": 9.3322,
      "step": 3690
    },
    {
      "epoch": 0.721180148495506,
      "grad_norm": 10.02996826171875,
      "learning_rate": 3.863131367060615e-06,
      "loss": 10.0922,
      "step": 3691
    },
    {
      "epoch": 0.7213755373192653,
      "grad_norm": 8.229723930358887,
      "learning_rate": 3.858135733080779e-06,
      "loss": 9.2091,
      "step": 3692
    },
    {
      "epoch": 0.7215709261430246,
      "grad_norm": 12.047175407409668,
      "learning_rate": 3.8531425591082605e-06,
      "loss": 9.8282,
      "step": 3693
    },
    {
      "epoch": 0.7217663149667839,
      "grad_norm": 7.275228023529053,
      "learning_rate": 3.848151847142988e-06,
      "loss": 9.1976,
      "step": 3694
    },
    {
      "epoch": 0.7219617037905431,
      "grad_norm": 6.717846393585205,
      "learning_rate": 3.84316359918389e-06,
      "loss": 8.3496,
      "step": 3695
    },
    {
      "epoch": 0.7221570926143025,
      "grad_norm": 8.532776832580566,
      "learning_rate": 3.838177817228922e-06,
      "loss": 9.1765,
      "step": 3696
    },
    {
      "epoch": 0.7223524814380617,
      "grad_norm": 20.730764389038086,
      "learning_rate": 3.833194503275038e-06,
      "loss": 9.5346,
      "step": 3697
    },
    {
      "epoch": 0.722547870261821,
      "grad_norm": 9.638972282409668,
      "learning_rate": 3.828213659318215e-06,
      "loss": 9.4929,
      "step": 3698
    },
    {
      "epoch": 0.7227432590855803,
      "grad_norm": 8.68529987335205,
      "learning_rate": 3.8232352873534295e-06,
      "loss": 9.7288,
      "step": 3699
    },
    {
      "epoch": 0.7229386479093396,
      "grad_norm": 11.741350173950195,
      "learning_rate": 3.818259389374682e-06,
      "loss": 8.9153,
      "step": 3700
    },
    {
      "epoch": 0.7231340367330988,
      "grad_norm": 9.967458724975586,
      "learning_rate": 3.8132859673749688e-06,
      "loss": 8.623,
      "step": 3701
    },
    {
      "epoch": 0.7233294255568582,
      "grad_norm": 7.636810302734375,
      "learning_rate": 3.8083150233462984e-06,
      "loss": 8.3056,
      "step": 3702
    },
    {
      "epoch": 0.7235248143806174,
      "grad_norm": 8.557602882385254,
      "learning_rate": 3.8033465592796946e-06,
      "loss": 9.1016,
      "step": 3703
    },
    {
      "epoch": 0.7237202032043767,
      "grad_norm": 8.045110702514648,
      "learning_rate": 3.7983805771651728e-06,
      "loss": 9.2546,
      "step": 3704
    },
    {
      "epoch": 0.723915592028136,
      "grad_norm": 9.116978645324707,
      "learning_rate": 3.7934170789917725e-06,
      "loss": 9.1902,
      "step": 3705
    },
    {
      "epoch": 0.7241109808518953,
      "grad_norm": 6.71596097946167,
      "learning_rate": 3.78845606674752e-06,
      "loss": 8.4867,
      "step": 3706
    },
    {
      "epoch": 0.7243063696756545,
      "grad_norm": 8.542150497436523,
      "learning_rate": 3.783497542419463e-06,
      "loss": 9.0346,
      "step": 3707
    },
    {
      "epoch": 0.7245017584994138,
      "grad_norm": 10.2990083694458,
      "learning_rate": 3.778541507993636e-06,
      "loss": 9.4014,
      "step": 3708
    },
    {
      "epoch": 0.7246971473231731,
      "grad_norm": 9.195205688476562,
      "learning_rate": 3.7735879654550935e-06,
      "loss": 8.6709,
      "step": 3709
    },
    {
      "epoch": 0.7248925361469324,
      "grad_norm": 10.924129486083984,
      "learning_rate": 3.7686369167878735e-06,
      "loss": 8.9261,
      "step": 3710
    },
    {
      "epoch": 0.7250879249706916,
      "grad_norm": 9.022870063781738,
      "learning_rate": 3.7636883639750343e-06,
      "loss": 9.8402,
      "step": 3711
    },
    {
      "epoch": 0.725283313794451,
      "grad_norm": 98.3774185180664,
      "learning_rate": 3.7587423089986174e-06,
      "loss": 8.5148,
      "step": 3712
    },
    {
      "epoch": 0.7254787026182102,
      "grad_norm": 6.8416924476623535,
      "learning_rate": 3.753798753839677e-06,
      "loss": 8.3697,
      "step": 3713
    },
    {
      "epoch": 0.7256740914419695,
      "grad_norm": 12.198902130126953,
      "learning_rate": 3.7488577004782578e-06,
      "loss": 9.4243,
      "step": 3714
    },
    {
      "epoch": 0.7258694802657288,
      "grad_norm": 7.990806579589844,
      "learning_rate": 3.7439191508934013e-06,
      "loss": 8.7914,
      "step": 3715
    },
    {
      "epoch": 0.7260648690894881,
      "grad_norm": 9.815138816833496,
      "learning_rate": 3.7389831070631577e-06,
      "loss": 8.9791,
      "step": 3716
    },
    {
      "epoch": 0.7262602579132473,
      "grad_norm": 9.926175117492676,
      "learning_rate": 3.7340495709645585e-06,
      "loss": 9.9698,
      "step": 3717
    },
    {
      "epoch": 0.7264556467370067,
      "grad_norm": 8.747008323669434,
      "learning_rate": 3.7291185445736445e-06,
      "loss": 9.4764,
      "step": 3718
    },
    {
      "epoch": 0.7266510355607659,
      "grad_norm": 6.5735697746276855,
      "learning_rate": 3.724190029865439e-06,
      "loss": 8.7215,
      "step": 3719
    },
    {
      "epoch": 0.7268464243845252,
      "grad_norm": 11.640028953552246,
      "learning_rate": 3.7192640288139704e-06,
      "loss": 10.2233,
      "step": 3720
    },
    {
      "epoch": 0.7270418132082845,
      "grad_norm": 7.5807204246521,
      "learning_rate": 3.7143405433922486e-06,
      "loss": 9.4529,
      "step": 3721
    },
    {
      "epoch": 0.7272372020320438,
      "grad_norm": 11.326850891113281,
      "learning_rate": 3.7094195755722906e-06,
      "loss": 9.1891,
      "step": 3722
    },
    {
      "epoch": 0.727432590855803,
      "grad_norm": 8.540494918823242,
      "learning_rate": 3.7045011273250898e-06,
      "loss": 9.5362,
      "step": 3723
    },
    {
      "epoch": 0.7276279796795623,
      "grad_norm": 8.911155700683594,
      "learning_rate": 3.6995852006206424e-06,
      "loss": 9.799,
      "step": 3724
    },
    {
      "epoch": 0.7278233685033216,
      "grad_norm": 9.42534065246582,
      "learning_rate": 3.6946717974279245e-06,
      "loss": 9.7036,
      "step": 3725
    },
    {
      "epoch": 0.7280187573270809,
      "grad_norm": 8.128520011901855,
      "learning_rate": 3.689760919714912e-06,
      "loss": 9.2025,
      "step": 3726
    },
    {
      "epoch": 0.7282141461508401,
      "grad_norm": 9.171087265014648,
      "learning_rate": 3.6848525694485627e-06,
      "loss": 9.3736,
      "step": 3727
    },
    {
      "epoch": 0.7284095349745995,
      "grad_norm": 9.77519416809082,
      "learning_rate": 3.679946748594818e-06,
      "loss": 9.7459,
      "step": 3728
    },
    {
      "epoch": 0.7286049237983587,
      "grad_norm": 8.036867141723633,
      "learning_rate": 3.675043459118619e-06,
      "loss": 9.031,
      "step": 3729
    },
    {
      "epoch": 0.728800312622118,
      "grad_norm": 8.540367126464844,
      "learning_rate": 3.670142702983878e-06,
      "loss": 8.9552,
      "step": 3730
    },
    {
      "epoch": 0.7289957014458773,
      "grad_norm": 10.316926956176758,
      "learning_rate": 3.6652444821535072e-06,
      "loss": 8.7552,
      "step": 3731
    },
    {
      "epoch": 0.7291910902696366,
      "grad_norm": 8.742179870605469,
      "learning_rate": 3.6603487985893894e-06,
      "loss": 8.6493,
      "step": 3732
    },
    {
      "epoch": 0.7293864790933958,
      "grad_norm": 10.511765480041504,
      "learning_rate": 3.6554556542524046e-06,
      "loss": 8.7671,
      "step": 3733
    },
    {
      "epoch": 0.7295818679171552,
      "grad_norm": 19.103931427001953,
      "learning_rate": 3.6505650511024014e-06,
      "loss": 9.5827,
      "step": 3734
    },
    {
      "epoch": 0.7297772567409144,
      "grad_norm": 9.42544174194336,
      "learning_rate": 3.645676991098227e-06,
      "loss": 9.5682,
      "step": 3735
    },
    {
      "epoch": 0.7299726455646737,
      "grad_norm": 8.16747760772705,
      "learning_rate": 3.6407914761976928e-06,
      "loss": 9.0845,
      "step": 3736
    },
    {
      "epoch": 0.730168034388433,
      "grad_norm": 9.376529693603516,
      "learning_rate": 3.635908508357606e-06,
      "loss": 9.4353,
      "step": 3737
    },
    {
      "epoch": 0.7303634232121923,
      "grad_norm": 8.061734199523926,
      "learning_rate": 3.6310280895337402e-06,
      "loss": 8.2525,
      "step": 3738
    },
    {
      "epoch": 0.7305588120359515,
      "grad_norm": 10.184969902038574,
      "learning_rate": 3.626150221680863e-06,
      "loss": 9.6473,
      "step": 3739
    },
    {
      "epoch": 0.7307542008597109,
      "grad_norm": 6.46053409576416,
      "learning_rate": 3.6212749067527054e-06,
      "loss": 8.5041,
      "step": 3740
    },
    {
      "epoch": 0.7309495896834701,
      "grad_norm": 8.612481117248535,
      "learning_rate": 3.6164021467019817e-06,
      "loss": 9.8489,
      "step": 3741
    },
    {
      "epoch": 0.7311449785072294,
      "grad_norm": 11.279267311096191,
      "learning_rate": 3.6115319434803897e-06,
      "loss": 10.6624,
      "step": 3742
    },
    {
      "epoch": 0.7313403673309886,
      "grad_norm": 8.220184326171875,
      "learning_rate": 3.606664299038589e-06,
      "loss": 9.1821,
      "step": 3743
    },
    {
      "epoch": 0.731535756154748,
      "grad_norm": 9.151845932006836,
      "learning_rate": 3.6017992153262303e-06,
      "loss": 9.3378,
      "step": 3744
    },
    {
      "epoch": 0.7317311449785072,
      "grad_norm": 10.015589714050293,
      "learning_rate": 3.5969366942919237e-06,
      "loss": 10.7843,
      "step": 3745
    },
    {
      "epoch": 0.7319265338022665,
      "grad_norm": 8.539863586425781,
      "learning_rate": 3.5920767378832666e-06,
      "loss": 9.346,
      "step": 3746
    },
    {
      "epoch": 0.7321219226260258,
      "grad_norm": 7.310202121734619,
      "learning_rate": 3.5872193480468155e-06,
      "loss": 9.3841,
      "step": 3747
    },
    {
      "epoch": 0.7323173114497851,
      "grad_norm": 9.648876190185547,
      "learning_rate": 3.5823645267281126e-06,
      "loss": 9.2589,
      "step": 3748
    },
    {
      "epoch": 0.7325127002735443,
      "grad_norm": 9.810677528381348,
      "learning_rate": 3.5775122758716585e-06,
      "loss": 9.8274,
      "step": 3749
    },
    {
      "epoch": 0.7327080890973037,
      "grad_norm": 8.219881057739258,
      "learning_rate": 3.572662597420935e-06,
      "loss": 9.3783,
      "step": 3750
    },
    {
      "epoch": 0.7329034779210629,
      "grad_norm": 9.856924057006836,
      "learning_rate": 3.567815493318385e-06,
      "loss": 9.4114,
      "step": 3751
    },
    {
      "epoch": 0.7330988667448222,
      "grad_norm": 7.9009833335876465,
      "learning_rate": 3.562970965505429e-06,
      "loss": 9.3978,
      "step": 3752
    },
    {
      "epoch": 0.7332942555685815,
      "grad_norm": 8.5271577835083,
      "learning_rate": 3.558129015922448e-06,
      "loss": 8.2685,
      "step": 3753
    },
    {
      "epoch": 0.7334896443923408,
      "grad_norm": 15.44652271270752,
      "learning_rate": 3.5532896465087897e-06,
      "loss": 9.7942,
      "step": 3754
    },
    {
      "epoch": 0.7336850332161,
      "grad_norm": 9.352770805358887,
      "learning_rate": 3.5484528592027778e-06,
      "loss": 9.0612,
      "step": 3755
    },
    {
      "epoch": 0.7338804220398594,
      "grad_norm": 11.316088676452637,
      "learning_rate": 3.543618655941691e-06,
      "loss": 10.4106,
      "step": 3756
    },
    {
      "epoch": 0.7340758108636186,
      "grad_norm": 13.240620613098145,
      "learning_rate": 3.5387870386617827e-06,
      "loss": 8.9332,
      "step": 3757
    },
    {
      "epoch": 0.7342711996873779,
      "grad_norm": 13.1504487991333,
      "learning_rate": 3.5339580092982594e-06,
      "loss": 11.0741,
      "step": 3758
    },
    {
      "epoch": 0.7344665885111371,
      "grad_norm": 11.847900390625,
      "learning_rate": 3.5291315697853044e-06,
      "loss": 9.0539,
      "step": 3759
    },
    {
      "epoch": 0.7346619773348965,
      "grad_norm": 8.007011413574219,
      "learning_rate": 3.5243077220560496e-06,
      "loss": 9.2799,
      "step": 3760
    },
    {
      "epoch": 0.7348573661586557,
      "grad_norm": 11.476548194885254,
      "learning_rate": 3.5194864680426023e-06,
      "loss": 8.8405,
      "step": 3761
    },
    {
      "epoch": 0.735052754982415,
      "grad_norm": 10.305270195007324,
      "learning_rate": 3.5146678096760166e-06,
      "loss": 9.647,
      "step": 3762
    },
    {
      "epoch": 0.7352481438061743,
      "grad_norm": 9.828678131103516,
      "learning_rate": 3.509851748886325e-06,
      "loss": 10.0625,
      "step": 3763
    },
    {
      "epoch": 0.7354435326299336,
      "grad_norm": 7.253856658935547,
      "learning_rate": 3.505038287602499e-06,
      "loss": 8.6707,
      "step": 3764
    },
    {
      "epoch": 0.7356389214536928,
      "grad_norm": 9.606746673583984,
      "learning_rate": 3.500227427752487e-06,
      "loss": 10.9174,
      "step": 3765
    },
    {
      "epoch": 0.7358343102774522,
      "grad_norm": 8.862897872924805,
      "learning_rate": 3.4954191712631856e-06,
      "loss": 9.2135,
      "step": 3766
    },
    {
      "epoch": 0.7360296991012114,
      "grad_norm": 7.542580604553223,
      "learning_rate": 3.4906135200604464e-06,
      "loss": 8.3897,
      "step": 3767
    },
    {
      "epoch": 0.7362250879249707,
      "grad_norm": 6.870659351348877,
      "learning_rate": 3.485810476069089e-06,
      "loss": 8.4792,
      "step": 3768
    },
    {
      "epoch": 0.73642047674873,
      "grad_norm": 8.340803146362305,
      "learning_rate": 3.4810100412128743e-06,
      "loss": 9.2354,
      "step": 3769
    },
    {
      "epoch": 0.7366158655724893,
      "grad_norm": 6.958797931671143,
      "learning_rate": 3.476212217414533e-06,
      "loss": 8.385,
      "step": 3770
    },
    {
      "epoch": 0.7368112543962485,
      "grad_norm": 9.973862648010254,
      "learning_rate": 3.471417006595735e-06,
      "loss": 9.4786,
      "step": 3771
    },
    {
      "epoch": 0.7370066432200079,
      "grad_norm": 8.524601936340332,
      "learning_rate": 3.4666244106771196e-06,
      "loss": 8.7931,
      "step": 3772
    },
    {
      "epoch": 0.7372020320437671,
      "grad_norm": 8.614114761352539,
      "learning_rate": 3.46183443157826e-06,
      "loss": 8.5033,
      "step": 3773
    },
    {
      "epoch": 0.7373974208675264,
      "grad_norm": 10.846492767333984,
      "learning_rate": 3.457047071217703e-06,
      "loss": 9.0408,
      "step": 3774
    },
    {
      "epoch": 0.7375928096912857,
      "grad_norm": 9.433976173400879,
      "learning_rate": 3.4522623315129244e-06,
      "loss": 9.3712,
      "step": 3775
    },
    {
      "epoch": 0.737788198515045,
      "grad_norm": 6.5583109855651855,
      "learning_rate": 3.4474802143803687e-06,
      "loss": 8.9908,
      "step": 3776
    },
    {
      "epoch": 0.7379835873388042,
      "grad_norm": 9.018953323364258,
      "learning_rate": 3.442700721735417e-06,
      "loss": 8.8893,
      "step": 3777
    },
    {
      "epoch": 0.7381789761625634,
      "grad_norm": 6.55804443359375,
      "learning_rate": 3.437923855492409e-06,
      "loss": 9.0761,
      "step": 3778
    },
    {
      "epoch": 0.7383743649863228,
      "grad_norm": 10.90300178527832,
      "learning_rate": 3.4331496175646263e-06,
      "loss": 9.2355,
      "step": 3779
    },
    {
      "epoch": 0.738569753810082,
      "grad_norm": 12.411234855651855,
      "learning_rate": 3.428378009864296e-06,
      "loss": 10.3317,
      "step": 3780
    },
    {
      "epoch": 0.7387651426338413,
      "grad_norm": 7.479711532592773,
      "learning_rate": 3.423609034302601e-06,
      "loss": 9.0381,
      "step": 3781
    },
    {
      "epoch": 0.7389605314576007,
      "grad_norm": 8.085942268371582,
      "learning_rate": 3.4188426927896567e-06,
      "loss": 9.7472,
      "step": 3782
    },
    {
      "epoch": 0.7391559202813599,
      "grad_norm": 7.482016086578369,
      "learning_rate": 3.41407898723454e-06,
      "loss": 9.0756,
      "step": 3783
    },
    {
      "epoch": 0.7393513091051191,
      "grad_norm": 6.062737941741943,
      "learning_rate": 3.409317919545254e-06,
      "loss": 8.9825,
      "step": 3784
    },
    {
      "epoch": 0.7395466979288785,
      "grad_norm": 9.465961456298828,
      "learning_rate": 3.4045594916287616e-06,
      "loss": 9.1356,
      "step": 3785
    },
    {
      "epoch": 0.7397420867526378,
      "grad_norm": 8.177811622619629,
      "learning_rate": 3.399803705390955e-06,
      "loss": 9.8667,
      "step": 3786
    },
    {
      "epoch": 0.739937475576397,
      "grad_norm": 6.569251537322998,
      "learning_rate": 3.3950505627366793e-06,
      "loss": 8.4834,
      "step": 3787
    },
    {
      "epoch": 0.7401328644001564,
      "grad_norm": 9.74722957611084,
      "learning_rate": 3.3903000655697094e-06,
      "loss": 10.6275,
      "step": 3788
    },
    {
      "epoch": 0.7403282532239156,
      "grad_norm": 7.447836875915527,
      "learning_rate": 3.3855522157927756e-06,
      "loss": 9.1656,
      "step": 3789
    },
    {
      "epoch": 0.7405236420476748,
      "grad_norm": 8.01734447479248,
      "learning_rate": 3.380807015307529e-06,
      "loss": 8.9998,
      "step": 3790
    },
    {
      "epoch": 0.7407190308714342,
      "grad_norm": 8.140630722045898,
      "learning_rate": 3.3760644660145794e-06,
      "loss": 9.3295,
      "step": 3791
    },
    {
      "epoch": 0.7409144196951934,
      "grad_norm": 6.896175861358643,
      "learning_rate": 3.37132456981346e-06,
      "loss": 7.9847,
      "step": 3792
    },
    {
      "epoch": 0.7411098085189527,
      "grad_norm": 7.749749660491943,
      "learning_rate": 3.366587328602644e-06,
      "loss": 8.6011,
      "step": 3793
    },
    {
      "epoch": 0.741305197342712,
      "grad_norm": 11.753620147705078,
      "learning_rate": 3.36185274427955e-06,
      "loss": 8.9515,
      "step": 3794
    },
    {
      "epoch": 0.7415005861664713,
      "grad_norm": 9.068345069885254,
      "learning_rate": 3.357120818740519e-06,
      "loss": 9.6614,
      "step": 3795
    },
    {
      "epoch": 0.7416959749902305,
      "grad_norm": 7.815624713897705,
      "learning_rate": 3.3523915538808406e-06,
      "loss": 8.1013,
      "step": 3796
    },
    {
      "epoch": 0.7418913638139898,
      "grad_norm": 9.315130233764648,
      "learning_rate": 3.3476649515947257e-06,
      "loss": 8.381,
      "step": 3797
    },
    {
      "epoch": 0.7420867526377491,
      "grad_norm": 7.379053115844727,
      "learning_rate": 3.3429410137753327e-06,
      "loss": 9.1761,
      "step": 3798
    },
    {
      "epoch": 0.7422821414615084,
      "grad_norm": 11.974115371704102,
      "learning_rate": 3.338219742314738e-06,
      "loss": 9.9621,
      "step": 3799
    },
    {
      "epoch": 0.7424775302852676,
      "grad_norm": 8.75280475616455,
      "learning_rate": 3.333501139103964e-06,
      "loss": 9.0321,
      "step": 3800
    },
    {
      "epoch": 0.742672919109027,
      "grad_norm": 14.717581748962402,
      "learning_rate": 3.328785206032952e-06,
      "loss": 8.8125,
      "step": 3801
    },
    {
      "epoch": 0.7428683079327862,
      "grad_norm": 12.672874450683594,
      "learning_rate": 3.3240719449905846e-06,
      "loss": 9.2577,
      "step": 3802
    },
    {
      "epoch": 0.7430636967565455,
      "grad_norm": 7.265257358551025,
      "learning_rate": 3.3193613578646633e-06,
      "loss": 7.7047,
      "step": 3803
    },
    {
      "epoch": 0.7432590855803048,
      "grad_norm": 6.224616527557373,
      "learning_rate": 3.3146534465419322e-06,
      "loss": 7.5593,
      "step": 3804
    },
    {
      "epoch": 0.7434544744040641,
      "grad_norm": 9.308873176574707,
      "learning_rate": 3.3099482129080518e-06,
      "loss": 9.8408,
      "step": 3805
    },
    {
      "epoch": 0.7436498632278233,
      "grad_norm": 7.382295608520508,
      "learning_rate": 3.30524565884761e-06,
      "loss": 9.6591,
      "step": 3806
    },
    {
      "epoch": 0.7438452520515827,
      "grad_norm": 10.499371528625488,
      "learning_rate": 3.300545786244134e-06,
      "loss": 8.9012,
      "step": 3807
    },
    {
      "epoch": 0.7440406408753419,
      "grad_norm": 7.243477821350098,
      "learning_rate": 3.2958485969800603e-06,
      "loss": 8.7174,
      "step": 3808
    },
    {
      "epoch": 0.7442360296991012,
      "grad_norm": 8.12458324432373,
      "learning_rate": 3.291154092936768e-06,
      "loss": 8.6591,
      "step": 3809
    },
    {
      "epoch": 0.7444314185228605,
      "grad_norm": 8.82094669342041,
      "learning_rate": 3.286462275994543e-06,
      "loss": 8.1543,
      "step": 3810
    },
    {
      "epoch": 0.7446268073466198,
      "grad_norm": 9.012158393859863,
      "learning_rate": 3.281773148032611e-06,
      "loss": 9.7201,
      "step": 3811
    },
    {
      "epoch": 0.744822196170379,
      "grad_norm": 9.375688552856445,
      "learning_rate": 3.277086710929106e-06,
      "loss": 9.6693,
      "step": 3812
    },
    {
      "epoch": 0.7450175849941383,
      "grad_norm": 9.3660888671875,
      "learning_rate": 3.2724029665611e-06,
      "loss": 8.9164,
      "step": 3813
    },
    {
      "epoch": 0.7452129738178976,
      "grad_norm": 7.924816608428955,
      "learning_rate": 3.2677219168045694e-06,
      "loss": 9.2943,
      "step": 3814
    },
    {
      "epoch": 0.7454083626416569,
      "grad_norm": 9.506559371948242,
      "learning_rate": 3.2630435635344283e-06,
      "loss": 10.2879,
      "step": 3815
    },
    {
      "epoch": 0.7456037514654161,
      "grad_norm": 10.148906707763672,
      "learning_rate": 3.2583679086244946e-06,
      "loss": 9.9814,
      "step": 3816
    },
    {
      "epoch": 0.7457991402891755,
      "grad_norm": 9.202022552490234,
      "learning_rate": 3.2536949539475195e-06,
      "loss": 9.4508,
      "step": 3817
    },
    {
      "epoch": 0.7459945291129347,
      "grad_norm": 10.081830978393555,
      "learning_rate": 3.249024701375165e-06,
      "loss": 9.2082,
      "step": 3818
    },
    {
      "epoch": 0.746189917936694,
      "grad_norm": 12.018476486206055,
      "learning_rate": 3.244357152778007e-06,
      "loss": 9.1257,
      "step": 3819
    },
    {
      "epoch": 0.7463853067604533,
      "grad_norm": 6.693393707275391,
      "learning_rate": 3.2396923100255515e-06,
      "loss": 8.8083,
      "step": 3820
    },
    {
      "epoch": 0.7465806955842126,
      "grad_norm": 6.632091045379639,
      "learning_rate": 3.235030174986209e-06,
      "loss": 7.9511,
      "step": 3821
    },
    {
      "epoch": 0.7467760844079718,
      "grad_norm": 9.492447853088379,
      "learning_rate": 3.230370749527306e-06,
      "loss": 8.7927,
      "step": 3822
    },
    {
      "epoch": 0.7469714732317312,
      "grad_norm": 8.027215957641602,
      "learning_rate": 3.225714035515094e-06,
      "loss": 9.1603,
      "step": 3823
    },
    {
      "epoch": 0.7471668620554904,
      "grad_norm": 6.63563871383667,
      "learning_rate": 3.221060034814728e-06,
      "loss": 7.9951,
      "step": 3824
    },
    {
      "epoch": 0.7473622508792497,
      "grad_norm": 8.573164939880371,
      "learning_rate": 3.216408749290276e-06,
      "loss": 8.796,
      "step": 3825
    },
    {
      "epoch": 0.747557639703009,
      "grad_norm": 7.453122615814209,
      "learning_rate": 3.2117601808047294e-06,
      "loss": 9.077,
      "step": 3826
    },
    {
      "epoch": 0.7477530285267683,
      "grad_norm": 8.522750854492188,
      "learning_rate": 3.207114331219978e-06,
      "loss": 9.0829,
      "step": 3827
    },
    {
      "epoch": 0.7479484173505275,
      "grad_norm": 7.745052814483643,
      "learning_rate": 3.202471202396835e-06,
      "loss": 9.6801,
      "step": 3828
    },
    {
      "epoch": 0.7481438061742869,
      "grad_norm": 8.938319206237793,
      "learning_rate": 3.1978307961950107e-06,
      "loss": 9.351,
      "step": 3829
    },
    {
      "epoch": 0.7483391949980461,
      "grad_norm": 8.661677360534668,
      "learning_rate": 3.1931931144731386e-06,
      "loss": 8.8951,
      "step": 3830
    },
    {
      "epoch": 0.7485345838218054,
      "grad_norm": 9.34511947631836,
      "learning_rate": 3.188558159088748e-06,
      "loss": 8.8357,
      "step": 3831
    },
    {
      "epoch": 0.7487299726455646,
      "grad_norm": 7.346662521362305,
      "learning_rate": 3.1839259318982885e-06,
      "loss": 8.201,
      "step": 3832
    },
    {
      "epoch": 0.748925361469324,
      "grad_norm": 6.691265106201172,
      "learning_rate": 3.1792964347571053e-06,
      "loss": 9.089,
      "step": 3833
    },
    {
      "epoch": 0.7491207502930832,
      "grad_norm": 8.465272903442383,
      "learning_rate": 3.174669669519461e-06,
      "loss": 8.5986,
      "step": 3834
    },
    {
      "epoch": 0.7493161391168425,
      "grad_norm": 7.529587268829346,
      "learning_rate": 3.1700456380385124e-06,
      "loss": 9.1974,
      "step": 3835
    },
    {
      "epoch": 0.7495115279406018,
      "grad_norm": 10.19723892211914,
      "learning_rate": 3.1654243421663355e-06,
      "loss": 9.8661,
      "step": 3836
    },
    {
      "epoch": 0.7497069167643611,
      "grad_norm": 7.2738213539123535,
      "learning_rate": 3.1608057837538976e-06,
      "loss": 9.3369,
      "step": 3837
    },
    {
      "epoch": 0.7499023055881203,
      "grad_norm": 9.646924018859863,
      "learning_rate": 3.156189964651073e-06,
      "loss": 8.6084,
      "step": 3838
    },
    {
      "epoch": 0.7500976944118797,
      "grad_norm": 10.57065486907959,
      "learning_rate": 3.1515768867066464e-06,
      "loss": 9.9245,
      "step": 3839
    },
    {
      "epoch": 0.7502930832356389,
      "grad_norm": 15.623647689819336,
      "learning_rate": 3.1469665517682923e-06,
      "loss": 10.3266,
      "step": 3840
    },
    {
      "epoch": 0.7504884720593982,
      "grad_norm": 8.014330863952637,
      "learning_rate": 3.1423589616825987e-06,
      "loss": 8.9361,
      "step": 3841
    },
    {
      "epoch": 0.7506838608831575,
      "grad_norm": 6.676121234893799,
      "learning_rate": 3.1377541182950445e-06,
      "loss": 8.8935,
      "step": 3842
    },
    {
      "epoch": 0.7508792497069168,
      "grad_norm": 8.338479042053223,
      "learning_rate": 3.1331520234500155e-06,
      "loss": 8.4392,
      "step": 3843
    },
    {
      "epoch": 0.751074638530676,
      "grad_norm": 9.977602005004883,
      "learning_rate": 3.1285526789907906e-06,
      "loss": 9.0042,
      "step": 3844
    },
    {
      "epoch": 0.7512700273544354,
      "grad_norm": 7.855495452880859,
      "learning_rate": 3.123956086759553e-06,
      "loss": 8.7187,
      "step": 3845
    },
    {
      "epoch": 0.7514654161781946,
      "grad_norm": 9.590307235717773,
      "learning_rate": 3.119362248597377e-06,
      "loss": 9.383,
      "step": 3846
    },
    {
      "epoch": 0.7516608050019539,
      "grad_norm": 7.904482364654541,
      "learning_rate": 3.1147711663442427e-06,
      "loss": 8.3656,
      "step": 3847
    },
    {
      "epoch": 0.7518561938257131,
      "grad_norm": 7.658161163330078,
      "learning_rate": 3.1101828418390147e-06,
      "loss": 8.2054,
      "step": 3848
    },
    {
      "epoch": 0.7520515826494725,
      "grad_norm": 8.585644721984863,
      "learning_rate": 3.1055972769194655e-06,
      "loss": 9.3294,
      "step": 3849
    },
    {
      "epoch": 0.7522469714732317,
      "grad_norm": 7.512440204620361,
      "learning_rate": 3.1010144734222535e-06,
      "loss": 8.4882,
      "step": 3850
    },
    {
      "epoch": 0.752442360296991,
      "grad_norm": 9.79820728302002,
      "learning_rate": 3.09643443318293e-06,
      "loss": 9.4896,
      "step": 3851
    },
    {
      "epoch": 0.7526377491207503,
      "grad_norm": 21.057519912719727,
      "learning_rate": 3.0918571580359504e-06,
      "loss": 9.8927,
      "step": 3852
    },
    {
      "epoch": 0.7528331379445096,
      "grad_norm": 12.370172500610352,
      "learning_rate": 3.087282649814648e-06,
      "loss": 9.0297,
      "step": 3853
    },
    {
      "epoch": 0.7530285267682688,
      "grad_norm": 10.342479705810547,
      "learning_rate": 3.0827109103512643e-06,
      "loss": 9.0344,
      "step": 3854
    },
    {
      "epoch": 0.7532239155920282,
      "grad_norm": 10.107954978942871,
      "learning_rate": 3.0781419414769132e-06,
      "loss": 8.8846,
      "step": 3855
    },
    {
      "epoch": 0.7534193044157874,
      "grad_norm": 7.770096778869629,
      "learning_rate": 3.0735757450216177e-06,
      "loss": 9.1557,
      "step": 3856
    },
    {
      "epoch": 0.7536146932395467,
      "grad_norm": 7.809630870819092,
      "learning_rate": 3.069012322814272e-06,
      "loss": 7.8859,
      "step": 3857
    },
    {
      "epoch": 0.753810082063306,
      "grad_norm": 8.783084869384766,
      "learning_rate": 3.064451676682678e-06,
      "loss": 9.449,
      "step": 3858
    },
    {
      "epoch": 0.7540054708870653,
      "grad_norm": 7.793107032775879,
      "learning_rate": 3.059893808453508e-06,
      "loss": 9.1583,
      "step": 3859
    },
    {
      "epoch": 0.7542008597108245,
      "grad_norm": 6.793116092681885,
      "learning_rate": 3.0553387199523354e-06,
      "loss": 8.1216,
      "step": 3860
    },
    {
      "epoch": 0.7543962485345839,
      "grad_norm": 8.21915340423584,
      "learning_rate": 3.0507864130036103e-06,
      "loss": 8.977,
      "step": 3861
    },
    {
      "epoch": 0.7545916373583431,
      "grad_norm": 8.23648738861084,
      "learning_rate": 3.046236889430677e-06,
      "loss": 9.1412,
      "step": 3862
    },
    {
      "epoch": 0.7547870261821024,
      "grad_norm": 8.978711128234863,
      "learning_rate": 3.0416901510557593e-06,
      "loss": 8.8821,
      "step": 3863
    },
    {
      "epoch": 0.7549824150058617,
      "grad_norm": 8.333256721496582,
      "learning_rate": 3.037146199699964e-06,
      "loss": 9.7569,
      "step": 3864
    },
    {
      "epoch": 0.755177803829621,
      "grad_norm": 6.864106178283691,
      "learning_rate": 3.0326050371832894e-06,
      "loss": 8.6864,
      "step": 3865
    },
    {
      "epoch": 0.7553731926533802,
      "grad_norm": 8.051876068115234,
      "learning_rate": 3.028066665324607e-06,
      "loss": 9.2071,
      "step": 3866
    },
    {
      "epoch": 0.7555685814771395,
      "grad_norm": 6.9381794929504395,
      "learning_rate": 3.023531085941682e-06,
      "loss": 9.3013,
      "step": 3867
    },
    {
      "epoch": 0.7557639703008988,
      "grad_norm": 9.048749923706055,
      "learning_rate": 3.0189983008511483e-06,
      "loss": 9.4292,
      "step": 3868
    },
    {
      "epoch": 0.7559593591246581,
      "grad_norm": 8.086410522460938,
      "learning_rate": 3.0144683118685336e-06,
      "loss": 9.2434,
      "step": 3869
    },
    {
      "epoch": 0.7561547479484173,
      "grad_norm": 7.529863357543945,
      "learning_rate": 3.009941120808232e-06,
      "loss": 9.6011,
      "step": 3870
    },
    {
      "epoch": 0.7563501367721767,
      "grad_norm": 8.00220775604248,
      "learning_rate": 3.0054167294835314e-06,
      "loss": 8.9942,
      "step": 3871
    },
    {
      "epoch": 0.7565455255959359,
      "grad_norm": 8.991710662841797,
      "learning_rate": 3.0008951397065832e-06,
      "loss": 9.8394,
      "step": 3872
    },
    {
      "epoch": 0.7567409144196952,
      "grad_norm": 7.8540472984313965,
      "learning_rate": 2.996376353288433e-06,
      "loss": 9.798,
      "step": 3873
    },
    {
      "epoch": 0.7569363032434545,
      "grad_norm": 9.14909839630127,
      "learning_rate": 2.991860372038986e-06,
      "loss": 9.1226,
      "step": 3874
    },
    {
      "epoch": 0.7571316920672138,
      "grad_norm": 8.964995384216309,
      "learning_rate": 2.987347197767041e-06,
      "loss": 8.612,
      "step": 3875
    },
    {
      "epoch": 0.757327080890973,
      "grad_norm": 9.87065601348877,
      "learning_rate": 2.9828368322802614e-06,
      "loss": 8.6584,
      "step": 3876
    },
    {
      "epoch": 0.7575224697147324,
      "grad_norm": 8.337159156799316,
      "learning_rate": 2.978329277385186e-06,
      "loss": 9.5367,
      "step": 3877
    },
    {
      "epoch": 0.7577178585384916,
      "grad_norm": 22.911712646484375,
      "learning_rate": 2.973824534887234e-06,
      "loss": 9.3709,
      "step": 3878
    },
    {
      "epoch": 0.7579132473622509,
      "grad_norm": 7.285062313079834,
      "learning_rate": 2.969322606590691e-06,
      "loss": 8.3937,
      "step": 3879
    },
    {
      "epoch": 0.7581086361860102,
      "grad_norm": 8.002433776855469,
      "learning_rate": 2.9648234942987242e-06,
      "loss": 8.646,
      "step": 3880
    },
    {
      "epoch": 0.7583040250097695,
      "grad_norm": 9.95106315612793,
      "learning_rate": 2.960327199813362e-06,
      "loss": 9.9894,
      "step": 3881
    },
    {
      "epoch": 0.7584994138335287,
      "grad_norm": 8.543545722961426,
      "learning_rate": 2.955833724935516e-06,
      "loss": 8.9347,
      "step": 3882
    },
    {
      "epoch": 0.758694802657288,
      "grad_norm": 24.709081649780273,
      "learning_rate": 2.951343071464956e-06,
      "loss": 9.0592,
      "step": 3883
    },
    {
      "epoch": 0.7588901914810473,
      "grad_norm": 6.559577941894531,
      "learning_rate": 2.946855241200335e-06,
      "loss": 7.936,
      "step": 3884
    },
    {
      "epoch": 0.7590855803048066,
      "grad_norm": 9.748425483703613,
      "learning_rate": 2.94237023593916e-06,
      "loss": 9.623,
      "step": 3885
    },
    {
      "epoch": 0.7592809691285658,
      "grad_norm": 7.7878336906433105,
      "learning_rate": 2.9378880574778245e-06,
      "loss": 9.0149,
      "step": 3886
    },
    {
      "epoch": 0.7594763579523252,
      "grad_norm": 6.624049663543701,
      "learning_rate": 2.9334087076115714e-06,
      "loss": 8.8067,
      "step": 3887
    },
    {
      "epoch": 0.7596717467760844,
      "grad_norm": 7.3495354652404785,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 8.3652,
      "step": 3888
    },
    {
      "epoch": 0.7598671355998436,
      "grad_norm": 8.031523704528809,
      "learning_rate": 2.9244585008396696e-06,
      "loss": 8.8399,
      "step": 3889
    },
    {
      "epoch": 0.760062524423603,
      "grad_norm": 8.215229034423828,
      "learning_rate": 2.9199876475188528e-06,
      "loss": 8.9035,
      "step": 3890
    },
    {
      "epoch": 0.7602579132473622,
      "grad_norm": 9.209456443786621,
      "learning_rate": 2.915519629962793e-06,
      "loss": 9.2322,
      "step": 3891
    },
    {
      "epoch": 0.7604533020711215,
      "grad_norm": 7.7869873046875,
      "learning_rate": 2.911054449961067e-06,
      "loss": 8.837,
      "step": 3892
    },
    {
      "epoch": 0.7606486908948809,
      "grad_norm": 8.22655963897705,
      "learning_rate": 2.906592109302123e-06,
      "loss": 9.6905,
      "step": 3893
    },
    {
      "epoch": 0.7608440797186401,
      "grad_norm": 10.398638725280762,
      "learning_rate": 2.9021326097732616e-06,
      "loss": 8.9069,
      "step": 3894
    },
    {
      "epoch": 0.7610394685423993,
      "grad_norm": 8.627163887023926,
      "learning_rate": 2.8976759531606556e-06,
      "loss": 9.115,
      "step": 3895
    },
    {
      "epoch": 0.7612348573661587,
      "grad_norm": 10.054001808166504,
      "learning_rate": 2.893222141249329e-06,
      "loss": 8.7211,
      "step": 3896
    },
    {
      "epoch": 0.761430246189918,
      "grad_norm": 8.275874137878418,
      "learning_rate": 2.8887711758231783e-06,
      "loss": 8.9959,
      "step": 3897
    },
    {
      "epoch": 0.7616256350136772,
      "grad_norm": 8.254229545593262,
      "learning_rate": 2.8843230586649463e-06,
      "loss": 9.1554,
      "step": 3898
    },
    {
      "epoch": 0.7618210238374366,
      "grad_norm": 8.058757781982422,
      "learning_rate": 2.879877791556248e-06,
      "loss": 8.6234,
      "step": 3899
    },
    {
      "epoch": 0.7620164126611958,
      "grad_norm": 8.41688060760498,
      "learning_rate": 2.8754353762775465e-06,
      "loss": 9.6577,
      "step": 3900
    },
    {
      "epoch": 0.762211801484955,
      "grad_norm": 50.58344268798828,
      "learning_rate": 2.8709958146081717e-06,
      "loss": 9.4878,
      "step": 3901
    },
    {
      "epoch": 0.7624071903087143,
      "grad_norm": 9.422719955444336,
      "learning_rate": 2.866559108326302e-06,
      "loss": 9.6191,
      "step": 3902
    },
    {
      "epoch": 0.7626025791324736,
      "grad_norm": 7.204468250274658,
      "learning_rate": 2.8621252592089752e-06,
      "loss": 8.2942,
      "step": 3903
    },
    {
      "epoch": 0.7627979679562329,
      "grad_norm": 7.632490158081055,
      "learning_rate": 2.8576942690320907e-06,
      "loss": 8.7558,
      "step": 3904
    },
    {
      "epoch": 0.7629933567799921,
      "grad_norm": 7.333845138549805,
      "learning_rate": 2.853266139570391e-06,
      "loss": 8.8135,
      "step": 3905
    },
    {
      "epoch": 0.7631887456037515,
      "grad_norm": 6.49276065826416,
      "learning_rate": 2.848840872597485e-06,
      "loss": 7.6965,
      "step": 3906
    },
    {
      "epoch": 0.7633841344275107,
      "grad_norm": 7.868816375732422,
      "learning_rate": 2.8444184698858247e-06,
      "loss": 7.6861,
      "step": 3907
    },
    {
      "epoch": 0.76357952325127,
      "grad_norm": 9.188851356506348,
      "learning_rate": 2.8399989332067258e-06,
      "loss": 8.7012,
      "step": 3908
    },
    {
      "epoch": 0.7637749120750293,
      "grad_norm": 9.501800537109375,
      "learning_rate": 2.8355822643303433e-06,
      "loss": 8.4651,
      "step": 3909
    },
    {
      "epoch": 0.7639703008987886,
      "grad_norm": 8.46778678894043,
      "learning_rate": 2.8311684650256964e-06,
      "loss": 7.8761,
      "step": 3910
    },
    {
      "epoch": 0.7641656897225478,
      "grad_norm": 9.765848159790039,
      "learning_rate": 2.8267575370606427e-06,
      "loss": 9.5767,
      "step": 3911
    },
    {
      "epoch": 0.7643610785463072,
      "grad_norm": 11.269289016723633,
      "learning_rate": 2.8223494822019015e-06,
      "loss": 9.2451,
      "step": 3912
    },
    {
      "epoch": 0.7645564673700664,
      "grad_norm": 8.891103744506836,
      "learning_rate": 2.81794430221503e-06,
      "loss": 9.3999,
      "step": 3913
    },
    {
      "epoch": 0.7647518561938257,
      "grad_norm": 9.411876678466797,
      "learning_rate": 2.8135419988644453e-06,
      "loss": 9.6113,
      "step": 3914
    },
    {
      "epoch": 0.764947245017585,
      "grad_norm": 8.9310884475708,
      "learning_rate": 2.8091425739134036e-06,
      "loss": 9.8384,
      "step": 3915
    },
    {
      "epoch": 0.7651426338413443,
      "grad_norm": 9.01473617553711,
      "learning_rate": 2.804746029124009e-06,
      "loss": 8.0194,
      "step": 3916
    },
    {
      "epoch": 0.7653380226651035,
      "grad_norm": 9.225128173828125,
      "learning_rate": 2.8003523662572195e-06,
      "loss": 8.8305,
      "step": 3917
    },
    {
      "epoch": 0.7655334114888629,
      "grad_norm": 13.47047233581543,
      "learning_rate": 2.7959615870728276e-06,
      "loss": 9.7281,
      "step": 3918
    },
    {
      "epoch": 0.7657288003126221,
      "grad_norm": 9.079336166381836,
      "learning_rate": 2.7915736933294813e-06,
      "loss": 9.0856,
      "step": 3919
    },
    {
      "epoch": 0.7659241891363814,
      "grad_norm": 9.223864555358887,
      "learning_rate": 2.7871886867846642e-06,
      "loss": 9.1781,
      "step": 3920
    },
    {
      "epoch": 0.7661195779601406,
      "grad_norm": 8.404414176940918,
      "learning_rate": 2.7828065691947137e-06,
      "loss": 9.6851,
      "step": 3921
    },
    {
      "epoch": 0.7663149667839,
      "grad_norm": 6.931270122528076,
      "learning_rate": 2.7784273423147966e-06,
      "loss": 8.5956,
      "step": 3922
    },
    {
      "epoch": 0.7665103556076592,
      "grad_norm": 9.345423698425293,
      "learning_rate": 2.774051007898936e-06,
      "loss": 9.2563,
      "step": 3923
    },
    {
      "epoch": 0.7667057444314185,
      "grad_norm": 39.78207015991211,
      "learning_rate": 2.7696775676999833e-06,
      "loss": 9.5094,
      "step": 3924
    },
    {
      "epoch": 0.7669011332551778,
      "grad_norm": 9.660195350646973,
      "learning_rate": 2.765307023469643e-06,
      "loss": 9.2362,
      "step": 3925
    },
    {
      "epoch": 0.7670965220789371,
      "grad_norm": 13.873888969421387,
      "learning_rate": 2.760939376958448e-06,
      "loss": 8.566,
      "step": 3926
    },
    {
      "epoch": 0.7672919109026963,
      "grad_norm": 6.7762651443481445,
      "learning_rate": 2.7565746299157826e-06,
      "loss": 9.5676,
      "step": 3927
    },
    {
      "epoch": 0.7674872997264557,
      "grad_norm": 10.185317039489746,
      "learning_rate": 2.7522127840898603e-06,
      "loss": 9.5498,
      "step": 3928
    },
    {
      "epoch": 0.7676826885502149,
      "grad_norm": 6.987491130828857,
      "learning_rate": 2.747853841227732e-06,
      "loss": 8.372,
      "step": 3929
    },
    {
      "epoch": 0.7678780773739742,
      "grad_norm": 8.690187454223633,
      "learning_rate": 2.743497803075298e-06,
      "loss": 9.0992,
      "step": 3930
    },
    {
      "epoch": 0.7680734661977335,
      "grad_norm": 7.142638683319092,
      "learning_rate": 2.7391446713772775e-06,
      "loss": 8.3566,
      "step": 3931
    },
    {
      "epoch": 0.7682688550214928,
      "grad_norm": 7.855913162231445,
      "learning_rate": 2.7347944478772436e-06,
      "loss": 9.3046,
      "step": 3932
    },
    {
      "epoch": 0.768464243845252,
      "grad_norm": 10.175846099853516,
      "learning_rate": 2.7304471343175876e-06,
      "loss": 9.1877,
      "step": 3933
    },
    {
      "epoch": 0.7686596326690114,
      "grad_norm": 8.158573150634766,
      "learning_rate": 2.726102732439552e-06,
      "loss": 8.2,
      "step": 3934
    },
    {
      "epoch": 0.7688550214927706,
      "grad_norm": 13.254216194152832,
      "learning_rate": 2.7217612439831974e-06,
      "loss": 9.6704,
      "step": 3935
    },
    {
      "epoch": 0.7690504103165299,
      "grad_norm": 12.06765365600586,
      "learning_rate": 2.71742267068743e-06,
      "loss": 9.2111,
      "step": 3936
    },
    {
      "epoch": 0.7692457991402891,
      "grad_norm": 7.126428127288818,
      "learning_rate": 2.713087014289978e-06,
      "loss": 8.5633,
      "step": 3937
    },
    {
      "epoch": 0.7694411879640485,
      "grad_norm": 8.78329849243164,
      "learning_rate": 2.7087542765274135e-06,
      "loss": 8.7111,
      "step": 3938
    },
    {
      "epoch": 0.7696365767878077,
      "grad_norm": 7.155397415161133,
      "learning_rate": 2.704424459135123e-06,
      "loss": 8.6173,
      "step": 3939
    },
    {
      "epoch": 0.769831965611567,
      "grad_norm": 8.362483978271484,
      "learning_rate": 2.7000975638473425e-06,
      "loss": 9.1443,
      "step": 3940
    },
    {
      "epoch": 0.7700273544353263,
      "grad_norm": 10.86987590789795,
      "learning_rate": 2.695773592397123e-06,
      "loss": 9.1629,
      "step": 3941
    },
    {
      "epoch": 0.7702227432590856,
      "grad_norm": 9.431424140930176,
      "learning_rate": 2.6914525465163466e-06,
      "loss": 7.8835,
      "step": 3942
    },
    {
      "epoch": 0.7704181320828448,
      "grad_norm": 12.760275840759277,
      "learning_rate": 2.6871344279357335e-06,
      "loss": 8.8893,
      "step": 3943
    },
    {
      "epoch": 0.7706135209066042,
      "grad_norm": 13.065872192382812,
      "learning_rate": 2.682819238384816e-06,
      "loss": 9.2098,
      "step": 3944
    },
    {
      "epoch": 0.7708089097303634,
      "grad_norm": 7.924666404724121,
      "learning_rate": 2.678506979591969e-06,
      "loss": 9.6815,
      "step": 3945
    },
    {
      "epoch": 0.7710042985541227,
      "grad_norm": 8.195263862609863,
      "learning_rate": 2.6741976532843795e-06,
      "loss": 8.47,
      "step": 3946
    },
    {
      "epoch": 0.771199687377882,
      "grad_norm": 8.898924827575684,
      "learning_rate": 2.6698912611880735e-06,
      "loss": 8.9015,
      "step": 3947
    },
    {
      "epoch": 0.7713950762016413,
      "grad_norm": 6.993349075317383,
      "learning_rate": 2.6655878050278874e-06,
      "loss": 8.9547,
      "step": 3948
    },
    {
      "epoch": 0.7715904650254005,
      "grad_norm": 8.768937110900879,
      "learning_rate": 2.6612872865274954e-06,
      "loss": 9.3573,
      "step": 3949
    },
    {
      "epoch": 0.7717858538491599,
      "grad_norm": 7.907637119293213,
      "learning_rate": 2.656989707409383e-06,
      "loss": 8.6754,
      "step": 3950
    },
    {
      "epoch": 0.7719812426729191,
      "grad_norm": 7.222424507141113,
      "learning_rate": 2.6526950693948692e-06,
      "loss": 9.0668,
      "step": 3951
    },
    {
      "epoch": 0.7721766314966784,
      "grad_norm": 7.868619918823242,
      "learning_rate": 2.648403374204086e-06,
      "loss": 9.5197,
      "step": 3952
    },
    {
      "epoch": 0.7723720203204377,
      "grad_norm": 8.713985443115234,
      "learning_rate": 2.6441146235559933e-06,
      "loss": 9.4808,
      "step": 3953
    },
    {
      "epoch": 0.772567409144197,
      "grad_norm": 6.500898361206055,
      "learning_rate": 2.6398288191683687e-06,
      "loss": 7.9375,
      "step": 3954
    },
    {
      "epoch": 0.7727627979679562,
      "grad_norm": 8.31683349609375,
      "learning_rate": 2.635545962757806e-06,
      "loss": 8.9627,
      "step": 3955
    },
    {
      "epoch": 0.7729581867917155,
      "grad_norm": 6.714325904846191,
      "learning_rate": 2.6312660560397273e-06,
      "loss": 9.4997,
      "step": 3956
    },
    {
      "epoch": 0.7731535756154748,
      "grad_norm": 9.260298728942871,
      "learning_rate": 2.6269891007283664e-06,
      "loss": 9.3548,
      "step": 3957
    },
    {
      "epoch": 0.7733489644392341,
      "grad_norm": 8.693593978881836,
      "learning_rate": 2.6227150985367735e-06,
      "loss": 10.1707,
      "step": 3958
    },
    {
      "epoch": 0.7735443532629933,
      "grad_norm": 7.7326765060424805,
      "learning_rate": 2.618444051176824e-06,
      "loss": 8.2939,
      "step": 3959
    },
    {
      "epoch": 0.7737397420867527,
      "grad_norm": 10.119888305664062,
      "learning_rate": 2.6141759603592033e-06,
      "loss": 9.2571,
      "step": 3960
    },
    {
      "epoch": 0.7739351309105119,
      "grad_norm": 10.321942329406738,
      "learning_rate": 2.6099108277934105e-06,
      "loss": 9.8407,
      "step": 3961
    },
    {
      "epoch": 0.7741305197342712,
      "grad_norm": 11.165168762207031,
      "learning_rate": 2.6056486551877704e-06,
      "loss": 8.3103,
      "step": 3962
    },
    {
      "epoch": 0.7743259085580305,
      "grad_norm": 10.171852111816406,
      "learning_rate": 2.601389444249408e-06,
      "loss": 9.9649,
      "step": 3963
    },
    {
      "epoch": 0.7745212973817898,
      "grad_norm": 10.053415298461914,
      "learning_rate": 2.597133196684277e-06,
      "loss": 9.0758,
      "step": 3964
    },
    {
      "epoch": 0.774716686205549,
      "grad_norm": 9.017916679382324,
      "learning_rate": 2.592879914197131e-06,
      "loss": 8.4977,
      "step": 3965
    },
    {
      "epoch": 0.7749120750293084,
      "grad_norm": 26.458412170410156,
      "learning_rate": 2.5886295984915456e-06,
      "loss": 9.6598,
      "step": 3966
    },
    {
      "epoch": 0.7751074638530676,
      "grad_norm": 9.767171859741211,
      "learning_rate": 2.584382251269899e-06,
      "loss": 8.8856,
      "step": 3967
    },
    {
      "epoch": 0.7753028526768269,
      "grad_norm": 8.08504867553711,
      "learning_rate": 2.5801378742333928e-06,
      "loss": 8.772,
      "step": 3968
    },
    {
      "epoch": 0.7754982415005862,
      "grad_norm": 11.8588228225708,
      "learning_rate": 2.5758964690820254e-06,
      "loss": 8.7757,
      "step": 3969
    },
    {
      "epoch": 0.7756936303243455,
      "grad_norm": 7.2053985595703125,
      "learning_rate": 2.571658037514616e-06,
      "loss": 9.1364,
      "step": 3970
    },
    {
      "epoch": 0.7758890191481047,
      "grad_norm": 8.632712364196777,
      "learning_rate": 2.5674225812287834e-06,
      "loss": 8.9427,
      "step": 3971
    },
    {
      "epoch": 0.776084407971864,
      "grad_norm": 7.743727207183838,
      "learning_rate": 2.563190101920965e-06,
      "loss": 9.5031,
      "step": 3972
    },
    {
      "epoch": 0.7762797967956233,
      "grad_norm": 9.737512588500977,
      "learning_rate": 2.5589606012863968e-06,
      "loss": 7.5548,
      "step": 3973
    },
    {
      "epoch": 0.7764751856193826,
      "grad_norm": 9.05333423614502,
      "learning_rate": 2.5547340810191223e-06,
      "loss": 9.6562,
      "step": 3974
    },
    {
      "epoch": 0.7766705744431418,
      "grad_norm": 7.593080997467041,
      "learning_rate": 2.550510542811999e-06,
      "loss": 8.6516,
      "step": 3975
    },
    {
      "epoch": 0.7768659632669012,
      "grad_norm": 9.689908027648926,
      "learning_rate": 2.5462899883566815e-06,
      "loss": 9.416,
      "step": 3976
    },
    {
      "epoch": 0.7770613520906604,
      "grad_norm": 12.020401000976562,
      "learning_rate": 2.5420724193436377e-06,
      "loss": 10.2451,
      "step": 3977
    },
    {
      "epoch": 0.7772567409144197,
      "grad_norm": 8.075044631958008,
      "learning_rate": 2.5378578374621286e-06,
      "loss": 8.8815,
      "step": 3978
    },
    {
      "epoch": 0.777452129738179,
      "grad_norm": 9.765203475952148,
      "learning_rate": 2.533646244400231e-06,
      "loss": 8.5774,
      "step": 3979
    },
    {
      "epoch": 0.7776475185619383,
      "grad_norm": 6.740596771240234,
      "learning_rate": 2.529437641844813e-06,
      "loss": 8.7272,
      "step": 3980
    },
    {
      "epoch": 0.7778429073856975,
      "grad_norm": 8.368038177490234,
      "learning_rate": 2.525232031481557e-06,
      "loss": 8.9325,
      "step": 3981
    },
    {
      "epoch": 0.7780382962094569,
      "grad_norm": 9.283183097839355,
      "learning_rate": 2.5210294149949332e-06,
      "loss": 10.0498,
      "step": 3982
    },
    {
      "epoch": 0.7782336850332161,
      "grad_norm": 8.498453140258789,
      "learning_rate": 2.5168297940682273e-06,
      "loss": 8.8612,
      "step": 3983
    },
    {
      "epoch": 0.7784290738569754,
      "grad_norm": 8.46799087524414,
      "learning_rate": 2.5126331703835117e-06,
      "loss": 8.133,
      "step": 3984
    },
    {
      "epoch": 0.7786244626807347,
      "grad_norm": 6.731637477874756,
      "learning_rate": 2.5084395456216694e-06,
      "loss": 8.3384,
      "step": 3985
    },
    {
      "epoch": 0.778819851504494,
      "grad_norm": 10.091724395751953,
      "learning_rate": 2.5042489214623754e-06,
      "loss": 9.0973,
      "step": 3986
    },
    {
      "epoch": 0.7790152403282532,
      "grad_norm": 7.505545139312744,
      "learning_rate": 2.5000612995841024e-06,
      "loss": 8.7713,
      "step": 3987
    },
    {
      "epoch": 0.7792106291520126,
      "grad_norm": 9.824227333068848,
      "learning_rate": 2.4958766816641265e-06,
      "loss": 8.9991,
      "step": 3988
    },
    {
      "epoch": 0.7794060179757718,
      "grad_norm": 7.5289692878723145,
      "learning_rate": 2.4916950693785136e-06,
      "loss": 9.1619,
      "step": 3989
    },
    {
      "epoch": 0.779601406799531,
      "grad_norm": 17.181425094604492,
      "learning_rate": 2.4875164644021343e-06,
      "loss": 9.0135,
      "step": 3990
    },
    {
      "epoch": 0.7797967956232903,
      "grad_norm": 7.876916408538818,
      "learning_rate": 2.4833408684086434e-06,
      "loss": 7.9555,
      "step": 3991
    },
    {
      "epoch": 0.7799921844470497,
      "grad_norm": 7.679806709289551,
      "learning_rate": 2.4791682830705034e-06,
      "loss": 9.2439,
      "step": 3992
    },
    {
      "epoch": 0.7801875732708089,
      "grad_norm": 6.449591636657715,
      "learning_rate": 2.474998710058958e-06,
      "loss": 8.5342,
      "step": 3993
    },
    {
      "epoch": 0.7803829620945681,
      "grad_norm": 7.2947797775268555,
      "learning_rate": 2.470832151044056e-06,
      "loss": 8.8409,
      "step": 3994
    },
    {
      "epoch": 0.7805783509183275,
      "grad_norm": 28.776031494140625,
      "learning_rate": 2.4666686076946278e-06,
      "loss": 9.1977,
      "step": 3995
    },
    {
      "epoch": 0.7807737397420867,
      "grad_norm": 39.56930160522461,
      "learning_rate": 2.4625080816783075e-06,
      "loss": 9.3758,
      "step": 3996
    },
    {
      "epoch": 0.780969128565846,
      "grad_norm": 8.051578521728516,
      "learning_rate": 2.4583505746615112e-06,
      "loss": 8.1979,
      "step": 3997
    },
    {
      "epoch": 0.7811645173896054,
      "grad_norm": 6.23033332824707,
      "learning_rate": 2.454196088309453e-06,
      "loss": 8.2484,
      "step": 3998
    },
    {
      "epoch": 0.7813599062133646,
      "grad_norm": 8.523757934570312,
      "learning_rate": 2.450044624286131e-06,
      "loss": 9.0294,
      "step": 3999
    },
    {
      "epoch": 0.7815552950371238,
      "grad_norm": 11.340201377868652,
      "learning_rate": 2.445896184254334e-06,
      "loss": 9.384,
      "step": 4000
    },
    {
      "epoch": 0.7817506838608832,
      "grad_norm": 9.148540496826172,
      "learning_rate": 2.441750769875645e-06,
      "loss": 9.3063,
      "step": 4001
    },
    {
      "epoch": 0.7819460726846424,
      "grad_norm": 9.013167381286621,
      "learning_rate": 2.437608382810427e-06,
      "loss": 8.689,
      "step": 4002
    },
    {
      "epoch": 0.7821414615084017,
      "grad_norm": 8.351414680480957,
      "learning_rate": 2.4334690247178396e-06,
      "loss": 8.9728,
      "step": 4003
    },
    {
      "epoch": 0.782336850332161,
      "grad_norm": 12.632786750793457,
      "learning_rate": 2.429332697255817e-06,
      "loss": 9.2797,
      "step": 4004
    },
    {
      "epoch": 0.7825322391559203,
      "grad_norm": 8.268104553222656,
      "learning_rate": 2.425199402081095e-06,
      "loss": 8.5112,
      "step": 4005
    },
    {
      "epoch": 0.7827276279796795,
      "grad_norm": 6.804723262786865,
      "learning_rate": 2.4210691408491805e-06,
      "loss": 9.2314,
      "step": 4006
    },
    {
      "epoch": 0.7829230168034389,
      "grad_norm": 6.3445916175842285,
      "learning_rate": 2.416941915214377e-06,
      "loss": 8.6576,
      "step": 4007
    },
    {
      "epoch": 0.7831184056271981,
      "grad_norm": 16.956613540649414,
      "learning_rate": 2.412817726829759e-06,
      "loss": 9.0176,
      "step": 4008
    },
    {
      "epoch": 0.7833137944509574,
      "grad_norm": 8.851213455200195,
      "learning_rate": 2.4086965773471993e-06,
      "loss": 8.9511,
      "step": 4009
    },
    {
      "epoch": 0.7835091832747166,
      "grad_norm": 8.128167152404785,
      "learning_rate": 2.40457846841734e-06,
      "loss": 9.7655,
      "step": 4010
    },
    {
      "epoch": 0.783704572098476,
      "grad_norm": 7.9139533042907715,
      "learning_rate": 2.4004634016896176e-06,
      "loss": 8.9192,
      "step": 4011
    },
    {
      "epoch": 0.7838999609222352,
      "grad_norm": 23.174137115478516,
      "learning_rate": 2.3963513788122416e-06,
      "loss": 9.2223,
      "step": 4012
    },
    {
      "epoch": 0.7840953497459945,
      "grad_norm": 9.148961067199707,
      "learning_rate": 2.392242401432202e-06,
      "loss": 8.3294,
      "step": 4013
    },
    {
      "epoch": 0.7842907385697538,
      "grad_norm": 10.631275177001953,
      "learning_rate": 2.3881364711952757e-06,
      "loss": 8.4833,
      "step": 4014
    },
    {
      "epoch": 0.7844861273935131,
      "grad_norm": 8.675843238830566,
      "learning_rate": 2.384033589746012e-06,
      "loss": 9.5845,
      "step": 4015
    },
    {
      "epoch": 0.7846815162172723,
      "grad_norm": 10.582771301269531,
      "learning_rate": 2.379933758727746e-06,
      "loss": 8.7382,
      "step": 4016
    },
    {
      "epoch": 0.7848769050410317,
      "grad_norm": 30.143577575683594,
      "learning_rate": 2.375836979782582e-06,
      "loss": 8.6157,
      "step": 4017
    },
    {
      "epoch": 0.7850722938647909,
      "grad_norm": 10.265090942382812,
      "learning_rate": 2.371743254551414e-06,
      "loss": 9.9724,
      "step": 4018
    },
    {
      "epoch": 0.7852676826885502,
      "grad_norm": 9.800621032714844,
      "learning_rate": 2.3676525846738996e-06,
      "loss": 8.9021,
      "step": 4019
    },
    {
      "epoch": 0.7854630715123095,
      "grad_norm": 7.986928939819336,
      "learning_rate": 2.363564971788486e-06,
      "loss": 8.969,
      "step": 4020
    },
    {
      "epoch": 0.7856584603360688,
      "grad_norm": 9.971089363098145,
      "learning_rate": 2.3594804175323806e-06,
      "loss": 9.3363,
      "step": 4021
    },
    {
      "epoch": 0.785853849159828,
      "grad_norm": 10.663592338562012,
      "learning_rate": 2.355398923541582e-06,
      "loss": 8.7179,
      "step": 4022
    },
    {
      "epoch": 0.7860492379835874,
      "grad_norm": 10.637864112854004,
      "learning_rate": 2.351320491450849e-06,
      "loss": 9.0681,
      "step": 4023
    },
    {
      "epoch": 0.7862446268073466,
      "grad_norm": 13.780783653259277,
      "learning_rate": 2.3472451228937254e-06,
      "loss": 9.7322,
      "step": 4024
    },
    {
      "epoch": 0.7864400156311059,
      "grad_norm": 7.653302192687988,
      "learning_rate": 2.343172819502522e-06,
      "loss": 9.7462,
      "step": 4025
    },
    {
      "epoch": 0.7866354044548651,
      "grad_norm": 18.18355941772461,
      "learning_rate": 2.3391035829083173e-06,
      "loss": 8.2701,
      "step": 4026
    },
    {
      "epoch": 0.7868307932786245,
      "grad_norm": 8.42194938659668,
      "learning_rate": 2.3350374147409748e-06,
      "loss": 8.2877,
      "step": 4027
    },
    {
      "epoch": 0.7870261821023837,
      "grad_norm": 9.77413558959961,
      "learning_rate": 2.330974316629113e-06,
      "loss": 8.9961,
      "step": 4028
    },
    {
      "epoch": 0.787221570926143,
      "grad_norm": 9.971951484680176,
      "learning_rate": 2.3269142902001374e-06,
      "loss": 8.6043,
      "step": 4029
    },
    {
      "epoch": 0.7874169597499023,
      "grad_norm": 6.719539642333984,
      "learning_rate": 2.3228573370802064e-06,
      "loss": 8.1629,
      "step": 4030
    },
    {
      "epoch": 0.7876123485736616,
      "grad_norm": 7.909749507904053,
      "learning_rate": 2.3188034588942633e-06,
      "loss": 9.08,
      "step": 4031
    },
    {
      "epoch": 0.7878077373974208,
      "grad_norm": 8.593505859375,
      "learning_rate": 2.314752657266005e-06,
      "loss": 8.6831,
      "step": 4032
    },
    {
      "epoch": 0.7880031262211802,
      "grad_norm": 7.045526504516602,
      "learning_rate": 2.31070493381791e-06,
      "loss": 7.8853,
      "step": 4033
    },
    {
      "epoch": 0.7881985150449394,
      "grad_norm": 8.301514625549316,
      "learning_rate": 2.306660290171211e-06,
      "loss": 8.8027,
      "step": 4034
    },
    {
      "epoch": 0.7883939038686987,
      "grad_norm": 8.381290435791016,
      "learning_rate": 2.302618727945919e-06,
      "loss": 8.9424,
      "step": 4035
    },
    {
      "epoch": 0.788589292692458,
      "grad_norm": 10.212125778198242,
      "learning_rate": 2.298580248760799e-06,
      "loss": 9.5702,
      "step": 4036
    },
    {
      "epoch": 0.7887846815162173,
      "grad_norm": 11.792108535766602,
      "learning_rate": 2.294544854233395e-06,
      "loss": 9.9829,
      "step": 4037
    },
    {
      "epoch": 0.7889800703399765,
      "grad_norm": 9.178725242614746,
      "learning_rate": 2.2905125459800027e-06,
      "loss": 9.7524,
      "step": 4038
    },
    {
      "epoch": 0.7891754591637359,
      "grad_norm": 10.572028160095215,
      "learning_rate": 2.286483325615685e-06,
      "loss": 9.0691,
      "step": 4039
    },
    {
      "epoch": 0.7893708479874951,
      "grad_norm": 7.9066338539123535,
      "learning_rate": 2.282457194754275e-06,
      "loss": 8.8206,
      "step": 4040
    },
    {
      "epoch": 0.7895662368112544,
      "grad_norm": 7.298490524291992,
      "learning_rate": 2.2784341550083577e-06,
      "loss": 8.3582,
      "step": 4041
    },
    {
      "epoch": 0.7897616256350137,
      "grad_norm": 11.32341194152832,
      "learning_rate": 2.2744142079892906e-06,
      "loss": 9.8477,
      "step": 4042
    },
    {
      "epoch": 0.789957014458773,
      "grad_norm": 7.7698845863342285,
      "learning_rate": 2.2703973553071834e-06,
      "loss": 8.897,
      "step": 4043
    },
    {
      "epoch": 0.7901524032825322,
      "grad_norm": 8.147603988647461,
      "learning_rate": 2.2663835985709138e-06,
      "loss": 9.0695,
      "step": 4044
    },
    {
      "epoch": 0.7903477921062915,
      "grad_norm": 7.221194267272949,
      "learning_rate": 2.2623729393881113e-06,
      "loss": 9.3573,
      "step": 4045
    },
    {
      "epoch": 0.7905431809300508,
      "grad_norm": 7.137025356292725,
      "learning_rate": 2.258365379365174e-06,
      "loss": 8.1276,
      "step": 4046
    },
    {
      "epoch": 0.7907385697538101,
      "grad_norm": 8.200143814086914,
      "learning_rate": 2.25436092010725e-06,
      "loss": 9.598,
      "step": 4047
    },
    {
      "epoch": 0.7909339585775693,
      "grad_norm": 8.565221786499023,
      "learning_rate": 2.250359563218255e-06,
      "loss": 8.8693,
      "step": 4048
    },
    {
      "epoch": 0.7911293474013287,
      "grad_norm": 12.058979034423828,
      "learning_rate": 2.24636131030085e-06,
      "loss": 9.9144,
      "step": 4049
    },
    {
      "epoch": 0.7913247362250879,
      "grad_norm": 9.160593032836914,
      "learning_rate": 2.242366162956465e-06,
      "loss": 9.7148,
      "step": 4050
    },
    {
      "epoch": 0.7915201250488472,
      "grad_norm": 11.761293411254883,
      "learning_rate": 2.2383741227852796e-06,
      "loss": 8.9634,
      "step": 4051
    },
    {
      "epoch": 0.7917155138726065,
      "grad_norm": 8.902557373046875,
      "learning_rate": 2.2343851913862258e-06,
      "loss": 8.9805,
      "step": 4052
    },
    {
      "epoch": 0.7919109026963658,
      "grad_norm": 12.234577178955078,
      "learning_rate": 2.2303993703569993e-06,
      "loss": 9.676,
      "step": 4053
    },
    {
      "epoch": 0.792106291520125,
      "grad_norm": 29.982467651367188,
      "learning_rate": 2.226416661294041e-06,
      "loss": 8.1496,
      "step": 4054
    },
    {
      "epoch": 0.7923016803438844,
      "grad_norm": 12.731390953063965,
      "learning_rate": 2.222437065792554e-06,
      "loss": 8.9748,
      "step": 4055
    },
    {
      "epoch": 0.7924970691676436,
      "grad_norm": 7.898471832275391,
      "learning_rate": 2.218460585446486e-06,
      "loss": 8.7759,
      "step": 4056
    },
    {
      "epoch": 0.7926924579914029,
      "grad_norm": 8.303677558898926,
      "learning_rate": 2.2144872218485435e-06,
      "loss": 8.8412,
      "step": 4057
    },
    {
      "epoch": 0.7928878468151622,
      "grad_norm": 9.355086326599121,
      "learning_rate": 2.210516976590179e-06,
      "loss": 9.176,
      "step": 4058
    },
    {
      "epoch": 0.7930832356389215,
      "grad_norm": 8.416228294372559,
      "learning_rate": 2.2065498512616036e-06,
      "loss": 8.2258,
      "step": 4059
    },
    {
      "epoch": 0.7932786244626807,
      "grad_norm": 9.133707046508789,
      "learning_rate": 2.202585847451768e-06,
      "loss": 8.8832,
      "step": 4060
    },
    {
      "epoch": 0.79347401328644,
      "grad_norm": 7.949434757232666,
      "learning_rate": 2.198624966748385e-06,
      "loss": 9.0004,
      "step": 4061
    },
    {
      "epoch": 0.7936694021101993,
      "grad_norm": 7.470557689666748,
      "learning_rate": 2.1946672107379052e-06,
      "loss": 8.2045,
      "step": 4062
    },
    {
      "epoch": 0.7938647909339586,
      "grad_norm": 8.743266105651855,
      "learning_rate": 2.1907125810055373e-06,
      "loss": 9.0787,
      "step": 4063
    },
    {
      "epoch": 0.7940601797577178,
      "grad_norm": 8.27902603149414,
      "learning_rate": 2.1867610791352313e-06,
      "loss": 9.0252,
      "step": 4064
    },
    {
      "epoch": 0.7942555685814772,
      "grad_norm": 8.170581817626953,
      "learning_rate": 2.1828127067096825e-06,
      "loss": 7.8556,
      "step": 4065
    },
    {
      "epoch": 0.7944509574052364,
      "grad_norm": 8.815306663513184,
      "learning_rate": 2.1788674653103436e-06,
      "loss": 9.4213,
      "step": 4066
    },
    {
      "epoch": 0.7946463462289957,
      "grad_norm": 6.542332649230957,
      "learning_rate": 2.174925356517401e-06,
      "loss": 7.9175,
      "step": 4067
    },
    {
      "epoch": 0.794841735052755,
      "grad_norm": 8.297599792480469,
      "learning_rate": 2.1709863819097952e-06,
      "loss": 9.1062,
      "step": 4068
    },
    {
      "epoch": 0.7950371238765143,
      "grad_norm": 9.112894058227539,
      "learning_rate": 2.1670505430652044e-06,
      "loss": 7.9776,
      "step": 4069
    },
    {
      "epoch": 0.7952325127002735,
      "grad_norm": 7.934484958648682,
      "learning_rate": 2.1631178415600572e-06,
      "loss": 8.5803,
      "step": 4070
    },
    {
      "epoch": 0.7954279015240329,
      "grad_norm": 7.044009208679199,
      "learning_rate": 2.1591882789695207e-06,
      "loss": 7.5398,
      "step": 4071
    },
    {
      "epoch": 0.7956232903477921,
      "grad_norm": 10.852298736572266,
      "learning_rate": 2.1552618568675088e-06,
      "loss": 7.795,
      "step": 4072
    },
    {
      "epoch": 0.7958186791715514,
      "grad_norm": 7.112560749053955,
      "learning_rate": 2.1513385768266713e-06,
      "loss": 8.4551,
      "step": 4073
    },
    {
      "epoch": 0.7960140679953107,
      "grad_norm": 7.160801887512207,
      "learning_rate": 2.1474184404184105e-06,
      "loss": 8.8399,
      "step": 4074
    },
    {
      "epoch": 0.79620945681907,
      "grad_norm": 7.856019973754883,
      "learning_rate": 2.1435014492128547e-06,
      "loss": 8.5651,
      "step": 4075
    },
    {
      "epoch": 0.7964048456428292,
      "grad_norm": 11.128726959228516,
      "learning_rate": 2.139587604778889e-06,
      "loss": 9.135,
      "step": 4076
    },
    {
      "epoch": 0.7966002344665886,
      "grad_norm": 11.092827796936035,
      "learning_rate": 2.1356769086841233e-06,
      "loss": 8.9956,
      "step": 4077
    },
    {
      "epoch": 0.7967956232903478,
      "grad_norm": 6.319994926452637,
      "learning_rate": 2.131769362494913e-06,
      "loss": 8.4058,
      "step": 4078
    },
    {
      "epoch": 0.7969910121141071,
      "grad_norm": 7.189726829528809,
      "learning_rate": 2.1278649677763553e-06,
      "loss": 8.8374,
      "step": 4079
    },
    {
      "epoch": 0.7971864009378663,
      "grad_norm": 7.846137046813965,
      "learning_rate": 2.1239637260922773e-06,
      "loss": 8.5696,
      "step": 4080
    },
    {
      "epoch": 0.7973817897616257,
      "grad_norm": 8.4998197555542,
      "learning_rate": 2.1200656390052523e-06,
      "loss": 9.4478,
      "step": 4081
    },
    {
      "epoch": 0.7975771785853849,
      "grad_norm": 8.599392890930176,
      "learning_rate": 2.116170708076579e-06,
      "loss": 9.4599,
      "step": 4082
    },
    {
      "epoch": 0.7977725674091442,
      "grad_norm": 9.487367630004883,
      "learning_rate": 2.112278934866305e-06,
      "loss": 8.6577,
      "step": 4083
    },
    {
      "epoch": 0.7979679562329035,
      "grad_norm": 6.9012885093688965,
      "learning_rate": 2.1083903209332004e-06,
      "loss": 8.4376,
      "step": 4084
    },
    {
      "epoch": 0.7981633450566628,
      "grad_norm": 7.9485297203063965,
      "learning_rate": 2.1045048678347814e-06,
      "loss": 8.9016,
      "step": 4085
    },
    {
      "epoch": 0.798358733880422,
      "grad_norm": 9.573652267456055,
      "learning_rate": 2.100622577127286e-06,
      "loss": 8.789,
      "step": 4086
    },
    {
      "epoch": 0.7985541227041814,
      "grad_norm": 8.524712562561035,
      "learning_rate": 2.0967434503656993e-06,
      "loss": 9.0548,
      "step": 4087
    },
    {
      "epoch": 0.7987495115279406,
      "grad_norm": 6.874051094055176,
      "learning_rate": 2.0928674891037247e-06,
      "loss": 9.174,
      "step": 4088
    },
    {
      "epoch": 0.7989449003516998,
      "grad_norm": 7.7614264488220215,
      "learning_rate": 2.0889946948938123e-06,
      "loss": 8.8263,
      "step": 4089
    },
    {
      "epoch": 0.7991402891754592,
      "grad_norm": 8.428397178649902,
      "learning_rate": 2.0851250692871327e-06,
      "loss": 8.4872,
      "step": 4090
    },
    {
      "epoch": 0.7993356779992185,
      "grad_norm": 7.815609931945801,
      "learning_rate": 2.081258613833589e-06,
      "loss": 9.7076,
      "step": 4091
    },
    {
      "epoch": 0.7995310668229777,
      "grad_norm": 8.944595336914062,
      "learning_rate": 2.0773953300818204e-06,
      "loss": 9.3888,
      "step": 4092
    },
    {
      "epoch": 0.7997264556467371,
      "grad_norm": 8.181797981262207,
      "learning_rate": 2.073535219579189e-06,
      "loss": 9.0652,
      "step": 4093
    },
    {
      "epoch": 0.7999218444704963,
      "grad_norm": 12.623723983764648,
      "learning_rate": 2.0696782838717934e-06,
      "loss": 9.017,
      "step": 4094
    },
    {
      "epoch": 0.8001172332942555,
      "grad_norm": 8.842370986938477,
      "learning_rate": 2.0658245245044515e-06,
      "loss": 9.6341,
      "step": 4095
    },
    {
      "epoch": 0.8003126221180149,
      "grad_norm": 10.332437515258789,
      "learning_rate": 2.0619739430207162e-06,
      "loss": 9.6929,
      "step": 4096
    },
    {
      "epoch": 0.8005080109417742,
      "grad_norm": 7.728264808654785,
      "learning_rate": 2.0581265409628613e-06,
      "loss": 8.5661,
      "step": 4097
    },
    {
      "epoch": 0.8007033997655334,
      "grad_norm": 8.843079566955566,
      "learning_rate": 2.0542823198718954e-06,
      "loss": 8.7008,
      "step": 4098
    },
    {
      "epoch": 0.8008987885892926,
      "grad_norm": 7.340389728546143,
      "learning_rate": 2.050441281287544e-06,
      "loss": 9.6193,
      "step": 4099
    },
    {
      "epoch": 0.801094177413052,
      "grad_norm": 8.187955856323242,
      "learning_rate": 2.046603426748267e-06,
      "loss": 9.2346,
      "step": 4100
    },
    {
      "epoch": 0.8012895662368112,
      "grad_norm": 13.1972017288208,
      "learning_rate": 2.0427687577912404e-06,
      "loss": 10.442,
      "step": 4101
    },
    {
      "epoch": 0.8014849550605705,
      "grad_norm": 7.015872478485107,
      "learning_rate": 2.038937275952372e-06,
      "loss": 7.6382,
      "step": 4102
    },
    {
      "epoch": 0.8016803438843298,
      "grad_norm": 10.845377922058105,
      "learning_rate": 2.0351089827662827e-06,
      "loss": 9.1876,
      "step": 4103
    },
    {
      "epoch": 0.8018757327080891,
      "grad_norm": 6.643104553222656,
      "learning_rate": 2.0312838797663315e-06,
      "loss": 8.6192,
      "step": 4104
    },
    {
      "epoch": 0.8020711215318483,
      "grad_norm": 8.04154109954834,
      "learning_rate": 2.027461968484583e-06,
      "loss": 7.2225,
      "step": 4105
    },
    {
      "epoch": 0.8022665103556077,
      "grad_norm": 9.899718284606934,
      "learning_rate": 2.023643250451838e-06,
      "loss": 9.4718,
      "step": 4106
    },
    {
      "epoch": 0.8024618991793669,
      "grad_norm": 6.92594051361084,
      "learning_rate": 2.019827727197605e-06,
      "loss": 8.9539,
      "step": 4107
    },
    {
      "epoch": 0.8026572880031262,
      "grad_norm": 15.220681190490723,
      "learning_rate": 2.0160154002501265e-06,
      "loss": 9.6917,
      "step": 4108
    },
    {
      "epoch": 0.8028526768268855,
      "grad_norm": 6.64687442779541,
      "learning_rate": 2.012206271136353e-06,
      "loss": 8.9374,
      "step": 4109
    },
    {
      "epoch": 0.8030480656506448,
      "grad_norm": 6.912804126739502,
      "learning_rate": 2.0084003413819585e-06,
      "loss": 8.6066,
      "step": 4110
    },
    {
      "epoch": 0.803243454474404,
      "grad_norm": 8.238934516906738,
      "learning_rate": 2.0045976125113407e-06,
      "loss": 8.3299,
      "step": 4111
    },
    {
      "epoch": 0.8034388432981634,
      "grad_norm": 9.851848602294922,
      "learning_rate": 2.000798086047604e-06,
      "loss": 9.4563,
      "step": 4112
    },
    {
      "epoch": 0.8036342321219226,
      "grad_norm": 8.415313720703125,
      "learning_rate": 1.997001763512584e-06,
      "loss": 8.1643,
      "step": 4113
    },
    {
      "epoch": 0.8038296209456819,
      "grad_norm": 86.34687042236328,
      "learning_rate": 1.9932086464268187e-06,
      "loss": 8.9568,
      "step": 4114
    },
    {
      "epoch": 0.8040250097694411,
      "grad_norm": 7.504203796386719,
      "learning_rate": 1.989418736309575e-06,
      "loss": 9.5223,
      "step": 4115
    },
    {
      "epoch": 0.8042203985932005,
      "grad_norm": 9.638359069824219,
      "learning_rate": 1.985632034678825e-06,
      "loss": 9.614,
      "step": 4116
    },
    {
      "epoch": 0.8044157874169597,
      "grad_norm": 10.84245491027832,
      "learning_rate": 1.981848543051265e-06,
      "loss": 9.5363,
      "step": 4117
    },
    {
      "epoch": 0.804611176240719,
      "grad_norm": 7.719939231872559,
      "learning_rate": 1.9780682629422954e-06,
      "loss": 9.0556,
      "step": 4118
    },
    {
      "epoch": 0.8048065650644783,
      "grad_norm": 8.364896774291992,
      "learning_rate": 1.9742911958660406e-06,
      "loss": 8.1381,
      "step": 4119
    },
    {
      "epoch": 0.8050019538882376,
      "grad_norm": 11.49855899810791,
      "learning_rate": 1.9705173433353297e-06,
      "loss": 9.4236,
      "step": 4120
    },
    {
      "epoch": 0.8051973427119968,
      "grad_norm": 8.680183410644531,
      "learning_rate": 1.966746706861712e-06,
      "loss": 9.4808,
      "step": 4121
    },
    {
      "epoch": 0.8053927315357562,
      "grad_norm": 8.259016990661621,
      "learning_rate": 1.962979287955441e-06,
      "loss": 9.5783,
      "step": 4122
    },
    {
      "epoch": 0.8055881203595154,
      "grad_norm": 7.60164213180542,
      "learning_rate": 1.9592150881254836e-06,
      "loss": 8.1726,
      "step": 4123
    },
    {
      "epoch": 0.8057835091832747,
      "grad_norm": 8.234026908874512,
      "learning_rate": 1.9554541088795244e-06,
      "loss": 9.0773,
      "step": 4124
    },
    {
      "epoch": 0.805978898007034,
      "grad_norm": 10.081842422485352,
      "learning_rate": 1.951696351723945e-06,
      "loss": 8.6069,
      "step": 4125
    },
    {
      "epoch": 0.8061742868307933,
      "grad_norm": 10.049712181091309,
      "learning_rate": 1.9479418181638508e-06,
      "loss": 9.2187,
      "step": 4126
    },
    {
      "epoch": 0.8063696756545525,
      "grad_norm": 9.004876136779785,
      "learning_rate": 1.9441905097030434e-06,
      "loss": 7.5274,
      "step": 4127
    },
    {
      "epoch": 0.8065650644783119,
      "grad_norm": 8.803363800048828,
      "learning_rate": 1.9404424278440424e-06,
      "loss": 8.1947,
      "step": 4128
    },
    {
      "epoch": 0.8067604533020711,
      "grad_norm": 8.290931701660156,
      "learning_rate": 1.9366975740880676e-06,
      "loss": 8.6446,
      "step": 4129
    },
    {
      "epoch": 0.8069558421258304,
      "grad_norm": 7.384932041168213,
      "learning_rate": 1.932955949935054e-06,
      "loss": 9.1818,
      "step": 4130
    },
    {
      "epoch": 0.8071512309495897,
      "grad_norm": 12.507975578308105,
      "learning_rate": 1.9292175568836314e-06,
      "loss": 10.3594,
      "step": 4131
    },
    {
      "epoch": 0.807346619773349,
      "grad_norm": 11.068806648254395,
      "learning_rate": 1.9254823964311488e-06,
      "loss": 9.9343,
      "step": 4132
    },
    {
      "epoch": 0.8075420085971082,
      "grad_norm": 11.15834903717041,
      "learning_rate": 1.9217504700736487e-06,
      "loss": 8.3411,
      "step": 4133
    },
    {
      "epoch": 0.8077373974208675,
      "grad_norm": 12.640612602233887,
      "learning_rate": 1.91802177930589e-06,
      "loss": 9.8704,
      "step": 4134
    },
    {
      "epoch": 0.8079327862446268,
      "grad_norm": 7.127387046813965,
      "learning_rate": 1.9142963256213234e-06,
      "loss": 8.5134,
      "step": 4135
    },
    {
      "epoch": 0.8081281750683861,
      "grad_norm": 8.432134628295898,
      "learning_rate": 1.910574110512108e-06,
      "loss": 8.7671,
      "step": 4136
    },
    {
      "epoch": 0.8083235638921453,
      "grad_norm": 11.820125579833984,
      "learning_rate": 1.9068551354691124e-06,
      "loss": 8.6807,
      "step": 4137
    },
    {
      "epoch": 0.8085189527159047,
      "grad_norm": 7.877838611602783,
      "learning_rate": 1.903139401981895e-06,
      "loss": 8.9795,
      "step": 4138
    },
    {
      "epoch": 0.8087143415396639,
      "grad_norm": 8.861102104187012,
      "learning_rate": 1.8994269115387276e-06,
      "loss": 9.2973,
      "step": 4139
    },
    {
      "epoch": 0.8089097303634232,
      "grad_norm": 8.142215728759766,
      "learning_rate": 1.8957176656265741e-06,
      "loss": 8.6584,
      "step": 4140
    },
    {
      "epoch": 0.8091051191871825,
      "grad_norm": 7.580594539642334,
      "learning_rate": 1.892011665731106e-06,
      "loss": 8.9637,
      "step": 4141
    },
    {
      "epoch": 0.8093005080109418,
      "grad_norm": 10.086572647094727,
      "learning_rate": 1.8883089133366882e-06,
      "loss": 9.515,
      "step": 4142
    },
    {
      "epoch": 0.809495896834701,
      "grad_norm": 6.808967590332031,
      "learning_rate": 1.8846094099263911e-06,
      "loss": 8.3363,
      "step": 4143
    },
    {
      "epoch": 0.8096912856584604,
      "grad_norm": 7.257555961608887,
      "learning_rate": 1.880913156981977e-06,
      "loss": 8.6286,
      "step": 4144
    },
    {
      "epoch": 0.8098866744822196,
      "grad_norm": 8.123153686523438,
      "learning_rate": 1.8772201559839143e-06,
      "loss": 8.6086,
      "step": 4145
    },
    {
      "epoch": 0.8100820633059789,
      "grad_norm": 10.420025825500488,
      "learning_rate": 1.8735304084113593e-06,
      "loss": 10.4163,
      "step": 4146
    },
    {
      "epoch": 0.8102774521297382,
      "grad_norm": 9.424251556396484,
      "learning_rate": 1.8698439157421767e-06,
      "loss": 9.763,
      "step": 4147
    },
    {
      "epoch": 0.8104728409534975,
      "grad_norm": 7.444462776184082,
      "learning_rate": 1.8661606794529174e-06,
      "loss": 8.2321,
      "step": 4148
    },
    {
      "epoch": 0.8106682297772567,
      "grad_norm": 12.281042098999023,
      "learning_rate": 1.8624807010188305e-06,
      "loss": 10.1341,
      "step": 4149
    },
    {
      "epoch": 0.810863618601016,
      "grad_norm": 6.998084545135498,
      "learning_rate": 1.858803981913866e-06,
      "loss": 9.0192,
      "step": 4150
    },
    {
      "epoch": 0.8110590074247753,
      "grad_norm": 7.253018379211426,
      "learning_rate": 1.8551305236106588e-06,
      "loss": 8.8867,
      "step": 4151
    },
    {
      "epoch": 0.8112543962485346,
      "grad_norm": 7.112781524658203,
      "learning_rate": 1.8514603275805487e-06,
      "loss": 8.7034,
      "step": 4152
    },
    {
      "epoch": 0.8114497850722938,
      "grad_norm": 7.53427267074585,
      "learning_rate": 1.8477933952935568e-06,
      "loss": 8.161,
      "step": 4153
    },
    {
      "epoch": 0.8116451738960532,
      "grad_norm": 11.889721870422363,
      "learning_rate": 1.84412972821841e-06,
      "loss": 9.7866,
      "step": 4154
    },
    {
      "epoch": 0.8118405627198124,
      "grad_norm": 11.897488594055176,
      "learning_rate": 1.840469327822515e-06,
      "loss": 9.418,
      "step": 4155
    },
    {
      "epoch": 0.8120359515435717,
      "grad_norm": 10.787230491638184,
      "learning_rate": 1.8368121955719808e-06,
      "loss": 10.2132,
      "step": 4156
    },
    {
      "epoch": 0.812231340367331,
      "grad_norm": 10.199027061462402,
      "learning_rate": 1.8331583329315972e-06,
      "loss": 8.7687,
      "step": 4157
    },
    {
      "epoch": 0.8124267291910903,
      "grad_norm": 9.7492094039917,
      "learning_rate": 1.8295077413648555e-06,
      "loss": 9.6211,
      "step": 4158
    },
    {
      "epoch": 0.8126221180148495,
      "grad_norm": 7.9797515869140625,
      "learning_rate": 1.825860422333925e-06,
      "loss": 8.9543,
      "step": 4159
    },
    {
      "epoch": 0.8128175068386089,
      "grad_norm": 15.008012771606445,
      "learning_rate": 1.8222163772996749e-06,
      "loss": 8.6499,
      "step": 4160
    },
    {
      "epoch": 0.8130128956623681,
      "grad_norm": 8.104180335998535,
      "learning_rate": 1.8185756077216576e-06,
      "loss": 8.5021,
      "step": 4161
    },
    {
      "epoch": 0.8132082844861274,
      "grad_norm": 7.260696887969971,
      "learning_rate": 1.814938115058109e-06,
      "loss": 8.7586,
      "step": 4162
    },
    {
      "epoch": 0.8134036733098867,
      "grad_norm": 7.8021159172058105,
      "learning_rate": 1.811303900765966e-06,
      "loss": 8.4024,
      "step": 4163
    },
    {
      "epoch": 0.813599062133646,
      "grad_norm": 11.577256202697754,
      "learning_rate": 1.8076729663008374e-06,
      "loss": 8.7474,
      "step": 4164
    },
    {
      "epoch": 0.8137944509574052,
      "grad_norm": 12.749056816101074,
      "learning_rate": 1.804045313117031e-06,
      "loss": 9.4589,
      "step": 4165
    },
    {
      "epoch": 0.8139898397811646,
      "grad_norm": 16.994075775146484,
      "learning_rate": 1.800420942667529e-06,
      "loss": 8.5454,
      "step": 4166
    },
    {
      "epoch": 0.8141852286049238,
      "grad_norm": 9.551435470581055,
      "learning_rate": 1.796799856404009e-06,
      "loss": 9.6272,
      "step": 4167
    },
    {
      "epoch": 0.8143806174286831,
      "grad_norm": 8.51735782623291,
      "learning_rate": 1.7931820557768241e-06,
      "loss": 7.9052,
      "step": 4168
    },
    {
      "epoch": 0.8145760062524423,
      "grad_norm": 9.26491928100586,
      "learning_rate": 1.78956754223502e-06,
      "loss": 9.768,
      "step": 4169
    },
    {
      "epoch": 0.8147713950762017,
      "grad_norm": 9.60155200958252,
      "learning_rate": 1.7859563172263183e-06,
      "loss": 9.4273,
      "step": 4170
    },
    {
      "epoch": 0.8149667838999609,
      "grad_norm": 9.635828971862793,
      "learning_rate": 1.78234838219713e-06,
      "loss": 9.2051,
      "step": 4171
    },
    {
      "epoch": 0.8151621727237202,
      "grad_norm": 7.847221851348877,
      "learning_rate": 1.7787437385925398e-06,
      "loss": 8.7587,
      "step": 4172
    },
    {
      "epoch": 0.8153575615474795,
      "grad_norm": 8.79216480255127,
      "learning_rate": 1.7751423878563267e-06,
      "loss": 8.9006,
      "step": 4173
    },
    {
      "epoch": 0.8155529503712388,
      "grad_norm": 8.901490211486816,
      "learning_rate": 1.771544331430939e-06,
      "loss": 8.9884,
      "step": 4174
    },
    {
      "epoch": 0.815748339194998,
      "grad_norm": 8.927732467651367,
      "learning_rate": 1.7679495707575078e-06,
      "loss": 8.4172,
      "step": 4175
    },
    {
      "epoch": 0.8159437280187574,
      "grad_norm": 9.58761215209961,
      "learning_rate": 1.7643581072758509e-06,
      "loss": 9.075,
      "step": 4176
    },
    {
      "epoch": 0.8161391168425166,
      "grad_norm": 8.125791549682617,
      "learning_rate": 1.7607699424244583e-06,
      "loss": 8.1893,
      "step": 4177
    },
    {
      "epoch": 0.8163345056662759,
      "grad_norm": 8.676210403442383,
      "learning_rate": 1.757185077640503e-06,
      "loss": 8.5555,
      "step": 4178
    },
    {
      "epoch": 0.8165298944900352,
      "grad_norm": 7.854306221008301,
      "learning_rate": 1.7536035143598328e-06,
      "loss": 9.0119,
      "step": 4179
    },
    {
      "epoch": 0.8167252833137945,
      "grad_norm": 8.398869514465332,
      "learning_rate": 1.7500252540169782e-06,
      "loss": 8.9472,
      "step": 4180
    },
    {
      "epoch": 0.8169206721375537,
      "grad_norm": 9.091753959655762,
      "learning_rate": 1.746450298045138e-06,
      "loss": 9.2992,
      "step": 4181
    },
    {
      "epoch": 0.8171160609613131,
      "grad_norm": 7.773633003234863,
      "learning_rate": 1.7428786478762005e-06,
      "loss": 8.4178,
      "step": 4182
    },
    {
      "epoch": 0.8173114497850723,
      "grad_norm": 9.261589050292969,
      "learning_rate": 1.7393103049407145e-06,
      "loss": 9.1313,
      "step": 4183
    },
    {
      "epoch": 0.8175068386088316,
      "grad_norm": 8.865499496459961,
      "learning_rate": 1.73574527066792e-06,
      "loss": 8.8908,
      "step": 4184
    },
    {
      "epoch": 0.8177022274325909,
      "grad_norm": 11.810985565185547,
      "learning_rate": 1.7321835464857162e-06,
      "loss": 9.0723,
      "step": 4185
    },
    {
      "epoch": 0.8178976162563502,
      "grad_norm": 13.720571517944336,
      "learning_rate": 1.7286251338206916e-06,
      "loss": 9.3866,
      "step": 4186
    },
    {
      "epoch": 0.8180930050801094,
      "grad_norm": 8.280376434326172,
      "learning_rate": 1.7250700340980942e-06,
      "loss": 8.0574,
      "step": 4187
    },
    {
      "epoch": 0.8182883939038686,
      "grad_norm": 7.536670207977295,
      "learning_rate": 1.721518248741858e-06,
      "loss": 8.0419,
      "step": 4188
    },
    {
      "epoch": 0.818483782727628,
      "grad_norm": 7.725442886352539,
      "learning_rate": 1.7179697791745797e-06,
      "loss": 8.6258,
      "step": 4189
    },
    {
      "epoch": 0.8186791715513873,
      "grad_norm": 8.78536319732666,
      "learning_rate": 1.7144246268175291e-06,
      "loss": 8.5599,
      "step": 4190
    },
    {
      "epoch": 0.8188745603751465,
      "grad_norm": 6.458619117736816,
      "learning_rate": 1.7108827930906568e-06,
      "loss": 7.56,
      "step": 4191
    },
    {
      "epoch": 0.8190699491989059,
      "grad_norm": 8.604883193969727,
      "learning_rate": 1.7073442794125694e-06,
      "loss": 9.3799,
      "step": 4192
    },
    {
      "epoch": 0.8192653380226651,
      "grad_norm": 9.05053997039795,
      "learning_rate": 1.7038090872005587e-06,
      "loss": 9.1484,
      "step": 4193
    },
    {
      "epoch": 0.8194607268464243,
      "grad_norm": 7.796328544616699,
      "learning_rate": 1.7002772178705717e-06,
      "loss": 9.0087,
      "step": 4194
    },
    {
      "epoch": 0.8196561156701837,
      "grad_norm": 8.378427505493164,
      "learning_rate": 1.6967486728372395e-06,
      "loss": 9.9526,
      "step": 4195
    },
    {
      "epoch": 0.819851504493943,
      "grad_norm": 13.551206588745117,
      "learning_rate": 1.693223453513847e-06,
      "loss": 9.6657,
      "step": 4196
    },
    {
      "epoch": 0.8200468933177022,
      "grad_norm": 9.289401054382324,
      "learning_rate": 1.689701561312359e-06,
      "loss": 9.4121,
      "step": 4197
    },
    {
      "epoch": 0.8202422821414616,
      "grad_norm": 8.325844764709473,
      "learning_rate": 1.6861829976433974e-06,
      "loss": 8.3365,
      "step": 4198
    },
    {
      "epoch": 0.8204376709652208,
      "grad_norm": 9.091864585876465,
      "learning_rate": 1.6826677639162635e-06,
      "loss": 9.9388,
      "step": 4199
    },
    {
      "epoch": 0.82063305978898,
      "grad_norm": 7.198275089263916,
      "learning_rate": 1.6791558615389103e-06,
      "loss": 8.9716,
      "step": 4200
    },
    {
      "epoch": 0.8208284486127394,
      "grad_norm": 9.158873558044434,
      "learning_rate": 1.675647291917969e-06,
      "loss": 9.7336,
      "step": 4201
    },
    {
      "epoch": 0.8210238374364986,
      "grad_norm": 8.986889839172363,
      "learning_rate": 1.6721420564587288e-06,
      "loss": 8.8077,
      "step": 4202
    },
    {
      "epoch": 0.8212192262602579,
      "grad_norm": 8.387405395507812,
      "learning_rate": 1.6686401565651423e-06,
      "loss": 9.0873,
      "step": 4203
    },
    {
      "epoch": 0.8214146150840171,
      "grad_norm": 10.184917449951172,
      "learning_rate": 1.665141593639834e-06,
      "loss": 9.9188,
      "step": 4204
    },
    {
      "epoch": 0.8216100039077765,
      "grad_norm": 8.070707321166992,
      "learning_rate": 1.6616463690840833e-06,
      "loss": 8.8431,
      "step": 4205
    },
    {
      "epoch": 0.8218053927315357,
      "grad_norm": 7.336175441741943,
      "learning_rate": 1.6581544842978391e-06,
      "loss": 8.619,
      "step": 4206
    },
    {
      "epoch": 0.822000781555295,
      "grad_norm": 6.866813659667969,
      "learning_rate": 1.6546659406797072e-06,
      "loss": 8.6299,
      "step": 4207
    },
    {
      "epoch": 0.8221961703790543,
      "grad_norm": 8.972613334655762,
      "learning_rate": 1.6511807396269607e-06,
      "loss": 9.2132,
      "step": 4208
    },
    {
      "epoch": 0.8223915592028136,
      "grad_norm": 10.178311347961426,
      "learning_rate": 1.6476988825355257e-06,
      "loss": 9.4726,
      "step": 4209
    },
    {
      "epoch": 0.8225869480265728,
      "grad_norm": 10.723286628723145,
      "learning_rate": 1.6442203707999994e-06,
      "loss": 9.7473,
      "step": 4210
    },
    {
      "epoch": 0.8227823368503322,
      "grad_norm": 7.932458400726318,
      "learning_rate": 1.6407452058136298e-06,
      "loss": 9.3409,
      "step": 4211
    },
    {
      "epoch": 0.8229777256740914,
      "grad_norm": 8.253729820251465,
      "learning_rate": 1.6372733889683323e-06,
      "loss": 9.2522,
      "step": 4212
    },
    {
      "epoch": 0.8231731144978507,
      "grad_norm": 7.544541835784912,
      "learning_rate": 1.6338049216546725e-06,
      "loss": 8.1359,
      "step": 4213
    },
    {
      "epoch": 0.82336850332161,
      "grad_norm": 9.513879776000977,
      "learning_rate": 1.6303398052618845e-06,
      "loss": 8.9512,
      "step": 4214
    },
    {
      "epoch": 0.8235638921453693,
      "grad_norm": 9.312100410461426,
      "learning_rate": 1.6268780411778529e-06,
      "loss": 9.0866,
      "step": 4215
    },
    {
      "epoch": 0.8237592809691285,
      "grad_norm": 7.766869068145752,
      "learning_rate": 1.6234196307891182e-06,
      "loss": 8.1376,
      "step": 4216
    },
    {
      "epoch": 0.8239546697928879,
      "grad_norm": 10.72817325592041,
      "learning_rate": 1.6199645754808867e-06,
      "loss": 10.2697,
      "step": 4217
    },
    {
      "epoch": 0.8241500586166471,
      "grad_norm": 7.453179836273193,
      "learning_rate": 1.6165128766370097e-06,
      "loss": 8.6152,
      "step": 4218
    },
    {
      "epoch": 0.8243454474404064,
      "grad_norm": 9.792024612426758,
      "learning_rate": 1.6130645356400054e-06,
      "loss": 8.3674,
      "step": 4219
    },
    {
      "epoch": 0.8245408362641657,
      "grad_norm": 8.387077331542969,
      "learning_rate": 1.609619553871037e-06,
      "loss": 8.6538,
      "step": 4220
    },
    {
      "epoch": 0.824736225087925,
      "grad_norm": 10.731857299804688,
      "learning_rate": 1.6061779327099303e-06,
      "loss": 9.5248,
      "step": 4221
    },
    {
      "epoch": 0.8249316139116842,
      "grad_norm": 7.778739929199219,
      "learning_rate": 1.6027396735351575e-06,
      "loss": 8.5324,
      "step": 4222
    },
    {
      "epoch": 0.8251270027354435,
      "grad_norm": 9.51624870300293,
      "learning_rate": 1.5993047777238525e-06,
      "loss": 8.9079,
      "step": 4223
    },
    {
      "epoch": 0.8253223915592028,
      "grad_norm": 8.001130104064941,
      "learning_rate": 1.5958732466517923e-06,
      "loss": 9.3273,
      "step": 4224
    },
    {
      "epoch": 0.8255177803829621,
      "grad_norm": 8.451061248779297,
      "learning_rate": 1.5924450816934177e-06,
      "loss": 9.2385,
      "step": 4225
    },
    {
      "epoch": 0.8257131692067213,
      "grad_norm": 6.511702537536621,
      "learning_rate": 1.5890202842218094e-06,
      "loss": 8.0991,
      "step": 4226
    },
    {
      "epoch": 0.8259085580304807,
      "grad_norm": 10.386144638061523,
      "learning_rate": 1.585598855608711e-06,
      "loss": 8.8742,
      "step": 4227
    },
    {
      "epoch": 0.8261039468542399,
      "grad_norm": 7.724105358123779,
      "learning_rate": 1.5821807972245073e-06,
      "loss": 9.5477,
      "step": 4228
    },
    {
      "epoch": 0.8262993356779992,
      "grad_norm": 7.820478439331055,
      "learning_rate": 1.5787661104382357e-06,
      "loss": 9.0195,
      "step": 4229
    },
    {
      "epoch": 0.8264947245017585,
      "grad_norm": 7.830831527709961,
      "learning_rate": 1.5753547966175875e-06,
      "loss": 8.3695,
      "step": 4230
    },
    {
      "epoch": 0.8266901133255178,
      "grad_norm": 8.081059455871582,
      "learning_rate": 1.5719468571288966e-06,
      "loss": 8.575,
      "step": 4231
    },
    {
      "epoch": 0.826885502149277,
      "grad_norm": 8.540038108825684,
      "learning_rate": 1.5685422933371518e-06,
      "loss": 8.8372,
      "step": 4232
    },
    {
      "epoch": 0.8270808909730364,
      "grad_norm": 7.683509826660156,
      "learning_rate": 1.5651411066059853e-06,
      "loss": 8.6626,
      "step": 4233
    },
    {
      "epoch": 0.8272762797967956,
      "grad_norm": 7.379648208618164,
      "learning_rate": 1.5617432982976776e-06,
      "loss": 8.4186,
      "step": 4234
    },
    {
      "epoch": 0.8274716686205549,
      "grad_norm": 8.802104949951172,
      "learning_rate": 1.5583488697731542e-06,
      "loss": 9.2311,
      "step": 4235
    },
    {
      "epoch": 0.8276670574443142,
      "grad_norm": 9.275979995727539,
      "learning_rate": 1.5549578223919938e-06,
      "loss": 9.0974,
      "step": 4236
    },
    {
      "epoch": 0.8278624462680735,
      "grad_norm": 12.322457313537598,
      "learning_rate": 1.5515701575124108e-06,
      "loss": 9.4169,
      "step": 4237
    },
    {
      "epoch": 0.8280578350918327,
      "grad_norm": 10.231083869934082,
      "learning_rate": 1.548185876491276e-06,
      "loss": 9.0337,
      "step": 4238
    },
    {
      "epoch": 0.828253223915592,
      "grad_norm": 8.195764541625977,
      "learning_rate": 1.5448049806840937e-06,
      "loss": 8.2788,
      "step": 4239
    },
    {
      "epoch": 0.8284486127393513,
      "grad_norm": 8.08511734008789,
      "learning_rate": 1.5414274714450228e-06,
      "loss": 8.8088,
      "step": 4240
    },
    {
      "epoch": 0.8286440015631106,
      "grad_norm": 58.29790115356445,
      "learning_rate": 1.5380533501268547e-06,
      "loss": 10.0793,
      "step": 4241
    },
    {
      "epoch": 0.8288393903868698,
      "grad_norm": 9.225850105285645,
      "learning_rate": 1.534682618081036e-06,
      "loss": 9.4692,
      "step": 4242
    },
    {
      "epoch": 0.8290347792106292,
      "grad_norm": 8.517762184143066,
      "learning_rate": 1.5313152766576434e-06,
      "loss": 9.4023,
      "step": 4243
    },
    {
      "epoch": 0.8292301680343884,
      "grad_norm": 25.837026596069336,
      "learning_rate": 1.527951327205407e-06,
      "loss": 9.359,
      "step": 4244
    },
    {
      "epoch": 0.8294255568581477,
      "grad_norm": 9.489827156066895,
      "learning_rate": 1.5245907710716912e-06,
      "loss": 8.9078,
      "step": 4245
    },
    {
      "epoch": 0.829620945681907,
      "grad_norm": 11.948223114013672,
      "learning_rate": 1.5212336096025005e-06,
      "loss": 9.2429,
      "step": 4246
    },
    {
      "epoch": 0.8298163345056663,
      "grad_norm": 8.366604804992676,
      "learning_rate": 1.5178798441424858e-06,
      "loss": 9.345,
      "step": 4247
    },
    {
      "epoch": 0.8300117233294255,
      "grad_norm": 8.934426307678223,
      "learning_rate": 1.5145294760349317e-06,
      "loss": 9.0068,
      "step": 4248
    },
    {
      "epoch": 0.8302071121531849,
      "grad_norm": 8.252093315124512,
      "learning_rate": 1.5111825066217668e-06,
      "loss": 9.1736,
      "step": 4249
    },
    {
      "epoch": 0.8304025009769441,
      "grad_norm": 8.491825103759766,
      "learning_rate": 1.5078389372435543e-06,
      "loss": 9.1984,
      "step": 4250
    },
    {
      "epoch": 0.8305978898007034,
      "grad_norm": 11.35191822052002,
      "learning_rate": 1.5044987692394997e-06,
      "loss": 9.3074,
      "step": 4251
    },
    {
      "epoch": 0.8307932786244627,
      "grad_norm": 8.606689453125,
      "learning_rate": 1.5011620039474407e-06,
      "loss": 8.9876,
      "step": 4252
    },
    {
      "epoch": 0.830988667448222,
      "grad_norm": 9.13046932220459,
      "learning_rate": 1.4978286427038602e-06,
      "loss": 9.5026,
      "step": 4253
    },
    {
      "epoch": 0.8311840562719812,
      "grad_norm": 10.701326370239258,
      "learning_rate": 1.4944986868438682e-06,
      "loss": 9.3942,
      "step": 4254
    },
    {
      "epoch": 0.8313794450957406,
      "grad_norm": 8.744029998779297,
      "learning_rate": 1.491172137701219e-06,
      "loss": 8.8568,
      "step": 4255
    },
    {
      "epoch": 0.8315748339194998,
      "grad_norm": 7.451327800750732,
      "learning_rate": 1.4878489966082942e-06,
      "loss": 8.6679,
      "step": 4256
    },
    {
      "epoch": 0.8317702227432591,
      "grad_norm": 8.913905143737793,
      "learning_rate": 1.4845292648961219e-06,
      "loss": 9.3168,
      "step": 4257
    },
    {
      "epoch": 0.8319656115670183,
      "grad_norm": 9.110503196716309,
      "learning_rate": 1.48121294389435e-06,
      "loss": 9.5957,
      "step": 4258
    },
    {
      "epoch": 0.8321610003907777,
      "grad_norm": 24.264135360717773,
      "learning_rate": 1.4779000349312754e-06,
      "loss": 9.6662,
      "step": 4259
    },
    {
      "epoch": 0.8323563892145369,
      "grad_norm": 7.600764751434326,
      "learning_rate": 1.4745905393338167e-06,
      "loss": 9.4055,
      "step": 4260
    },
    {
      "epoch": 0.8325517780382962,
      "grad_norm": 5.802551746368408,
      "learning_rate": 1.4712844584275275e-06,
      "loss": 7.2947,
      "step": 4261
    },
    {
      "epoch": 0.8327471668620555,
      "grad_norm": 13.387187957763672,
      "learning_rate": 1.4679817935366014e-06,
      "loss": 9.8224,
      "step": 4262
    },
    {
      "epoch": 0.8329425556858148,
      "grad_norm": 8.431346893310547,
      "learning_rate": 1.4646825459838532e-06,
      "loss": 9.5619,
      "step": 4263
    },
    {
      "epoch": 0.833137944509574,
      "grad_norm": 9.317984580993652,
      "learning_rate": 1.4613867170907391e-06,
      "loss": 9.7581,
      "step": 4264
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 7.618305683135986,
      "learning_rate": 1.4580943081773346e-06,
      "loss": 8.5974,
      "step": 4265
    },
    {
      "epoch": 0.8335287221570926,
      "grad_norm": 9.696817398071289,
      "learning_rate": 1.454805320562358e-06,
      "loss": 8.7606,
      "step": 4266
    },
    {
      "epoch": 0.8337241109808519,
      "grad_norm": 8.106192588806152,
      "learning_rate": 1.4515197555631467e-06,
      "loss": 9.3544,
      "step": 4267
    },
    {
      "epoch": 0.8339194998046112,
      "grad_norm": 7.084646224975586,
      "learning_rate": 1.4482376144956744e-06,
      "loss": 8.9587,
      "step": 4268
    },
    {
      "epoch": 0.8341148886283705,
      "grad_norm": 7.673435211181641,
      "learning_rate": 1.4449588986745367e-06,
      "loss": 9.1257,
      "step": 4269
    },
    {
      "epoch": 0.8343102774521297,
      "grad_norm": 9.534943580627441,
      "learning_rate": 1.441683609412965e-06,
      "loss": 9.3258,
      "step": 4270
    },
    {
      "epoch": 0.8345056662758891,
      "grad_norm": 7.976189613342285,
      "learning_rate": 1.438411748022811e-06,
      "loss": 9.1592,
      "step": 4271
    },
    {
      "epoch": 0.8347010550996483,
      "grad_norm": 9.561246871948242,
      "learning_rate": 1.435143315814561e-06,
      "loss": 9.6537,
      "step": 4272
    },
    {
      "epoch": 0.8348964439234076,
      "grad_norm": 7.933861255645752,
      "learning_rate": 1.4318783140973214e-06,
      "loss": 9.3873,
      "step": 4273
    },
    {
      "epoch": 0.8350918327471668,
      "grad_norm": 7.248596668243408,
      "learning_rate": 1.4286167441788234e-06,
      "loss": 8.6292,
      "step": 4274
    },
    {
      "epoch": 0.8352872215709262,
      "grad_norm": 8.570123672485352,
      "learning_rate": 1.4253586073654325e-06,
      "loss": 9.133,
      "step": 4275
    },
    {
      "epoch": 0.8354826103946854,
      "grad_norm": 16.415498733520508,
      "learning_rate": 1.4221039049621277e-06,
      "loss": 9.2063,
      "step": 4276
    },
    {
      "epoch": 0.8356779992184447,
      "grad_norm": 8.9085111618042,
      "learning_rate": 1.4188526382725255e-06,
      "loss": 8.7019,
      "step": 4277
    },
    {
      "epoch": 0.835873388042204,
      "grad_norm": 7.478800296783447,
      "learning_rate": 1.4156048085988528e-06,
      "loss": 8.8263,
      "step": 4278
    },
    {
      "epoch": 0.8360687768659633,
      "grad_norm": 9.699009895324707,
      "learning_rate": 1.4123604172419714e-06,
      "loss": 9.9962,
      "step": 4279
    },
    {
      "epoch": 0.8362641656897225,
      "grad_norm": 8.485151290893555,
      "learning_rate": 1.4091194655013552e-06,
      "loss": 9.8214,
      "step": 4280
    },
    {
      "epoch": 0.8364595545134819,
      "grad_norm": 9.319722175598145,
      "learning_rate": 1.4058819546751123e-06,
      "loss": 9.7082,
      "step": 4281
    },
    {
      "epoch": 0.8366549433372411,
      "grad_norm": 8.54001235961914,
      "learning_rate": 1.4026478860599602e-06,
      "loss": 9.5097,
      "step": 4282
    },
    {
      "epoch": 0.8368503321610004,
      "grad_norm": 148.36419677734375,
      "learning_rate": 1.3994172609512492e-06,
      "loss": 9.5145,
      "step": 4283
    },
    {
      "epoch": 0.8370457209847597,
      "grad_norm": 6.561718940734863,
      "learning_rate": 1.3961900806429396e-06,
      "loss": 8.4146,
      "step": 4284
    },
    {
      "epoch": 0.837241109808519,
      "grad_norm": 9.170931816101074,
      "learning_rate": 1.3929663464276222e-06,
      "loss": 9.1873,
      "step": 4285
    },
    {
      "epoch": 0.8374364986322782,
      "grad_norm": 82.37069702148438,
      "learning_rate": 1.3897460595965006e-06,
      "loss": 9.6036,
      "step": 4286
    },
    {
      "epoch": 0.8376318874560376,
      "grad_norm": 5.722506046295166,
      "learning_rate": 1.3865292214393977e-06,
      "loss": 8.3765,
      "step": 4287
    },
    {
      "epoch": 0.8378272762797968,
      "grad_norm": 8.180362701416016,
      "learning_rate": 1.3833158332447615e-06,
      "loss": 8.4419,
      "step": 4288
    },
    {
      "epoch": 0.838022665103556,
      "grad_norm": 6.897322654724121,
      "learning_rate": 1.3801058962996482e-06,
      "loss": 8.6013,
      "step": 4289
    },
    {
      "epoch": 0.8382180539273154,
      "grad_norm": 9.958105087280273,
      "learning_rate": 1.3768994118897427e-06,
      "loss": 10.2248,
      "step": 4290
    },
    {
      "epoch": 0.8384134427510747,
      "grad_norm": 7.8778767585754395,
      "learning_rate": 1.3736963812993365e-06,
      "loss": 9.15,
      "step": 4291
    },
    {
      "epoch": 0.8386088315748339,
      "grad_norm": 9.966565132141113,
      "learning_rate": 1.3704968058113487e-06,
      "loss": 8.7372,
      "step": 4292
    },
    {
      "epoch": 0.8388042203985931,
      "grad_norm": 7.740724563598633,
      "learning_rate": 1.3673006867073013e-06,
      "loss": 8.7413,
      "step": 4293
    },
    {
      "epoch": 0.8389996092223525,
      "grad_norm": 15.081512451171875,
      "learning_rate": 1.3641080252673467e-06,
      "loss": 9.7921,
      "step": 4294
    },
    {
      "epoch": 0.8391949980461118,
      "grad_norm": 9.276426315307617,
      "learning_rate": 1.3609188227702387e-06,
      "loss": 9.8176,
      "step": 4295
    },
    {
      "epoch": 0.839390386869871,
      "grad_norm": 8.647336959838867,
      "learning_rate": 1.3577330804933563e-06,
      "loss": 8.5131,
      "step": 4296
    },
    {
      "epoch": 0.8395857756936304,
      "grad_norm": 8.417224884033203,
      "learning_rate": 1.354550799712685e-06,
      "loss": 7.2477,
      "step": 4297
    },
    {
      "epoch": 0.8397811645173896,
      "grad_norm": 10.368106842041016,
      "learning_rate": 1.3513719817028302e-06,
      "loss": 8.8178,
      "step": 4298
    },
    {
      "epoch": 0.8399765533411488,
      "grad_norm": 7.813013553619385,
      "learning_rate": 1.3481966277370061e-06,
      "loss": 8.5611,
      "step": 4299
    },
    {
      "epoch": 0.8401719421649082,
      "grad_norm": 9.421406745910645,
      "learning_rate": 1.3450247390870374e-06,
      "loss": 8.6661,
      "step": 4300
    },
    {
      "epoch": 0.8403673309886674,
      "grad_norm": 10.088541984558105,
      "learning_rate": 1.3418563170233689e-06,
      "loss": 10.0167,
      "step": 4301
    },
    {
      "epoch": 0.8405627198124267,
      "grad_norm": 6.547402381896973,
      "learning_rate": 1.3386913628150477e-06,
      "loss": 8.4222,
      "step": 4302
    },
    {
      "epoch": 0.840758108636186,
      "grad_norm": 7.839051723480225,
      "learning_rate": 1.3355298777297399e-06,
      "loss": 8.9197,
      "step": 4303
    },
    {
      "epoch": 0.8409534974599453,
      "grad_norm": 9.462071418762207,
      "learning_rate": 1.3323718630337145e-06,
      "loss": 9.1845,
      "step": 4304
    },
    {
      "epoch": 0.8411488862837045,
      "grad_norm": 9.410916328430176,
      "learning_rate": 1.329217319991859e-06,
      "loss": 9.1187,
      "step": 4305
    },
    {
      "epoch": 0.8413442751074639,
      "grad_norm": 16.695587158203125,
      "learning_rate": 1.3260662498676613e-06,
      "loss": 8.5229,
      "step": 4306
    },
    {
      "epoch": 0.8415396639312231,
      "grad_norm": 8.594900131225586,
      "learning_rate": 1.3229186539232275e-06,
      "loss": 9.1592,
      "step": 4307
    },
    {
      "epoch": 0.8417350527549824,
      "grad_norm": 10.912017822265625,
      "learning_rate": 1.3197745334192624e-06,
      "loss": 9.6075,
      "step": 4308
    },
    {
      "epoch": 0.8419304415787418,
      "grad_norm": 7.126783847808838,
      "learning_rate": 1.3166338896150887e-06,
      "loss": 9.0325,
      "step": 4309
    },
    {
      "epoch": 0.842125830402501,
      "grad_norm": 7.9845757484436035,
      "learning_rate": 1.313496723768628e-06,
      "loss": 8.4876,
      "step": 4310
    },
    {
      "epoch": 0.8423212192262602,
      "grad_norm": 9.117425918579102,
      "learning_rate": 1.3103630371364151e-06,
      "loss": 9.6359,
      "step": 4311
    },
    {
      "epoch": 0.8425166080500195,
      "grad_norm": 9.479337692260742,
      "learning_rate": 1.3072328309735892e-06,
      "loss": 9.6399,
      "step": 4312
    },
    {
      "epoch": 0.8427119968737788,
      "grad_norm": 7.3381476402282715,
      "learning_rate": 1.30410610653389e-06,
      "loss": 8.1394,
      "step": 4313
    },
    {
      "epoch": 0.8429073856975381,
      "grad_norm": 8.921339988708496,
      "learning_rate": 1.3009828650696743e-06,
      "loss": 9.6048,
      "step": 4314
    },
    {
      "epoch": 0.8431027745212973,
      "grad_norm": 7.185836315155029,
      "learning_rate": 1.2978631078318927e-06,
      "loss": 9.2513,
      "step": 4315
    },
    {
      "epoch": 0.8432981633450567,
      "grad_norm": 8.77956485748291,
      "learning_rate": 1.2947468360701076e-06,
      "loss": 9.8427,
      "step": 4316
    },
    {
      "epoch": 0.8434935521688159,
      "grad_norm": 7.311408996582031,
      "learning_rate": 1.29163405103248e-06,
      "loss": 9.4744,
      "step": 4317
    },
    {
      "epoch": 0.8436889409925752,
      "grad_norm": 7.208104133605957,
      "learning_rate": 1.2885247539657808e-06,
      "loss": 8.9895,
      "step": 4318
    },
    {
      "epoch": 0.8438843298163345,
      "grad_norm": 9.21522045135498,
      "learning_rate": 1.2854189461153753e-06,
      "loss": 9.1812,
      "step": 4319
    },
    {
      "epoch": 0.8440797186400938,
      "grad_norm": 8.503985404968262,
      "learning_rate": 1.2823166287252397e-06,
      "loss": 9.2833,
      "step": 4320
    },
    {
      "epoch": 0.844275107463853,
      "grad_norm": 8.361055374145508,
      "learning_rate": 1.2792178030379454e-06,
      "loss": 9.3749,
      "step": 4321
    },
    {
      "epoch": 0.8444704962876124,
      "grad_norm": 6.587082862854004,
      "learning_rate": 1.2761224702946727e-06,
      "loss": 9.228,
      "step": 4322
    },
    {
      "epoch": 0.8446658851113716,
      "grad_norm": 7.459856986999512,
      "learning_rate": 1.2730306317351926e-06,
      "loss": 9.1024,
      "step": 4323
    },
    {
      "epoch": 0.8448612739351309,
      "grad_norm": 9.749171257019043,
      "learning_rate": 1.2699422885978886e-06,
      "loss": 9.5284,
      "step": 4324
    },
    {
      "epoch": 0.8450566627588902,
      "grad_norm": 9.829996109008789,
      "learning_rate": 1.2668574421197343e-06,
      "loss": 8.3387,
      "step": 4325
    },
    {
      "epoch": 0.8452520515826495,
      "grad_norm": 7.526081085205078,
      "learning_rate": 1.2637760935363053e-06,
      "loss": 8.2625,
      "step": 4326
    },
    {
      "epoch": 0.8454474404064087,
      "grad_norm": 9.518244743347168,
      "learning_rate": 1.2606982440817804e-06,
      "loss": 9.0902,
      "step": 4327
    },
    {
      "epoch": 0.845642829230168,
      "grad_norm": 7.615694522857666,
      "learning_rate": 1.2576238949889308e-06,
      "loss": 9.7948,
      "step": 4328
    },
    {
      "epoch": 0.8458382180539273,
      "grad_norm": 7.726398468017578,
      "learning_rate": 1.2545530474891332e-06,
      "loss": 8.0975,
      "step": 4329
    },
    {
      "epoch": 0.8460336068776866,
      "grad_norm": 6.965662956237793,
      "learning_rate": 1.251485702812353e-06,
      "loss": 8.5139,
      "step": 4330
    },
    {
      "epoch": 0.8462289957014458,
      "grad_norm": 7.9974799156188965,
      "learning_rate": 1.2484218621871603e-06,
      "loss": 9.1442,
      "step": 4331
    },
    {
      "epoch": 0.8464243845252052,
      "grad_norm": 10.21238899230957,
      "learning_rate": 1.2453615268407149e-06,
      "loss": 9.5565,
      "step": 4332
    },
    {
      "epoch": 0.8466197733489644,
      "grad_norm": 9.505382537841797,
      "learning_rate": 1.2423046979987797e-06,
      "loss": 9.3563,
      "step": 4333
    },
    {
      "epoch": 0.8468151621727237,
      "grad_norm": 6.460310935974121,
      "learning_rate": 1.239251376885706e-06,
      "loss": 8.7227,
      "step": 4334
    },
    {
      "epoch": 0.847010550996483,
      "grad_norm": 7.768892288208008,
      "learning_rate": 1.2362015647244485e-06,
      "loss": 8.3439,
      "step": 4335
    },
    {
      "epoch": 0.8472059398202423,
      "grad_norm": 8.403698921203613,
      "learning_rate": 1.2331552627365462e-06,
      "loss": 9.6446,
      "step": 4336
    },
    {
      "epoch": 0.8474013286440015,
      "grad_norm": 8.840633392333984,
      "learning_rate": 1.230112472142142e-06,
      "loss": 9.7222,
      "step": 4337
    },
    {
      "epoch": 0.8475967174677609,
      "grad_norm": 9.855066299438477,
      "learning_rate": 1.2270731941599667e-06,
      "loss": 9.3417,
      "step": 4338
    },
    {
      "epoch": 0.8477921062915201,
      "grad_norm": 8.718852043151855,
      "learning_rate": 1.2240374300073422e-06,
      "loss": 8.9743,
      "step": 4339
    },
    {
      "epoch": 0.8479874951152794,
      "grad_norm": 17.223560333251953,
      "learning_rate": 1.221005180900192e-06,
      "loss": 9.4087,
      "step": 4340
    },
    {
      "epoch": 0.8481828839390387,
      "grad_norm": 9.039057731628418,
      "learning_rate": 1.2179764480530197e-06,
      "loss": 8.351,
      "step": 4341
    },
    {
      "epoch": 0.848378272762798,
      "grad_norm": 8.74781608581543,
      "learning_rate": 1.2149512326789337e-06,
      "loss": 9.2977,
      "step": 4342
    },
    {
      "epoch": 0.8485736615865572,
      "grad_norm": 9.709748268127441,
      "learning_rate": 1.21192953598962e-06,
      "loss": 9.4355,
      "step": 4343
    },
    {
      "epoch": 0.8487690504103166,
      "grad_norm": 9.52278995513916,
      "learning_rate": 1.2089113591953684e-06,
      "loss": 9.4792,
      "step": 4344
    },
    {
      "epoch": 0.8489644392340758,
      "grad_norm": 6.382225036621094,
      "learning_rate": 1.2058967035050461e-06,
      "loss": 8.2168,
      "step": 4345
    },
    {
      "epoch": 0.8491598280578351,
      "grad_norm": 9.21106243133545,
      "learning_rate": 1.2028855701261223e-06,
      "loss": 10.1486,
      "step": 4346
    },
    {
      "epoch": 0.8493552168815943,
      "grad_norm": 9.709728240966797,
      "learning_rate": 1.1998779602646438e-06,
      "loss": 8.5562,
      "step": 4347
    },
    {
      "epoch": 0.8495506057053537,
      "grad_norm": 14.302094459533691,
      "learning_rate": 1.1968738751252562e-06,
      "loss": 9.4097,
      "step": 4348
    },
    {
      "epoch": 0.8497459945291129,
      "grad_norm": 7.293747425079346,
      "learning_rate": 1.1938733159111849e-06,
      "loss": 8.9235,
      "step": 4349
    },
    {
      "epoch": 0.8499413833528722,
      "grad_norm": 7.6194376945495605,
      "learning_rate": 1.1908762838242515e-06,
      "loss": 9.5107,
      "step": 4350
    },
    {
      "epoch": 0.8501367721766315,
      "grad_norm": 8.724333763122559,
      "learning_rate": 1.1878827800648585e-06,
      "loss": 9.5367,
      "step": 4351
    },
    {
      "epoch": 0.8503321610003908,
      "grad_norm": 9.795305252075195,
      "learning_rate": 1.1848928058319942e-06,
      "loss": 8.3142,
      "step": 4352
    },
    {
      "epoch": 0.85052754982415,
      "grad_norm": 7.095324516296387,
      "learning_rate": 1.1819063623232408e-06,
      "loss": 8.7944,
      "step": 4353
    },
    {
      "epoch": 0.8507229386479094,
      "grad_norm": 9.359949111938477,
      "learning_rate": 1.1789234507347568e-06,
      "loss": 8.3412,
      "step": 4354
    },
    {
      "epoch": 0.8509183274716686,
      "grad_norm": 10.278728485107422,
      "learning_rate": 1.1759440722612957e-06,
      "loss": 9.0669,
      "step": 4355
    },
    {
      "epoch": 0.8511137162954279,
      "grad_norm": 10.226433753967285,
      "learning_rate": 1.1729682280961873e-06,
      "loss": 9.0757,
      "step": 4356
    },
    {
      "epoch": 0.8513091051191872,
      "grad_norm": 10.047861099243164,
      "learning_rate": 1.1699959194313525e-06,
      "loss": 9.5237,
      "step": 4357
    },
    {
      "epoch": 0.8515044939429465,
      "grad_norm": 7.863394260406494,
      "learning_rate": 1.1670271474572903e-06,
      "loss": 8.8631,
      "step": 4358
    },
    {
      "epoch": 0.8516998827667057,
      "grad_norm": 7.909746170043945,
      "learning_rate": 1.1640619133630894e-06,
      "loss": 8.9935,
      "step": 4359
    },
    {
      "epoch": 0.8518952715904651,
      "grad_norm": 9.45473861694336,
      "learning_rate": 1.1611002183364152e-06,
      "loss": 9.7448,
      "step": 4360
    },
    {
      "epoch": 0.8520906604142243,
      "grad_norm": 7.1263837814331055,
      "learning_rate": 1.1581420635635198e-06,
      "loss": 9.5666,
      "step": 4361
    },
    {
      "epoch": 0.8522860492379836,
      "grad_norm": 8.597981452941895,
      "learning_rate": 1.1551874502292359e-06,
      "loss": 8.9688,
      "step": 4362
    },
    {
      "epoch": 0.8524814380617428,
      "grad_norm": 8.461832046508789,
      "learning_rate": 1.1522363795169788e-06,
      "loss": 9.1263,
      "step": 4363
    },
    {
      "epoch": 0.8526768268855022,
      "grad_norm": 8.200736999511719,
      "learning_rate": 1.149288852608743e-06,
      "loss": 9.6539,
      "step": 4364
    },
    {
      "epoch": 0.8528722157092614,
      "grad_norm": 9.481253623962402,
      "learning_rate": 1.1463448706851021e-06,
      "loss": 8.6186,
      "step": 4365
    },
    {
      "epoch": 0.8530676045330207,
      "grad_norm": 11.275646209716797,
      "learning_rate": 1.1434044349252172e-06,
      "loss": 9.864,
      "step": 4366
    },
    {
      "epoch": 0.85326299335678,
      "grad_norm": 7.154355049133301,
      "learning_rate": 1.1404675465068193e-06,
      "loss": 9.6745,
      "step": 4367
    },
    {
      "epoch": 0.8534583821805393,
      "grad_norm": 9.041812896728516,
      "learning_rate": 1.1375342066062277e-06,
      "loss": 9.1627,
      "step": 4368
    },
    {
      "epoch": 0.8536537710042985,
      "grad_norm": 7.8985090255737305,
      "learning_rate": 1.1346044163983327e-06,
      "loss": 8.8172,
      "step": 4369
    },
    {
      "epoch": 0.8538491598280579,
      "grad_norm": 10.522104263305664,
      "learning_rate": 1.1316781770566087e-06,
      "loss": 9.471,
      "step": 4370
    },
    {
      "epoch": 0.8540445486518171,
      "grad_norm": 6.28630256652832,
      "learning_rate": 1.128755489753105e-06,
      "loss": 8.1832,
      "step": 4371
    },
    {
      "epoch": 0.8542399374755764,
      "grad_norm": 7.923295497894287,
      "learning_rate": 1.1258363556584473e-06,
      "loss": 9.5306,
      "step": 4372
    },
    {
      "epoch": 0.8544353262993357,
      "grad_norm": 7.415898323059082,
      "learning_rate": 1.1229207759418381e-06,
      "loss": 8.9138,
      "step": 4373
    },
    {
      "epoch": 0.854630715123095,
      "grad_norm": 7.470461368560791,
      "learning_rate": 1.1200087517710623e-06,
      "loss": 9.5922,
      "step": 4374
    },
    {
      "epoch": 0.8548261039468542,
      "grad_norm": 8.525838851928711,
      "learning_rate": 1.11710028431247e-06,
      "loss": 8.5091,
      "step": 4375
    },
    {
      "epoch": 0.8550214927706136,
      "grad_norm": 8.140287399291992,
      "learning_rate": 1.1141953747309986e-06,
      "loss": 8.9586,
      "step": 4376
    },
    {
      "epoch": 0.8552168815943728,
      "grad_norm": 18.053083419799805,
      "learning_rate": 1.1112940241901504e-06,
      "loss": 9.1289,
      "step": 4377
    },
    {
      "epoch": 0.8554122704181321,
      "grad_norm": 8.162330627441406,
      "learning_rate": 1.1083962338520093e-06,
      "loss": 8.8091,
      "step": 4378
    },
    {
      "epoch": 0.8556076592418914,
      "grad_norm": 8.185535430908203,
      "learning_rate": 1.1055020048772268e-06,
      "loss": 9.3505,
      "step": 4379
    },
    {
      "epoch": 0.8558030480656507,
      "grad_norm": 9.759170532226562,
      "learning_rate": 1.102611338425036e-06,
      "loss": 9.4584,
      "step": 4380
    },
    {
      "epoch": 0.8559984368894099,
      "grad_norm": 10.33072566986084,
      "learning_rate": 1.0997242356532335e-06,
      "loss": 8.1816,
      "step": 4381
    },
    {
      "epoch": 0.8561938257131692,
      "grad_norm": 7.357562065124512,
      "learning_rate": 1.096840697718199e-06,
      "loss": 8.4596,
      "step": 4382
    },
    {
      "epoch": 0.8563892145369285,
      "grad_norm": 6.7860870361328125,
      "learning_rate": 1.0939607257748752e-06,
      "loss": 8.8797,
      "step": 4383
    },
    {
      "epoch": 0.8565846033606878,
      "grad_norm": 8.139361381530762,
      "learning_rate": 1.0910843209767797e-06,
      "loss": 9.3265,
      "step": 4384
    },
    {
      "epoch": 0.856779992184447,
      "grad_norm": 8.5925874710083,
      "learning_rate": 1.088211484476005e-06,
      "loss": 9.0923,
      "step": 4385
    },
    {
      "epoch": 0.8569753810082064,
      "grad_norm": 11.344609260559082,
      "learning_rate": 1.0853422174232087e-06,
      "loss": 9.6272,
      "step": 4386
    },
    {
      "epoch": 0.8571707698319656,
      "grad_norm": 7.951318740844727,
      "learning_rate": 1.0824765209676235e-06,
      "loss": 8.6183,
      "step": 4387
    },
    {
      "epoch": 0.8573661586557249,
      "grad_norm": 8.583785057067871,
      "learning_rate": 1.079614396257047e-06,
      "loss": 9.3787,
      "step": 4388
    },
    {
      "epoch": 0.8575615474794842,
      "grad_norm": 7.105406284332275,
      "learning_rate": 1.0767558444378535e-06,
      "loss": 8.2011,
      "step": 4389
    },
    {
      "epoch": 0.8577569363032435,
      "grad_norm": 6.523879051208496,
      "learning_rate": 1.0739008666549767e-06,
      "loss": 8.861,
      "step": 4390
    },
    {
      "epoch": 0.8579523251270027,
      "grad_norm": 7.175806522369385,
      "learning_rate": 1.0710494640519276e-06,
      "loss": 8.2861,
      "step": 4391
    },
    {
      "epoch": 0.8581477139507621,
      "grad_norm": 9.211763381958008,
      "learning_rate": 1.0682016377707772e-06,
      "loss": 9.3112,
      "step": 4392
    },
    {
      "epoch": 0.8583431027745213,
      "grad_norm": 7.219457149505615,
      "learning_rate": 1.065357388952174e-06,
      "loss": 9.6006,
      "step": 4393
    },
    {
      "epoch": 0.8585384915982806,
      "grad_norm": 8.839190483093262,
      "learning_rate": 1.0625167187353236e-06,
      "loss": 9.7189,
      "step": 4394
    },
    {
      "epoch": 0.8587338804220399,
      "grad_norm": 8.153175354003906,
      "learning_rate": 1.0596796282580058e-06,
      "loss": 8.6085,
      "step": 4395
    },
    {
      "epoch": 0.8589292692457992,
      "grad_norm": 7.035209655761719,
      "learning_rate": 1.0568461186565604e-06,
      "loss": 8.595,
      "step": 4396
    },
    {
      "epoch": 0.8591246580695584,
      "grad_norm": 11.23974609375,
      "learning_rate": 1.0540161910658964e-06,
      "loss": 8.1272,
      "step": 4397
    },
    {
      "epoch": 0.8593200468933178,
      "grad_norm": 7.754348278045654,
      "learning_rate": 1.0511898466194903e-06,
      "loss": 9.4844,
      "step": 4398
    },
    {
      "epoch": 0.859515435717077,
      "grad_norm": 11.255935668945312,
      "learning_rate": 1.0483670864493777e-06,
      "loss": 8.4559,
      "step": 4399
    },
    {
      "epoch": 0.8597108245408362,
      "grad_norm": 7.187273025512695,
      "learning_rate": 1.0455479116861645e-06,
      "loss": 8.0356,
      "step": 4400
    },
    {
      "epoch": 0.8599062133645955,
      "grad_norm": 10.443448066711426,
      "learning_rate": 1.0427323234590137e-06,
      "loss": 9.0796,
      "step": 4401
    },
    {
      "epoch": 0.8601016021883549,
      "grad_norm": 9.212592124938965,
      "learning_rate": 1.0399203228956611e-06,
      "loss": 10.1266,
      "step": 4402
    },
    {
      "epoch": 0.8602969910121141,
      "grad_norm": 10.160746574401855,
      "learning_rate": 1.0371119111223949e-06,
      "loss": 8.8189,
      "step": 4403
    },
    {
      "epoch": 0.8604923798358733,
      "grad_norm": 8.360414505004883,
      "learning_rate": 1.034307089264076e-06,
      "loss": 9.6335,
      "step": 4404
    },
    {
      "epoch": 0.8606877686596327,
      "grad_norm": 8.571639060974121,
      "learning_rate": 1.0315058584441185e-06,
      "loss": 9.2224,
      "step": 4405
    },
    {
      "epoch": 0.860883157483392,
      "grad_norm": 9.940386772155762,
      "learning_rate": 1.0287082197845055e-06,
      "loss": 9.1021,
      "step": 4406
    },
    {
      "epoch": 0.8610785463071512,
      "grad_norm": 9.578882217407227,
      "learning_rate": 1.0259141744057755e-06,
      "loss": 9.4857,
      "step": 4407
    },
    {
      "epoch": 0.8612739351309106,
      "grad_norm": 7.491061210632324,
      "learning_rate": 1.0231237234270331e-06,
      "loss": 9.269,
      "step": 4408
    },
    {
      "epoch": 0.8614693239546698,
      "grad_norm": 8.327766418457031,
      "learning_rate": 1.0203368679659399e-06,
      "loss": 8.8025,
      "step": 4409
    },
    {
      "epoch": 0.861664712778429,
      "grad_norm": 8.449031829833984,
      "learning_rate": 1.017553609138715e-06,
      "loss": 8.0793,
      "step": 4410
    },
    {
      "epoch": 0.8618601016021884,
      "grad_norm": 7.012770652770996,
      "learning_rate": 1.0147739480601448e-06,
      "loss": 8.1449,
      "step": 4411
    },
    {
      "epoch": 0.8620554904259476,
      "grad_norm": 7.9110918045043945,
      "learning_rate": 1.0119978858435653e-06,
      "loss": 9.2474,
      "step": 4412
    },
    {
      "epoch": 0.8622508792497069,
      "grad_norm": 8.513326644897461,
      "learning_rate": 1.0092254236008793e-06,
      "loss": 8.8102,
      "step": 4413
    },
    {
      "epoch": 0.8624462680734662,
      "grad_norm": 64.09504699707031,
      "learning_rate": 1.0064565624425404e-06,
      "loss": 8.6384,
      "step": 4414
    },
    {
      "epoch": 0.8626416568972255,
      "grad_norm": 8.766754150390625,
      "learning_rate": 1.0036913034775675e-06,
      "loss": 8.5842,
      "step": 4415
    },
    {
      "epoch": 0.8628370457209847,
      "grad_norm": 9.33322525024414,
      "learning_rate": 1.0009296478135289e-06,
      "loss": 8.9067,
      "step": 4416
    },
    {
      "epoch": 0.863032434544744,
      "grad_norm": 9.218510627746582,
      "learning_rate": 9.981715965565565e-07,
      "loss": 9.0773,
      "step": 4417
    },
    {
      "epoch": 0.8632278233685033,
      "grad_norm": 8.177020072937012,
      "learning_rate": 9.95417150811333e-07,
      "loss": 7.9319,
      "step": 4418
    },
    {
      "epoch": 0.8634232121922626,
      "grad_norm": 8.058493614196777,
      "learning_rate": 9.926663116811019e-07,
      "loss": 8.6709,
      "step": 4419
    },
    {
      "epoch": 0.8636186010160218,
      "grad_norm": 8.650046348571777,
      "learning_rate": 9.89919080267656e-07,
      "loss": 9.4331,
      "step": 4420
    },
    {
      "epoch": 0.8638139898397812,
      "grad_norm": 11.028565406799316,
      "learning_rate": 9.87175457671351e-07,
      "loss": 9.8931,
      "step": 4421
    },
    {
      "epoch": 0.8640093786635404,
      "grad_norm": 9.88052749633789,
      "learning_rate": 9.84435444991092e-07,
      "loss": 9.7764,
      "step": 4422
    },
    {
      "epoch": 0.8642047674872997,
      "grad_norm": 8.127789497375488,
      "learning_rate": 9.816990433243356e-07,
      "loss": 9.4421,
      "step": 4423
    },
    {
      "epoch": 0.864400156311059,
      "grad_norm": 7.8543219566345215,
      "learning_rate": 9.789662537671007e-07,
      "loss": 8.0521,
      "step": 4424
    },
    {
      "epoch": 0.8645955451348183,
      "grad_norm": 8.689035415649414,
      "learning_rate": 9.762370774139484e-07,
      "loss": 9.1581,
      "step": 4425
    },
    {
      "epoch": 0.8647909339585775,
      "grad_norm": 9.804941177368164,
      "learning_rate": 9.735115153580055e-07,
      "loss": 10.1786,
      "step": 4426
    },
    {
      "epoch": 0.8649863227823369,
      "grad_norm": 7.875881195068359,
      "learning_rate": 9.707895686909385e-07,
      "loss": 8.8083,
      "step": 4427
    },
    {
      "epoch": 0.8651817116060961,
      "grad_norm": 7.706095218658447,
      "learning_rate": 9.680712385029755e-07,
      "loss": 9.2423,
      "step": 4428
    },
    {
      "epoch": 0.8653771004298554,
      "grad_norm": 12.053943634033203,
      "learning_rate": 9.65356525882888e-07,
      "loss": 9.502,
      "step": 4429
    },
    {
      "epoch": 0.8655724892536147,
      "grad_norm": 10.30573844909668,
      "learning_rate": 9.626454319180078e-07,
      "loss": 9.3767,
      "step": 4430
    },
    {
      "epoch": 0.865767878077374,
      "grad_norm": 9.071868896484375,
      "learning_rate": 9.599379576942069e-07,
      "loss": 8.9809,
      "step": 4431
    },
    {
      "epoch": 0.8659632669011332,
      "grad_norm": 7.747514724731445,
      "learning_rate": 9.572341042959177e-07,
      "loss": 8.4974,
      "step": 4432
    },
    {
      "epoch": 0.8661586557248926,
      "grad_norm": 10.477039337158203,
      "learning_rate": 9.545338728061125e-07,
      "loss": 9.1711,
      "step": 4433
    },
    {
      "epoch": 0.8663540445486518,
      "grad_norm": 12.359572410583496,
      "learning_rate": 9.518372643063234e-07,
      "loss": 9.9854,
      "step": 4434
    },
    {
      "epoch": 0.8665494333724111,
      "grad_norm": 7.519041061401367,
      "learning_rate": 9.491442798766226e-07,
      "loss": 8.6956,
      "step": 4435
    },
    {
      "epoch": 0.8667448221961703,
      "grad_norm": 12.082423210144043,
      "learning_rate": 9.464549205956319e-07,
      "loss": 8.4,
      "step": 4436
    },
    {
      "epoch": 0.8669402110199297,
      "grad_norm": 10.535531997680664,
      "learning_rate": 9.437691875405275e-07,
      "loss": 9.6666,
      "step": 4437
    },
    {
      "epoch": 0.8671355998436889,
      "grad_norm": 18.231430053710938,
      "learning_rate": 9.410870817870254e-07,
      "loss": 10.2696,
      "step": 4438
    },
    {
      "epoch": 0.8673309886674482,
      "grad_norm": 10.223472595214844,
      "learning_rate": 9.384086044093954e-07,
      "loss": 9.7457,
      "step": 4439
    },
    {
      "epoch": 0.8675263774912075,
      "grad_norm": 9.341796875,
      "learning_rate": 9.35733756480447e-07,
      "loss": 8.7086,
      "step": 4440
    },
    {
      "epoch": 0.8677217663149668,
      "grad_norm": 6.032013416290283,
      "learning_rate": 9.330625390715442e-07,
      "loss": 8.1348,
      "step": 4441
    },
    {
      "epoch": 0.867917155138726,
      "grad_norm": 8.410514831542969,
      "learning_rate": 9.30394953252588e-07,
      "loss": 8.7077,
      "step": 4442
    },
    {
      "epoch": 0.8681125439624854,
      "grad_norm": 7.853433609008789,
      "learning_rate": 9.277310000920337e-07,
      "loss": 8.4718,
      "step": 4443
    },
    {
      "epoch": 0.8683079327862446,
      "grad_norm": 8.205535888671875,
      "learning_rate": 9.250706806568744e-07,
      "loss": 9.3656,
      "step": 4444
    },
    {
      "epoch": 0.8685033216100039,
      "grad_norm": 8.704334259033203,
      "learning_rate": 9.224139960126521e-07,
      "loss": 9.2046,
      "step": 4445
    },
    {
      "epoch": 0.8686987104337632,
      "grad_norm": 8.048930168151855,
      "learning_rate": 9.197609472234493e-07,
      "loss": 9.0922,
      "step": 4446
    },
    {
      "epoch": 0.8688940992575225,
      "grad_norm": 7.903109550476074,
      "learning_rate": 9.171115353518978e-07,
      "loss": 8.3976,
      "step": 4447
    },
    {
      "epoch": 0.8690894880812817,
      "grad_norm": 101.25208282470703,
      "learning_rate": 9.144657614591679e-07,
      "loss": 9.5005,
      "step": 4448
    },
    {
      "epoch": 0.8692848769050411,
      "grad_norm": 7.834874629974365,
      "learning_rate": 9.118236266049707e-07,
      "loss": 8.7355,
      "step": 4449
    },
    {
      "epoch": 0.8694802657288003,
      "grad_norm": 6.803002834320068,
      "learning_rate": 9.091851318475687e-07,
      "loss": 7.4237,
      "step": 4450
    },
    {
      "epoch": 0.8696756545525596,
      "grad_norm": 9.04742431640625,
      "learning_rate": 9.065502782437574e-07,
      "loss": 8.9199,
      "step": 4451
    },
    {
      "epoch": 0.8698710433763188,
      "grad_norm": 7.7430548667907715,
      "learning_rate": 9.039190668488795e-07,
      "loss": 9.5888,
      "step": 4452
    },
    {
      "epoch": 0.8700664322000782,
      "grad_norm": 9.58332633972168,
      "learning_rate": 9.012914987168143e-07,
      "loss": 9.9074,
      "step": 4453
    },
    {
      "epoch": 0.8702618210238374,
      "grad_norm": 28.31788444519043,
      "learning_rate": 8.986675748999885e-07,
      "loss": 8.6878,
      "step": 4454
    },
    {
      "epoch": 0.8704572098475967,
      "grad_norm": 7.615555286407471,
      "learning_rate": 8.960472964493594e-07,
      "loss": 9.6356,
      "step": 4455
    },
    {
      "epoch": 0.870652598671356,
      "grad_norm": 6.998612403869629,
      "learning_rate": 8.934306644144363e-07,
      "loss": 8.8061,
      "step": 4456
    },
    {
      "epoch": 0.8708479874951153,
      "grad_norm": 7.023011207580566,
      "learning_rate": 8.908176798432555e-07,
      "loss": 7.6729,
      "step": 4457
    },
    {
      "epoch": 0.8710433763188745,
      "grad_norm": 6.8855485916137695,
      "learning_rate": 8.88208343782404e-07,
      "loss": 8.3022,
      "step": 4458
    },
    {
      "epoch": 0.8712387651426339,
      "grad_norm": 9.884055137634277,
      "learning_rate": 8.856026572769971e-07,
      "loss": 9.3856,
      "step": 4459
    },
    {
      "epoch": 0.8714341539663931,
      "grad_norm": 9.977845191955566,
      "learning_rate": 8.830006213706965e-07,
      "loss": 10.5723,
      "step": 4460
    },
    {
      "epoch": 0.8716295427901524,
      "grad_norm": 6.782526969909668,
      "learning_rate": 8.804022371056975e-07,
      "loss": 8.438,
      "step": 4461
    },
    {
      "epoch": 0.8718249316139117,
      "grad_norm": 9.667460441589355,
      "learning_rate": 8.77807505522732e-07,
      "loss": 9.2985,
      "step": 4462
    },
    {
      "epoch": 0.872020320437671,
      "grad_norm": 9.0558500289917,
      "learning_rate": 8.752164276610731e-07,
      "loss": 9.3276,
      "step": 4463
    },
    {
      "epoch": 0.8722157092614302,
      "grad_norm": 8.599926948547363,
      "learning_rate": 8.72629004558525e-07,
      "loss": 8.5314,
      "step": 4464
    },
    {
      "epoch": 0.8724110980851896,
      "grad_norm": 8.152388572692871,
      "learning_rate": 8.700452372514345e-07,
      "loss": 8.797,
      "step": 4465
    },
    {
      "epoch": 0.8726064869089488,
      "grad_norm": 9.223265647888184,
      "learning_rate": 8.674651267746769e-07,
      "loss": 8.859,
      "step": 4466
    },
    {
      "epoch": 0.8728018757327081,
      "grad_norm": 9.665291786193848,
      "learning_rate": 8.648886741616691e-07,
      "loss": 9.4288,
      "step": 4467
    },
    {
      "epoch": 0.8729972645564674,
      "grad_norm": 7.774353504180908,
      "learning_rate": 8.623158804443598e-07,
      "loss": 8.5911,
      "step": 4468
    },
    {
      "epoch": 0.8731926533802267,
      "grad_norm": 10.168339729309082,
      "learning_rate": 8.597467466532327e-07,
      "loss": 8.9385,
      "step": 4469
    },
    {
      "epoch": 0.8733880422039859,
      "grad_norm": 7.361769676208496,
      "learning_rate": 8.571812738173047e-07,
      "loss": 8.7944,
      "step": 4470
    },
    {
      "epoch": 0.8735834310277452,
      "grad_norm": 8.525863647460938,
      "learning_rate": 8.546194629641302e-07,
      "loss": 8.6798,
      "step": 4471
    },
    {
      "epoch": 0.8737788198515045,
      "grad_norm": 6.9871087074279785,
      "learning_rate": 8.520613151197899e-07,
      "loss": 8.3553,
      "step": 4472
    },
    {
      "epoch": 0.8739742086752638,
      "grad_norm": 8.88776969909668,
      "learning_rate": 8.49506831308905e-07,
      "loss": 9.5342,
      "step": 4473
    },
    {
      "epoch": 0.874169597499023,
      "grad_norm": 6.036227703094482,
      "learning_rate": 8.469560125546239e-07,
      "loss": 8.7762,
      "step": 4474
    },
    {
      "epoch": 0.8743649863227824,
      "grad_norm": 9.236881256103516,
      "learning_rate": 8.444088598786282e-07,
      "loss": 9.0996,
      "step": 4475
    },
    {
      "epoch": 0.8745603751465416,
      "grad_norm": 8.032596588134766,
      "learning_rate": 8.418653743011329e-07,
      "loss": 9.2407,
      "step": 4476
    },
    {
      "epoch": 0.8747557639703009,
      "grad_norm": 8.645380973815918,
      "learning_rate": 8.39325556840881e-07,
      "loss": 8.2006,
      "step": 4477
    },
    {
      "epoch": 0.8749511527940602,
      "grad_norm": 6.382596969604492,
      "learning_rate": 8.367894085151518e-07,
      "loss": 7.9927,
      "step": 4478
    },
    {
      "epoch": 0.8751465416178195,
      "grad_norm": 6.3066253662109375,
      "learning_rate": 8.342569303397474e-07,
      "loss": 8.2247,
      "step": 4479
    },
    {
      "epoch": 0.8753419304415787,
      "grad_norm": 8.22442626953125,
      "learning_rate": 8.317281233290064e-07,
      "loss": 8.3415,
      "step": 4480
    },
    {
      "epoch": 0.8755373192653381,
      "grad_norm": 10.735391616821289,
      "learning_rate": 8.29202988495793e-07,
      "loss": 10.3575,
      "step": 4481
    },
    {
      "epoch": 0.8757327080890973,
      "grad_norm": 8.855950355529785,
      "learning_rate": 8.266815268515049e-07,
      "loss": 9.7522,
      "step": 4482
    },
    {
      "epoch": 0.8759280969128566,
      "grad_norm": 9.162856101989746,
      "learning_rate": 8.241637394060619e-07,
      "loss": 9.6998,
      "step": 4483
    },
    {
      "epoch": 0.8761234857366159,
      "grad_norm": 9.004673957824707,
      "learning_rate": 8.216496271679198e-07,
      "loss": 8.7978,
      "step": 4484
    },
    {
      "epoch": 0.8763188745603752,
      "grad_norm": 7.955142021179199,
      "learning_rate": 8.191391911440549e-07,
      "loss": 9.3476,
      "step": 4485
    },
    {
      "epoch": 0.8765142633841344,
      "grad_norm": 9.796892166137695,
      "learning_rate": 8.166324323399788e-07,
      "loss": 8.8969,
      "step": 4486
    },
    {
      "epoch": 0.8767096522078938,
      "grad_norm": 10.60122013092041,
      "learning_rate": 8.141293517597238e-07,
      "loss": 9.199,
      "step": 4487
    },
    {
      "epoch": 0.876905041031653,
      "grad_norm": 9.102415084838867,
      "learning_rate": 8.1162995040585e-07,
      "loss": 10.0809,
      "step": 4488
    },
    {
      "epoch": 0.8771004298554123,
      "grad_norm": 8.153091430664062,
      "learning_rate": 8.091342292794491e-07,
      "loss": 9.1124,
      "step": 4489
    },
    {
      "epoch": 0.8772958186791715,
      "grad_norm": 11.209802627563477,
      "learning_rate": 8.066421893801302e-07,
      "loss": 9.4476,
      "step": 4490
    },
    {
      "epoch": 0.8774912075029309,
      "grad_norm": 12.699365615844727,
      "learning_rate": 8.041538317060371e-07,
      "loss": 8.7293,
      "step": 4491
    },
    {
      "epoch": 0.8776865963266901,
      "grad_norm": 7.22158145904541,
      "learning_rate": 8.016691572538315e-07,
      "loss": 9.8362,
      "step": 4492
    },
    {
      "epoch": 0.8778819851504494,
      "grad_norm": 10.028066635131836,
      "learning_rate": 7.99188167018704e-07,
      "loss": 8.9034,
      "step": 4493
    },
    {
      "epoch": 0.8780773739742087,
      "grad_norm": 9.291180610656738,
      "learning_rate": 7.967108619943664e-07,
      "loss": 9.7576,
      "step": 4494
    },
    {
      "epoch": 0.878272762797968,
      "grad_norm": 9.38831615447998,
      "learning_rate": 7.942372431730594e-07,
      "loss": 9.0136,
      "step": 4495
    },
    {
      "epoch": 0.8784681516217272,
      "grad_norm": 8.535799026489258,
      "learning_rate": 7.917673115455415e-07,
      "loss": 9.3348,
      "step": 4496
    },
    {
      "epoch": 0.8786635404454866,
      "grad_norm": 10.475357055664062,
      "learning_rate": 7.893010681010982e-07,
      "loss": 9.7507,
      "step": 4497
    },
    {
      "epoch": 0.8788589292692458,
      "grad_norm": 7.551560401916504,
      "learning_rate": 7.868385138275358e-07,
      "loss": 9.5255,
      "step": 4498
    },
    {
      "epoch": 0.879054318093005,
      "grad_norm": 18.42361831665039,
      "learning_rate": 7.843796497111844e-07,
      "loss": 8.8792,
      "step": 4499
    },
    {
      "epoch": 0.8792497069167644,
      "grad_norm": 11.153907775878906,
      "learning_rate": 7.81924476736896e-07,
      "loss": 8.9052,
      "step": 4500
    },
    {
      "epoch": 0.8794450957405237,
      "grad_norm": 5.823765277862549,
      "learning_rate": 7.79472995888041e-07,
      "loss": 8.4478,
      "step": 4501
    },
    {
      "epoch": 0.8796404845642829,
      "grad_norm": 8.497440338134766,
      "learning_rate": 7.77025208146518e-07,
      "loss": 9.1186,
      "step": 4502
    },
    {
      "epoch": 0.8798358733880423,
      "grad_norm": 7.643743515014648,
      "learning_rate": 7.745811144927362e-07,
      "loss": 7.929,
      "step": 4503
    },
    {
      "epoch": 0.8800312622118015,
      "grad_norm": 14.62929916381836,
      "learning_rate": 7.72140715905636e-07,
      "loss": 9.1529,
      "step": 4504
    },
    {
      "epoch": 0.8802266510355607,
      "grad_norm": 11.714295387268066,
      "learning_rate": 7.697040133626698e-07,
      "loss": 10.5666,
      "step": 4505
    },
    {
      "epoch": 0.88042203985932,
      "grad_norm": 8.457733154296875,
      "learning_rate": 7.67271007839816e-07,
      "loss": 9.4585,
      "step": 4506
    },
    {
      "epoch": 0.8806174286830794,
      "grad_norm": 7.4943366050720215,
      "learning_rate": 7.648417003115637e-07,
      "loss": 9.0691,
      "step": 4507
    },
    {
      "epoch": 0.8808128175068386,
      "grad_norm": 10.112883567810059,
      "learning_rate": 7.624160917509316e-07,
      "loss": 8.9789,
      "step": 4508
    },
    {
      "epoch": 0.8810082063305978,
      "grad_norm": 10.249805450439453,
      "learning_rate": 7.599941831294477e-07,
      "loss": 9.7161,
      "step": 4509
    },
    {
      "epoch": 0.8812035951543572,
      "grad_norm": 7.2333083152771,
      "learning_rate": 7.57575975417163e-07,
      "loss": 9.1904,
      "step": 4510
    },
    {
      "epoch": 0.8813989839781164,
      "grad_norm": 10.90276050567627,
      "learning_rate": 7.551614695826414e-07,
      "loss": 9.4712,
      "step": 4511
    },
    {
      "epoch": 0.8815943728018757,
      "grad_norm": 9.032613754272461,
      "learning_rate": 7.527506665929729e-07,
      "loss": 8.7427,
      "step": 4512
    },
    {
      "epoch": 0.881789761625635,
      "grad_norm": 7.372929096221924,
      "learning_rate": 7.50343567413755e-07,
      "loss": 9.654,
      "step": 4513
    },
    {
      "epoch": 0.8819851504493943,
      "grad_norm": 8.577502250671387,
      "learning_rate": 7.479401730091074e-07,
      "loss": 9.232,
      "step": 4514
    },
    {
      "epoch": 0.8821805392731535,
      "grad_norm": 8.171577453613281,
      "learning_rate": 7.455404843416636e-07,
      "loss": 8.0958,
      "step": 4515
    },
    {
      "epoch": 0.8823759280969129,
      "grad_norm": 10.150381088256836,
      "learning_rate": 7.431445023725736e-07,
      "loss": 9.2528,
      "step": 4516
    },
    {
      "epoch": 0.8825713169206721,
      "grad_norm": 9.050222396850586,
      "learning_rate": 7.40752228061502e-07,
      "loss": 8.2561,
      "step": 4517
    },
    {
      "epoch": 0.8827667057444314,
      "grad_norm": 8.562286376953125,
      "learning_rate": 7.383636623666302e-07,
      "loss": 9.0159,
      "step": 4518
    },
    {
      "epoch": 0.8829620945681907,
      "grad_norm": 9.105795860290527,
      "learning_rate": 7.359788062446527e-07,
      "loss": 9.1769,
      "step": 4519
    },
    {
      "epoch": 0.88315748339195,
      "grad_norm": 8.349502563476562,
      "learning_rate": 7.335976606507766e-07,
      "loss": 9.4373,
      "step": 4520
    },
    {
      "epoch": 0.8833528722157092,
      "grad_norm": 9.987401008605957,
      "learning_rate": 7.312202265387269e-07,
      "loss": 8.829,
      "step": 4521
    },
    {
      "epoch": 0.8835482610394686,
      "grad_norm": 8.053004264831543,
      "learning_rate": 7.288465048607373e-07,
      "loss": 8.9076,
      "step": 4522
    },
    {
      "epoch": 0.8837436498632278,
      "grad_norm": 9.053923606872559,
      "learning_rate": 7.264764965675597e-07,
      "loss": 8.3517,
      "step": 4523
    },
    {
      "epoch": 0.8839390386869871,
      "grad_norm": 8.437475204467773,
      "learning_rate": 7.241102026084534e-07,
      "loss": 8.5202,
      "step": 4524
    },
    {
      "epoch": 0.8841344275107463,
      "grad_norm": 12.07729721069336,
      "learning_rate": 7.21747623931196e-07,
      "loss": 9.1904,
      "step": 4525
    },
    {
      "epoch": 0.8843298163345057,
      "grad_norm": 7.962643146514893,
      "learning_rate": 7.193887614820694e-07,
      "loss": 10.0814,
      "step": 4526
    },
    {
      "epoch": 0.8845252051582649,
      "grad_norm": 6.445322036743164,
      "learning_rate": 7.170336162058755e-07,
      "loss": 8.7302,
      "step": 4527
    },
    {
      "epoch": 0.8847205939820242,
      "grad_norm": 9.202902793884277,
      "learning_rate": 7.146821890459188e-07,
      "loss": 9.811,
      "step": 4528
    },
    {
      "epoch": 0.8849159828057835,
      "grad_norm": 7.17478609085083,
      "learning_rate": 7.123344809440236e-07,
      "loss": 8.906,
      "step": 4529
    },
    {
      "epoch": 0.8851113716295428,
      "grad_norm": 7.45881986618042,
      "learning_rate": 7.099904928405155e-07,
      "loss": 9.1534,
      "step": 4530
    },
    {
      "epoch": 0.885306760453302,
      "grad_norm": 9.53923511505127,
      "learning_rate": 7.076502256742379e-07,
      "loss": 9.2339,
      "step": 4531
    },
    {
      "epoch": 0.8855021492770614,
      "grad_norm": 8.295525550842285,
      "learning_rate": 7.053136803825389e-07,
      "loss": 8.5461,
      "step": 4532
    },
    {
      "epoch": 0.8856975381008206,
      "grad_norm": 8.039118766784668,
      "learning_rate": 7.029808579012765e-07,
      "loss": 9.0157,
      "step": 4533
    },
    {
      "epoch": 0.8858929269245799,
      "grad_norm": 8.169330596923828,
      "learning_rate": 7.00651759164821e-07,
      "loss": 8.5258,
      "step": 4534
    },
    {
      "epoch": 0.8860883157483392,
      "grad_norm": 8.783629417419434,
      "learning_rate": 6.983263851060451e-07,
      "loss": 9.4092,
      "step": 4535
    },
    {
      "epoch": 0.8862837045720985,
      "grad_norm": 9.435042381286621,
      "learning_rate": 6.960047366563372e-07,
      "loss": 9.1183,
      "step": 4536
    },
    {
      "epoch": 0.8864790933958577,
      "grad_norm": 9.772560119628906,
      "learning_rate": 6.936868147455866e-07,
      "loss": 9.5689,
      "step": 4537
    },
    {
      "epoch": 0.8866744822196171,
      "grad_norm": 10.278083801269531,
      "learning_rate": 6.913726203021954e-07,
      "loss": 9.6147,
      "step": 4538
    },
    {
      "epoch": 0.8868698710433763,
      "grad_norm": 8.931450843811035,
      "learning_rate": 6.890621542530684e-07,
      "loss": 9.0847,
      "step": 4539
    },
    {
      "epoch": 0.8870652598671356,
      "grad_norm": 11.48383903503418,
      "learning_rate": 6.867554175236202e-07,
      "loss": 9.6914,
      "step": 4540
    },
    {
      "epoch": 0.8872606486908948,
      "grad_norm": 9.49150276184082,
      "learning_rate": 6.844524110377692e-07,
      "loss": 9.1198,
      "step": 4541
    },
    {
      "epoch": 0.8874560375146542,
      "grad_norm": 10.23884391784668,
      "learning_rate": 6.821531357179434e-07,
      "loss": 8.8346,
      "step": 4542
    },
    {
      "epoch": 0.8876514263384134,
      "grad_norm": 7.415590286254883,
      "learning_rate": 6.798575924850715e-07,
      "loss": 8.6566,
      "step": 4543
    },
    {
      "epoch": 0.8878468151621727,
      "grad_norm": 7.1067633628845215,
      "learning_rate": 6.775657822585913e-07,
      "loss": 9.1218,
      "step": 4544
    },
    {
      "epoch": 0.888042203985932,
      "grad_norm": 8.757865905761719,
      "learning_rate": 6.752777059564431e-07,
      "loss": 8.2709,
      "step": 4545
    },
    {
      "epoch": 0.8882375928096913,
      "grad_norm": 8.671647071838379,
      "learning_rate": 6.729933644950726e-07,
      "loss": 8.8263,
      "step": 4546
    },
    {
      "epoch": 0.8884329816334505,
      "grad_norm": 9.775559425354004,
      "learning_rate": 6.707127587894302e-07,
      "loss": 8.3645,
      "step": 4547
    },
    {
      "epoch": 0.8886283704572099,
      "grad_norm": 8.119025230407715,
      "learning_rate": 6.684358897529675e-07,
      "loss": 8.2647,
      "step": 4548
    },
    {
      "epoch": 0.8888237592809691,
      "grad_norm": 8.645852088928223,
      "learning_rate": 6.661627582976438e-07,
      "loss": 9.4793,
      "step": 4549
    },
    {
      "epoch": 0.8890191481047284,
      "grad_norm": 8.791229248046875,
      "learning_rate": 6.638933653339153e-07,
      "loss": 9.5033,
      "step": 4550
    },
    {
      "epoch": 0.8892145369284877,
      "grad_norm": 9.361802101135254,
      "learning_rate": 6.616277117707493e-07,
      "loss": 9.4369,
      "step": 4551
    },
    {
      "epoch": 0.889409925752247,
      "grad_norm": 8.829449653625488,
      "learning_rate": 6.593657985156055e-07,
      "loss": 9.1888,
      "step": 4552
    },
    {
      "epoch": 0.8896053145760062,
      "grad_norm": 7.402151107788086,
      "learning_rate": 6.571076264744547e-07,
      "loss": 8.5052,
      "step": 4553
    },
    {
      "epoch": 0.8898007033997656,
      "grad_norm": 8.041218757629395,
      "learning_rate": 6.548531965517601e-07,
      "loss": 8.4622,
      "step": 4554
    },
    {
      "epoch": 0.8899960922235248,
      "grad_norm": 8.33278751373291,
      "learning_rate": 6.52602509650495e-07,
      "loss": 8.6969,
      "step": 4555
    },
    {
      "epoch": 0.8901914810472841,
      "grad_norm": 6.467602729797363,
      "learning_rate": 6.503555666721273e-07,
      "loss": 9.3724,
      "step": 4556
    },
    {
      "epoch": 0.8903868698710434,
      "grad_norm": 7.473740577697754,
      "learning_rate": 6.481123685166291e-07,
      "loss": 8.0025,
      "step": 4557
    },
    {
      "epoch": 0.8905822586948027,
      "grad_norm": 8.307002067565918,
      "learning_rate": 6.458729160824706e-07,
      "loss": 9.0213,
      "step": 4558
    },
    {
      "epoch": 0.8907776475185619,
      "grad_norm": 9.312746047973633,
      "learning_rate": 6.436372102666188e-07,
      "loss": 9.2817,
      "step": 4559
    },
    {
      "epoch": 0.8909730363423212,
      "grad_norm": 9.156469345092773,
      "learning_rate": 6.414052519645475e-07,
      "loss": 9.197,
      "step": 4560
    },
    {
      "epoch": 0.8911684251660805,
      "grad_norm": 8.96273136138916,
      "learning_rate": 6.391770420702215e-07,
      "loss": 10.073,
      "step": 4561
    },
    {
      "epoch": 0.8913638139898398,
      "grad_norm": 8.047943115234375,
      "learning_rate": 6.369525814761124e-07,
      "loss": 9.3223,
      "step": 4562
    },
    {
      "epoch": 0.891559202813599,
      "grad_norm": 7.415777206420898,
      "learning_rate": 6.347318710731809e-07,
      "loss": 9.2714,
      "step": 4563
    },
    {
      "epoch": 0.8917545916373584,
      "grad_norm": 8.42771053314209,
      "learning_rate": 6.325149117508955e-07,
      "loss": 9.1933,
      "step": 4564
    },
    {
      "epoch": 0.8919499804611176,
      "grad_norm": 7.360374450683594,
      "learning_rate": 6.303017043972126e-07,
      "loss": 8.4876,
      "step": 4565
    },
    {
      "epoch": 0.8921453692848769,
      "grad_norm": 8.370670318603516,
      "learning_rate": 6.280922498985942e-07,
      "loss": 8.844,
      "step": 4566
    },
    {
      "epoch": 0.8923407581086362,
      "grad_norm": 7.138886451721191,
      "learning_rate": 6.258865491399913e-07,
      "loss": 8.1337,
      "step": 4567
    },
    {
      "epoch": 0.8925361469323955,
      "grad_norm": 9.302095413208008,
      "learning_rate": 6.236846030048605e-07,
      "loss": 9.2156,
      "step": 4568
    },
    {
      "epoch": 0.8927315357561547,
      "grad_norm": 8.165783882141113,
      "learning_rate": 6.214864123751463e-07,
      "loss": 9.412,
      "step": 4569
    },
    {
      "epoch": 0.8929269245799141,
      "grad_norm": 10.003043174743652,
      "learning_rate": 6.192919781312933e-07,
      "loss": 9.6109,
      "step": 4570
    },
    {
      "epoch": 0.8931223134036733,
      "grad_norm": 9.300692558288574,
      "learning_rate": 6.171013011522409e-07,
      "loss": 9.0853,
      "step": 4571
    },
    {
      "epoch": 0.8933177022274326,
      "grad_norm": 13.417641639709473,
      "learning_rate": 6.149143823154213e-07,
      "loss": 9.0996,
      "step": 4572
    },
    {
      "epoch": 0.8935130910511919,
      "grad_norm": 10.26941204071045,
      "learning_rate": 6.127312224967652e-07,
      "loss": 9.5182,
      "step": 4573
    },
    {
      "epoch": 0.8937084798749512,
      "grad_norm": 7.442628383636475,
      "learning_rate": 6.105518225706952e-07,
      "loss": 8.868,
      "step": 4574
    },
    {
      "epoch": 0.8939038686987104,
      "grad_norm": 8.16067123413086,
      "learning_rate": 6.083761834101298e-07,
      "loss": 8.9479,
      "step": 4575
    },
    {
      "epoch": 0.8940992575224697,
      "grad_norm": 7.069525718688965,
      "learning_rate": 6.062043058864775e-07,
      "loss": 8.8524,
      "step": 4576
    },
    {
      "epoch": 0.894294646346229,
      "grad_norm": 8.341736793518066,
      "learning_rate": 6.040361908696457e-07,
      "loss": 8.4742,
      "step": 4577
    },
    {
      "epoch": 0.8944900351699883,
      "grad_norm": 10.101537704467773,
      "learning_rate": 6.018718392280299e-07,
      "loss": 9.0462,
      "step": 4578
    },
    {
      "epoch": 0.8946854239937475,
      "grad_norm": 7.973488807678223,
      "learning_rate": 5.997112518285208e-07,
      "loss": 9.04,
      "step": 4579
    },
    {
      "epoch": 0.8948808128175069,
      "grad_norm": 6.7695722579956055,
      "learning_rate": 5.975544295364998e-07,
      "loss": 8.5619,
      "step": 4580
    },
    {
      "epoch": 0.8950762016412661,
      "grad_norm": 7.932429313659668,
      "learning_rate": 5.954013732158436e-07,
      "loss": 8.5523,
      "step": 4581
    },
    {
      "epoch": 0.8952715904650254,
      "grad_norm": 10.644766807556152,
      "learning_rate": 5.932520837289135e-07,
      "loss": 10.0759,
      "step": 4582
    },
    {
      "epoch": 0.8954669792887847,
      "grad_norm": 8.38106918334961,
      "learning_rate": 5.91106561936573e-07,
      "loss": 9.2323,
      "step": 4583
    },
    {
      "epoch": 0.895662368112544,
      "grad_norm": 9.988547325134277,
      "learning_rate": 5.889648086981659e-07,
      "loss": 9.3092,
      "step": 4584
    },
    {
      "epoch": 0.8958577569363032,
      "grad_norm": 7.1622395515441895,
      "learning_rate": 5.868268248715292e-07,
      "loss": 7.0635,
      "step": 4585
    },
    {
      "epoch": 0.8960531457600626,
      "grad_norm": 11.381697654724121,
      "learning_rate": 5.84692611312997e-07,
      "loss": 9.1964,
      "step": 4586
    },
    {
      "epoch": 0.8962485345838218,
      "grad_norm": 10.23323917388916,
      "learning_rate": 5.825621688773842e-07,
      "loss": 9.7945,
      "step": 4587
    },
    {
      "epoch": 0.8964439234075811,
      "grad_norm": 21.31047248840332,
      "learning_rate": 5.804354984180017e-07,
      "loss": 8.9616,
      "step": 4588
    },
    {
      "epoch": 0.8966393122313404,
      "grad_norm": 8.889069557189941,
      "learning_rate": 5.783126007866435e-07,
      "loss": 8.7006,
      "step": 4589
    },
    {
      "epoch": 0.8968347010550997,
      "grad_norm": 8.523070335388184,
      "learning_rate": 5.761934768335997e-07,
      "loss": 9.2636,
      "step": 4590
    },
    {
      "epoch": 0.8970300898788589,
      "grad_norm": 17.558130264282227,
      "learning_rate": 5.740781274076423e-07,
      "loss": 9.6848,
      "step": 4591
    },
    {
      "epoch": 0.8972254787026183,
      "grad_norm": 7.022171497344971,
      "learning_rate": 5.719665533560381e-07,
      "loss": 8.2926,
      "step": 4592
    },
    {
      "epoch": 0.8974208675263775,
      "grad_norm": 12.051819801330566,
      "learning_rate": 5.698587555245339e-07,
      "loss": 9.8679,
      "step": 4593
    },
    {
      "epoch": 0.8976162563501368,
      "grad_norm": 8.585456848144531,
      "learning_rate": 5.67754734757372e-07,
      "loss": 9.2112,
      "step": 4594
    },
    {
      "epoch": 0.897811645173896,
      "grad_norm": 8.774697303771973,
      "learning_rate": 5.65654491897274e-07,
      "loss": 9.3123,
      "step": 4595
    },
    {
      "epoch": 0.8980070339976554,
      "grad_norm": 7.8391313552856445,
      "learning_rate": 5.635580277854558e-07,
      "loss": 8.8919,
      "step": 4596
    },
    {
      "epoch": 0.8982024228214146,
      "grad_norm": 11.301671981811523,
      "learning_rate": 5.614653432616158e-07,
      "loss": 9.3398,
      "step": 4597
    },
    {
      "epoch": 0.8983978116451738,
      "grad_norm": 7.452347755432129,
      "learning_rate": 5.593764391639367e-07,
      "loss": 9.0038,
      "step": 4598
    },
    {
      "epoch": 0.8985932004689332,
      "grad_norm": 6.5044121742248535,
      "learning_rate": 5.572913163290927e-07,
      "loss": 8.6604,
      "step": 4599
    },
    {
      "epoch": 0.8987885892926925,
      "grad_norm": 10.333758354187012,
      "learning_rate": 5.552099755922369e-07,
      "loss": 8.8245,
      "step": 4600
    },
    {
      "epoch": 0.8989839781164517,
      "grad_norm": 7.728987216949463,
      "learning_rate": 5.531324177870145e-07,
      "loss": 9.1096,
      "step": 4601
    },
    {
      "epoch": 0.8991793669402111,
      "grad_norm": 10.902887344360352,
      "learning_rate": 5.510586437455478e-07,
      "loss": 8.6363,
      "step": 4602
    },
    {
      "epoch": 0.8993747557639703,
      "grad_norm": 8.748449325561523,
      "learning_rate": 5.489886542984524e-07,
      "loss": 9.1152,
      "step": 4603
    },
    {
      "epoch": 0.8995701445877295,
      "grad_norm": 8.182178497314453,
      "learning_rate": 5.469224502748193e-07,
      "loss": 9.7359,
      "step": 4604
    },
    {
      "epoch": 0.8997655334114889,
      "grad_norm": 10.667359352111816,
      "learning_rate": 5.448600325022302e-07,
      "loss": 8.3748,
      "step": 4605
    },
    {
      "epoch": 0.8999609222352482,
      "grad_norm": 7.522793292999268,
      "learning_rate": 5.428014018067441e-07,
      "loss": 8.6727,
      "step": 4606
    },
    {
      "epoch": 0.9001563110590074,
      "grad_norm": 7.92480993270874,
      "learning_rate": 5.407465590129101e-07,
      "loss": 8.1705,
      "step": 4607
    },
    {
      "epoch": 0.9003516998827668,
      "grad_norm": 6.827234268188477,
      "learning_rate": 5.386955049437548e-07,
      "loss": 8.2938,
      "step": 4608
    },
    {
      "epoch": 0.900547088706526,
      "grad_norm": 7.39226770401001,
      "learning_rate": 5.366482404207896e-07,
      "loss": 9.7872,
      "step": 4609
    },
    {
      "epoch": 0.9007424775302852,
      "grad_norm": 7.509429454803467,
      "learning_rate": 5.346047662640075e-07,
      "loss": 8.5024,
      "step": 4610
    },
    {
      "epoch": 0.9009378663540446,
      "grad_norm": 9.19665813446045,
      "learning_rate": 5.325650832918816e-07,
      "loss": 9.1335,
      "step": 4611
    },
    {
      "epoch": 0.9011332551778038,
      "grad_norm": 10.296882629394531,
      "learning_rate": 5.305291923213718e-07,
      "loss": 9.4044,
      "step": 4612
    },
    {
      "epoch": 0.9013286440015631,
      "grad_norm": 8.83753776550293,
      "learning_rate": 5.284970941679113e-07,
      "loss": 9.4248,
      "step": 4613
    },
    {
      "epoch": 0.9015240328253223,
      "grad_norm": 8.993282318115234,
      "learning_rate": 5.264687896454224e-07,
      "loss": 8.6009,
      "step": 4614
    },
    {
      "epoch": 0.9017194216490817,
      "grad_norm": 6.61049747467041,
      "learning_rate": 5.244442795663007e-07,
      "loss": 7.7182,
      "step": 4615
    },
    {
      "epoch": 0.9019148104728409,
      "grad_norm": 7.919312477111816,
      "learning_rate": 5.224235647414277e-07,
      "loss": 9.915,
      "step": 4616
    },
    {
      "epoch": 0.9021101992966002,
      "grad_norm": 9.764341354370117,
      "learning_rate": 5.204066459801615e-07,
      "loss": 8.7336,
      "step": 4617
    },
    {
      "epoch": 0.9023055881203595,
      "grad_norm": 7.9409637451171875,
      "learning_rate": 5.183935240903415e-07,
      "loss": 8.6041,
      "step": 4618
    },
    {
      "epoch": 0.9025009769441188,
      "grad_norm": 8.285787582397461,
      "learning_rate": 5.163841998782837e-07,
      "loss": 9.4445,
      "step": 4619
    },
    {
      "epoch": 0.902696365767878,
      "grad_norm": 9.423076629638672,
      "learning_rate": 5.143786741487877e-07,
      "loss": 8.5114,
      "step": 4620
    },
    {
      "epoch": 0.9028917545916374,
      "grad_norm": 11.886407852172852,
      "learning_rate": 5.123769477051255e-07,
      "loss": 9.2892,
      "step": 4621
    },
    {
      "epoch": 0.9030871434153966,
      "grad_norm": 7.831228733062744,
      "learning_rate": 5.103790213490545e-07,
      "loss": 8.2329,
      "step": 4622
    },
    {
      "epoch": 0.9032825322391559,
      "grad_norm": 10.496557235717773,
      "learning_rate": 5.083848958808035e-07,
      "loss": 9.1138,
      "step": 4623
    },
    {
      "epoch": 0.9034779210629152,
      "grad_norm": 17.42132568359375,
      "learning_rate": 5.063945720990803e-07,
      "loss": 9.1258,
      "step": 4624
    },
    {
      "epoch": 0.9036733098866745,
      "grad_norm": 6.3549885749816895,
      "learning_rate": 5.044080508010751e-07,
      "loss": 8.6081,
      "step": 4625
    },
    {
      "epoch": 0.9038686987104337,
      "grad_norm": 8.271417617797852,
      "learning_rate": 5.02425332782448e-07,
      "loss": 8.8358,
      "step": 4626
    },
    {
      "epoch": 0.9040640875341931,
      "grad_norm": 9.527425765991211,
      "learning_rate": 5.004464188373426e-07,
      "loss": 8.5627,
      "step": 4627
    },
    {
      "epoch": 0.9042594763579523,
      "grad_norm": 7.8970232009887695,
      "learning_rate": 4.984713097583715e-07,
      "loss": 9.9096,
      "step": 4628
    },
    {
      "epoch": 0.9044548651817116,
      "grad_norm": 8.61898422241211,
      "learning_rate": 4.96500006336631e-07,
      "loss": 8.2139,
      "step": 4629
    },
    {
      "epoch": 0.9046502540054708,
      "grad_norm": 9.109868049621582,
      "learning_rate": 4.94532509361686e-07,
      "loss": 8.6737,
      "step": 4630
    },
    {
      "epoch": 0.9048456428292302,
      "grad_norm": 8.05453872680664,
      "learning_rate": 4.925688196215828e-07,
      "loss": 9.8056,
      "step": 4631
    },
    {
      "epoch": 0.9050410316529894,
      "grad_norm": 88.9920883178711,
      "learning_rate": 4.906089379028367e-07,
      "loss": 10.1364,
      "step": 4632
    },
    {
      "epoch": 0.9052364204767487,
      "grad_norm": 7.312097072601318,
      "learning_rate": 4.886528649904465e-07,
      "loss": 8.6661,
      "step": 4633
    },
    {
      "epoch": 0.905431809300508,
      "grad_norm": 8.211421966552734,
      "learning_rate": 4.867006016678744e-07,
      "loss": 8.9301,
      "step": 4634
    },
    {
      "epoch": 0.9056271981242673,
      "grad_norm": 12.64998722076416,
      "learning_rate": 4.84752148717067e-07,
      "loss": 9.0967,
      "step": 4635
    },
    {
      "epoch": 0.9058225869480265,
      "grad_norm": 7.412219047546387,
      "learning_rate": 4.828075069184379e-07,
      "loss": 7.99,
      "step": 4636
    },
    {
      "epoch": 0.9060179757717859,
      "grad_norm": 9.487683296203613,
      "learning_rate": 4.808666770508752e-07,
      "loss": 8.7177,
      "step": 4637
    },
    {
      "epoch": 0.9062133645955451,
      "grad_norm": 11.212496757507324,
      "learning_rate": 4.789296598917459e-07,
      "loss": 9.1199,
      "step": 4638
    },
    {
      "epoch": 0.9064087534193044,
      "grad_norm": 8.739277839660645,
      "learning_rate": 4.769964562168805e-07,
      "loss": 9.9834,
      "step": 4639
    },
    {
      "epoch": 0.9066041422430637,
      "grad_norm": 13.263821601867676,
      "learning_rate": 4.750670668005908e-07,
      "loss": 9.4101,
      "step": 4640
    },
    {
      "epoch": 0.906799531066823,
      "grad_norm": 9.429154396057129,
      "learning_rate": 4.7314149241565413e-07,
      "loss": 8.9467,
      "step": 4641
    },
    {
      "epoch": 0.9069949198905822,
      "grad_norm": 9.144912719726562,
      "learning_rate": 4.712197338333269e-07,
      "loss": 9.005,
      "step": 4642
    },
    {
      "epoch": 0.9071903087143416,
      "grad_norm": 8.525959014892578,
      "learning_rate": 4.693017918233278e-07,
      "loss": 8.4626,
      "step": 4643
    },
    {
      "epoch": 0.9073856975381008,
      "grad_norm": 11.013566017150879,
      "learning_rate": 4.67387667153858e-07,
      "loss": 9.1099,
      "step": 4644
    },
    {
      "epoch": 0.9075810863618601,
      "grad_norm": 9.597681999206543,
      "learning_rate": 4.6547736059157857e-07,
      "loss": 9.493,
      "step": 4645
    },
    {
      "epoch": 0.9077764751856194,
      "grad_norm": 6.093164443969727,
      "learning_rate": 4.6357087290162973e-07,
      "loss": 7.5166,
      "step": 4646
    },
    {
      "epoch": 0.9079718640093787,
      "grad_norm": 10.783944129943848,
      "learning_rate": 4.6166820484761954e-07,
      "loss": 9.2103,
      "step": 4647
    },
    {
      "epoch": 0.9081672528331379,
      "grad_norm": 7.36179256439209,
      "learning_rate": 4.597693571916229e-07,
      "loss": 8.1978,
      "step": 4648
    },
    {
      "epoch": 0.9083626416568972,
      "grad_norm": 10.97280216217041,
      "learning_rate": 4.5787433069418796e-07,
      "loss": 10.0678,
      "step": 4649
    },
    {
      "epoch": 0.9085580304806565,
      "grad_norm": 8.537384986877441,
      "learning_rate": 4.5598312611433324e-07,
      "loss": 9.7343,
      "step": 4650
    },
    {
      "epoch": 0.9087534193044158,
      "grad_norm": 7.580404281616211,
      "learning_rate": 4.540957442095428e-07,
      "loss": 8.5973,
      "step": 4651
    },
    {
      "epoch": 0.908948808128175,
      "grad_norm": 8.775703430175781,
      "learning_rate": 4.522121857357742e-07,
      "loss": 8.0289,
      "step": 4652
    },
    {
      "epoch": 0.9091441969519344,
      "grad_norm": 9.910072326660156,
      "learning_rate": 4.503324514474483e-07,
      "loss": 9.032,
      "step": 4653
    },
    {
      "epoch": 0.9093395857756936,
      "grad_norm": 13.489008903503418,
      "learning_rate": 4.484565420974596e-07,
      "loss": 9.2559,
      "step": 4654
    },
    {
      "epoch": 0.9095349745994529,
      "grad_norm": 7.9361467361450195,
      "learning_rate": 4.4658445843716704e-07,
      "loss": 8.5171,
      "step": 4655
    },
    {
      "epoch": 0.9097303634232122,
      "grad_norm": 9.057096481323242,
      "learning_rate": 4.4471620121639634e-07,
      "loss": 8.3678,
      "step": 4656
    },
    {
      "epoch": 0.9099257522469715,
      "grad_norm": 9.839863777160645,
      "learning_rate": 4.428517711834468e-07,
      "loss": 8.7929,
      "step": 4657
    },
    {
      "epoch": 0.9101211410707307,
      "grad_norm": 7.448708534240723,
      "learning_rate": 4.4099116908507545e-07,
      "loss": 8.6383,
      "step": 4658
    },
    {
      "epoch": 0.9103165298944901,
      "grad_norm": 8.156187057495117,
      "learning_rate": 4.391343956665151e-07,
      "loss": 8.5996,
      "step": 4659
    },
    {
      "epoch": 0.9105119187182493,
      "grad_norm": 7.864421844482422,
      "learning_rate": 4.372814516714596e-07,
      "loss": 9.6516,
      "step": 4660
    },
    {
      "epoch": 0.9107073075420086,
      "grad_norm": 9.196493148803711,
      "learning_rate": 4.3543233784207106e-07,
      "loss": 9.4084,
      "step": 4661
    },
    {
      "epoch": 0.9109026963657679,
      "grad_norm": 7.9403910636901855,
      "learning_rate": 4.335870549189758e-07,
      "loss": 8.8431,
      "step": 4662
    },
    {
      "epoch": 0.9110980851895272,
      "grad_norm": 7.183925628662109,
      "learning_rate": 4.317456036412682e-07,
      "loss": 8.0112,
      "step": 4663
    },
    {
      "epoch": 0.9112934740132864,
      "grad_norm": 10.258187294006348,
      "learning_rate": 4.29907984746506e-07,
      "loss": 7.8186,
      "step": 4664
    },
    {
      "epoch": 0.9114888628370457,
      "grad_norm": 8.958974838256836,
      "learning_rate": 4.2807419897071376e-07,
      "loss": 8.9419,
      "step": 4665
    },
    {
      "epoch": 0.911684251660805,
      "grad_norm": 8.800867080688477,
      "learning_rate": 4.262442470483774e-07,
      "loss": 9.1317,
      "step": 4666
    },
    {
      "epoch": 0.9118796404845643,
      "grad_norm": 7.859527111053467,
      "learning_rate": 4.244181297124528e-07,
      "loss": 8.9888,
      "step": 4667
    },
    {
      "epoch": 0.9120750293083235,
      "grad_norm": 8.100984573364258,
      "learning_rate": 4.225958476943537e-07,
      "loss": 9.5326,
      "step": 4668
    },
    {
      "epoch": 0.9122704181320829,
      "grad_norm": 8.952960968017578,
      "learning_rate": 4.2077740172396074e-07,
      "loss": 9.0005,
      "step": 4669
    },
    {
      "epoch": 0.9124658069558421,
      "grad_norm": 7.81179141998291,
      "learning_rate": 4.189627925296202e-07,
      "loss": 7.9352,
      "step": 4670
    },
    {
      "epoch": 0.9126611957796014,
      "grad_norm": 7.670047760009766,
      "learning_rate": 4.171520208381363e-07,
      "loss": 9.8982,
      "step": 4671
    },
    {
      "epoch": 0.9128565846033607,
      "grad_norm": 8.656058311462402,
      "learning_rate": 4.153450873747822e-07,
      "loss": 8.8703,
      "step": 4672
    },
    {
      "epoch": 0.91305197342712,
      "grad_norm": 7.943114757537842,
      "learning_rate": 4.135419928632889e-07,
      "loss": 8.4271,
      "step": 4673
    },
    {
      "epoch": 0.9132473622508792,
      "grad_norm": 7.806142807006836,
      "learning_rate": 4.117427380258532e-07,
      "loss": 9.5021,
      "step": 4674
    },
    {
      "epoch": 0.9134427510746386,
      "grad_norm": 9.020642280578613,
      "learning_rate": 4.0994732358313085e-07,
      "loss": 8.1661,
      "step": 4675
    },
    {
      "epoch": 0.9136381398983978,
      "grad_norm": 35.45269012451172,
      "learning_rate": 4.0815575025424436e-07,
      "loss": 10.1253,
      "step": 4676
    },
    {
      "epoch": 0.9138335287221571,
      "grad_norm": 10.363656997680664,
      "learning_rate": 4.063680187567698e-07,
      "loss": 8.9976,
      "step": 4677
    },
    {
      "epoch": 0.9140289175459164,
      "grad_norm": 7.983797073364258,
      "learning_rate": 4.0458412980675323e-07,
      "loss": 9.0811,
      "step": 4678
    },
    {
      "epoch": 0.9142243063696757,
      "grad_norm": 8.363759994506836,
      "learning_rate": 4.028040841186942e-07,
      "loss": 8.5338,
      "step": 4679
    },
    {
      "epoch": 0.9144196951934349,
      "grad_norm": 8.386238098144531,
      "learning_rate": 4.0102788240555914e-07,
      "loss": 9.381,
      "step": 4680
    },
    {
      "epoch": 0.9146150840171943,
      "grad_norm": 11.326103210449219,
      "learning_rate": 3.992555253787711e-07,
      "loss": 9.0722,
      "step": 4681
    },
    {
      "epoch": 0.9148104728409535,
      "grad_norm": 8.53385066986084,
      "learning_rate": 3.974870137482123e-07,
      "loss": 9.4061,
      "step": 4682
    },
    {
      "epoch": 0.9150058616647128,
      "grad_norm": 10.785691261291504,
      "learning_rate": 3.957223482222294e-07,
      "loss": 9.9972,
      "step": 4683
    },
    {
      "epoch": 0.915201250488472,
      "grad_norm": 8.604122161865234,
      "learning_rate": 3.9396152950762267e-07,
      "loss": 8.727,
      "step": 4684
    },
    {
      "epoch": 0.9153966393122314,
      "grad_norm": 9.557936668395996,
      "learning_rate": 3.922045583096579e-07,
      "loss": 9.3909,
      "step": 4685
    },
    {
      "epoch": 0.9155920281359906,
      "grad_norm": 8.236830711364746,
      "learning_rate": 3.904514353320543e-07,
      "loss": 9.4076,
      "step": 4686
    },
    {
      "epoch": 0.9157874169597499,
      "grad_norm": 8.740063667297363,
      "learning_rate": 3.887021612769937e-07,
      "loss": 8.691,
      "step": 4687
    },
    {
      "epoch": 0.9159828057835092,
      "grad_norm": 8.937166213989258,
      "learning_rate": 3.869567368451121e-07,
      "loss": 9.3647,
      "step": 4688
    },
    {
      "epoch": 0.9161781946072685,
      "grad_norm": 8.27723217010498,
      "learning_rate": 3.852151627355094e-07,
      "loss": 7.8112,
      "step": 4689
    },
    {
      "epoch": 0.9163735834310277,
      "grad_norm": 56.198974609375,
      "learning_rate": 3.834774396457375e-07,
      "loss": 7.8735,
      "step": 4690
    },
    {
      "epoch": 0.9165689722547871,
      "grad_norm": 10.308794021606445,
      "learning_rate": 3.817435682718096e-07,
      "loss": 9.0237,
      "step": 4691
    },
    {
      "epoch": 0.9167643610785463,
      "grad_norm": 10.48629379272461,
      "learning_rate": 3.8001354930819467e-07,
      "loss": 9.9133,
      "step": 4692
    },
    {
      "epoch": 0.9169597499023056,
      "grad_norm": 7.921257019042969,
      "learning_rate": 3.7828738344782067e-07,
      "loss": 7.5941,
      "step": 4693
    },
    {
      "epoch": 0.9171551387260649,
      "grad_norm": 6.670938968658447,
      "learning_rate": 3.76565071382069e-07,
      "loss": 8.4198,
      "step": 4694
    },
    {
      "epoch": 0.9173505275498242,
      "grad_norm": 8.655702590942383,
      "learning_rate": 3.748466138007778e-07,
      "loss": 8.5202,
      "step": 4695
    },
    {
      "epoch": 0.9175459163735834,
      "grad_norm": 9.957679748535156,
      "learning_rate": 3.7313201139224544e-07,
      "loss": 9.579,
      "step": 4696
    },
    {
      "epoch": 0.9177413051973428,
      "grad_norm": 9.686959266662598,
      "learning_rate": 3.714212648432203e-07,
      "loss": 10.1259,
      "step": 4697
    },
    {
      "epoch": 0.917936694021102,
      "grad_norm": 9.702391624450684,
      "learning_rate": 3.6971437483891427e-07,
      "loss": 9.2411,
      "step": 4698
    },
    {
      "epoch": 0.9181320828448613,
      "grad_norm": 7.698165416717529,
      "learning_rate": 3.6801134206298386e-07,
      "loss": 9.0435,
      "step": 4699
    },
    {
      "epoch": 0.9183274716686206,
      "grad_norm": 11.09498119354248,
      "learning_rate": 3.6631216719755225e-07,
      "loss": 9.8762,
      "step": 4700
    },
    {
      "epoch": 0.9185228604923799,
      "grad_norm": 9.750531196594238,
      "learning_rate": 3.6461685092318735e-07,
      "loss": 9.7316,
      "step": 4701
    },
    {
      "epoch": 0.9187182493161391,
      "grad_norm": 69.85716247558594,
      "learning_rate": 3.6292539391891814e-07,
      "loss": 10.0667,
      "step": 4702
    },
    {
      "epoch": 0.9189136381398983,
      "grad_norm": 7.222912311553955,
      "learning_rate": 3.6123779686222493e-07,
      "loss": 9.3583,
      "step": 4703
    },
    {
      "epoch": 0.9191090269636577,
      "grad_norm": 7.434391498565674,
      "learning_rate": 3.595540604290437e-07,
      "loss": 9.436,
      "step": 4704
    },
    {
      "epoch": 0.919304415787417,
      "grad_norm": 9.064720153808594,
      "learning_rate": 3.578741852937617e-07,
      "loss": 9.2944,
      "step": 4705
    },
    {
      "epoch": 0.9194998046111762,
      "grad_norm": 7.21022367477417,
      "learning_rate": 3.561981721292218e-07,
      "loss": 8.2714,
      "step": 4706
    },
    {
      "epoch": 0.9196951934349356,
      "grad_norm": 7.854347229003906,
      "learning_rate": 3.5452602160672033e-07,
      "loss": 9.0663,
      "step": 4707
    },
    {
      "epoch": 0.9198905822586948,
      "grad_norm": 9.648550987243652,
      "learning_rate": 3.5285773439600377e-07,
      "loss": 10.1147,
      "step": 4708
    },
    {
      "epoch": 0.920085971082454,
      "grad_norm": 8.285013198852539,
      "learning_rate": 3.5119331116527434e-07,
      "loss": 9.0837,
      "step": 4709
    },
    {
      "epoch": 0.9202813599062134,
      "grad_norm": 11.687288284301758,
      "learning_rate": 3.4953275258118314e-07,
      "loss": 9.547,
      "step": 4710
    },
    {
      "epoch": 0.9204767487299726,
      "grad_norm": 20.341156005859375,
      "learning_rate": 3.478760593088382e-07,
      "loss": 9.4041,
      "step": 4711
    },
    {
      "epoch": 0.9206721375537319,
      "grad_norm": 11.171713829040527,
      "learning_rate": 3.462232320117942e-07,
      "loss": 7.7676,
      "step": 4712
    },
    {
      "epoch": 0.9208675263774913,
      "grad_norm": 9.766027450561523,
      "learning_rate": 3.445742713520617e-07,
      "loss": 9.4757,
      "step": 4713
    },
    {
      "epoch": 0.9210629152012505,
      "grad_norm": 8.181708335876465,
      "learning_rate": 3.429291779900989e-07,
      "loss": 8.9366,
      "step": 4714
    },
    {
      "epoch": 0.9212583040250097,
      "grad_norm": 14.08752155303955,
      "learning_rate": 3.4128795258481896e-07,
      "loss": 9.5898,
      "step": 4715
    },
    {
      "epoch": 0.9214536928487691,
      "grad_norm": 7.5151591300964355,
      "learning_rate": 3.396505957935814e-07,
      "loss": 7.3947,
      "step": 4716
    },
    {
      "epoch": 0.9216490816725283,
      "grad_norm": 9.194290161132812,
      "learning_rate": 3.380171082721995e-07,
      "loss": 9.9539,
      "step": 4717
    },
    {
      "epoch": 0.9218444704962876,
      "grad_norm": 8.23127269744873,
      "learning_rate": 3.3638749067493534e-07,
      "loss": 9.0621,
      "step": 4718
    },
    {
      "epoch": 0.9220398593200468,
      "grad_norm": 8.573027610778809,
      "learning_rate": 3.3476174365450255e-07,
      "loss": 8.5441,
      "step": 4719
    },
    {
      "epoch": 0.9222352481438062,
      "grad_norm": 10.570746421813965,
      "learning_rate": 3.331398678620623e-07,
      "loss": 9.3848,
      "step": 4720
    },
    {
      "epoch": 0.9224306369675654,
      "grad_norm": 8.211080551147461,
      "learning_rate": 3.3152186394722506e-07,
      "loss": 8.2236,
      "step": 4721
    },
    {
      "epoch": 0.9226260257913247,
      "grad_norm": 8.340023040771484,
      "learning_rate": 3.299077325580535e-07,
      "loss": 9.4421,
      "step": 4722
    },
    {
      "epoch": 0.922821414615084,
      "grad_norm": 7.6149678230285645,
      "learning_rate": 3.2829747434105606e-07,
      "loss": 8.7159,
      "step": 4723
    },
    {
      "epoch": 0.9230168034388433,
      "grad_norm": 8.387945175170898,
      "learning_rate": 3.2669108994119236e-07,
      "loss": 9.4417,
      "step": 4724
    },
    {
      "epoch": 0.9232121922626025,
      "grad_norm": 8.78463363647461,
      "learning_rate": 3.250885800018677e-07,
      "loss": 9.0099,
      "step": 4725
    },
    {
      "epoch": 0.9234075810863619,
      "grad_norm": 7.261391639709473,
      "learning_rate": 3.234899451649398e-07,
      "loss": 9.0144,
      "step": 4726
    },
    {
      "epoch": 0.9236029699101211,
      "grad_norm": 10.1168212890625,
      "learning_rate": 3.2189518607070756e-07,
      "loss": 9.5387,
      "step": 4727
    },
    {
      "epoch": 0.9237983587338804,
      "grad_norm": 8.67878246307373,
      "learning_rate": 3.203043033579256e-07,
      "loss": 9.395,
      "step": 4728
    },
    {
      "epoch": 0.9239937475576397,
      "grad_norm": 7.543828010559082,
      "learning_rate": 3.1871729766378867e-07,
      "loss": 8.5915,
      "step": 4729
    },
    {
      "epoch": 0.924189136381399,
      "grad_norm": 7.960564613342285,
      "learning_rate": 3.171341696239438e-07,
      "loss": 9.3239,
      "step": 4730
    },
    {
      "epoch": 0.9243845252051582,
      "grad_norm": 9.259940147399902,
      "learning_rate": 3.1555491987248056e-07,
      "loss": 8.4018,
      "step": 4731
    },
    {
      "epoch": 0.9245799140289176,
      "grad_norm": 8.424381256103516,
      "learning_rate": 3.139795490419417e-07,
      "loss": 8.2413,
      "step": 4732
    },
    {
      "epoch": 0.9247753028526768,
      "grad_norm": 9.10590648651123,
      "learning_rate": 3.1240805776330816e-07,
      "loss": 8.8184,
      "step": 4733
    },
    {
      "epoch": 0.9249706916764361,
      "grad_norm": 6.929506778717041,
      "learning_rate": 3.10840446666012e-07,
      "loss": 8.5184,
      "step": 4734
    },
    {
      "epoch": 0.9251660805001954,
      "grad_norm": 11.383835792541504,
      "learning_rate": 3.09276716377932e-07,
      "loss": 9.9644,
      "step": 4735
    },
    {
      "epoch": 0.9253614693239547,
      "grad_norm": 7.0921406745910645,
      "learning_rate": 3.077168675253872e-07,
      "loss": 9.0621,
      "step": 4736
    },
    {
      "epoch": 0.9255568581477139,
      "grad_norm": 9.944376945495605,
      "learning_rate": 3.0616090073314896e-07,
      "loss": 9.0666,
      "step": 4737
    },
    {
      "epoch": 0.9257522469714732,
      "grad_norm": 7.62268590927124,
      "learning_rate": 3.046088166244276e-07,
      "loss": 8.8974,
      "step": 4738
    },
    {
      "epoch": 0.9259476357952325,
      "grad_norm": 9.146649360656738,
      "learning_rate": 3.0306061582088374e-07,
      "loss": 9.1993,
      "step": 4739
    },
    {
      "epoch": 0.9261430246189918,
      "grad_norm": 9.146553993225098,
      "learning_rate": 3.015162989426168e-07,
      "loss": 9.1427,
      "step": 4740
    },
    {
      "epoch": 0.926338413442751,
      "grad_norm": 7.872930526733398,
      "learning_rate": 2.9997586660817643e-07,
      "loss": 9.728,
      "step": 4741
    },
    {
      "epoch": 0.9265338022665104,
      "grad_norm": 7.268552780151367,
      "learning_rate": 2.984393194345514e-07,
      "loss": 8.4245,
      "step": 4742
    },
    {
      "epoch": 0.9267291910902696,
      "grad_norm": 9.872431755065918,
      "learning_rate": 2.969066580371782e-07,
      "loss": 8.7397,
      "step": 4743
    },
    {
      "epoch": 0.9269245799140289,
      "grad_norm": 9.95816707611084,
      "learning_rate": 2.9537788302993477e-07,
      "loss": 8.6281,
      "step": 4744
    },
    {
      "epoch": 0.9271199687377882,
      "grad_norm": 6.963932991027832,
      "learning_rate": 2.9385299502514233e-07,
      "loss": 8.7163,
      "step": 4745
    },
    {
      "epoch": 0.9273153575615475,
      "grad_norm": 10.237529754638672,
      "learning_rate": 2.923319946335668e-07,
      "loss": 8.6967,
      "step": 4746
    },
    {
      "epoch": 0.9275107463853067,
      "grad_norm": 8.188759803771973,
      "learning_rate": 2.9081488246441527e-07,
      "loss": 8.4961,
      "step": 4747
    },
    {
      "epoch": 0.9277061352090661,
      "grad_norm": 9.989144325256348,
      "learning_rate": 2.8930165912533945e-07,
      "loss": 8.1715,
      "step": 4748
    },
    {
      "epoch": 0.9279015240328253,
      "grad_norm": 7.172133445739746,
      "learning_rate": 2.877923252224302e-07,
      "loss": 8.7734,
      "step": 4749
    },
    {
      "epoch": 0.9280969128565846,
      "grad_norm": 9.931110382080078,
      "learning_rate": 2.8628688136022487e-07,
      "loss": 9.472,
      "step": 4750
    },
    {
      "epoch": 0.9282923016803439,
      "grad_norm": 8.37661075592041,
      "learning_rate": 2.847853281416979e-07,
      "loss": 9.0718,
      "step": 4751
    },
    {
      "epoch": 0.9284876905041032,
      "grad_norm": 8.126204490661621,
      "learning_rate": 2.832876661682715e-07,
      "loss": 8.2512,
      "step": 4752
    },
    {
      "epoch": 0.9286830793278624,
      "grad_norm": 7.955578327178955,
      "learning_rate": 2.817938960398014e-07,
      "loss": 9.0037,
      "step": 4753
    },
    {
      "epoch": 0.9288784681516217,
      "grad_norm": 9.936150550842285,
      "learning_rate": 2.803040183545935e-07,
      "loss": 8.9849,
      "step": 4754
    },
    {
      "epoch": 0.929073856975381,
      "grad_norm": 10.026246070861816,
      "learning_rate": 2.78818033709386e-07,
      "loss": 9.4893,
      "step": 4755
    },
    {
      "epoch": 0.9292692457991403,
      "grad_norm": 9.931303024291992,
      "learning_rate": 2.7733594269936606e-07,
      "loss": 9.9154,
      "step": 4756
    },
    {
      "epoch": 0.9294646346228995,
      "grad_norm": 7.731067657470703,
      "learning_rate": 2.7585774591815217e-07,
      "loss": 8.77,
      "step": 4757
    },
    {
      "epoch": 0.9296600234466589,
      "grad_norm": 7.984607219696045,
      "learning_rate": 2.7438344395781305e-07,
      "loss": 8.161,
      "step": 4758
    },
    {
      "epoch": 0.9298554122704181,
      "grad_norm": 9.72499942779541,
      "learning_rate": 2.7291303740884955e-07,
      "loss": 9.1158,
      "step": 4759
    },
    {
      "epoch": 0.9300508010941774,
      "grad_norm": 8.763854026794434,
      "learning_rate": 2.714465268602051e-07,
      "loss": 9.2098,
      "step": 4760
    },
    {
      "epoch": 0.9302461899179367,
      "grad_norm": 15.150201797485352,
      "learning_rate": 2.6998391289926316e-07,
      "loss": 9.5308,
      "step": 4761
    },
    {
      "epoch": 0.930441578741696,
      "grad_norm": 9.16210651397705,
      "learning_rate": 2.6852519611184623e-07,
      "loss": 9.9065,
      "step": 4762
    },
    {
      "epoch": 0.9306369675654552,
      "grad_norm": 8.150128364562988,
      "learning_rate": 2.6707037708221584e-07,
      "loss": 8.2702,
      "step": 4763
    },
    {
      "epoch": 0.9308323563892146,
      "grad_norm": 6.748828411102295,
      "learning_rate": 2.656194563930714e-07,
      "loss": 9.3573,
      "step": 4764
    },
    {
      "epoch": 0.9310277452129738,
      "grad_norm": 6.6904683113098145,
      "learning_rate": 2.6417243462555364e-07,
      "loss": 8.2785,
      "step": 4765
    },
    {
      "epoch": 0.9312231340367331,
      "grad_norm": 6.414731025695801,
      "learning_rate": 2.627293123592367e-07,
      "loss": 8.253,
      "step": 4766
    },
    {
      "epoch": 0.9314185228604924,
      "grad_norm": 10.10960578918457,
      "learning_rate": 2.6129009017213826e-07,
      "loss": 9.0605,
      "step": 4767
    },
    {
      "epoch": 0.9316139116842517,
      "grad_norm": 7.682331562042236,
      "learning_rate": 2.5985476864070935e-07,
      "loss": 8.9442,
      "step": 4768
    },
    {
      "epoch": 0.9318093005080109,
      "grad_norm": 9.000534057617188,
      "learning_rate": 2.584233483398435e-07,
      "loss": 9.626,
      "step": 4769
    },
    {
      "epoch": 0.9320046893317703,
      "grad_norm": 9.852577209472656,
      "learning_rate": 2.5699582984286655e-07,
      "loss": 8.8923,
      "step": 4770
    },
    {
      "epoch": 0.9322000781555295,
      "grad_norm": 8.212091445922852,
      "learning_rate": 2.555722137215455e-07,
      "loss": 8.4139,
      "step": 4771
    },
    {
      "epoch": 0.9323954669792888,
      "grad_norm": 7.885027885437012,
      "learning_rate": 2.541525005460821e-07,
      "loss": 8.9835,
      "step": 4772
    },
    {
      "epoch": 0.932590855803048,
      "grad_norm": 8.578592300415039,
      "learning_rate": 2.5273669088511696e-07,
      "loss": 9.0651,
      "step": 4773
    },
    {
      "epoch": 0.9327862446268074,
      "grad_norm": 7.461894989013672,
      "learning_rate": 2.5132478530572436e-07,
      "loss": 7.6183,
      "step": 4774
    },
    {
      "epoch": 0.9329816334505666,
      "grad_norm": 6.767593860626221,
      "learning_rate": 2.4991678437341515e-07,
      "loss": 8.8828,
      "step": 4775
    },
    {
      "epoch": 0.9331770222743259,
      "grad_norm": 8.581154823303223,
      "learning_rate": 2.485126886521394e-07,
      "loss": 9.6972,
      "step": 4776
    },
    {
      "epoch": 0.9333724110980852,
      "grad_norm": 10.012948036193848,
      "learning_rate": 2.471124987042806e-07,
      "loss": 9.1716,
      "step": 4777
    },
    {
      "epoch": 0.9335677999218445,
      "grad_norm": 8.47640323638916,
      "learning_rate": 2.4571621509065893e-07,
      "loss": 8.9597,
      "step": 4778
    },
    {
      "epoch": 0.9337631887456037,
      "grad_norm": 8.2051362991333,
      "learning_rate": 2.443238383705271e-07,
      "loss": 8.9585,
      "step": 4779
    },
    {
      "epoch": 0.9339585775693631,
      "grad_norm": 7.979532718658447,
      "learning_rate": 2.429353691015779e-07,
      "loss": 8.8593,
      "step": 4780
    },
    {
      "epoch": 0.9341539663931223,
      "grad_norm": 8.390580177307129,
      "learning_rate": 2.4155080783993423e-07,
      "loss": 9.2431,
      "step": 4781
    },
    {
      "epoch": 0.9343493552168816,
      "grad_norm": 8.807951927185059,
      "learning_rate": 2.4017015514015806e-07,
      "loss": 8.272,
      "step": 4782
    },
    {
      "epoch": 0.9345447440406409,
      "grad_norm": 11.47719955444336,
      "learning_rate": 2.387934115552404e-07,
      "loss": 9.1402,
      "step": 4783
    },
    {
      "epoch": 0.9347401328644002,
      "grad_norm": 8.282925605773926,
      "learning_rate": 2.374205776366134e-07,
      "loss": 8.0761,
      "step": 4784
    },
    {
      "epoch": 0.9349355216881594,
      "grad_norm": 7.644189834594727,
      "learning_rate": 2.3605165393413841e-07,
      "loss": 8.367,
      "step": 4785
    },
    {
      "epoch": 0.9351309105119188,
      "grad_norm": 8.098487854003906,
      "learning_rate": 2.3468664099611128e-07,
      "loss": 8.7649,
      "step": 4786
    },
    {
      "epoch": 0.935326299335678,
      "grad_norm": 7.511486053466797,
      "learning_rate": 2.333255393692613e-07,
      "loss": 8.7467,
      "step": 4787
    },
    {
      "epoch": 0.9355216881594373,
      "grad_norm": 27.118160247802734,
      "learning_rate": 2.3196834959875348e-07,
      "loss": 10.1981,
      "step": 4788
    },
    {
      "epoch": 0.9357170769831966,
      "grad_norm": 6.575469493865967,
      "learning_rate": 2.3061507222818303e-07,
      "loss": 8.2403,
      "step": 4789
    },
    {
      "epoch": 0.9359124658069559,
      "grad_norm": 7.317230701446533,
      "learning_rate": 2.292657077995819e-07,
      "loss": 9.0534,
      "step": 4790
    },
    {
      "epoch": 0.9361078546307151,
      "grad_norm": 11.103191375732422,
      "learning_rate": 2.2792025685341113e-07,
      "loss": 8.3513,
      "step": 4791
    },
    {
      "epoch": 0.9363032434544744,
      "grad_norm": 7.332716941833496,
      "learning_rate": 2.2657871992856407e-07,
      "loss": 8.6779,
      "step": 4792
    },
    {
      "epoch": 0.9364986322782337,
      "grad_norm": 10.763957023620605,
      "learning_rate": 2.2524109756236977e-07,
      "loss": 9.7406,
      "step": 4793
    },
    {
      "epoch": 0.936694021101993,
      "grad_norm": 8.400541305541992,
      "learning_rate": 2.2390739029058639e-07,
      "loss": 9.0757,
      "step": 4794
    },
    {
      "epoch": 0.9368894099257522,
      "grad_norm": 6.764805793762207,
      "learning_rate": 2.2257759864740768e-07,
      "loss": 8.4885,
      "step": 4795
    },
    {
      "epoch": 0.9370847987495116,
      "grad_norm": 7.343725204467773,
      "learning_rate": 2.2125172316545206e-07,
      "loss": 8.0188,
      "step": 4796
    },
    {
      "epoch": 0.9372801875732708,
      "grad_norm": 8.75682544708252,
      "learning_rate": 2.1992976437577807e-07,
      "loss": 8.8639,
      "step": 4797
    },
    {
      "epoch": 0.93747557639703,
      "grad_norm": 6.371266841888428,
      "learning_rate": 2.186117228078688e-07,
      "loss": 7.8495,
      "step": 4798
    },
    {
      "epoch": 0.9376709652207894,
      "grad_norm": 10.427175521850586,
      "learning_rate": 2.1729759898964197e-07,
      "loss": 8.9847,
      "step": 4799
    },
    {
      "epoch": 0.9378663540445487,
      "grad_norm": 7.948103904724121,
      "learning_rate": 2.1598739344744323e-07,
      "loss": 8.7709,
      "step": 4800
    },
    {
      "epoch": 0.9380617428683079,
      "grad_norm": 7.93149995803833,
      "learning_rate": 2.146811067060539e-07,
      "loss": 9.1597,
      "step": 4801
    },
    {
      "epoch": 0.9382571316920673,
      "grad_norm": 9.593384742736816,
      "learning_rate": 2.133787392886788e-07,
      "loss": 8.5564,
      "step": 4802
    },
    {
      "epoch": 0.9384525205158265,
      "grad_norm": 7.262961387634277,
      "learning_rate": 2.1208029171695844e-07,
      "loss": 8.4633,
      "step": 4803
    },
    {
      "epoch": 0.9386479093395858,
      "grad_norm": 8.87141227722168,
      "learning_rate": 2.1078576451096123e-07,
      "loss": 10.1814,
      "step": 4804
    },
    {
      "epoch": 0.9388432981633451,
      "grad_norm": 7.445680618286133,
      "learning_rate": 2.0949515818918465e-07,
      "loss": 9.209,
      "step": 4805
    },
    {
      "epoch": 0.9390386869871044,
      "grad_norm": 8.99807357788086,
      "learning_rate": 2.082084732685574e-07,
      "loss": 9.5791,
      "step": 4806
    },
    {
      "epoch": 0.9392340758108636,
      "grad_norm": 7.861563205718994,
      "learning_rate": 2.0692571026443618e-07,
      "loss": 8.1887,
      "step": 4807
    },
    {
      "epoch": 0.9394294646346228,
      "grad_norm": 6.860263824462891,
      "learning_rate": 2.0564686969060887e-07,
      "loss": 7.9136,
      "step": 4808
    },
    {
      "epoch": 0.9396248534583822,
      "grad_norm": 8.453475952148438,
      "learning_rate": 2.0437195205928906e-07,
      "loss": 9.5563,
      "step": 4809
    },
    {
      "epoch": 0.9398202422821414,
      "grad_norm": 8.56126594543457,
      "learning_rate": 2.0310095788112162e-07,
      "loss": 9.7925,
      "step": 4810
    },
    {
      "epoch": 0.9400156311059007,
      "grad_norm": 7.933896064758301,
      "learning_rate": 2.018338876651782e-07,
      "loss": 8.5813,
      "step": 4811
    },
    {
      "epoch": 0.94021101992966,
      "grad_norm": 8.358501434326172,
      "learning_rate": 2.0057074191896176e-07,
      "loss": 8.4015,
      "step": 4812
    },
    {
      "epoch": 0.9404064087534193,
      "grad_norm": 12.378544807434082,
      "learning_rate": 1.9931152114839868e-07,
      "loss": 9.8506,
      "step": 4813
    },
    {
      "epoch": 0.9406017975771785,
      "grad_norm": 14.021265029907227,
      "learning_rate": 1.9805622585785001e-07,
      "loss": 10.5824,
      "step": 4814
    },
    {
      "epoch": 0.9407971864009379,
      "grad_norm": 6.565126419067383,
      "learning_rate": 1.9680485655009573e-07,
      "loss": 8.7138,
      "step": 4815
    },
    {
      "epoch": 0.9409925752246971,
      "grad_norm": 7.936339855194092,
      "learning_rate": 1.955574137263516e-07,
      "loss": 9.2013,
      "step": 4816
    },
    {
      "epoch": 0.9411879640484564,
      "grad_norm": 9.885173797607422,
      "learning_rate": 1.9431389788625576e-07,
      "loss": 10.3527,
      "step": 4817
    },
    {
      "epoch": 0.9413833528722158,
      "grad_norm": 9.101330757141113,
      "learning_rate": 1.930743095278753e-07,
      "loss": 9.6166,
      "step": 4818
    },
    {
      "epoch": 0.941578741695975,
      "grad_norm": 7.584497928619385,
      "learning_rate": 1.9183864914770312e-07,
      "loss": 9.5762,
      "step": 4819
    },
    {
      "epoch": 0.9417741305197342,
      "grad_norm": 9.555315971374512,
      "learning_rate": 1.9060691724065995e-07,
      "loss": 10.0103,
      "step": 4820
    },
    {
      "epoch": 0.9419695193434936,
      "grad_norm": 8.41330337524414,
      "learning_rate": 1.8937911430009338e-07,
      "loss": 9.634,
      "step": 4821
    },
    {
      "epoch": 0.9421649081672528,
      "grad_norm": 9.359399795532227,
      "learning_rate": 1.8815524081777559e-07,
      "loss": 9.9597,
      "step": 4822
    },
    {
      "epoch": 0.9423602969910121,
      "grad_norm": 9.938153266906738,
      "learning_rate": 1.869352972839067e-07,
      "loss": 9.2335,
      "step": 4823
    },
    {
      "epoch": 0.9425556858147714,
      "grad_norm": 11.01329517364502,
      "learning_rate": 1.8571928418711137e-07,
      "loss": 9.3612,
      "step": 4824
    },
    {
      "epoch": 0.9427510746385307,
      "grad_norm": 8.913078308105469,
      "learning_rate": 1.8450720201444116e-07,
      "loss": 8.643,
      "step": 4825
    },
    {
      "epoch": 0.9429464634622899,
      "grad_norm": 9.12736701965332,
      "learning_rate": 1.8329905125137215e-07,
      "loss": 8.3215,
      "step": 4826
    },
    {
      "epoch": 0.9431418522860492,
      "grad_norm": 7.773070812225342,
      "learning_rate": 1.8209483238180726e-07,
      "loss": 9.3525,
      "step": 4827
    },
    {
      "epoch": 0.9433372411098085,
      "grad_norm": 6.683079719543457,
      "learning_rate": 1.8089454588807066e-07,
      "loss": 8.8364,
      "step": 4828
    },
    {
      "epoch": 0.9435326299335678,
      "grad_norm": 7.532355785369873,
      "learning_rate": 1.7969819225091888e-07,
      "loss": 9.1896,
      "step": 4829
    },
    {
      "epoch": 0.943728018757327,
      "grad_norm": 8.087076187133789,
      "learning_rate": 1.7850577194952535e-07,
      "loss": 8.8348,
      "step": 4830
    },
    {
      "epoch": 0.9439234075810864,
      "grad_norm": 7.5128607749938965,
      "learning_rate": 1.773172854614913e-07,
      "loss": 8.2127,
      "step": 4831
    },
    {
      "epoch": 0.9441187964048456,
      "grad_norm": 6.844386100769043,
      "learning_rate": 1.7613273326284374e-07,
      "loss": 7.821,
      "step": 4832
    },
    {
      "epoch": 0.9443141852286049,
      "grad_norm": 8.602503776550293,
      "learning_rate": 1.7495211582803208e-07,
      "loss": 9.7718,
      "step": 4833
    },
    {
      "epoch": 0.9445095740523642,
      "grad_norm": 47.355567932128906,
      "learning_rate": 1.7377543362993132e-07,
      "loss": 9.2821,
      "step": 4834
    },
    {
      "epoch": 0.9447049628761235,
      "grad_norm": 8.016357421875,
      "learning_rate": 1.7260268713983675e-07,
      "loss": 8.796,
      "step": 4835
    },
    {
      "epoch": 0.9449003516998827,
      "grad_norm": 11.182827949523926,
      "learning_rate": 1.714338768274726e-07,
      "loss": 9.1631,
      "step": 4836
    },
    {
      "epoch": 0.9450957405236421,
      "grad_norm": 7.747939109802246,
      "learning_rate": 1.7026900316098217e-07,
      "loss": 9.759,
      "step": 4837
    },
    {
      "epoch": 0.9452911293474013,
      "grad_norm": 7.514883995056152,
      "learning_rate": 1.6910806660693445e-07,
      "loss": 8.0,
      "step": 4838
    },
    {
      "epoch": 0.9454865181711606,
      "grad_norm": 6.813232898712158,
      "learning_rate": 1.6795106763031867e-07,
      "loss": 8.5752,
      "step": 4839
    },
    {
      "epoch": 0.9456819069949199,
      "grad_norm": 9.245199203491211,
      "learning_rate": 1.6679800669455072e-07,
      "loss": 9.3135,
      "step": 4840
    },
    {
      "epoch": 0.9458772958186792,
      "grad_norm": 14.45501708984375,
      "learning_rate": 1.6564888426146674e-07,
      "loss": 8.9624,
      "step": 4841
    },
    {
      "epoch": 0.9460726846424384,
      "grad_norm": 9.25383472442627,
      "learning_rate": 1.6450370079132637e-07,
      "loss": 8.9619,
      "step": 4842
    },
    {
      "epoch": 0.9462680734661977,
      "grad_norm": 7.9999003410339355,
      "learning_rate": 1.6336245674280937e-07,
      "loss": 8.6831,
      "step": 4843
    },
    {
      "epoch": 0.946463462289957,
      "grad_norm": 11.17634391784668,
      "learning_rate": 1.6222515257302118e-07,
      "loss": 9.9972,
      "step": 4844
    },
    {
      "epoch": 0.9466588511137163,
      "grad_norm": 7.812686920166016,
      "learning_rate": 1.6109178873748855e-07,
      "loss": 8.5462,
      "step": 4845
    },
    {
      "epoch": 0.9468542399374755,
      "grad_norm": 7.749689102172852,
      "learning_rate": 1.59962365690155e-07,
      "loss": 9.3361,
      "step": 4846
    },
    {
      "epoch": 0.9470496287612349,
      "grad_norm": 7.081580638885498,
      "learning_rate": 1.5883688388339314e-07,
      "loss": 8.4492,
      "step": 4847
    },
    {
      "epoch": 0.9472450175849941,
      "grad_norm": 9.793357849121094,
      "learning_rate": 1.5771534376799125e-07,
      "loss": 9.6255,
      "step": 4848
    },
    {
      "epoch": 0.9474404064087534,
      "grad_norm": 12.532551765441895,
      "learning_rate": 1.5659774579316222e-07,
      "loss": 10.3395,
      "step": 4849
    },
    {
      "epoch": 0.9476357952325127,
      "grad_norm": 8.414477348327637,
      "learning_rate": 1.554840904065369e-07,
      "loss": 9.4367,
      "step": 4850
    },
    {
      "epoch": 0.947831184056272,
      "grad_norm": 6.619372367858887,
      "learning_rate": 1.5437437805417178e-07,
      "loss": 8.6572,
      "step": 4851
    },
    {
      "epoch": 0.9480265728800312,
      "grad_norm": 6.143033504486084,
      "learning_rate": 1.532686091805391e-07,
      "loss": 8.9362,
      "step": 4852
    },
    {
      "epoch": 0.9482219617037906,
      "grad_norm": 8.467583656311035,
      "learning_rate": 1.521667842285346e-07,
      "loss": 8.3852,
      "step": 4853
    },
    {
      "epoch": 0.9484173505275498,
      "grad_norm": 7.625029563903809,
      "learning_rate": 1.510689036394719e-07,
      "loss": 8.4104,
      "step": 4854
    },
    {
      "epoch": 0.9486127393513091,
      "grad_norm": 6.924171447753906,
      "learning_rate": 1.4997496785308818e-07,
      "loss": 8.7127,
      "step": 4855
    },
    {
      "epoch": 0.9488081281750684,
      "grad_norm": 9.200094223022461,
      "learning_rate": 1.4888497730753627e-07,
      "loss": 8.9581,
      "step": 4856
    },
    {
      "epoch": 0.9490035169988277,
      "grad_norm": 9.852042198181152,
      "learning_rate": 1.4779893243939358e-07,
      "loss": 9.4307,
      "step": 4857
    },
    {
      "epoch": 0.9491989058225869,
      "grad_norm": 8.711514472961426,
      "learning_rate": 1.467168336836544e-07,
      "loss": 8.9601,
      "step": 4858
    },
    {
      "epoch": 0.9493942946463463,
      "grad_norm": 8.813514709472656,
      "learning_rate": 1.4563868147373095e-07,
      "loss": 8.983,
      "step": 4859
    },
    {
      "epoch": 0.9495896834701055,
      "grad_norm": 7.923171520233154,
      "learning_rate": 1.4456447624145998e-07,
      "loss": 8.7049,
      "step": 4860
    },
    {
      "epoch": 0.9497850722938648,
      "grad_norm": 8.254159927368164,
      "learning_rate": 1.4349421841709067e-07,
      "loss": 7.9401,
      "step": 4861
    },
    {
      "epoch": 0.949980461117624,
      "grad_norm": 6.768784999847412,
      "learning_rate": 1.4242790842929567e-07,
      "loss": 8.5365,
      "step": 4862
    },
    {
      "epoch": 0.9501758499413834,
      "grad_norm": 7.572927474975586,
      "learning_rate": 1.4136554670516667e-07,
      "loss": 8.3918,
      "step": 4863
    },
    {
      "epoch": 0.9503712387651426,
      "grad_norm": 11.043039321899414,
      "learning_rate": 1.4030713367021108e-07,
      "loss": 9.5904,
      "step": 4864
    },
    {
      "epoch": 0.9505666275889019,
      "grad_norm": 9.595534324645996,
      "learning_rate": 1.3925266974835538e-07,
      "loss": 9.3853,
      "step": 4865
    },
    {
      "epoch": 0.9507620164126612,
      "grad_norm": 7.735306739807129,
      "learning_rate": 1.3820215536194725e-07,
      "loss": 8.9673,
      "step": 4866
    },
    {
      "epoch": 0.9509574052364205,
      "grad_norm": 9.947039604187012,
      "learning_rate": 1.37155590931749e-07,
      "loss": 9.6789,
      "step": 4867
    },
    {
      "epoch": 0.9511527940601797,
      "grad_norm": 7.425265312194824,
      "learning_rate": 1.3611297687694312e-07,
      "loss": 9.7541,
      "step": 4868
    },
    {
      "epoch": 0.9513481828839391,
      "grad_norm": 7.681835651397705,
      "learning_rate": 1.350743136151267e-07,
      "loss": 10.2013,
      "step": 4869
    },
    {
      "epoch": 0.9515435717076983,
      "grad_norm": 7.884913921356201,
      "learning_rate": 1.3403960156232022e-07,
      "loss": 8.9761,
      "step": 4870
    },
    {
      "epoch": 0.9517389605314576,
      "grad_norm": 7.007309913635254,
      "learning_rate": 1.3300884113295442e-07,
      "loss": 8.6048,
      "step": 4871
    },
    {
      "epoch": 0.9519343493552169,
      "grad_norm": 7.391115188598633,
      "learning_rate": 1.319820327398824e-07,
      "loss": 8.8546,
      "step": 4872
    },
    {
      "epoch": 0.9521297381789762,
      "grad_norm": 8.182116508483887,
      "learning_rate": 1.3095917679437408e-07,
      "loss": 9.6435,
      "step": 4873
    },
    {
      "epoch": 0.9523251270027354,
      "grad_norm": 10.858007431030273,
      "learning_rate": 1.2994027370611173e-07,
      "loss": 9.0315,
      "step": 4874
    },
    {
      "epoch": 0.9525205158264948,
      "grad_norm": 8.331448554992676,
      "learning_rate": 1.2892532388320113e-07,
      "loss": 9.6667,
      "step": 4875
    },
    {
      "epoch": 0.952715904650254,
      "grad_norm": 7.683496952056885,
      "learning_rate": 1.2791432773215928e-07,
      "loss": 7.4806,
      "step": 4876
    },
    {
      "epoch": 0.9529112934740133,
      "grad_norm": 8.721888542175293,
      "learning_rate": 1.2690728565792232e-07,
      "loss": 9.2332,
      "step": 4877
    },
    {
      "epoch": 0.9531066822977726,
      "grad_norm": 7.155975341796875,
      "learning_rate": 1.259041980638398e-07,
      "loss": 8.8735,
      "step": 4878
    },
    {
      "epoch": 0.9533020711215319,
      "grad_norm": 8.131467819213867,
      "learning_rate": 1.2490506535168368e-07,
      "loss": 8.2475,
      "step": 4879
    },
    {
      "epoch": 0.9534974599452911,
      "grad_norm": 8.241068840026855,
      "learning_rate": 1.2390988792163272e-07,
      "loss": 9.6212,
      "step": 4880
    },
    {
      "epoch": 0.9536928487690504,
      "grad_norm": 11.174493789672852,
      "learning_rate": 1.229186661722903e-07,
      "loss": 9.0657,
      "step": 4881
    },
    {
      "epoch": 0.9538882375928097,
      "grad_norm": 13.578412055969238,
      "learning_rate": 1.2193140050066887e-07,
      "loss": 8.9438,
      "step": 4882
    },
    {
      "epoch": 0.954083626416569,
      "grad_norm": 6.047148704528809,
      "learning_rate": 1.2094809130220097e-07,
      "loss": 8.6007,
      "step": 4883
    },
    {
      "epoch": 0.9542790152403282,
      "grad_norm": 6.912959098815918,
      "learning_rate": 1.1996873897073158e-07,
      "loss": 7.6259,
      "step": 4884
    },
    {
      "epoch": 0.9544744040640876,
      "grad_norm": 9.91042423248291,
      "learning_rate": 1.1899334389852135e-07,
      "loss": 9.8637,
      "step": 4885
    },
    {
      "epoch": 0.9546697928878468,
      "grad_norm": 6.9881768226623535,
      "learning_rate": 1.1802190647624667e-07,
      "loss": 8.7643,
      "step": 4886
    },
    {
      "epoch": 0.9548651817116061,
      "grad_norm": 8.107124328613281,
      "learning_rate": 1.1705442709299852e-07,
      "loss": 8.6659,
      "step": 4887
    },
    {
      "epoch": 0.9550605705353654,
      "grad_norm": 8.125308990478516,
      "learning_rate": 1.1609090613628249e-07,
      "loss": 8.6355,
      "step": 4888
    },
    {
      "epoch": 0.9552559593591247,
      "grad_norm": 7.603143692016602,
      "learning_rate": 1.1513134399201765e-07,
      "loss": 9.0061,
      "step": 4889
    },
    {
      "epoch": 0.9554513481828839,
      "grad_norm": 10.955138206481934,
      "learning_rate": 1.1417574104453877e-07,
      "loss": 8.4504,
      "step": 4890
    },
    {
      "epoch": 0.9556467370066433,
      "grad_norm": 9.054905891418457,
      "learning_rate": 1.1322409767659526e-07,
      "loss": 8.9073,
      "step": 4891
    },
    {
      "epoch": 0.9558421258304025,
      "grad_norm": 8.954079627990723,
      "learning_rate": 1.122764142693511e-07,
      "loss": 9.7326,
      "step": 4892
    },
    {
      "epoch": 0.9560375146541618,
      "grad_norm": 7.77183198928833,
      "learning_rate": 1.1133269120237933e-07,
      "loss": 8.7141,
      "step": 4893
    },
    {
      "epoch": 0.9562329034779211,
      "grad_norm": 8.300443649291992,
      "learning_rate": 1.1039292885367315e-07,
      "loss": 9.2519,
      "step": 4894
    },
    {
      "epoch": 0.9564282923016804,
      "grad_norm": 9.766745567321777,
      "learning_rate": 1.0945712759963478e-07,
      "loss": 9.4207,
      "step": 4895
    },
    {
      "epoch": 0.9566236811254396,
      "grad_norm": 8.486104965209961,
      "learning_rate": 1.0852528781508442e-07,
      "loss": 9.179,
      "step": 4896
    },
    {
      "epoch": 0.9568190699491989,
      "grad_norm": 7.6745429039001465,
      "learning_rate": 1.0759740987325017e-07,
      "loss": 9.4932,
      "step": 4897
    },
    {
      "epoch": 0.9570144587729582,
      "grad_norm": 9.365317344665527,
      "learning_rate": 1.0667349414577588e-07,
      "loss": 8.9638,
      "step": 4898
    },
    {
      "epoch": 0.9572098475967175,
      "grad_norm": 28.076303482055664,
      "learning_rate": 1.0575354100272106e-07,
      "loss": 9.2634,
      "step": 4899
    },
    {
      "epoch": 0.9574052364204767,
      "grad_norm": 9.686136245727539,
      "learning_rate": 1.0483755081255209e-07,
      "loss": 8.7086,
      "step": 4900
    },
    {
      "epoch": 0.9576006252442361,
      "grad_norm": 8.524884223937988,
      "learning_rate": 1.0392552394215438e-07,
      "loss": 9.3499,
      "step": 4901
    },
    {
      "epoch": 0.9577960140679953,
      "grad_norm": 7.558854103088379,
      "learning_rate": 1.0301746075682128e-07,
      "loss": 9.2694,
      "step": 4902
    },
    {
      "epoch": 0.9579914028917546,
      "grad_norm": 16.856613159179688,
      "learning_rate": 1.0211336162026187e-07,
      "loss": 8.7741,
      "step": 4903
    },
    {
      "epoch": 0.9581867917155139,
      "grad_norm": 8.923369407653809,
      "learning_rate": 1.0121322689459422e-07,
      "loss": 8.8212,
      "step": 4904
    },
    {
      "epoch": 0.9583821805392732,
      "grad_norm": 6.723689079284668,
      "learning_rate": 1.003170569403511e-07,
      "loss": 8.2187,
      "step": 4905
    },
    {
      "epoch": 0.9585775693630324,
      "grad_norm": 9.944154739379883,
      "learning_rate": 9.942485211647645e-08,
      "loss": 9.0431,
      "step": 4906
    },
    {
      "epoch": 0.9587729581867918,
      "grad_norm": 9.219122886657715,
      "learning_rate": 9.853661278032556e-08,
      "loss": 9.4427,
      "step": 4907
    },
    {
      "epoch": 0.958968347010551,
      "grad_norm": 8.637513160705566,
      "learning_rate": 9.765233928766493e-08,
      "loss": 7.9597,
      "step": 4908
    },
    {
      "epoch": 0.9591637358343102,
      "grad_norm": 6.749722957611084,
      "learning_rate": 9.677203199267682e-08,
      "loss": 8.2952,
      "step": 4909
    },
    {
      "epoch": 0.9593591246580696,
      "grad_norm": 22.5472412109375,
      "learning_rate": 9.589569124794918e-08,
      "loss": 8.8987,
      "step": 4910
    },
    {
      "epoch": 0.9595545134818289,
      "grad_norm": 7.759912490844727,
      "learning_rate": 9.502331740448233e-08,
      "loss": 9.3352,
      "step": 4911
    },
    {
      "epoch": 0.9597499023055881,
      "grad_norm": 8.938230514526367,
      "learning_rate": 9.415491081169237e-08,
      "loss": 9.2042,
      "step": 4912
    },
    {
      "epoch": 0.9599452911293475,
      "grad_norm": 8.625635147094727,
      "learning_rate": 9.329047181740103e-08,
      "loss": 9.4016,
      "step": 4913
    },
    {
      "epoch": 0.9601406799531067,
      "grad_norm": 8.247848510742188,
      "learning_rate": 9.243000076784359e-08,
      "loss": 8.846,
      "step": 4914
    },
    {
      "epoch": 0.960336068776866,
      "grad_norm": 7.982619285583496,
      "learning_rate": 9.157349800766435e-08,
      "loss": 8.2125,
      "step": 4915
    },
    {
      "epoch": 0.9605314576006252,
      "grad_norm": 10.335287094116211,
      "learning_rate": 9.072096387992113e-08,
      "loss": 8.868,
      "step": 4916
    },
    {
      "epoch": 0.9607268464243846,
      "grad_norm": 8.667943954467773,
      "learning_rate": 8.987239872607856e-08,
      "loss": 9.4956,
      "step": 4917
    },
    {
      "epoch": 0.9609222352481438,
      "grad_norm": 11.45176887512207,
      "learning_rate": 8.902780288601587e-08,
      "loss": 10.3252,
      "step": 4918
    },
    {
      "epoch": 0.961117624071903,
      "grad_norm": 7.204813480377197,
      "learning_rate": 8.818717669801691e-08,
      "loss": 8.4043,
      "step": 4919
    },
    {
      "epoch": 0.9613130128956624,
      "grad_norm": 8.619112014770508,
      "learning_rate": 8.735052049878012e-08,
      "loss": 9.1276,
      "step": 4920
    },
    {
      "epoch": 0.9615084017194216,
      "grad_norm": 10.103872299194336,
      "learning_rate": 8.651783462341078e-08,
      "loss": 9.4011,
      "step": 4921
    },
    {
      "epoch": 0.9617037905431809,
      "grad_norm": 8.542306900024414,
      "learning_rate": 8.568911940542879e-08,
      "loss": 8.6925,
      "step": 4922
    },
    {
      "epoch": 0.9618991793669402,
      "grad_norm": 11.84721851348877,
      "learning_rate": 8.48643751767575e-08,
      "loss": 9.5716,
      "step": 4923
    },
    {
      "epoch": 0.9620945681906995,
      "grad_norm": 7.118066787719727,
      "learning_rate": 8.404360226773267e-08,
      "loss": 8.3886,
      "step": 4924
    },
    {
      "epoch": 0.9622899570144587,
      "grad_norm": 6.681229114532471,
      "learning_rate": 8.322680100710023e-08,
      "loss": 8.4012,
      "step": 4925
    },
    {
      "epoch": 0.9624853458382181,
      "grad_norm": 8.622283935546875,
      "learning_rate": 8.241397172201405e-08,
      "loss": 9.1808,
      "step": 4926
    },
    {
      "epoch": 0.9626807346619773,
      "grad_norm": 12.33980941772461,
      "learning_rate": 8.160511473803701e-08,
      "loss": 8.9697,
      "step": 4927
    },
    {
      "epoch": 0.9628761234857366,
      "grad_norm": 8.637188911437988,
      "learning_rate": 8.080023037914331e-08,
      "loss": 9.0236,
      "step": 4928
    },
    {
      "epoch": 0.963071512309496,
      "grad_norm": 10.9049072265625,
      "learning_rate": 7.999931896771284e-08,
      "loss": 9.8126,
      "step": 4929
    },
    {
      "epoch": 0.9632669011332552,
      "grad_norm": 7.670109748840332,
      "learning_rate": 7.920238082453457e-08,
      "loss": 8.4132,
      "step": 4930
    },
    {
      "epoch": 0.9634622899570144,
      "grad_norm": 17.906023025512695,
      "learning_rate": 7.840941626880871e-08,
      "loss": 9.3315,
      "step": 4931
    },
    {
      "epoch": 0.9636576787807737,
      "grad_norm": 6.756670951843262,
      "learning_rate": 7.762042561814009e-08,
      "loss": 8.0376,
      "step": 4932
    },
    {
      "epoch": 0.963853067604533,
      "grad_norm": 8.018085479736328,
      "learning_rate": 7.683540918854593e-08,
      "loss": 9.0674,
      "step": 4933
    },
    {
      "epoch": 0.9640484564282923,
      "grad_norm": 7.3927788734436035,
      "learning_rate": 7.605436729444915e-08,
      "loss": 9.0571,
      "step": 4934
    },
    {
      "epoch": 0.9642438452520515,
      "grad_norm": 9.25537109375,
      "learning_rate": 7.527730024868062e-08,
      "loss": 9.1265,
      "step": 4935
    },
    {
      "epoch": 0.9644392340758109,
      "grad_norm": 9.464190483093262,
      "learning_rate": 7.450420836248029e-08,
      "loss": 9.5791,
      "step": 4936
    },
    {
      "epoch": 0.9646346228995701,
      "grad_norm": 8.062870979309082,
      "learning_rate": 7.373509194549599e-08,
      "loss": 9.1729,
      "step": 4937
    },
    {
      "epoch": 0.9648300117233294,
      "grad_norm": 8.228679656982422,
      "learning_rate": 7.296995130578243e-08,
      "loss": 8.4348,
      "step": 4938
    },
    {
      "epoch": 0.9650254005470887,
      "grad_norm": 8.645241737365723,
      "learning_rate": 7.220878674980225e-08,
      "loss": 9.5973,
      "step": 4939
    },
    {
      "epoch": 0.965220789370848,
      "grad_norm": 7.164260387420654,
      "learning_rate": 7.145159858242379e-08,
      "loss": 8.5819,
      "step": 4940
    },
    {
      "epoch": 0.9654161781946072,
      "grad_norm": 8.89592456817627,
      "learning_rate": 7.069838710692666e-08,
      "loss": 9.1587,
      "step": 4941
    },
    {
      "epoch": 0.9656115670183666,
      "grad_norm": 9.88491439819336,
      "learning_rate": 6.994915262499513e-08,
      "loss": 8.8914,
      "step": 4942
    },
    {
      "epoch": 0.9658069558421258,
      "grad_norm": 11.146038055419922,
      "learning_rate": 6.920389543671913e-08,
      "loss": 9.8806,
      "step": 4943
    },
    {
      "epoch": 0.9660023446658851,
      "grad_norm": 8.83292293548584,
      "learning_rate": 6.846261584059988e-08,
      "loss": 8.9843,
      "step": 4944
    },
    {
      "epoch": 0.9661977334896444,
      "grad_norm": 8.25486946105957,
      "learning_rate": 6.772531413353989e-08,
      "loss": 7.6047,
      "step": 4945
    },
    {
      "epoch": 0.9663931223134037,
      "grad_norm": 7.636203289031982,
      "learning_rate": 6.699199061085515e-08,
      "loss": 9.1152,
      "step": 4946
    },
    {
      "epoch": 0.9665885111371629,
      "grad_norm": 8.359781265258789,
      "learning_rate": 6.626264556626071e-08,
      "loss": 8.99,
      "step": 4947
    },
    {
      "epoch": 0.9667838999609223,
      "grad_norm": 8.841544151306152,
      "learning_rate": 6.553727929188403e-08,
      "loss": 8.7749,
      "step": 4948
    },
    {
      "epoch": 0.9669792887846815,
      "grad_norm": 8.096626281738281,
      "learning_rate": 6.4815892078256e-08,
      "loss": 9.272,
      "step": 4949
    },
    {
      "epoch": 0.9671746776084408,
      "grad_norm": 8.442158699035645,
      "learning_rate": 6.409848421431553e-08,
      "loss": 9.2086,
      "step": 4950
    },
    {
      "epoch": 0.9673700664322,
      "grad_norm": 14.558295249938965,
      "learning_rate": 6.338505598740608e-08,
      "loss": 9.3925,
      "step": 4951
    },
    {
      "epoch": 0.9675654552559594,
      "grad_norm": 8.240745544433594,
      "learning_rate": 6.267560768327796e-08,
      "loss": 8.2754,
      "step": 4952
    },
    {
      "epoch": 0.9677608440797186,
      "grad_norm": 8.812482833862305,
      "learning_rate": 6.197013958608611e-08,
      "loss": 8.1315,
      "step": 4953
    },
    {
      "epoch": 0.9679562329034779,
      "grad_norm": 7.7007155418396,
      "learning_rate": 6.126865197839449e-08,
      "loss": 9.099,
      "step": 4954
    },
    {
      "epoch": 0.9681516217272372,
      "grad_norm": 7.753499984741211,
      "learning_rate": 6.057114514116947e-08,
      "loss": 8.4374,
      "step": 4955
    },
    {
      "epoch": 0.9683470105509965,
      "grad_norm": 7.419507026672363,
      "learning_rate": 5.987761935378422e-08,
      "loss": 8.917,
      "step": 4956
    },
    {
      "epoch": 0.9685423993747557,
      "grad_norm": 7.506134986877441,
      "learning_rate": 5.918807489401768e-08,
      "loss": 9.2826,
      "step": 4957
    },
    {
      "epoch": 0.9687377881985151,
      "grad_norm": 7.759176254272461,
      "learning_rate": 5.8502512038053347e-08,
      "loss": 8.6572,
      "step": 4958
    },
    {
      "epoch": 0.9689331770222743,
      "grad_norm": 8.795549392700195,
      "learning_rate": 5.782093106048159e-08,
      "loss": 9.0967,
      "step": 4959
    },
    {
      "epoch": 0.9691285658460336,
      "grad_norm": 8.271385192871094,
      "learning_rate": 5.714333223429402e-08,
      "loss": 8.7773,
      "step": 4960
    },
    {
      "epoch": 0.9693239546697929,
      "grad_norm": 7.687859058380127,
      "learning_rate": 5.6469715830893556e-08,
      "loss": 8.9301,
      "step": 4961
    },
    {
      "epoch": 0.9695193434935522,
      "grad_norm": 11.383176803588867,
      "learning_rate": 5.5800082120082143e-08,
      "loss": 9.0446,
      "step": 4962
    },
    {
      "epoch": 0.9697147323173114,
      "grad_norm": 9.46120548248291,
      "learning_rate": 5.5134431370069685e-08,
      "loss": 9.951,
      "step": 4963
    },
    {
      "epoch": 0.9699101211410708,
      "grad_norm": 9.806598663330078,
      "learning_rate": 5.447276384746958e-08,
      "loss": 8.7027,
      "step": 4964
    },
    {
      "epoch": 0.97010550996483,
      "grad_norm": 8.101316452026367,
      "learning_rate": 5.3815079817300944e-08,
      "loss": 8.0066,
      "step": 4965
    },
    {
      "epoch": 0.9703008987885893,
      "grad_norm": 9.437013626098633,
      "learning_rate": 5.316137954298528e-08,
      "loss": 9.5399,
      "step": 4966
    },
    {
      "epoch": 0.9704962876123485,
      "grad_norm": 11.781790733337402,
      "learning_rate": 5.251166328635093e-08,
      "loss": 9.9803,
      "step": 4967
    },
    {
      "epoch": 0.9706916764361079,
      "grad_norm": 8.37763786315918,
      "learning_rate": 5.186593130762974e-08,
      "loss": 8.733,
      "step": 4968
    },
    {
      "epoch": 0.9708870652598671,
      "grad_norm": 7.655576705932617,
      "learning_rate": 5.122418386545591e-08,
      "loss": 8.9133,
      "step": 4969
    },
    {
      "epoch": 0.9710824540836264,
      "grad_norm": 7.902202129364014,
      "learning_rate": 5.058642121687163e-08,
      "loss": 8.4583,
      "step": 4970
    },
    {
      "epoch": 0.9712778429073857,
      "grad_norm": 8.114288330078125,
      "learning_rate": 4.995264361731811e-08,
      "loss": 8.8281,
      "step": 4971
    },
    {
      "epoch": 0.971473231731145,
      "grad_norm": 9.332640647888184,
      "learning_rate": 4.9322851320644514e-08,
      "loss": 9.5133,
      "step": 4972
    },
    {
      "epoch": 0.9716686205549042,
      "grad_norm": 8.73367691040039,
      "learning_rate": 4.8697044579101295e-08,
      "loss": 7.8143,
      "step": 4973
    },
    {
      "epoch": 0.9718640093786636,
      "grad_norm": 10.668346405029297,
      "learning_rate": 4.8075223643343494e-08,
      "loss": 10.1966,
      "step": 4974
    },
    {
      "epoch": 0.9720593982024228,
      "grad_norm": 11.458003044128418,
      "learning_rate": 4.745738876243078e-08,
      "loss": 8.8785,
      "step": 4975
    },
    {
      "epoch": 0.9722547870261821,
      "grad_norm": 7.601654052734375,
      "learning_rate": 4.684354018382409e-08,
      "loss": 7.8087,
      "step": 4976
    },
    {
      "epoch": 0.9724501758499414,
      "grad_norm": 8.562968254089355,
      "learning_rate": 4.623367815339008e-08,
      "loss": 9.594,
      "step": 4977
    },
    {
      "epoch": 0.9726455646737007,
      "grad_norm": 7.967249870300293,
      "learning_rate": 4.562780291539559e-08,
      "loss": 8.6013,
      "step": 4978
    },
    {
      "epoch": 0.9728409534974599,
      "grad_norm": 10.01623821258545,
      "learning_rate": 4.5025914712514276e-08,
      "loss": 9.4209,
      "step": 4979
    },
    {
      "epoch": 0.9730363423212193,
      "grad_norm": 6.892254829406738,
      "learning_rate": 4.4428013785819954e-08,
      "loss": 9.9811,
      "step": 4980
    },
    {
      "epoch": 0.9732317311449785,
      "grad_norm": 7.683324813842773,
      "learning_rate": 4.3834100374791075e-08,
      "loss": 8.6577,
      "step": 4981
    },
    {
      "epoch": 0.9734271199687378,
      "grad_norm": 8.40973949432373,
      "learning_rate": 4.3244174717308464e-08,
      "loss": 9.6727,
      "step": 4982
    },
    {
      "epoch": 0.9736225087924971,
      "grad_norm": 8.25219440460205,
      "learning_rate": 4.2658237049655325e-08,
      "loss": 8.2001,
      "step": 4983
    },
    {
      "epoch": 0.9738178976162564,
      "grad_norm": 7.207298755645752,
      "learning_rate": 4.2076287606517275e-08,
      "loss": 8.6362,
      "step": 4984
    },
    {
      "epoch": 0.9740132864400156,
      "grad_norm": 11.115377426147461,
      "learning_rate": 4.149832662098452e-08,
      "loss": 9.2928,
      "step": 4985
    },
    {
      "epoch": 0.9742086752637749,
      "grad_norm": 8.779725074768066,
      "learning_rate": 4.0924354324546335e-08,
      "loss": 8.5166,
      "step": 4986
    },
    {
      "epoch": 0.9744040640875342,
      "grad_norm": 10.163342475891113,
      "learning_rate": 4.035437094709882e-08,
      "loss": 10.2392,
      "step": 4987
    },
    {
      "epoch": 0.9745994529112935,
      "grad_norm": 7.949179172515869,
      "learning_rate": 3.9788376716936025e-08,
      "loss": 8.439,
      "step": 4988
    },
    {
      "epoch": 0.9747948417350527,
      "grad_norm": 6.725371360778809,
      "learning_rate": 3.922637186075884e-08,
      "loss": 8.2117,
      "step": 4989
    },
    {
      "epoch": 0.9749902305588121,
      "grad_norm": 8.798787117004395,
      "learning_rate": 3.866835660366386e-08,
      "loss": 9.3537,
      "step": 4990
    },
    {
      "epoch": 0.9751856193825713,
      "grad_norm": 12.254704475402832,
      "learning_rate": 3.811433116915675e-08,
      "loss": 9.1208,
      "step": 4991
    },
    {
      "epoch": 0.9753810082063306,
      "grad_norm": 8.929437637329102,
      "learning_rate": 3.7564295779141115e-08,
      "loss": 9.5874,
      "step": 4992
    },
    {
      "epoch": 0.9755763970300899,
      "grad_norm": 7.432225227355957,
      "learning_rate": 3.701825065392184e-08,
      "loss": 8.3311,
      "step": 4993
    },
    {
      "epoch": 0.9757717858538492,
      "grad_norm": 9.744684219360352,
      "learning_rate": 3.647619601220953e-08,
      "loss": 9.9649,
      "step": 4994
    },
    {
      "epoch": 0.9759671746776084,
      "grad_norm": 7.140275955200195,
      "learning_rate": 3.593813207111052e-08,
      "loss": 8.2258,
      "step": 4995
    },
    {
      "epoch": 0.9761625635013678,
      "grad_norm": 10.876626014709473,
      "learning_rate": 3.540405904613908e-08,
      "loss": 9.3578,
      "step": 4996
    },
    {
      "epoch": 0.976357952325127,
      "grad_norm": 7.864413738250732,
      "learning_rate": 3.487397715120633e-08,
      "loss": 9.3616,
      "step": 4997
    },
    {
      "epoch": 0.9765533411488863,
      "grad_norm": 7.5049638748168945,
      "learning_rate": 3.4347886598626865e-08,
      "loss": 9.0532,
      "step": 4998
    },
    {
      "epoch": 0.9767487299726456,
      "grad_norm": 15.398046493530273,
      "learning_rate": 3.382578759911659e-08,
      "loss": 9.5909,
      "step": 4999
    },
    {
      "epoch": 0.9769441187964049,
      "grad_norm": 7.960659503936768,
      "learning_rate": 3.330768036179266e-08,
      "loss": 9.2443,
      "step": 5000
    },
    {
      "epoch": 0.9771395076201641,
      "grad_norm": 10.92277717590332,
      "learning_rate": 3.279356509417242e-08,
      "loss": 8.6086,
      "step": 5001
    },
    {
      "epoch": 0.9773348964439235,
      "grad_norm": 7.6711039543151855,
      "learning_rate": 3.228344200217559e-08,
      "loss": 9.291,
      "step": 5002
    },
    {
      "epoch": 0.9775302852676827,
      "grad_norm": 6.7085795402526855,
      "learning_rate": 3.177731129012207e-08,
      "loss": 8.5472,
      "step": 5003
    },
    {
      "epoch": 0.977725674091442,
      "grad_norm": 8.609145164489746,
      "learning_rate": 3.127517316073303e-08,
      "loss": 9.4066,
      "step": 5004
    },
    {
      "epoch": 0.9779210629152012,
      "grad_norm": 7.209899425506592,
      "learning_rate": 3.07770278151287e-08,
      "loss": 8.5165,
      "step": 5005
    },
    {
      "epoch": 0.9781164517389606,
      "grad_norm": 9.901558876037598,
      "learning_rate": 3.0282875452835034e-08,
      "loss": 8.6032,
      "step": 5006
    },
    {
      "epoch": 0.9783118405627198,
      "grad_norm": 8.552497863769531,
      "learning_rate": 2.9792716271772605e-08,
      "loss": 9.3627,
      "step": 5007
    },
    {
      "epoch": 0.978507229386479,
      "grad_norm": 9.150060653686523,
      "learning_rate": 2.9306550468266582e-08,
      "loss": 9.2595,
      "step": 5008
    },
    {
      "epoch": 0.9787026182102384,
      "grad_norm": 7.365716934204102,
      "learning_rate": 2.8824378237042318e-08,
      "loss": 9.1018,
      "step": 5009
    },
    {
      "epoch": 0.9788980070339977,
      "grad_norm": 7.778922080993652,
      "learning_rate": 2.8346199771221995e-08,
      "loss": 8.7822,
      "step": 5010
    },
    {
      "epoch": 0.9790933958577569,
      "grad_norm": 8.252825736999512,
      "learning_rate": 2.7872015262334628e-08,
      "loss": 9.1626,
      "step": 5011
    },
    {
      "epoch": 0.9792887846815163,
      "grad_norm": 10.327372550964355,
      "learning_rate": 2.7401824900302743e-08,
      "loss": 9.7747,
      "step": 5012
    },
    {
      "epoch": 0.9794841735052755,
      "grad_norm": 11.820197105407715,
      "learning_rate": 2.6935628873453468e-08,
      "loss": 9.6602,
      "step": 5013
    },
    {
      "epoch": 0.9796795623290347,
      "grad_norm": 83.55939483642578,
      "learning_rate": 2.6473427368511883e-08,
      "loss": 10.351,
      "step": 5014
    },
    {
      "epoch": 0.9798749511527941,
      "grad_norm": 12.766364097595215,
      "learning_rate": 2.6015220570605458e-08,
      "loss": 9.6129,
      "step": 5015
    },
    {
      "epoch": 0.9800703399765534,
      "grad_norm": 10.860968589782715,
      "learning_rate": 2.5561008663259613e-08,
      "loss": 9.4036,
      "step": 5016
    },
    {
      "epoch": 0.9802657288003126,
      "grad_norm": 9.887362480163574,
      "learning_rate": 2.5110791828399928e-08,
      "loss": 8.7792,
      "step": 5017
    },
    {
      "epoch": 0.980461117624072,
      "grad_norm": 9.302169799804688,
      "learning_rate": 2.466457024635327e-08,
      "loss": 8.8441,
      "step": 5018
    },
    {
      "epoch": 0.9806565064478312,
      "grad_norm": 9.141999244689941,
      "learning_rate": 2.4222344095844452e-08,
      "loss": 8.8272,
      "step": 5019
    },
    {
      "epoch": 0.9808518952715904,
      "grad_norm": 9.363895416259766,
      "learning_rate": 2.3784113553998454e-08,
      "loss": 9.4373,
      "step": 5020
    },
    {
      "epoch": 0.9810472840953497,
      "grad_norm": 11.084437370300293,
      "learning_rate": 2.334987879634043e-08,
      "loss": 9.3966,
      "step": 5021
    },
    {
      "epoch": 0.981242672919109,
      "grad_norm": 14.619644165039062,
      "learning_rate": 2.2919639996795695e-08,
      "loss": 9.4309,
      "step": 5022
    },
    {
      "epoch": 0.9814380617428683,
      "grad_norm": 7.190587520599365,
      "learning_rate": 2.2493397327686404e-08,
      "loss": 8.5346,
      "step": 5023
    },
    {
      "epoch": 0.9816334505666275,
      "grad_norm": 9.363481521606445,
      "learning_rate": 2.207115095973822e-08,
      "loss": 9.0853,
      "step": 5024
    },
    {
      "epoch": 0.9818288393903869,
      "grad_norm": 9.095197677612305,
      "learning_rate": 2.1652901062072517e-08,
      "loss": 8.4736,
      "step": 5025
    },
    {
      "epoch": 0.9820242282141461,
      "grad_norm": 8.333402633666992,
      "learning_rate": 2.1238647802213075e-08,
      "loss": 9.3244,
      "step": 5026
    },
    {
      "epoch": 0.9822196170379054,
      "grad_norm": 26.559803009033203,
      "learning_rate": 2.082839134607828e-08,
      "loss": 8.7026,
      "step": 5027
    },
    {
      "epoch": 0.9824150058616647,
      "grad_norm": 11.467344284057617,
      "learning_rate": 2.0422131857991133e-08,
      "loss": 9.5603,
      "step": 5028
    },
    {
      "epoch": 0.982610394685424,
      "grad_norm": 8.219799995422363,
      "learning_rate": 2.0019869500669255e-08,
      "loss": 9.8476,
      "step": 5029
    },
    {
      "epoch": 0.9828057835091832,
      "grad_norm": 13.808085441589355,
      "learning_rate": 1.9621604435232645e-08,
      "loss": 8.0784,
      "step": 5030
    },
    {
      "epoch": 0.9830011723329426,
      "grad_norm": 8.613082885742188,
      "learning_rate": 1.922733682119815e-08,
      "loss": 8.7472,
      "step": 5031
    },
    {
      "epoch": 0.9831965611567018,
      "grad_norm": 21.057920455932617,
      "learning_rate": 1.8837066816482785e-08,
      "loss": 8.6952,
      "step": 5032
    },
    {
      "epoch": 0.9833919499804611,
      "grad_norm": 9.525463104248047,
      "learning_rate": 1.845079457740151e-08,
      "loss": 8.8848,
      "step": 5033
    },
    {
      "epoch": 0.9835873388042204,
      "grad_norm": 9.992208480834961,
      "learning_rate": 1.8068520258667233e-08,
      "loss": 8.8986,
      "step": 5034
    },
    {
      "epoch": 0.9837827276279797,
      "grad_norm": 14.523115158081055,
      "learning_rate": 1.7690244013394142e-08,
      "loss": 9.344,
      "step": 5035
    },
    {
      "epoch": 0.9839781164517389,
      "grad_norm": 8.562339782714844,
      "learning_rate": 1.7315965993093264e-08,
      "loss": 9.6286,
      "step": 5036
    },
    {
      "epoch": 0.9841735052754983,
      "grad_norm": 8.526247024536133,
      "learning_rate": 1.6945686347674683e-08,
      "loss": 9.0399,
      "step": 5037
    },
    {
      "epoch": 0.9843688940992575,
      "grad_norm": 10.577889442443848,
      "learning_rate": 1.657940522544643e-08,
      "loss": 9.2172,
      "step": 5038
    },
    {
      "epoch": 0.9845642829230168,
      "grad_norm": 9.276185035705566,
      "learning_rate": 1.6217122773115603e-08,
      "loss": 7.9005,
      "step": 5039
    },
    {
      "epoch": 0.984759671746776,
      "grad_norm": 7.138728141784668,
      "learning_rate": 1.585883913578834e-08,
      "loss": 8.8503,
      "step": 5040
    },
    {
      "epoch": 0.9849550605705354,
      "grad_norm": 7.980425834655762,
      "learning_rate": 1.5504554456966525e-08,
      "loss": 7.087,
      "step": 5041
    },
    {
      "epoch": 0.9851504493942946,
      "grad_norm": 11.567042350769043,
      "learning_rate": 1.5154268878554422e-08,
      "loss": 9.2517,
      "step": 5042
    },
    {
      "epoch": 0.9853458382180539,
      "grad_norm": 8.005668640136719,
      "learning_rate": 1.4807982540850919e-08,
      "loss": 9.2925,
      "step": 5043
    },
    {
      "epoch": 0.9855412270418132,
      "grad_norm": 9.739407539367676,
      "learning_rate": 1.4465695582553951e-08,
      "loss": 9.5434,
      "step": 5044
    },
    {
      "epoch": 0.9857366158655725,
      "grad_norm": 8.46045970916748,
      "learning_rate": 1.4127408140761633e-08,
      "loss": 9.3594,
      "step": 5045
    },
    {
      "epoch": 0.9859320046893317,
      "grad_norm": 6.429606914520264,
      "learning_rate": 1.3793120350967804e-08,
      "loss": 8.3392,
      "step": 5046
    },
    {
      "epoch": 0.9861273935130911,
      "grad_norm": 9.116524696350098,
      "learning_rate": 1.3462832347065358e-08,
      "loss": 9.0163,
      "step": 5047
    },
    {
      "epoch": 0.9863227823368503,
      "grad_norm": 8.624211311340332,
      "learning_rate": 1.3136544261344031e-08,
      "loss": 9.3884,
      "step": 5048
    },
    {
      "epoch": 0.9865181711606096,
      "grad_norm": 8.262917518615723,
      "learning_rate": 1.2814256224493727e-08,
      "loss": 10.0031,
      "step": 5049
    },
    {
      "epoch": 0.9867135599843689,
      "grad_norm": 7.221947193145752,
      "learning_rate": 1.2495968365598964e-08,
      "loss": 8.3295,
      "step": 5050
    },
    {
      "epoch": 0.9869089488081282,
      "grad_norm": 7.57441520690918,
      "learning_rate": 1.2181680812145546e-08,
      "loss": 9.3922,
      "step": 5051
    },
    {
      "epoch": 0.9871043376318874,
      "grad_norm": 7.57441520690918,
      "learning_rate": 1.2181680812145546e-08,
      "loss": 9.2995,
      "step": 5052
    },
    {
      "epoch": 0.9872997264556468,
      "grad_norm": 8.46338939666748,
      "learning_rate": 1.1871393690014998e-08,
      "loss": 8.8303,
      "step": 5053
    },
    {
      "epoch": 0.987495115279406,
      "grad_norm": 9.874073028564453,
      "learning_rate": 1.1565107123486797e-08,
      "loss": 8.1657,
      "step": 5054
    },
    {
      "epoch": 0.9876905041031653,
      "grad_norm": 10.480962753295898,
      "learning_rate": 1.1262821235237253e-08,
      "loss": 9.8809,
      "step": 5055
    },
    {
      "epoch": 0.9878858929269245,
      "grad_norm": 9.544706344604492,
      "learning_rate": 1.0964536146342852e-08,
      "loss": 9.468,
      "step": 5056
    },
    {
      "epoch": 0.9880812817506839,
      "grad_norm": 9.673017501831055,
      "learning_rate": 1.0670251976275803e-08,
      "loss": 8.9854,
      "step": 5057
    },
    {
      "epoch": 0.9882766705744431,
      "grad_norm": 56.37615966796875,
      "learning_rate": 1.0379968842905153e-08,
      "loss": 9.9719,
      "step": 5058
    },
    {
      "epoch": 0.9884720593982024,
      "grad_norm": 7.474615573883057,
      "learning_rate": 1.0093686862499008e-08,
      "loss": 9.6783,
      "step": 5059
    },
    {
      "epoch": 0.9886674482219617,
      "grad_norm": 7.644552230834961,
      "learning_rate": 9.811406149721203e-09,
      "loss": 9.8834,
      "step": 5060
    },
    {
      "epoch": 0.988862837045721,
      "grad_norm": 9.560027122497559,
      "learning_rate": 9.533126817635741e-09,
      "loss": 9.0952,
      "step": 5061
    },
    {
      "epoch": 0.9890582258694802,
      "grad_norm": 8.46674919128418,
      "learning_rate": 9.25884897770013e-09,
      "loss": 8.96,
      "step": 5062
    },
    {
      "epoch": 0.9892536146932396,
      "grad_norm": 9.916620254516602,
      "learning_rate": 8.98857273977316e-09,
      "loss": 8.7321,
      "step": 5063
    },
    {
      "epoch": 0.9894490035169988,
      "grad_norm": 8.093708038330078,
      "learning_rate": 8.722298212107128e-09,
      "loss": 8.8145,
      "step": 5064
    },
    {
      "epoch": 0.9896443923407581,
      "grad_norm": 6.903542995452881,
      "learning_rate": 8.460025501353386e-09,
      "loss": 7.712,
      "step": 5065
    },
    {
      "epoch": 0.9898397811645174,
      "grad_norm": 7.556875705718994,
      "learning_rate": 8.201754712561238e-09,
      "loss": 8.5463,
      "step": 5066
    },
    {
      "epoch": 0.9900351699882767,
      "grad_norm": 7.805353164672852,
      "learning_rate": 7.947485949175715e-09,
      "loss": 8.5967,
      "step": 5067
    },
    {
      "epoch": 0.9902305588120359,
      "grad_norm": 8.18213176727295,
      "learning_rate": 7.697219313039794e-09,
      "loss": 8.5686,
      "step": 5068
    },
    {
      "epoch": 0.9904259476357953,
      "grad_norm": 9.124711990356445,
      "learning_rate": 7.450954904392182e-09,
      "loss": 9.3234,
      "step": 5069
    },
    {
      "epoch": 0.9906213364595545,
      "grad_norm": 8.662633895874023,
      "learning_rate": 7.2086928218706446e-09,
      "loss": 7.7065,
      "step": 5070
    },
    {
      "epoch": 0.9908167252833138,
      "grad_norm": 7.768984794616699,
      "learning_rate": 6.970433162507562e-09,
      "loss": 8.3612,
      "step": 5071
    },
    {
      "epoch": 0.9910121141070731,
      "grad_norm": 8.390610694885254,
      "learning_rate": 6.7361760217332654e-09,
      "loss": 8.2334,
      "step": 5072
    },
    {
      "epoch": 0.9912075029308324,
      "grad_norm": 7.7365312576293945,
      "learning_rate": 6.505921493374923e-09,
      "loss": 8.4446,
      "step": 5073
    },
    {
      "epoch": 0.9914028917545916,
      "grad_norm": 8.441004753112793,
      "learning_rate": 6.279669669657651e-09,
      "loss": 9.3394,
      "step": 5074
    },
    {
      "epoch": 0.9915982805783509,
      "grad_norm": 10.001994132995605,
      "learning_rate": 6.057420641202294e-09,
      "loss": 8.4434,
      "step": 5075
    },
    {
      "epoch": 0.9917936694021102,
      "grad_norm": 8.94590950012207,
      "learning_rate": 5.839174497025424e-09,
      "loss": 8.8657,
      "step": 5076
    },
    {
      "epoch": 0.9919890582258695,
      "grad_norm": 8.118600845336914,
      "learning_rate": 5.624931324541561e-09,
      "loss": 8.8006,
      "step": 5077
    },
    {
      "epoch": 0.9921844470496287,
      "grad_norm": 10.194698333740234,
      "learning_rate": 5.414691209563172e-09,
      "loss": 9.4743,
      "step": 5078
    },
    {
      "epoch": 0.9923798358733881,
      "grad_norm": 8.967790603637695,
      "learning_rate": 5.208454236296234e-09,
      "loss": 9.1476,
      "step": 5079
    },
    {
      "epoch": 0.9925752246971473,
      "grad_norm": 7.312407970428467,
      "learning_rate": 5.006220487345781e-09,
      "loss": 8.651,
      "step": 5080
    },
    {
      "epoch": 0.9927706135209066,
      "grad_norm": 6.785313606262207,
      "learning_rate": 4.807990043712574e-09,
      "loss": 7.4504,
      "step": 5081
    },
    {
      "epoch": 0.9929660023446659,
      "grad_norm": 10.432636260986328,
      "learning_rate": 4.613762984794212e-09,
      "loss": 8.8484,
      "step": 5082
    },
    {
      "epoch": 0.9931613911684252,
      "grad_norm": 8.083576202392578,
      "learning_rate": 4.423539388385134e-09,
      "loss": 9.9983,
      "step": 5083
    },
    {
      "epoch": 0.9933567799921844,
      "grad_norm": 7.688929080963135,
      "learning_rate": 4.237319330674394e-09,
      "loss": 9.4568,
      "step": 5084
    },
    {
      "epoch": 0.9935521688159438,
      "grad_norm": 8.144380569458008,
      "learning_rate": 4.055102886250106e-09,
      "loss": 8.8193,
      "step": 5085
    },
    {
      "epoch": 0.993747557639703,
      "grad_norm": 10.60041332244873,
      "learning_rate": 3.876890128095001e-09,
      "loss": 8.3073,
      "step": 5086
    },
    {
      "epoch": 0.9939429464634623,
      "grad_norm": 6.922112464904785,
      "learning_rate": 3.702681127589758e-09,
      "loss": 8.6703,
      "step": 5087
    },
    {
      "epoch": 0.9941383352872216,
      "grad_norm": 7.471693992614746,
      "learning_rate": 3.532475954508563e-09,
      "loss": 8.9103,
      "step": 5088
    },
    {
      "epoch": 0.9943337241109809,
      "grad_norm": 10.693000793457031,
      "learning_rate": 3.3662746770268816e-09,
      "loss": 9.3337,
      "step": 5089
    },
    {
      "epoch": 0.9945291129347401,
      "grad_norm": 9.791496276855469,
      "learning_rate": 3.204077361710356e-09,
      "loss": 9.2034,
      "step": 5090
    },
    {
      "epoch": 0.9947245017584995,
      "grad_norm": 9.332125663757324,
      "learning_rate": 3.0458840735259065e-09,
      "loss": 8.6006,
      "step": 5091
    },
    {
      "epoch": 0.9949198905822587,
      "grad_norm": 8.35640811920166,
      "learning_rate": 2.8916948758339613e-09,
      "loss": 9.2044,
      "step": 5092
    },
    {
      "epoch": 0.995115279406018,
      "grad_norm": 9.399685859680176,
      "learning_rate": 2.7415098303928966e-09,
      "loss": 8.4741,
      "step": 5093
    },
    {
      "epoch": 0.9953106682297772,
      "grad_norm": 7.687415599822998,
      "learning_rate": 2.5953289973568164e-09,
      "loss": 8.0641,
      "step": 5094
    },
    {
      "epoch": 0.9955060570535366,
      "grad_norm": 9.725567817687988,
      "learning_rate": 2.4531524352744417e-09,
      "loss": 8.7922,
      "step": 5095
    },
    {
      "epoch": 0.9957014458772958,
      "grad_norm": 8.47549819946289,
      "learning_rate": 2.3149802010913323e-09,
      "loss": 9.0966,
      "step": 5096
    },
    {
      "epoch": 0.9958968347010551,
      "grad_norm": 8.40412425994873,
      "learning_rate": 2.180812350152106e-09,
      "loss": 10.0462,
      "step": 5097
    },
    {
      "epoch": 0.9960922235248144,
      "grad_norm": 9.487896919250488,
      "learning_rate": 2.0506489361937776e-09,
      "loss": 9.3278,
      "step": 5098
    },
    {
      "epoch": 0.9962876123485737,
      "grad_norm": 7.810408592224121,
      "learning_rate": 1.9244900113501996e-09,
      "loss": 8.3744,
      "step": 5099
    },
    {
      "epoch": 0.9964830011723329,
      "grad_norm": 11.756729125976562,
      "learning_rate": 1.8023356261531732e-09,
      "loss": 9.7681,
      "step": 5100
    },
    {
      "epoch": 0.9966783899960923,
      "grad_norm": 8.442209243774414,
      "learning_rate": 1.6841858295291169e-09,
      "loss": 9.5311,
      "step": 5101
    },
    {
      "epoch": 0.9968737788198515,
      "grad_norm": 9.346763610839844,
      "learning_rate": 1.5700406688001768e-09,
      "loss": 9.0829,
      "step": 5102
    },
    {
      "epoch": 0.9970691676436108,
      "grad_norm": 7.547781944274902,
      "learning_rate": 1.4599001896842268e-09,
      "loss": 7.8833,
      "step": 5103
    },
    {
      "epoch": 0.9972645564673701,
      "grad_norm": 8.288264274597168,
      "learning_rate": 1.3537644362981995e-09,
      "loss": 9.1846,
      "step": 5104
    },
    {
      "epoch": 0.9974599452911294,
      "grad_norm": 6.996792793273926,
      "learning_rate": 1.2516334511514238e-09,
      "loss": 8.4772,
      "step": 5105
    },
    {
      "epoch": 0.9976553341148886,
      "grad_norm": 9.186427116394043,
      "learning_rate": 1.1535072751500676e-09,
      "loss": 9.8799,
      "step": 5106
    },
    {
      "epoch": 0.997850722938648,
      "grad_norm": 7.869745254516602,
      "learning_rate": 1.0593859475982459e-09,
      "loss": 9.2626,
      "step": 5107
    },
    {
      "epoch": 0.9980461117624072,
      "grad_norm": 7.204585552215576,
      "learning_rate": 9.692695061924717e-10,
      "loss": 8.7089,
      "step": 5108
    },
    {
      "epoch": 0.9982415005861665,
      "grad_norm": 9.824396133422852,
      "learning_rate": 8.831579870283158e-10,
      "loss": 9.2606,
      "step": 5109
    },
    {
      "epoch": 0.9984368894099257,
      "grad_norm": 8.864141464233398,
      "learning_rate": 8.010514245970769e-10,
      "loss": 8.4175,
      "step": 5110
    },
    {
      "epoch": 0.9986322782336851,
      "grad_norm": 9.267828941345215,
      "learning_rate": 7.22949851783561e-10,
      "loss": 9.4053,
      "step": 5111
    },
    {
      "epoch": 0.9988276670574443,
      "grad_norm": 7.050810813903809,
      "learning_rate": 6.48853299869412e-10,
      "loss": 8.1441,
      "step": 5112
    },
    {
      "epoch": 0.9990230558812035,
      "grad_norm": 9.182863235473633,
      "learning_rate": 5.78761798534222e-10,
      "loss": 9.0271,
      "step": 5113
    },
    {
      "epoch": 0.9992184447049629,
      "grad_norm": 17.024173736572266,
      "learning_rate": 5.126753758499803e-10,
      "loss": 10.0235,
      "step": 5114
    },
    {
      "epoch": 0.9994138335287222,
      "grad_norm": 9.310667991638184,
      "learning_rate": 4.5059405828884464e-10,
      "loss": 8.7047,
      "step": 5115
    },
    {
      "epoch": 0.9996092223524814,
      "grad_norm": 8.904376029968262,
      "learning_rate": 3.9251787071425963e-10,
      "loss": 8.4758,
      "step": 5116
    },
    {
      "epoch": 0.9998046111762408,
      "grad_norm": 6.933563709259033,
      "learning_rate": 3.384468363876181e-10,
      "loss": 8.654,
      "step": 5117
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.946910858154297,
      "learning_rate": 2.8838097696715083e-10,
      "loss": 9.0272,
      "step": 5118
    },
    {
      "epoch": 1.0,
      "step": 5118,
      "total_flos": 4.1041189744830054e+17,
      "train_loss": 9.726896564119167,
      "train_runtime": 62185.3542,
      "train_samples_per_second": 2.634,
      "train_steps_per_second": 0.082
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5118,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.1041189744830054e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}