{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 15.0,
  "eval_steps": 500,
  "global_step": 1755,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008547008547008548,
      "grad_norm": 2.3799763053026064,
      "learning_rate": 4.999995994512315e-06,
      "loss": 0.955,
      "step": 1
    },
    {
      "epoch": 0.017094017094017096,
      "grad_norm": 2.494033963310767,
      "learning_rate": 4.999983978062096e-06,
      "loss": 0.8287,
      "step": 2
    },
    {
      "epoch": 0.02564102564102564,
      "grad_norm": 2.201030171157517,
      "learning_rate": 4.999963950687846e-06,
      "loss": 0.9251,
      "step": 3
    },
    {
      "epoch": 0.03418803418803419,
      "grad_norm": 2.146415568844464,
      "learning_rate": 4.999935912453743e-06,
      "loss": 0.969,
      "step": 4
    },
    {
      "epoch": 0.042735042735042736,
      "grad_norm": 1.647445561456696,
      "learning_rate": 4.999899863449631e-06,
      "loss": 0.8622,
      "step": 5
    },
    {
      "epoch": 0.05128205128205128,
      "grad_norm": 1.6809896297017388,
      "learning_rate": 4.999855803791026e-06,
      "loss": 0.8451,
      "step": 6
    },
    {
      "epoch": 0.05982905982905983,
      "grad_norm": 1.5793650947747724,
      "learning_rate": 4.9998037336191115e-06,
      "loss": 0.8039,
      "step": 7
    },
    {
      "epoch": 0.06837606837606838,
      "grad_norm": 1.5869313841843558,
      "learning_rate": 4.9997436531007415e-06,
      "loss": 0.8729,
      "step": 8
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 1.597287174144046,
      "learning_rate": 4.999675562428437e-06,
      "loss": 0.8276,
      "step": 9
    },
    {
      "epoch": 0.08547008547008547,
      "grad_norm": 1.491091630883272,
      "learning_rate": 4.999599461820387e-06,
      "loss": 0.7985,
      "step": 10
    },
    {
      "epoch": 0.09401709401709402,
      "grad_norm": 1.4611263163645851,
      "learning_rate": 4.999515351520447e-06,
      "loss": 0.8432,
      "step": 11
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 1.5771492301572616,
      "learning_rate": 4.9994232317981405e-06,
      "loss": 0.9565,
      "step": 12
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 1.470477557035158,
      "learning_rate": 4.999323102948655e-06,
      "loss": 1.1248,
      "step": 13
    },
    {
      "epoch": 0.11965811965811966,
      "grad_norm": 1.1911774620668554,
      "learning_rate": 4.999214965292841e-06,
      "loss": 0.6717,
      "step": 14
    },
    {
      "epoch": 0.1282051282051282,
      "grad_norm": 1.3432981249016747,
      "learning_rate": 4.999098819177214e-06,
      "loss": 0.7159,
      "step": 15
    },
    {
      "epoch": 0.13675213675213677,
      "grad_norm": 1.3091938255082172,
      "learning_rate": 4.998974664973953e-06,
      "loss": 0.6966,
      "step": 16
    },
    {
      "epoch": 0.1452991452991453,
      "grad_norm": 1.4536201204300017,
      "learning_rate": 4.998842503080894e-06,
      "loss": 0.9169,
      "step": 17
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 1.2252488705593612,
      "learning_rate": 4.998702333921538e-06,
      "loss": 0.7068,
      "step": 18
    },
    {
      "epoch": 0.1623931623931624,
      "grad_norm": 1.1196522934351878,
      "learning_rate": 4.99855415794504e-06,
      "loss": 0.7144,
      "step": 19
    },
    {
      "epoch": 0.17094017094017094,
      "grad_norm": 1.4171269229549868,
      "learning_rate": 4.998397975626213e-06,
      "loss": 0.9539,
      "step": 20
    },
    {
      "epoch": 0.1794871794871795,
      "grad_norm": 1.2794264496219363,
      "learning_rate": 4.998233787465529e-06,
      "loss": 0.7078,
      "step": 21
    },
    {
      "epoch": 0.18803418803418803,
      "grad_norm": 1.3730284575172316,
      "learning_rate": 4.998061593989108e-06,
      "loss": 0.8275,
      "step": 22
    },
    {
      "epoch": 0.19658119658119658,
      "grad_norm": 1.4940223968438926,
      "learning_rate": 4.997881395748727e-06,
      "loss": 0.8149,
      "step": 23
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 1.127435221045007,
      "learning_rate": 4.99769319332181e-06,
      "loss": 0.7661,
      "step": 24
    },
    {
      "epoch": 0.21367521367521367,
      "grad_norm": 1.4374573099245262,
      "learning_rate": 4.997496987311431e-06,
      "loss": 0.8096,
      "step": 25
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1.0383554614188748,
      "learning_rate": 4.997292778346312e-06,
      "loss": 0.6898,
      "step": 26
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 1.2454500227279588,
      "learning_rate": 4.9970805670808174e-06,
      "loss": 0.8476,
      "step": 27
    },
    {
      "epoch": 0.23931623931623933,
      "grad_norm": 1.3895431899110833,
      "learning_rate": 4.996860354194954e-06,
      "loss": 0.9476,
      "step": 28
    },
    {
      "epoch": 0.24786324786324787,
      "grad_norm": 1.2147090185053293,
      "learning_rate": 4.996632140394372e-06,
      "loss": 0.9277,
      "step": 29
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 1.201270924823543,
      "learning_rate": 4.996395926410354e-06,
      "loss": 0.7633,
      "step": 30
    },
    {
      "epoch": 0.26495726495726496,
      "grad_norm": 1.1116608939922576,
      "learning_rate": 4.996151712999826e-06,
      "loss": 0.8779,
      "step": 31
    },
    {
      "epoch": 0.27350427350427353,
      "grad_norm": 1.2255856149469802,
      "learning_rate": 4.995899500945341e-06,
      "loss": 0.6816,
      "step": 32
    },
    {
      "epoch": 0.28205128205128205,
      "grad_norm": 0.9559499921214146,
      "learning_rate": 4.995639291055084e-06,
      "loss": 0.7282,
      "step": 33
    },
    {
      "epoch": 0.2905982905982906,
      "grad_norm": 0.9736062146544777,
      "learning_rate": 4.99537108416287e-06,
      "loss": 0.706,
      "step": 34
    },
    {
      "epoch": 0.29914529914529914,
      "grad_norm": 1.1387327794469184,
      "learning_rate": 4.995094881128138e-06,
      "loss": 0.6738,
      "step": 35
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 1.248397867664662,
      "learning_rate": 4.994810682835951e-06,
      "loss": 0.7774,
      "step": 36
    },
    {
      "epoch": 0.3162393162393162,
      "grad_norm": 1.1882850864655046,
      "learning_rate": 4.99451849019699e-06,
      "loss": 0.5986,
      "step": 37
    },
    {
      "epoch": 0.3247863247863248,
      "grad_norm": 1.0587607675254118,
      "learning_rate": 4.994218304147556e-06,
      "loss": 0.7882,
      "step": 38
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.0022539150956298,
      "learning_rate": 4.993910125649561e-06,
      "loss": 0.8321,
      "step": 39
    },
    {
      "epoch": 0.3418803418803419,
      "grad_norm": 1.1127228219625331,
      "learning_rate": 4.993593955690529e-06,
      "loss": 0.6737,
      "step": 40
    },
    {
      "epoch": 0.3504273504273504,
      "grad_norm": 1.114475532217064,
      "learning_rate": 4.9932697952835925e-06,
      "loss": 0.7492,
      "step": 41
    },
    {
      "epoch": 0.358974358974359,
      "grad_norm": 1.04367701725325,
      "learning_rate": 4.992937645467489e-06,
      "loss": 0.8464,
      "step": 42
    },
    {
      "epoch": 0.36752136752136755,
      "grad_norm": 1.0727745473359473,
      "learning_rate": 4.992597507306552e-06,
      "loss": 1.0087,
      "step": 43
    },
    {
      "epoch": 0.37606837606837606,
      "grad_norm": 0.9660036788448103,
      "learning_rate": 4.992249381890722e-06,
      "loss": 0.7119,
      "step": 44
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 1.2125533839718292,
      "learning_rate": 4.991893270335526e-06,
      "loss": 1.0206,
      "step": 45
    },
    {
      "epoch": 0.39316239316239315,
      "grad_norm": 1.1273464632844163,
      "learning_rate": 4.9915291737820836e-06,
      "loss": 0.6818,
      "step": 46
    },
    {
      "epoch": 0.4017094017094017,
      "grad_norm": 1.0554418415896571,
      "learning_rate": 4.991157093397104e-06,
      "loss": 0.8175,
      "step": 47
    },
    {
      "epoch": 0.41025641025641024,
      "grad_norm": 1.0700579311552258,
      "learning_rate": 4.990777030372877e-06,
      "loss": 0.6876,
      "step": 48
    },
    {
      "epoch": 0.4188034188034188,
      "grad_norm": 1.0637403108101426,
      "learning_rate": 4.990388985927273e-06,
      "loss": 0.8423,
      "step": 49
    },
    {
      "epoch": 0.42735042735042733,
      "grad_norm": 0.9688437819111178,
      "learning_rate": 4.989992961303738e-06,
      "loss": 0.7794,
      "step": 50
    },
    {
      "epoch": 0.4358974358974359,
      "grad_norm": 1.0114236396805474,
      "learning_rate": 4.989588957771289e-06,
      "loss": 0.6794,
      "step": 51
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 1.0131329274100702,
      "learning_rate": 4.989176976624511e-06,
      "loss": 0.5956,
      "step": 52
    },
    {
      "epoch": 0.452991452991453,
      "grad_norm": 1.0333055990414326,
      "learning_rate": 4.988757019183553e-06,
      "loss": 0.6177,
      "step": 53
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 1.148102486098422,
      "learning_rate": 4.988329086794122e-06,
      "loss": 0.7744,
      "step": 54
    },
    {
      "epoch": 0.4700854700854701,
      "grad_norm": 1.3052841083061404,
      "learning_rate": 4.9878931808274796e-06,
      "loss": 0.644,
      "step": 55
    },
    {
      "epoch": 0.47863247863247865,
      "grad_norm": 1.0116165469923584,
      "learning_rate": 4.98744930268044e-06,
      "loss": 0.6745,
      "step": 56
    },
    {
      "epoch": 0.48717948717948717,
      "grad_norm": 1.086387211887458,
      "learning_rate": 4.986997453775361e-06,
      "loss": 0.736,
      "step": 57
    },
    {
      "epoch": 0.49572649572649574,
      "grad_norm": 0.9988882124617762,
      "learning_rate": 4.986537635560144e-06,
      "loss": 0.6586,
      "step": 58
    },
    {
      "epoch": 0.5042735042735043,
      "grad_norm": 1.0656418155038476,
      "learning_rate": 4.986069849508223e-06,
      "loss": 0.7999,
      "step": 59
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 1.1201336257983052,
      "learning_rate": 4.9855940971185705e-06,
      "loss": 0.7822,
      "step": 60
    },
    {
      "epoch": 0.5213675213675214,
      "grad_norm": 1.1598381779546625,
      "learning_rate": 4.985110379915681e-06,
      "loss": 0.764,
      "step": 61
    },
    {
      "epoch": 0.5299145299145299,
      "grad_norm": 1.0921954470681912,
      "learning_rate": 4.984618699449573e-06,
      "loss": 0.6829,
      "step": 62
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 0.9783029683124532,
      "learning_rate": 4.984119057295783e-06,
      "loss": 0.6498,
      "step": 63
    },
    {
      "epoch": 0.5470085470085471,
      "grad_norm": 1.076901000673284,
      "learning_rate": 4.983611455055359e-06,
      "loss": 0.667,
      "step": 64
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 1.2457502019896778,
      "learning_rate": 4.983095894354858e-06,
      "loss": 0.8111,
      "step": 65
    },
    {
      "epoch": 0.5641025641025641,
      "grad_norm": 1.0896425595072872,
      "learning_rate": 4.982572376846336e-06,
      "loss": 0.6329,
      "step": 66
    },
    {
      "epoch": 0.5726495726495726,
      "grad_norm": 0.9232081104327517,
      "learning_rate": 4.982040904207348e-06,
      "loss": 0.8773,
      "step": 67
    },
    {
      "epoch": 0.5811965811965812,
      "grad_norm": 0.9422862400984294,
      "learning_rate": 4.98150147814094e-06,
      "loss": 0.8834,
      "step": 68
    },
    {
      "epoch": 0.5897435897435898,
      "grad_norm": 1.1595435361807889,
      "learning_rate": 4.980954100375642e-06,
      "loss": 0.6244,
      "step": 69
    },
    {
      "epoch": 0.5982905982905983,
      "grad_norm": 1.146463819945862,
      "learning_rate": 4.980398772665468e-06,
      "loss": 0.6958,
      "step": 70
    },
    {
      "epoch": 0.6068376068376068,
      "grad_norm": 1.1240734279946885,
      "learning_rate": 4.979835496789904e-06,
      "loss": 0.7953,
      "step": 71
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 1.1655040421601977,
      "learning_rate": 4.979264274553906e-06,
      "loss": 0.9859,
      "step": 72
    },
    {
      "epoch": 0.6239316239316239,
      "grad_norm": 0.9972089985683658,
      "learning_rate": 4.97868510778789e-06,
      "loss": 0.7303,
      "step": 73
    },
    {
      "epoch": 0.6324786324786325,
      "grad_norm": 1.1284011355004735,
      "learning_rate": 4.978097998347737e-06,
      "loss": 0.861,
      "step": 74
    },
    {
      "epoch": 0.6410256410256411,
      "grad_norm": 1.050649795863996,
      "learning_rate": 4.977502948114772e-06,
      "loss": 0.7585,
      "step": 75
    },
    {
      "epoch": 0.6495726495726496,
      "grad_norm": 1.2055242119481449,
      "learning_rate": 4.9768999589957675e-06,
      "loss": 0.7491,
      "step": 76
    },
    {
      "epoch": 0.6581196581196581,
      "grad_norm": 1.045945265764331,
      "learning_rate": 4.976289032922937e-06,
      "loss": 0.8598,
      "step": 77
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 1.0154676826015707,
      "learning_rate": 4.975670171853926e-06,
      "loss": 0.9274,
      "step": 78
    },
    {
      "epoch": 0.6752136752136753,
      "grad_norm": 1.0849056329321192,
      "learning_rate": 4.975043377771806e-06,
      "loss": 0.6739,
      "step": 79
    },
    {
      "epoch": 0.6837606837606838,
      "grad_norm": 0.918648215949127,
      "learning_rate": 4.9744086526850724e-06,
      "loss": 0.723,
      "step": 80
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 1.269473123433632,
      "learning_rate": 4.973765998627628e-06,
      "loss": 0.9916,
      "step": 81
    },
    {
      "epoch": 0.7008547008547008,
      "grad_norm": 1.0609779535228994,
      "learning_rate": 4.97311541765879e-06,
      "loss": 0.8092,
      "step": 82
    },
    {
      "epoch": 0.7094017094017094,
      "grad_norm": 1.1075178803628833,
      "learning_rate": 4.972456911863273e-06,
      "loss": 0.8491,
      "step": 83
    },
    {
      "epoch": 0.717948717948718,
      "grad_norm": 1.1177775363091615,
      "learning_rate": 4.971790483351186e-06,
      "loss": 0.7181,
      "step": 84
    },
    {
      "epoch": 0.7264957264957265,
      "grad_norm": 1.19435718577341,
      "learning_rate": 4.971116134258026e-06,
      "loss": 0.7371,
      "step": 85
    },
    {
      "epoch": 0.7350427350427351,
      "grad_norm": 0.9344635761516649,
      "learning_rate": 4.97043386674467e-06,
      "loss": 0.8082,
      "step": 86
    },
    {
      "epoch": 0.7435897435897436,
      "grad_norm": 1.1047970864696917,
      "learning_rate": 4.969743682997372e-06,
      "loss": 0.6576,
      "step": 87
    },
    {
      "epoch": 0.7521367521367521,
      "grad_norm": 1.1492319889376637,
      "learning_rate": 4.969045585227747e-06,
      "loss": 0.7261,
      "step": 88
    },
    {
      "epoch": 0.7606837606837606,
      "grad_norm": 0.8540587595768252,
      "learning_rate": 4.968339575672773e-06,
      "loss": 0.7566,
      "step": 89
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 1.435559330737617,
      "learning_rate": 4.967625656594782e-06,
      "loss": 0.8515,
      "step": 90
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 1.0885688071755686,
      "learning_rate": 4.966903830281449e-06,
      "loss": 0.7278,
      "step": 91
    },
    {
      "epoch": 0.7863247863247863,
      "grad_norm": 1.0793773087579055,
      "learning_rate": 4.966174099045784e-06,
      "loss": 0.7646,
      "step": 92
    },
    {
      "epoch": 0.7948717948717948,
      "grad_norm": 1.192286892157816,
      "learning_rate": 4.9654364652261345e-06,
      "loss": 0.8218,
      "step": 93
    },
    {
      "epoch": 0.8034188034188035,
      "grad_norm": 0.9421147050497265,
      "learning_rate": 4.964690931186165e-06,
      "loss": 0.7428,
      "step": 94
    },
    {
      "epoch": 0.811965811965812,
      "grad_norm": 1.0300568563597792,
      "learning_rate": 4.963937499314857e-06,
      "loss": 0.9733,
      "step": 95
    },
    {
      "epoch": 0.8205128205128205,
      "grad_norm": 1.1507829257449622,
      "learning_rate": 4.963176172026501e-06,
      "loss": 0.8656,
      "step": 96
    },
    {
      "epoch": 0.8290598290598291,
      "grad_norm": 1.0732730564082225,
      "learning_rate": 4.962406951760687e-06,
      "loss": 0.6834,
      "step": 97
    },
    {
      "epoch": 0.8376068376068376,
      "grad_norm": 1.272187919122158,
      "learning_rate": 4.961629840982296e-06,
      "loss": 0.5806,
      "step": 98
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 1.0209482085094501,
      "learning_rate": 4.9608448421814944e-06,
      "loss": 0.593,
      "step": 99
    },
    {
      "epoch": 0.8547008547008547,
      "grad_norm": 1.070116471178533,
      "learning_rate": 4.960051957873726e-06,
      "loss": 0.7731,
      "step": 100
    },
    {
      "epoch": 0.8632478632478633,
      "grad_norm": 1.1286783767982944,
      "learning_rate": 4.959251190599699e-06,
      "loss": 0.9449,
      "step": 101
    },
    {
      "epoch": 0.8717948717948718,
      "grad_norm": 0.9613784562622458,
      "learning_rate": 4.958442542925385e-06,
      "loss": 0.5717,
      "step": 102
    },
    {
      "epoch": 0.8803418803418803,
      "grad_norm": 1.0834987740649913,
      "learning_rate": 4.9576260174420085e-06,
      "loss": 0.6655,
      "step": 103
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 1.1508010697308189,
      "learning_rate": 4.956801616766033e-06,
      "loss": 0.7036,
      "step": 104
    },
    {
      "epoch": 0.8974358974358975,
      "grad_norm": 0.857235837591077,
      "learning_rate": 4.955969343539162e-06,
      "loss": 0.7836,
      "step": 105
    },
    {
      "epoch": 0.905982905982906,
      "grad_norm": 0.9861436603055789,
      "learning_rate": 4.955129200428323e-06,
      "loss": 0.6834,
      "step": 106
    },
    {
      "epoch": 0.9145299145299145,
      "grad_norm": 1.0827086944967481,
      "learning_rate": 4.9542811901256615e-06,
      "loss": 0.7214,
      "step": 107
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 1.0999297259841632,
      "learning_rate": 4.953425315348534e-06,
      "loss": 0.6893,
      "step": 108
    },
    {
      "epoch": 0.9316239316239316,
      "grad_norm": 1.22960169622494,
      "learning_rate": 4.952561578839498e-06,
      "loss": 0.7697,
      "step": 109
    },
    {
      "epoch": 0.9401709401709402,
      "grad_norm": 1.0578878999566572,
      "learning_rate": 4.9516899833663e-06,
      "loss": 0.7045,
      "step": 110
    },
    {
      "epoch": 0.9487179487179487,
      "grad_norm": 1.1846780679296018,
      "learning_rate": 4.950810531721874e-06,
      "loss": 0.7342,
      "step": 111
    },
    {
      "epoch": 0.9572649572649573,
      "grad_norm": 1.1912506798196463,
      "learning_rate": 4.949923226724325e-06,
      "loss": 0.8233,
      "step": 112
    },
    {
      "epoch": 0.9658119658119658,
      "grad_norm": 1.1252852469907122,
      "learning_rate": 4.949028071216926e-06,
      "loss": 0.624,
      "step": 113
    },
    {
      "epoch": 0.9743589743589743,
      "grad_norm": 1.1517000451843173,
      "learning_rate": 4.948125068068102e-06,
      "loss": 0.544,
      "step": 114
    },
    {
      "epoch": 0.9829059829059829,
      "grad_norm": 1.060592964482793,
      "learning_rate": 4.94721422017143e-06,
      "loss": 0.959,
      "step": 115
    },
    {
      "epoch": 0.9914529914529915,
      "grad_norm": 1.0042046551638841,
      "learning_rate": 4.946295530445621e-06,
      "loss": 0.7238,
      "step": 116
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0001249481240222,
      "learning_rate": 4.9453690018345144e-06,
      "loss": 0.619,
      "step": 117
    },
    {
      "epoch": 1.0085470085470085,
      "grad_norm": 0.9380046535719442,
      "learning_rate": 4.94443463730707e-06,
      "loss": 0.6444,
      "step": 118
    },
    {
      "epoch": 1.017094017094017,
      "grad_norm": 1.0293011172402107,
      "learning_rate": 4.943492439857357e-06,
      "loss": 0.7716,
      "step": 119
    },
    {
      "epoch": 1.0256410256410255,
      "grad_norm": 0.9855467678686244,
      "learning_rate": 4.942542412504543e-06,
      "loss": 0.7719,
      "step": 120
    },
    {
      "epoch": 1.0341880341880343,
      "grad_norm": 1.1807086649027472,
      "learning_rate": 4.9415845582928866e-06,
      "loss": 0.6453,
      "step": 121
    },
    {
      "epoch": 1.0427350427350428,
      "grad_norm": 1.1780514473878632,
      "learning_rate": 4.940618880291725e-06,
      "loss": 0.9361,
      "step": 122
    },
    {
      "epoch": 1.0512820512820513,
      "grad_norm": 1.0138477857952197,
      "learning_rate": 4.9396453815954695e-06,
      "loss": 0.5969,
      "step": 123
    },
    {
      "epoch": 1.0598290598290598,
      "grad_norm": 0.9411679288409218,
      "learning_rate": 4.938664065323588e-06,
      "loss": 0.6425,
      "step": 124
    },
    {
      "epoch": 1.0683760683760684,
      "grad_norm": 1.0955596194799122,
      "learning_rate": 4.937674934620601e-06,
      "loss": 0.7439,
      "step": 125
    },
    {
      "epoch": 1.0769230769230769,
      "grad_norm": 1.2132702603532166,
      "learning_rate": 4.9366779926560705e-06,
      "loss": 0.7427,
      "step": 126
    },
    {
      "epoch": 1.0854700854700854,
      "grad_norm": 0.8719242281374018,
      "learning_rate": 4.935673242624585e-06,
      "loss": 0.5546,
      "step": 127
    },
    {
      "epoch": 1.0940170940170941,
      "grad_norm": 0.913689131986586,
      "learning_rate": 4.934660687745758e-06,
      "loss": 0.6433,
      "step": 128
    },
    {
      "epoch": 1.1025641025641026,
      "grad_norm": 1.0562719427443,
      "learning_rate": 4.93364033126421e-06,
      "loss": 0.6447,
      "step": 129
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 1.1406761768492368,
      "learning_rate": 4.93261217644956e-06,
      "loss": 0.7865,
      "step": 130
    },
    {
      "epoch": 1.1196581196581197,
      "grad_norm": 1.0287132018048986,
      "learning_rate": 4.931576226596418e-06,
      "loss": 0.8902,
      "step": 131
    },
    {
      "epoch": 1.1282051282051282,
      "grad_norm": 1.0679366877890835,
      "learning_rate": 4.930532485024372e-06,
      "loss": 0.5658,
      "step": 132
    },
    {
      "epoch": 1.1367521367521367,
      "grad_norm": 1.1236922170795136,
      "learning_rate": 4.929480955077976e-06,
      "loss": 0.5752,
      "step": 133
    },
    {
      "epoch": 1.1452991452991452,
      "grad_norm": 0.8853014304117948,
      "learning_rate": 4.928421640126742e-06,
      "loss": 0.5763,
      "step": 134
    },
    {
      "epoch": 1.1538461538461537,
      "grad_norm": 0.9775295301218139,
      "learning_rate": 4.927354543565131e-06,
      "loss": 0.6277,
      "step": 135
    },
    {
      "epoch": 1.1623931623931625,
      "grad_norm": 0.9579962790913475,
      "learning_rate": 4.926279668812533e-06,
      "loss": 0.8088,
      "step": 136
    },
    {
      "epoch": 1.170940170940171,
      "grad_norm": 1.1435685522698529,
      "learning_rate": 4.925197019313269e-06,
      "loss": 0.697,
      "step": 137
    },
    {
      "epoch": 1.1794871794871795,
      "grad_norm": 0.870824019840728,
      "learning_rate": 4.9241065985365695e-06,
      "loss": 0.6292,
      "step": 138
    },
    {
      "epoch": 1.188034188034188,
      "grad_norm": 1.0983119201797829,
      "learning_rate": 4.923008409976568e-06,
      "loss": 0.7311,
      "step": 139
    },
    {
      "epoch": 1.1965811965811965,
      "grad_norm": 1.0305388966879632,
      "learning_rate": 4.921902457152289e-06,
      "loss": 0.7046,
      "step": 140
    },
    {
      "epoch": 1.205128205128205,
      "grad_norm": 1.119605051218424,
      "learning_rate": 4.920788743607636e-06,
      "loss": 0.843,
      "step": 141
    },
    {
      "epoch": 1.2136752136752136,
      "grad_norm": 0.915923341125377,
      "learning_rate": 4.919667272911383e-06,
      "loss": 0.6354,
      "step": 142
    },
    {
      "epoch": 1.2222222222222223,
      "grad_norm": 0.9108503566123858,
      "learning_rate": 4.91853804865716e-06,
      "loss": 0.7452,
      "step": 143
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 0.9004822052884224,
      "learning_rate": 4.917401074463441e-06,
      "loss": 0.7132,
      "step": 144
    },
    {
      "epoch": 1.2393162393162394,
      "grad_norm": 1.0063673480230306,
      "learning_rate": 4.916256353973535e-06,
      "loss": 0.6438,
      "step": 145
    },
    {
      "epoch": 1.2478632478632479,
      "grad_norm": 0.9421746838568855,
      "learning_rate": 4.915103890855574e-06,
      "loss": 0.578,
      "step": 146
    },
    {
      "epoch": 1.2564102564102564,
      "grad_norm": 1.0676892185226952,
      "learning_rate": 4.913943688802497e-06,
      "loss": 0.6855,
      "step": 147
    },
    {
      "epoch": 1.264957264957265,
      "grad_norm": 1.1188578607492226,
      "learning_rate": 4.912775751532047e-06,
      "loss": 0.7427,
      "step": 148
    },
    {
      "epoch": 1.2735042735042734,
      "grad_norm": 0.9469435476035265,
      "learning_rate": 4.91160008278675e-06,
      "loss": 0.5891,
      "step": 149
    },
    {
      "epoch": 1.282051282051282,
      "grad_norm": 1.0167095611518875,
      "learning_rate": 4.9104166863339065e-06,
      "loss": 0.7139,
      "step": 150
    },
    {
      "epoch": 1.2905982905982907,
      "grad_norm": 1.0458260632626932,
      "learning_rate": 4.90922556596558e-06,
      "loss": 0.874,
      "step": 151
    },
    {
      "epoch": 1.2991452991452992,
      "grad_norm": 1.0453297709360865,
      "learning_rate": 4.908026725498586e-06,
      "loss": 0.5972,
      "step": 152
    },
    {
      "epoch": 1.3076923076923077,
      "grad_norm": 1.0265238458385364,
      "learning_rate": 4.9068201687744774e-06,
      "loss": 0.6141,
      "step": 153
    },
    {
      "epoch": 1.3162393162393162,
      "grad_norm": 0.9359503022694231,
      "learning_rate": 4.905605899659532e-06,
      "loss": 0.5768,
      "step": 154
    },
    {
      "epoch": 1.3247863247863247,
      "grad_norm": 1.1684603854006186,
      "learning_rate": 4.90438392204474e-06,
      "loss": 0.7656,
      "step": 155
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 1.0501902958929783,
      "learning_rate": 4.903154239845798e-06,
      "loss": 0.7487,
      "step": 156
    },
    {
      "epoch": 1.341880341880342,
      "grad_norm": 1.1842195482922369,
      "learning_rate": 4.901916857003084e-06,
      "loss": 0.8917,
      "step": 157
    },
    {
      "epoch": 1.3504273504273505,
      "grad_norm": 0.9216838395103608,
      "learning_rate": 4.9006717774816585e-06,
      "loss": 0.9301,
      "step": 158
    },
    {
      "epoch": 1.358974358974359,
      "grad_norm": 1.1773273648490519,
      "learning_rate": 4.8994190052712406e-06,
      "loss": 0.4604,
      "step": 159
    },
    {
      "epoch": 1.3675213675213675,
      "grad_norm": 0.9974738324004961,
      "learning_rate": 4.898158544386201e-06,
      "loss": 0.8768,
      "step": 160
    },
    {
      "epoch": 1.376068376068376,
      "grad_norm": 0.9555997535951641,
      "learning_rate": 4.896890398865548e-06,
      "loss": 0.5342,
      "step": 161
    },
    {
      "epoch": 1.3846153846153846,
      "grad_norm": 1.0211237936640918,
      "learning_rate": 4.895614572772916e-06,
      "loss": 0.5775,
      "step": 162
    },
    {
      "epoch": 1.393162393162393,
      "grad_norm": 1.0818255776569934,
      "learning_rate": 4.894331070196548e-06,
      "loss": 0.7447,
      "step": 163
    },
    {
      "epoch": 1.4017094017094016,
      "grad_norm": 0.9298860880587865,
      "learning_rate": 4.893039895249288e-06,
      "loss": 0.5779,
      "step": 164
    },
    {
      "epoch": 1.4102564102564101,
      "grad_norm": 1.0062273244598126,
      "learning_rate": 4.8917410520685635e-06,
      "loss": 0.6862,
      "step": 165
    },
    {
      "epoch": 1.4188034188034189,
      "grad_norm": 1.0998490770749472,
      "learning_rate": 4.890434544816375e-06,
      "loss": 0.6407,
      "step": 166
    },
    {
      "epoch": 1.4273504273504274,
      "grad_norm": 1.0565419203304869,
      "learning_rate": 4.889120377679282e-06,
      "loss": 0.6379,
      "step": 167
    },
    {
      "epoch": 1.435897435897436,
      "grad_norm": 1.0792399462153217,
      "learning_rate": 4.887798554868388e-06,
      "loss": 0.6028,
      "step": 168
    },
    {
      "epoch": 1.4444444444444444,
      "grad_norm": 1.0475389343787085,
      "learning_rate": 4.88646908061933e-06,
      "loss": 0.586,
      "step": 169
    },
    {
      "epoch": 1.452991452991453,
      "grad_norm": 1.0323292401835535,
      "learning_rate": 4.885131959192262e-06,
      "loss": 0.6262,
      "step": 170
    },
    {
      "epoch": 1.4615384615384617,
      "grad_norm": 1.1577807260074058,
      "learning_rate": 4.883787194871841e-06,
      "loss": 0.6364,
      "step": 171
    },
    {
      "epoch": 1.4700854700854702,
      "grad_norm": 0.8278225378608461,
      "learning_rate": 4.882434791967219e-06,
      "loss": 0.5483,
      "step": 172
    },
    {
      "epoch": 1.4786324786324787,
      "grad_norm": 1.1193607717443066,
      "learning_rate": 4.881074754812021e-06,
      "loss": 0.5526,
      "step": 173
    },
    {
      "epoch": 1.4871794871794872,
      "grad_norm": 1.1417530862216305,
      "learning_rate": 4.879707087764336e-06,
      "loss": 0.6403,
      "step": 174
    },
    {
      "epoch": 1.4957264957264957,
      "grad_norm": 0.9659694011039861,
      "learning_rate": 4.878331795206705e-06,
      "loss": 0.7537,
      "step": 175
    },
    {
      "epoch": 1.5042735042735043,
      "grad_norm": 1.0693878337410088,
      "learning_rate": 4.876948881546101e-06,
      "loss": 0.6976,
      "step": 176
    },
    {
      "epoch": 1.5128205128205128,
      "grad_norm": 1.0498589499731368,
      "learning_rate": 4.875558351213918e-06,
      "loss": 0.7038,
      "step": 177
    },
    {
      "epoch": 1.5213675213675213,
      "grad_norm": 1.1001761941159536,
      "learning_rate": 4.874160208665958e-06,
      "loss": 0.4691,
      "step": 178
    },
    {
      "epoch": 1.5299145299145298,
      "grad_norm": 1.2397955357433883,
      "learning_rate": 4.872754458382416e-06,
      "loss": 0.6688,
      "step": 179
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.9672667050888256,
      "learning_rate": 4.8713411048678635e-06,
      "loss": 0.8776,
      "step": 180
    },
    {
      "epoch": 1.547008547008547,
      "grad_norm": 1.051939715180058,
      "learning_rate": 4.869920152651239e-06,
      "loss": 0.8041,
      "step": 181
    },
    {
      "epoch": 1.5555555555555556,
      "grad_norm": 1.3625309523663536,
      "learning_rate": 4.868491606285823e-06,
      "loss": 0.8096,
      "step": 182
    },
    {
      "epoch": 1.564102564102564,
      "grad_norm": 1.1734670133598417,
      "learning_rate": 4.86705547034924e-06,
      "loss": 0.6965,
      "step": 183
    },
    {
      "epoch": 1.5726495726495726,
      "grad_norm": 1.0740816983559884,
      "learning_rate": 4.865611749443428e-06,
      "loss": 0.6973,
      "step": 184
    },
    {
      "epoch": 1.5811965811965814,
      "grad_norm": 0.8760799984387524,
      "learning_rate": 4.864160448194632e-06,
      "loss": 0.5856,
      "step": 185
    },
    {
      "epoch": 1.5897435897435899,
      "grad_norm": 1.184991710617603,
      "learning_rate": 4.862701571253387e-06,
      "loss": 0.5975,
      "step": 186
    },
    {
      "epoch": 1.5982905982905984,
      "grad_norm": 0.9980320434419719,
      "learning_rate": 4.861235123294505e-06,
      "loss": 0.595,
      "step": 187
    },
    {
      "epoch": 1.606837606837607,
      "grad_norm": 1.2814407369839131,
      "learning_rate": 4.859761109017056e-06,
      "loss": 0.5688,
      "step": 188
    },
    {
      "epoch": 1.6153846153846154,
      "grad_norm": 1.0094832294279448,
      "learning_rate": 4.858279533144358e-06,
      "loss": 0.671,
      "step": 189
    },
    {
      "epoch": 1.623931623931624,
      "grad_norm": 1.1444371260857773,
      "learning_rate": 4.856790400423958e-06,
      "loss": 0.6731,
      "step": 190
    },
    {
      "epoch": 1.6324786324786325,
      "grad_norm": 1.1115974299621325,
      "learning_rate": 4.8552937156276185e-06,
      "loss": 0.5646,
      "step": 191
    },
    {
      "epoch": 1.641025641025641,
      "grad_norm": 1.03171495160948,
      "learning_rate": 4.8537894835513e-06,
      "loss": 0.973,
      "step": 192
    },
    {
      "epoch": 1.6495726495726495,
      "grad_norm": 0.8294802712494503,
      "learning_rate": 4.8522777090151505e-06,
      "loss": 0.5523,
      "step": 193
    },
    {
      "epoch": 1.658119658119658,
      "grad_norm": 0.9531056635949867,
      "learning_rate": 4.8507583968634845e-06,
      "loss": 0.6492,
      "step": 194
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 1.0522126505451395,
      "learning_rate": 4.849231551964771e-06,
      "loss": 0.6263,
      "step": 195
    },
    {
      "epoch": 1.6752136752136753,
      "grad_norm": 1.2954172102399524,
      "learning_rate": 4.847697179211618e-06,
      "loss": 0.5302,
      "step": 196
    },
    {
      "epoch": 1.6837606837606838,
      "grad_norm": 1.1470853197940476,
      "learning_rate": 4.8461552835207524e-06,
      "loss": 0.6199,
      "step": 197
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 0.8217311790990888,
      "learning_rate": 4.844605869833011e-06,
      "loss": 0.5702,
      "step": 198
    },
    {
      "epoch": 1.7008547008547008,
      "grad_norm": 1.0740164329626947,
      "learning_rate": 4.84304894311332e-06,
      "loss": 0.8951,
      "step": 199
    },
    {
      "epoch": 1.7094017094017095,
      "grad_norm": 1.0571598154246917,
      "learning_rate": 4.841484508350679e-06,
      "loss": 0.5311,
      "step": 200
    },
    {
      "epoch": 1.717948717948718,
      "grad_norm": 1.0253573731407604,
      "learning_rate": 4.839912570558148e-06,
      "loss": 0.5615,
      "step": 201
    },
    {
      "epoch": 1.7264957264957266,
      "grad_norm": 1.1096555314448195,
      "learning_rate": 4.838333134772828e-06,
      "loss": 0.5405,
      "step": 202
    },
    {
      "epoch": 1.735042735042735,
      "grad_norm": 1.1490180420824463,
      "learning_rate": 4.836746206055849e-06,
      "loss": 0.5965,
      "step": 203
    },
    {
      "epoch": 1.7435897435897436,
      "grad_norm": 1.248513432955595,
      "learning_rate": 4.835151789492348e-06,
      "loss": 0.6728,
      "step": 204
    },
    {
      "epoch": 1.7521367521367521,
      "grad_norm": 1.0730022039252907,
      "learning_rate": 4.83354989019146e-06,
      "loss": 0.5746,
      "step": 205
    },
    {
      "epoch": 1.7606837606837606,
      "grad_norm": 1.1292781253242146,
      "learning_rate": 4.831940513286293e-06,
      "loss": 0.7255,
      "step": 206
    },
    {
      "epoch": 1.7692307692307692,
      "grad_norm": 1.2114916668909483,
      "learning_rate": 4.83032366393392e-06,
      "loss": 0.6832,
      "step": 207
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 1.0038198857460197,
      "learning_rate": 4.828699347315357e-06,
      "loss": 0.763,
      "step": 208
    },
    {
      "epoch": 1.7863247863247862,
      "grad_norm": 1.2044564952881287,
      "learning_rate": 4.827067568635546e-06,
      "loss": 0.5676,
      "step": 209
    },
    {
      "epoch": 1.7948717948717947,
      "grad_norm": 1.1783964857396678,
      "learning_rate": 4.825428333123346e-06,
      "loss": 0.749,
      "step": 210
    },
    {
      "epoch": 1.8034188034188035,
      "grad_norm": 1.2129952692965869,
      "learning_rate": 4.823781646031505e-06,
      "loss": 0.744,
      "step": 211
    },
    {
      "epoch": 1.811965811965812,
      "grad_norm": 1.1997039791188244,
      "learning_rate": 4.822127512636652e-06,
      "loss": 0.4263,
      "step": 212
    },
    {
      "epoch": 1.8205128205128205,
      "grad_norm": 1.2801580897822806,
      "learning_rate": 4.820465938239274e-06,
      "loss": 0.7157,
      "step": 213
    },
    {
      "epoch": 1.8290598290598292,
      "grad_norm": 1.2201502360145773,
      "learning_rate": 4.8187969281637054e-06,
      "loss": 0.6941,
      "step": 214
    },
    {
      "epoch": 1.8376068376068377,
      "grad_norm": 1.0696558627600015,
      "learning_rate": 4.817120487758105e-06,
      "loss": 0.6869,
      "step": 215
    },
    {
      "epoch": 1.8461538461538463,
      "grad_norm": 0.9738340449626065,
      "learning_rate": 4.815436622394442e-06,
      "loss": 0.5687,
      "step": 216
    },
    {
      "epoch": 1.8547008547008548,
      "grad_norm": 1.0884107651937243,
      "learning_rate": 4.813745337468478e-06,
      "loss": 0.8293,
      "step": 217
    },
    {
      "epoch": 1.8632478632478633,
      "grad_norm": 1.0301313414697264,
      "learning_rate": 4.8120466383997486e-06,
      "loss": 0.631,
      "step": 218
    },
    {
      "epoch": 1.8717948717948718,
      "grad_norm": 1.162851200863082,
      "learning_rate": 4.81034053063155e-06,
      "loss": 0.6991,
      "step": 219
    },
    {
      "epoch": 1.8803418803418803,
      "grad_norm": 1.0191227737123674,
      "learning_rate": 4.8086270196309174e-06,
      "loss": 0.869,
      "step": 220
    },
    {
      "epoch": 1.8888888888888888,
      "grad_norm": 1.145852959901522,
      "learning_rate": 4.806906110888606e-06,
      "loss": 0.6711,
      "step": 221
    },
    {
      "epoch": 1.8974358974358974,
      "grad_norm": 1.0405069802093607,
      "learning_rate": 4.805177809919081e-06,
      "loss": 0.7719,
      "step": 222
    },
    {
      "epoch": 1.9059829059829059,
      "grad_norm": 1.187095805002808,
      "learning_rate": 4.803442122260494e-06,
      "loss": 0.7155,
      "step": 223
    },
    {
      "epoch": 1.9145299145299144,
      "grad_norm": 1.1479226660380026,
      "learning_rate": 4.801699053474663e-06,
      "loss": 0.5877,
      "step": 224
    },
    {
      "epoch": 1.9230769230769231,
      "grad_norm": 1.0558651526719405,
      "learning_rate": 4.799948609147061e-06,
      "loss": 0.5684,
      "step": 225
    },
    {
      "epoch": 1.9316239316239316,
      "grad_norm": 1.0038732472086436,
      "learning_rate": 4.798190794886795e-06,
      "loss": 0.7175,
      "step": 226
    },
    {
      "epoch": 1.9401709401709402,
      "grad_norm": 0.9248964570130919,
      "learning_rate": 4.796425616326588e-06,
      "loss": 0.5006,
      "step": 227
    },
    {
      "epoch": 1.9487179487179487,
      "grad_norm": 1.2300999080227526,
      "learning_rate": 4.79465307912276e-06,
      "loss": 0.6256,
      "step": 228
    },
    {
      "epoch": 1.9572649572649574,
      "grad_norm": 1.0417084674820174,
      "learning_rate": 4.792873188955213e-06,
      "loss": 0.8554,
      "step": 229
    },
    {
      "epoch": 1.965811965811966,
      "grad_norm": 1.2349753774243406,
      "learning_rate": 4.791085951527408e-06,
      "loss": 0.5971,
      "step": 230
    },
    {
      "epoch": 1.9743589743589745,
      "grad_norm": 0.9778954080850863,
      "learning_rate": 4.789291372566352e-06,
      "loss": 0.696,
      "step": 231
    },
    {
      "epoch": 1.982905982905983,
      "grad_norm": 1.0172802549418838,
      "learning_rate": 4.787489457822576e-06,
      "loss": 0.6935,
      "step": 232
    },
    {
      "epoch": 1.9914529914529915,
      "grad_norm": 1.1202625109201434,
      "learning_rate": 4.785680213070117e-06,
      "loss": 0.9207,
      "step": 233
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0820721665308195,
      "learning_rate": 4.783863644106502e-06,
      "loss": 0.6833,
      "step": 234
    },
    {
      "epoch": 2.0085470085470085,
      "grad_norm": 1.1358440696728398,
      "learning_rate": 4.782039756752728e-06,
      "loss": 0.618,
      "step": 235
    },
    {
      "epoch": 2.017094017094017,
      "grad_norm": 1.0457159806264371,
      "learning_rate": 4.780208556853239e-06,
      "loss": 0.4722,
      "step": 236
    },
    {
      "epoch": 2.0256410256410255,
      "grad_norm": 1.029809279617813,
      "learning_rate": 4.7783700502759145e-06,
      "loss": 0.656,
      "step": 237
    },
    {
      "epoch": 2.034188034188034,
      "grad_norm": 1.103878378087795,
      "learning_rate": 4.776524242912047e-06,
      "loss": 0.7855,
      "step": 238
    },
    {
      "epoch": 2.0427350427350426,
      "grad_norm": 1.0267266691810852,
      "learning_rate": 4.774671140676325e-06,
      "loss": 0.6245,
      "step": 239
    },
    {
      "epoch": 2.051282051282051,
      "grad_norm": 1.1132991520998483,
      "learning_rate": 4.77281074950681e-06,
      "loss": 0.9077,
      "step": 240
    },
    {
      "epoch": 2.0598290598290596,
      "grad_norm": 1.0667495151776363,
      "learning_rate": 4.77094307536492e-06,
      "loss": 0.5684,
      "step": 241
    },
    {
      "epoch": 2.0683760683760686,
      "grad_norm": 1.0054382245506719,
      "learning_rate": 4.769068124235413e-06,
      "loss": 0.4944,
      "step": 242
    },
    {
      "epoch": 2.076923076923077,
      "grad_norm": 0.9198876553952378,
      "learning_rate": 4.7671859021263635e-06,
      "loss": 0.7061,
      "step": 243
    },
    {
      "epoch": 2.0854700854700856,
      "grad_norm": 1.0023794715887067,
      "learning_rate": 4.765296415069146e-06,
      "loss": 0.6205,
      "step": 244
    },
    {
      "epoch": 2.094017094017094,
      "grad_norm": 1.0373648195951122,
      "learning_rate": 4.763399669118414e-06,
      "loss": 0.794,
      "step": 245
    },
    {
      "epoch": 2.1025641025641026,
      "grad_norm": 0.9929772876721678,
      "learning_rate": 4.761495670352081e-06,
      "loss": 0.5498,
      "step": 246
    },
    {
      "epoch": 2.111111111111111,
      "grad_norm": 1.0991873012161186,
      "learning_rate": 4.759584424871302e-06,
      "loss": 0.6007,
      "step": 247
    },
    {
      "epoch": 2.1196581196581197,
      "grad_norm": 0.9826836883278143,
      "learning_rate": 4.757665938800453e-06,
      "loss": 0.7296,
      "step": 248
    },
    {
      "epoch": 2.128205128205128,
      "grad_norm": 1.0917471368807863,
      "learning_rate": 4.755740218287113e-06,
      "loss": 0.3556,
      "step": 249
    },
    {
      "epoch": 2.1367521367521367,
      "grad_norm": 1.0825910621465624,
      "learning_rate": 4.753807269502041e-06,
      "loss": 0.664,
      "step": 250
    },
    {
      "epoch": 2.1452991452991452,
      "grad_norm": 0.8905203613513742,
      "learning_rate": 4.7518670986391576e-06,
      "loss": 0.4741,
      "step": 251
    },
    {
      "epoch": 2.1538461538461537,
      "grad_norm": 0.9127232740781913,
      "learning_rate": 4.749919711915531e-06,
      "loss": 0.6486,
      "step": 252
    },
    {
      "epoch": 2.1623931623931623,
      "grad_norm": 1.2654836138848309,
      "learning_rate": 4.747965115571345e-06,
      "loss": 0.6307,
      "step": 253
    },
    {
      "epoch": 2.1709401709401708,
      "grad_norm": 1.049141772610685,
      "learning_rate": 4.746003315869889e-06,
      "loss": 0.6238,
      "step": 254
    },
    {
      "epoch": 2.1794871794871793,
      "grad_norm": 1.2805011771546446,
      "learning_rate": 4.744034319097536e-06,
      "loss": 0.6115,
      "step": 255
    },
    {
      "epoch": 2.1880341880341883,
      "grad_norm": 1.0625960491407178,
      "learning_rate": 4.742058131563718e-06,
      "loss": 0.6182,
      "step": 256
    },
    {
      "epoch": 2.1965811965811968,
      "grad_norm": 1.1293658867526906,
      "learning_rate": 4.7400747596009125e-06,
      "loss": 0.5825,
      "step": 257
    },
    {
      "epoch": 2.2051282051282053,
      "grad_norm": 0.9590474269446825,
      "learning_rate": 4.738084209564617e-06,
      "loss": 0.5145,
      "step": 258
    },
    {
      "epoch": 2.213675213675214,
      "grad_norm": 0.8835052747120326,
      "learning_rate": 4.73608648783333e-06,
      "loss": 0.5783,
      "step": 259
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 1.0657600277339365,
      "learning_rate": 4.734081600808531e-06,
      "loss": 0.7102,
      "step": 260
    },
    {
      "epoch": 2.230769230769231,
      "grad_norm": 1.1192032409276582,
      "learning_rate": 4.73206955491466e-06,
      "loss": 0.5007,
      "step": 261
    },
    {
      "epoch": 2.2393162393162394,
      "grad_norm": 1.056716369944498,
      "learning_rate": 4.7300503565990985e-06,
      "loss": 0.5169,
      "step": 262
    },
    {
      "epoch": 2.247863247863248,
      "grad_norm": 0.9269001645142524,
      "learning_rate": 4.728024012332145e-06,
      "loss": 0.4769,
      "step": 263
    },
    {
      "epoch": 2.2564102564102564,
      "grad_norm": 1.0440329703988538,
      "learning_rate": 4.725990528606996e-06,
      "loss": 0.4783,
      "step": 264
    },
    {
      "epoch": 2.264957264957265,
      "grad_norm": 1.0105785765055395,
      "learning_rate": 4.723949911939728e-06,
      "loss": 0.618,
      "step": 265
    },
    {
      "epoch": 2.2735042735042734,
      "grad_norm": 1.1776085377093957,
      "learning_rate": 4.7219021688692725e-06,
      "loss": 0.6379,
      "step": 266
    },
    {
      "epoch": 2.282051282051282,
      "grad_norm": 1.2314287795636818,
      "learning_rate": 4.719847305957398e-06,
      "loss": 0.6684,
      "step": 267
    },
    {
      "epoch": 2.2905982905982905,
      "grad_norm": 1.1004054117724542,
      "learning_rate": 4.717785329788685e-06,
      "loss": 0.7569,
      "step": 268
    },
    {
      "epoch": 2.299145299145299,
      "grad_norm": 1.08262338980394,
      "learning_rate": 4.715716246970511e-06,
      "loss": 0.7849,
      "step": 269
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 0.9736885767195094,
      "learning_rate": 4.7136400641330245e-06,
      "loss": 0.6033,
      "step": 270
    },
    {
      "epoch": 2.316239316239316,
      "grad_norm": 0.9932485252912346,
      "learning_rate": 4.7115567879291265e-06,
      "loss": 0.5402,
      "step": 271
    },
    {
      "epoch": 2.324786324786325,
      "grad_norm": 0.8913116656138574,
      "learning_rate": 4.709466425034445e-06,
      "loss": 0.6844,
      "step": 272
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 1.0793309402123525,
      "learning_rate": 4.707368982147318e-06,
      "loss": 0.6513,
      "step": 273
    },
    {
      "epoch": 2.341880341880342,
      "grad_norm": 0.9562775991017588,
      "learning_rate": 4.705264465988771e-06,
      "loss": 0.6892,
      "step": 274
    },
    {
      "epoch": 2.3504273504273505,
      "grad_norm": 1.1089663383001975,
      "learning_rate": 4.703152883302498e-06,
      "loss": 0.4878,
      "step": 275
    },
    {
      "epoch": 2.358974358974359,
      "grad_norm": 0.9548004271587162,
      "learning_rate": 4.701034240854829e-06,
      "loss": 0.6004,
      "step": 276
    },
    {
      "epoch": 2.3675213675213675,
      "grad_norm": 1.151717398068187,
      "learning_rate": 4.6989085454347236e-06,
      "loss": 0.6388,
      "step": 277
    },
    {
      "epoch": 2.376068376068376,
      "grad_norm": 1.110544374011295,
      "learning_rate": 4.696775803853739e-06,
      "loss": 0.6309,
      "step": 278
    },
    {
      "epoch": 2.3846153846153846,
      "grad_norm": 1.018520072977354,
      "learning_rate": 4.694636022946012e-06,
      "loss": 0.6167,
      "step": 279
    },
    {
      "epoch": 2.393162393162393,
      "grad_norm": 1.0420288996844116,
      "learning_rate": 4.692489209568234e-06,
      "loss": 0.7685,
      "step": 280
    },
    {
      "epoch": 2.4017094017094016,
      "grad_norm": 1.08086419442703,
      "learning_rate": 4.690335370599633e-06,
      "loss": 0.6983,
      "step": 281
    },
    {
      "epoch": 2.41025641025641,
      "grad_norm": 0.8754314274596378,
      "learning_rate": 4.68817451294195e-06,
      "loss": 0.5258,
      "step": 282
    },
    {
      "epoch": 2.4188034188034186,
      "grad_norm": 0.9691136068755867,
      "learning_rate": 4.686006643519415e-06,
      "loss": 0.715,
      "step": 283
    },
    {
      "epoch": 2.427350427350427,
      "grad_norm": 1.0533303155651648,
      "learning_rate": 4.683831769278729e-06,
      "loss": 0.6357,
      "step": 284
    },
    {
      "epoch": 2.435897435897436,
      "grad_norm": 1.0091984594483059,
      "learning_rate": 4.681649897189036e-06,
      "loss": 0.6379,
      "step": 285
    },
    {
      "epoch": 2.4444444444444446,
      "grad_norm": 1.0015146496481933,
      "learning_rate": 4.679461034241906e-06,
      "loss": 0.6303,
      "step": 286
    },
    {
      "epoch": 2.452991452991453,
      "grad_norm": 1.0754076483476527,
      "learning_rate": 4.677265187451311e-06,
      "loss": 0.5038,
      "step": 287
    },
    {
      "epoch": 2.4615384615384617,
      "grad_norm": 0.9176429719414579,
      "learning_rate": 4.675062363853599e-06,
      "loss": 0.6666,
      "step": 288
    },
    {
      "epoch": 2.47008547008547,
      "grad_norm": 0.9478045791249746,
      "learning_rate": 4.672852570507476e-06,
      "loss": 0.4181,
      "step": 289
    },
    {
      "epoch": 2.4786324786324787,
      "grad_norm": 0.9852262013119069,
      "learning_rate": 4.670635814493985e-06,
      "loss": 0.5489,
      "step": 290
    },
    {
      "epoch": 2.4871794871794872,
      "grad_norm": 1.0021736330504551,
      "learning_rate": 4.668412102916474e-06,
      "loss": 0.5666,
      "step": 291
    },
    {
      "epoch": 2.4957264957264957,
      "grad_norm": 1.1968080215087806,
      "learning_rate": 4.666181442900584e-06,
      "loss": 0.7247,
      "step": 292
    },
    {
      "epoch": 2.5042735042735043,
      "grad_norm": 1.40060050962396,
      "learning_rate": 4.663943841594219e-06,
      "loss": 0.876,
      "step": 293
    },
    {
      "epoch": 2.5128205128205128,
      "grad_norm": 1.083324854920149,
      "learning_rate": 4.6616993061675275e-06,
      "loss": 0.5184,
      "step": 294
    },
    {
      "epoch": 2.5213675213675213,
      "grad_norm": 1.028281360150688,
      "learning_rate": 4.659447843812876e-06,
      "loss": 0.6421,
      "step": 295
    },
    {
      "epoch": 2.52991452991453,
      "grad_norm": 1.087837641665171,
      "learning_rate": 4.657189461744829e-06,
      "loss": 0.6551,
      "step": 296
    },
    {
      "epoch": 2.5384615384615383,
      "grad_norm": 1.0872114120511283,
      "learning_rate": 4.654924167200124e-06,
      "loss": 0.6856,
      "step": 297
    },
    {
      "epoch": 2.547008547008547,
      "grad_norm": 1.1629580209566888,
      "learning_rate": 4.652651967437647e-06,
      "loss": 0.5672,
      "step": 298
    },
    {
      "epoch": 2.5555555555555554,
      "grad_norm": 1.0863391277267664,
      "learning_rate": 4.650372869738415e-06,
      "loss": 0.4294,
      "step": 299
    },
    {
      "epoch": 2.564102564102564,
      "grad_norm": 1.1561933663636217,
      "learning_rate": 4.648086881405542e-06,
      "loss": 0.6269,
      "step": 300
    },
    {
      "epoch": 2.5726495726495724,
      "grad_norm": 1.0717857311905097,
      "learning_rate": 4.6457940097642315e-06,
      "loss": 0.4044,
      "step": 301
    },
    {
      "epoch": 2.5811965811965814,
      "grad_norm": 1.1707781764907386,
      "learning_rate": 4.643494262161735e-06,
      "loss": 0.5534,
      "step": 302
    },
    {
      "epoch": 2.58974358974359,
      "grad_norm": 1.1549910291770171,
      "learning_rate": 4.6411876459673435e-06,
      "loss": 0.6494,
      "step": 303
    },
    {
      "epoch": 2.5982905982905984,
      "grad_norm": 1.1094722653709168,
      "learning_rate": 4.638874168572355e-06,
      "loss": 0.64,
      "step": 304
    },
    {
      "epoch": 2.606837606837607,
      "grad_norm": 1.207505213396902,
      "learning_rate": 4.636553837390051e-06,
      "loss": 0.8608,
      "step": 305
    },
    {
      "epoch": 2.6153846153846154,
      "grad_norm": 1.045512641873836,
      "learning_rate": 4.634226659855681e-06,
      "loss": 0.6105,
      "step": 306
    },
    {
      "epoch": 2.623931623931624,
      "grad_norm": 0.8967372988993064,
      "learning_rate": 4.631892643426428e-06,
      "loss": 0.6645,
      "step": 307
    },
    {
      "epoch": 2.6324786324786325,
      "grad_norm": 1.0987505726387903,
      "learning_rate": 4.629551795581393e-06,
      "loss": 0.5404,
      "step": 308
    },
    {
      "epoch": 2.641025641025641,
      "grad_norm": 1.1466793203775922,
      "learning_rate": 4.627204123821563e-06,
      "loss": 0.7323,
      "step": 309
    },
    {
      "epoch": 2.6495726495726495,
      "grad_norm": 1.0888693227497082,
      "learning_rate": 4.624849635669797e-06,
      "loss": 0.744,
      "step": 310
    },
    {
      "epoch": 2.658119658119658,
      "grad_norm": 1.094040236503888,
      "learning_rate": 4.622488338670792e-06,
      "loss": 0.7397,
      "step": 311
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.133946860080225,
      "learning_rate": 4.620120240391065e-06,
      "loss": 0.4986,
      "step": 312
    },
    {
      "epoch": 2.6752136752136755,
      "grad_norm": 1.0023771702905808,
      "learning_rate": 4.617745348418928e-06,
      "loss": 0.8425,
      "step": 313
    },
    {
      "epoch": 2.683760683760684,
      "grad_norm": 1.0556369869801077,
      "learning_rate": 4.61536367036446e-06,
      "loss": 0.5125,
      "step": 314
    },
    {
      "epoch": 2.6923076923076925,
      "grad_norm": 1.1430379187000481,
      "learning_rate": 4.612975213859487e-06,
      "loss": 0.7373,
      "step": 315
    },
    {
      "epoch": 2.700854700854701,
      "grad_norm": 1.0652237090128354,
      "learning_rate": 4.6105799865575565e-06,
      "loss": 0.3925,
      "step": 316
    },
    {
      "epoch": 2.7094017094017095,
      "grad_norm": 0.9895034294385068,
      "learning_rate": 4.60817799613391e-06,
      "loss": 0.5777,
      "step": 317
    },
    {
      "epoch": 2.717948717948718,
      "grad_norm": 0.9336438508199297,
      "learning_rate": 4.605769250285462e-06,
      "loss": 0.6017,
      "step": 318
    },
    {
      "epoch": 2.7264957264957266,
      "grad_norm": 1.0611915631340922,
      "learning_rate": 4.603353756730775e-06,
      "loss": 0.8405,
      "step": 319
    },
    {
      "epoch": 2.735042735042735,
      "grad_norm": 1.0455572665815593,
      "learning_rate": 4.600931523210032e-06,
      "loss": 0.5468,
      "step": 320
    },
    {
      "epoch": 2.7435897435897436,
      "grad_norm": 0.9999670533186025,
      "learning_rate": 4.598502557485015e-06,
      "loss": 0.5728,
      "step": 321
    },
    {
      "epoch": 2.752136752136752,
      "grad_norm": 1.0769830544040775,
      "learning_rate": 4.5960668673390776e-06,
      "loss": 0.7857,
      "step": 322
    },
    {
      "epoch": 2.7606837606837606,
      "grad_norm": 0.934935206627796,
      "learning_rate": 4.59362446057712e-06,
      "loss": 0.6265,
      "step": 323
    },
    {
      "epoch": 2.769230769230769,
      "grad_norm": 1.0960881582165751,
      "learning_rate": 4.591175345025567e-06,
      "loss": 0.4837,
      "step": 324
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 0.9454844727455721,
      "learning_rate": 4.588719528532342e-06,
      "loss": 0.6177,
      "step": 325
    },
    {
      "epoch": 2.786324786324786,
      "grad_norm": 1.1553344447596179,
      "learning_rate": 4.586257018966837e-06,
      "loss": 0.8151,
      "step": 326
    },
    {
      "epoch": 2.7948717948717947,
      "grad_norm": 1.0896014725475482,
      "learning_rate": 4.583787824219894e-06,
      "loss": 0.6745,
      "step": 327
    },
    {
      "epoch": 2.8034188034188032,
      "grad_norm": 1.1258318879696727,
      "learning_rate": 4.5813119522037765e-06,
      "loss": 0.7511,
      "step": 328
    },
    {
      "epoch": 2.8119658119658117,
      "grad_norm": 1.048893301821594,
      "learning_rate": 4.578829410852145e-06,
      "loss": 0.8495,
      "step": 329
    },
    {
      "epoch": 2.8205128205128203,
      "grad_norm": 1.1543372343743354,
      "learning_rate": 4.5763402081200295e-06,
      "loss": 0.52,
      "step": 330
    },
    {
      "epoch": 2.8290598290598292,
      "grad_norm": 0.9769569666395369,
      "learning_rate": 4.573844351983807e-06,
      "loss": 0.5414,
      "step": 331
    },
    {
      "epoch": 2.8376068376068377,
      "grad_norm": 1.1088566573346965,
      "learning_rate": 4.571341850441175e-06,
      "loss": 0.545,
      "step": 332
    },
    {
      "epoch": 2.8461538461538463,
      "grad_norm": 1.144611010547972,
      "learning_rate": 4.568832711511125e-06,
      "loss": 0.6951,
      "step": 333
    },
    {
      "epoch": 2.8547008547008548,
      "grad_norm": 0.8252577047852332,
      "learning_rate": 4.566316943233916e-06,
      "loss": 0.5658,
      "step": 334
    },
    {
      "epoch": 2.8632478632478633,
      "grad_norm": 0.8953345634595089,
      "learning_rate": 4.56379455367105e-06,
      "loss": 0.4984,
      "step": 335
    },
    {
      "epoch": 2.871794871794872,
      "grad_norm": 1.1004516056302105,
      "learning_rate": 4.561265550905251e-06,
      "loss": 0.415,
      "step": 336
    },
    {
      "epoch": 2.8803418803418803,
      "grad_norm": 1.0095678234487127,
      "learning_rate": 4.558729943040427e-06,
      "loss": 0.569,
      "step": 337
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 0.9321134809739067,
      "learning_rate": 4.556187738201656e-06,
      "loss": 0.6132,
      "step": 338
    },
    {
      "epoch": 2.8974358974358974,
      "grad_norm": 1.0238817667487585,
      "learning_rate": 4.553638944535155e-06,
      "loss": 0.5555,
      "step": 339
    },
    {
      "epoch": 2.905982905982906,
      "grad_norm": 1.0428311557389862,
      "learning_rate": 4.551083570208251e-06,
      "loss": 0.6989,
      "step": 340
    },
    {
      "epoch": 2.9145299145299144,
      "grad_norm": 0.9124242004855897,
      "learning_rate": 4.548521623409364e-06,
      "loss": 0.47,
      "step": 341
    },
    {
      "epoch": 2.9230769230769234,
      "grad_norm": 1.2657698218051596,
      "learning_rate": 4.545953112347967e-06,
      "loss": 0.6867,
      "step": 342
    },
    {
      "epoch": 2.931623931623932,
      "grad_norm": 1.1422213876431186,
      "learning_rate": 4.543378045254575e-06,
      "loss": 0.5938,
      "step": 343
    },
    {
      "epoch": 2.9401709401709404,
      "grad_norm": 1.1667272144576992,
      "learning_rate": 4.540796430380706e-06,
      "loss": 0.5566,
      "step": 344
    },
    {
      "epoch": 2.948717948717949,
      "grad_norm": 1.2696464046225642,
      "learning_rate": 4.538208275998861e-06,
      "loss": 0.7144,
      "step": 345
    },
    {
      "epoch": 2.9572649572649574,
      "grad_norm": 1.0653410607116212,
      "learning_rate": 4.535613590402497e-06,
      "loss": 0.618,
      "step": 346
    },
    {
      "epoch": 2.965811965811966,
      "grad_norm": 1.0344003466892562,
      "learning_rate": 4.533012381905999e-06,
      "loss": 0.5135,
      "step": 347
    },
    {
      "epoch": 2.9743589743589745,
      "grad_norm": 1.0045172330544723,
      "learning_rate": 4.530404658844654e-06,
      "loss": 0.4556,
      "step": 348
    },
    {
      "epoch": 2.982905982905983,
      "grad_norm": 1.1125798982264483,
      "learning_rate": 4.527790429574623e-06,
      "loss": 0.714,
      "step": 349
    },
    {
      "epoch": 2.9914529914529915,
      "grad_norm": 1.0281797974354252,
      "learning_rate": 4.525169702472917e-06,
      "loss": 0.3698,
      "step": 350
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9244672248228908,
      "learning_rate": 4.522542485937369e-06,
      "loss": 0.4979,
      "step": 351
    },
    {
      "epoch": 3.0085470085470085,
      "grad_norm": 0.8672830130925646,
      "learning_rate": 4.519908788386605e-06,
      "loss": 0.5368,
      "step": 352
    },
    {
      "epoch": 3.017094017094017,
      "grad_norm": 0.8961300006092746,
      "learning_rate": 4.51726861826002e-06,
      "loss": 0.5831,
      "step": 353
    },
    {
      "epoch": 3.0256410256410255,
      "grad_norm": 1.055331807150709,
      "learning_rate": 4.514621984017748e-06,
      "loss": 0.4283,
      "step": 354
    },
    {
      "epoch": 3.034188034188034,
      "grad_norm": 0.922217785174904,
      "learning_rate": 4.511968894140639e-06,
      "loss": 0.7095,
      "step": 355
    },
    {
      "epoch": 3.0427350427350426,
      "grad_norm": 1.0000694441929876,
      "learning_rate": 4.509309357130227e-06,
      "loss": 0.5385,
      "step": 356
    },
    {
      "epoch": 3.051282051282051,
      "grad_norm": 0.9230262407432018,
      "learning_rate": 4.5066433815087076e-06,
      "loss": 0.4442,
      "step": 357
    },
    {
      "epoch": 3.0598290598290596,
      "grad_norm": 1.0391084822072165,
      "learning_rate": 4.503970975818905e-06,
      "loss": 0.5411,
      "step": 358
    },
    {
      "epoch": 3.0683760683760686,
      "grad_norm": 1.028051333795015,
      "learning_rate": 4.501292148624251e-06,
      "loss": 0.5172,
      "step": 359
    },
    {
      "epoch": 3.076923076923077,
      "grad_norm": 1.050209054930378,
      "learning_rate": 4.498606908508754e-06,
      "loss": 0.5412,
      "step": 360
    },
    {
      "epoch": 3.0854700854700856,
      "grad_norm": 1.0309848435184052,
      "learning_rate": 4.495915264076967e-06,
      "loss": 0.5168,
      "step": 361
    },
    {
      "epoch": 3.094017094017094,
      "grad_norm": 1.018331740357964,
      "learning_rate": 4.493217223953974e-06,
      "loss": 0.5317,
      "step": 362
    },
    {
      "epoch": 3.1025641025641026,
      "grad_norm": 1.0412643296482653,
      "learning_rate": 4.490512796785344e-06,
      "loss": 0.4551,
      "step": 363
    },
    {
      "epoch": 3.111111111111111,
      "grad_norm": 1.0473001504092734,
      "learning_rate": 4.48780199123712e-06,
      "loss": 0.4791,
      "step": 364
    },
    {
      "epoch": 3.1196581196581197,
      "grad_norm": 1.0841505339438793,
      "learning_rate": 4.485084815995778e-06,
      "loss": 0.467,
      "step": 365
    },
    {
      "epoch": 3.128205128205128,
      "grad_norm": 1.1191014630317282,
      "learning_rate": 4.482361279768209e-06,
      "loss": 0.5728,
      "step": 366
    },
    {
      "epoch": 3.1367521367521367,
      "grad_norm": 1.0474104646663822,
      "learning_rate": 4.479631391281685e-06,
      "loss": 0.8182,
      "step": 367
    },
    {
      "epoch": 3.1452991452991452,
      "grad_norm": 1.109664777957958,
      "learning_rate": 4.476895159283835e-06,
      "loss": 0.7313,
      "step": 368
    },
    {
      "epoch": 3.1538461538461537,
      "grad_norm": 0.9922516229814357,
      "learning_rate": 4.474152592542613e-06,
      "loss": 0.5853,
      "step": 369
    },
    {
      "epoch": 3.1623931623931623,
      "grad_norm": 1.0930241384311647,
      "learning_rate": 4.4714036998462715e-06,
      "loss": 0.7107,
      "step": 370
    },
    {
      "epoch": 3.1709401709401708,
      "grad_norm": 1.1835050143124834,
      "learning_rate": 4.4686484900033375e-06,
      "loss": 0.5722,
      "step": 371
    },
    {
      "epoch": 3.1794871794871793,
      "grad_norm": 0.9587380893066114,
      "learning_rate": 4.465886971842578e-06,
      "loss": 0.5348,
      "step": 372
    },
    {
      "epoch": 3.1880341880341883,
      "grad_norm": 1.069484300863465,
      "learning_rate": 4.463119154212972e-06,
      "loss": 0.5701,
      "step": 373
    },
    {
      "epoch": 3.1965811965811968,
      "grad_norm": 0.9644695900573369,
      "learning_rate": 4.46034504598369e-06,
      "loss": 0.5821,
      "step": 374
    },
    {
      "epoch": 3.2051282051282053,
      "grad_norm": 1.2138323456142817,
      "learning_rate": 4.457564656044056e-06,
      "loss": 0.4864,
      "step": 375
    },
    {
      "epoch": 3.213675213675214,
      "grad_norm": 1.0335758299400877,
      "learning_rate": 4.454777993303524e-06,
      "loss": 0.736,
      "step": 376
    },
    {
      "epoch": 3.2222222222222223,
      "grad_norm": 1.215687453793553,
      "learning_rate": 4.451985066691649e-06,
      "loss": 0.4861,
      "step": 377
    },
    {
      "epoch": 3.230769230769231,
      "grad_norm": 0.9760526489106304,
      "learning_rate": 4.449185885158056e-06,
      "loss": 0.473,
      "step": 378
    },
    {
      "epoch": 3.2393162393162394,
      "grad_norm": 1.025138689320582,
      "learning_rate": 4.446380457672417e-06,
      "loss": 0.438,
      "step": 379
    },
    {
      "epoch": 3.247863247863248,
      "grad_norm": 1.0304858209940453,
      "learning_rate": 4.443568793224415e-06,
      "loss": 0.5764,
      "step": 380
    },
    {
      "epoch": 3.2564102564102564,
      "grad_norm": 1.0641902860180872,
      "learning_rate": 4.44075090082372e-06,
      "loss": 0.5768,
      "step": 381
    },
    {
      "epoch": 3.264957264957265,
      "grad_norm": 1.0141681638042424,
      "learning_rate": 4.437926789499959e-06,
      "loss": 0.6206,
      "step": 382
    },
    {
      "epoch": 3.2735042735042734,
      "grad_norm": 0.9719514259271992,
      "learning_rate": 4.435096468302687e-06,
      "loss": 0.3911,
      "step": 383
    },
    {
      "epoch": 3.282051282051282,
      "grad_norm": 1.0928813834901747,
      "learning_rate": 4.432259946301355e-06,
      "loss": 0.5447,
      "step": 384
    },
    {
      "epoch": 3.2905982905982905,
      "grad_norm": 1.1490777184396737,
      "learning_rate": 4.429417232585288e-06,
      "loss": 0.5154,
      "step": 385
    },
    {
      "epoch": 3.299145299145299,
      "grad_norm": 1.1605361713285143,
      "learning_rate": 4.42656833626365e-06,
      "loss": 0.379,
      "step": 386
    },
    {
      "epoch": 3.3076923076923075,
      "grad_norm": 1.1628214583644674,
      "learning_rate": 4.423713266465415e-06,
      "loss": 0.5248,
      "step": 387
    },
    {
      "epoch": 3.316239316239316,
      "grad_norm": 1.1349331779893848,
      "learning_rate": 4.4208520323393425e-06,
      "loss": 0.4598,
      "step": 388
    },
    {
      "epoch": 3.324786324786325,
      "grad_norm": 1.1043916941033478,
      "learning_rate": 4.417984643053941e-06,
      "loss": 0.744,
      "step": 389
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.9604478327433834,
      "learning_rate": 4.415111107797445e-06,
      "loss": 0.5136,
      "step": 390
    },
    {
      "epoch": 3.341880341880342,
      "grad_norm": 1.280358040612093,
      "learning_rate": 4.412231435777784e-06,
      "loss": 0.5602,
      "step": 391
    },
    {
      "epoch": 3.3504273504273505,
      "grad_norm": 1.055925678568642,
      "learning_rate": 4.409345636222549e-06,
      "loss": 0.5936,
      "step": 392
    },
    {
      "epoch": 3.358974358974359,
      "grad_norm": 1.0454713545217449,
      "learning_rate": 4.406453718378968e-06,
      "loss": 0.7998,
      "step": 393
    },
    {
      "epoch": 3.3675213675213675,
      "grad_norm": 1.0350994528202249,
      "learning_rate": 4.4035556915138745e-06,
      "loss": 0.5932,
      "step": 394
    },
    {
      "epoch": 3.376068376068376,
      "grad_norm": 1.0672167525817227,
      "learning_rate": 4.400651564913676e-06,
      "loss": 0.6762,
      "step": 395
    },
    {
      "epoch": 3.3846153846153846,
      "grad_norm": 1.0783182827330902,
      "learning_rate": 4.397741347884329e-06,
      "loss": 0.623,
      "step": 396
    },
    {
      "epoch": 3.393162393162393,
      "grad_norm": 1.1809114761670625,
      "learning_rate": 4.394825049751303e-06,
      "loss": 0.7212,
      "step": 397
    },
    {
      "epoch": 3.4017094017094016,
      "grad_norm": 1.2022044534831018,
      "learning_rate": 4.391902679859557e-06,
      "loss": 0.3709,
      "step": 398
    },
    {
      "epoch": 3.41025641025641,
      "grad_norm": 1.1287039650189177,
      "learning_rate": 4.388974247573501e-06,
      "loss": 0.5523,
      "step": 399
    },
    {
      "epoch": 3.4188034188034186,
      "grad_norm": 1.1605193293531824,
      "learning_rate": 4.386039762276976e-06,
      "loss": 0.556,
      "step": 400
    },
    {
      "epoch": 3.427350427350427,
      "grad_norm": 1.127095779493485,
      "learning_rate": 4.3830992333732185e-06,
      "loss": 0.5659,
      "step": 401
    },
    {
      "epoch": 3.435897435897436,
      "grad_norm": 1.0387250899849618,
      "learning_rate": 4.3801526702848306e-06,
      "loss": 0.632,
      "step": 402
    },
    {
      "epoch": 3.4444444444444446,
      "grad_norm": 0.9521939601614868,
      "learning_rate": 4.377200082453748e-06,
      "loss": 0.4848,
      "step": 403
    },
    {
      "epoch": 3.452991452991453,
      "grad_norm": 1.0609804557798395,
      "learning_rate": 4.374241479341216e-06,
      "loss": 0.4597,
      "step": 404
    },
    {
      "epoch": 3.4615384615384617,
      "grad_norm": 1.122757058652312,
      "learning_rate": 4.3712768704277535e-06,
      "loss": 0.5509,
      "step": 405
    },
    {
      "epoch": 3.47008547008547,
      "grad_norm": 1.2171462597861757,
      "learning_rate": 4.368306265213122e-06,
      "loss": 0.5441,
      "step": 406
    },
    {
      "epoch": 3.4786324786324787,
      "grad_norm": 1.1815243719579964,
      "learning_rate": 4.365329673216301e-06,
      "loss": 0.4705,
      "step": 407
    },
    {
      "epoch": 3.4871794871794872,
      "grad_norm": 1.0455156764851956,
      "learning_rate": 4.3623471039754525e-06,
      "loss": 0.6764,
      "step": 408
    },
    {
      "epoch": 3.4957264957264957,
      "grad_norm": 1.0444224354943048,
      "learning_rate": 4.359358567047892e-06,
      "loss": 0.5449,
      "step": 409
    },
    {
      "epoch": 3.5042735042735043,
      "grad_norm": 0.9079323012608914,
      "learning_rate": 4.356364072010059e-06,
      "loss": 0.6122,
      "step": 410
    },
    {
      "epoch": 3.5128205128205128,
      "grad_norm": 1.1994820113729636,
      "learning_rate": 4.35336362845748e-06,
      "loss": 0.9035,
      "step": 411
    },
    {
      "epoch": 3.5213675213675213,
      "grad_norm": 1.0572483323899298,
      "learning_rate": 4.35035724600475e-06,
      "loss": 0.7419,
      "step": 412
    },
    {
      "epoch": 3.52991452991453,
      "grad_norm": 1.0994551505656394,
      "learning_rate": 4.347344934285492e-06,
      "loss": 0.7047,
      "step": 413
    },
    {
      "epoch": 3.5384615384615383,
      "grad_norm": 1.2445047273471745,
      "learning_rate": 4.3443267029523265e-06,
      "loss": 0.6336,
      "step": 414
    },
    {
      "epoch": 3.547008547008547,
      "grad_norm": 0.9353142553554201,
      "learning_rate": 4.3413025616768426e-06,
      "loss": 0.5763,
      "step": 415
    },
    {
      "epoch": 3.5555555555555554,
      "grad_norm": 1.0815380950652063,
      "learning_rate": 4.338272520149572e-06,
      "loss": 0.5757,
      "step": 416
    },
    {
      "epoch": 3.564102564102564,
      "grad_norm": 1.0211900635168667,
      "learning_rate": 4.335236588079949e-06,
      "loss": 0.5728,
      "step": 417
    },
    {
      "epoch": 3.5726495726495724,
      "grad_norm": 1.0205341832766097,
      "learning_rate": 4.332194775196282e-06,
      "loss": 0.6203,
      "step": 418
    },
    {
      "epoch": 3.5811965811965814,
      "grad_norm": 1.0250376149242053,
      "learning_rate": 4.329147091245729e-06,
      "loss": 0.4802,
      "step": 419
    },
    {
      "epoch": 3.58974358974359,
      "grad_norm": 1.0498223443754728,
      "learning_rate": 4.326093545994258e-06,
      "loss": 0.4724,
      "step": 420
    },
    {
      "epoch": 3.5982905982905984,
      "grad_norm": 1.126184117542039,
      "learning_rate": 4.3230341492266195e-06,
      "loss": 0.5114,
      "step": 421
    },
    {
      "epoch": 3.606837606837607,
      "grad_norm": 1.0434562063600874,
      "learning_rate": 4.3199689107463125e-06,
      "loss": 0.4119,
      "step": 422
    },
    {
      "epoch": 3.6153846153846154,
      "grad_norm": 1.0248635506729542,
      "learning_rate": 4.316897840375558e-06,
      "loss": 0.5205,
      "step": 423
    },
    {
      "epoch": 3.623931623931624,
      "grad_norm": 1.0077448554942794,
      "learning_rate": 4.313820947955265e-06,
      "loss": 0.4752,
      "step": 424
    },
    {
      "epoch": 3.6324786324786325,
      "grad_norm": 1.031230850742542,
      "learning_rate": 4.310738243344996e-06,
      "loss": 0.4966,
      "step": 425
    },
    {
      "epoch": 3.641025641025641,
      "grad_norm": 1.1227610381702842,
      "learning_rate": 4.307649736422939e-06,
      "loss": 0.3805,
      "step": 426
    },
    {
      "epoch": 3.6495726495726495,
      "grad_norm": 1.1094301736827121,
      "learning_rate": 4.304555437085876e-06,
      "loss": 0.5154,
      "step": 427
    },
    {
      "epoch": 3.658119658119658,
      "grad_norm": 1.1641744131192364,
      "learning_rate": 4.301455355249148e-06,
      "loss": 0.6265,
      "step": 428
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 1.3182527561357478,
      "learning_rate": 4.2983495008466285e-06,
      "loss": 0.5251,
      "step": 429
    },
    {
      "epoch": 3.6752136752136755,
      "grad_norm": 1.057755716798989,
      "learning_rate": 4.2952378838306855e-06,
      "loss": 0.549,
      "step": 430
    },
    {
      "epoch": 3.683760683760684,
      "grad_norm": 1.1167572928661555,
      "learning_rate": 4.292120514172154e-06,
      "loss": 0.5538,
      "step": 431
    },
    {
      "epoch": 3.6923076923076925,
      "grad_norm": 1.0189034773737762,
      "learning_rate": 4.288997401860303e-06,
      "loss": 0.4497,
      "step": 432
    },
    {
      "epoch": 3.700854700854701,
      "grad_norm": 1.1052813795847365,
      "learning_rate": 4.285868556902803e-06,
      "loss": 0.5662,
      "step": 433
    },
    {
      "epoch": 3.7094017094017095,
      "grad_norm": 1.1983962217285142,
      "learning_rate": 4.2827339893256935e-06,
      "loss": 0.5738,
      "step": 434
    },
    {
      "epoch": 3.717948717948718,
      "grad_norm": 1.217707027286218,
      "learning_rate": 4.279593709173352e-06,
      "loss": 0.583,
      "step": 435
    },
    {
      "epoch": 3.7264957264957266,
      "grad_norm": 1.0467283527547593,
      "learning_rate": 4.276447726508461e-06,
      "loss": 0.6682,
      "step": 436
    },
    {
      "epoch": 3.735042735042735,
      "grad_norm": 1.0574228677466284,
      "learning_rate": 4.273296051411978e-06,
      "loss": 0.6307,
      "step": 437
    },
    {
      "epoch": 3.7435897435897436,
      "grad_norm": 1.0940623257733564,
      "learning_rate": 4.2701386939830966e-06,
      "loss": 0.7574,
      "step": 438
    },
    {
      "epoch": 3.752136752136752,
      "grad_norm": 1.2037642109054294,
      "learning_rate": 4.2669756643392255e-06,
      "loss": 0.744,
      "step": 439
    },
    {
      "epoch": 3.7606837606837606,
      "grad_norm": 1.0771890780466298,
      "learning_rate": 4.263806972615943e-06,
      "loss": 0.3926,
      "step": 440
    },
    {
      "epoch": 3.769230769230769,
      "grad_norm": 1.1140250234730074,
      "learning_rate": 4.260632628966974e-06,
      "loss": 0.5387,
      "step": 441
    },
    {
      "epoch": 3.7777777777777777,
      "grad_norm": 1.0216104979859169,
      "learning_rate": 4.257452643564155e-06,
      "loss": 0.4784,
      "step": 442
    },
    {
      "epoch": 3.786324786324786,
      "grad_norm": 1.031163827042152,
      "learning_rate": 4.254267026597399e-06,
      "loss": 0.6873,
      "step": 443
    },
    {
      "epoch": 3.7948717948717947,
      "grad_norm": 1.182047554124205,
      "learning_rate": 4.251075788274667e-06,
      "loss": 0.306,
      "step": 444
    },
    {
      "epoch": 3.8034188034188032,
      "grad_norm": 1.1464751159617779,
      "learning_rate": 4.247878938821929e-06,
      "loss": 0.5585,
      "step": 445
    },
    {
      "epoch": 3.8119658119658117,
      "grad_norm": 1.0901425013917103,
      "learning_rate": 4.2446764884831404e-06,
      "loss": 0.6373,
      "step": 446
    },
    {
      "epoch": 3.8205128205128203,
      "grad_norm": 1.1012303502937792,
      "learning_rate": 4.2414684475202014e-06,
      "loss": 0.554,
      "step": 447
    },
    {
      "epoch": 3.8290598290598292,
      "grad_norm": 1.141830081568288,
      "learning_rate": 4.238254826212925e-06,
      "loss": 0.6192,
      "step": 448
    },
    {
      "epoch": 3.8376068376068377,
      "grad_norm": 1.11435215769542,
      "learning_rate": 4.2350356348590096e-06,
      "loss": 0.5138,
      "step": 449
    },
    {
      "epoch": 3.8461538461538463,
      "grad_norm": 1.0466376945625686,
      "learning_rate": 4.231810883773999e-06,
      "loss": 0.6832,
      "step": 450
    },
    {
      "epoch": 3.8547008547008548,
      "grad_norm": 1.1772526601190012,
      "learning_rate": 4.228580583291254e-06,
      "loss": 0.5216,
      "step": 451
    },
    {
      "epoch": 3.8632478632478633,
      "grad_norm": 1.0414249916352694,
      "learning_rate": 4.225344743761918e-06,
      "loss": 0.5568,
      "step": 452
    },
    {
      "epoch": 3.871794871794872,
      "grad_norm": 1.135999207400427,
      "learning_rate": 4.2221033755548835e-06,
      "loss": 0.4843,
      "step": 453
    },
    {
      "epoch": 3.8803418803418803,
      "grad_norm": 1.1714177527617442,
      "learning_rate": 4.218856489056758e-06,
      "loss": 0.3179,
      "step": 454
    },
    {
      "epoch": 3.888888888888889,
      "grad_norm": 0.9729054252414072,
      "learning_rate": 4.215604094671835e-06,
      "loss": 0.6261,
      "step": 455
    },
    {
      "epoch": 3.8974358974358974,
      "grad_norm": 1.066184341545607,
      "learning_rate": 4.2123462028220505e-06,
      "loss": 0.6724,
      "step": 456
    },
    {
      "epoch": 3.905982905982906,
      "grad_norm": 1.230748130789342,
      "learning_rate": 4.209082823946965e-06,
      "loss": 0.8234,
      "step": 457
    },
    {
      "epoch": 3.9145299145299144,
      "grad_norm": 1.1988530892117408,
      "learning_rate": 4.205813968503717e-06,
      "loss": 0.6273,
      "step": 458
    },
    {
      "epoch": 3.9230769230769234,
      "grad_norm": 1.0786203487111534,
      "learning_rate": 4.202539646966993e-06,
      "loss": 0.5313,
      "step": 459
    },
    {
      "epoch": 3.931623931623932,
      "grad_norm": 0.9160073002850568,
      "learning_rate": 4.1992598698289985e-06,
      "loss": 0.3638,
      "step": 460
    },
    {
      "epoch": 3.9401709401709404,
      "grad_norm": 1.1479053266981039,
      "learning_rate": 4.1959746475994175e-06,
      "loss": 0.6149,
      "step": 461
    },
    {
      "epoch": 3.948717948717949,
      "grad_norm": 0.9316297041443543,
      "learning_rate": 4.1926839908053855e-06,
      "loss": 0.5541,
      "step": 462
    },
    {
      "epoch": 3.9572649572649574,
      "grad_norm": 1.0830846381257961,
      "learning_rate": 4.189387909991448e-06,
      "loss": 0.5234,
      "step": 463
    },
    {
      "epoch": 3.965811965811966,
      "grad_norm": 1.1678471598237286,
      "learning_rate": 4.186086415719537e-06,
      "loss": 0.5652,
      "step": 464
    },
    {
      "epoch": 3.9743589743589745,
      "grad_norm": 0.9099147923682155,
      "learning_rate": 4.182779518568925e-06,
      "loss": 0.5946,
      "step": 465
    },
    {
      "epoch": 3.982905982905983,
      "grad_norm": 1.1141807669712536,
      "learning_rate": 4.179467229136205e-06,
      "loss": 0.3703,
      "step": 466
    },
    {
      "epoch": 3.9914529914529915,
      "grad_norm": 1.4764971597166958,
      "learning_rate": 4.176149558035241e-06,
      "loss": 0.6054,
      "step": 467
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.22721031943352,
      "learning_rate": 4.172826515897146e-06,
      "loss": 0.4847,
      "step": 468
    },
    {
      "epoch": 4.0085470085470085,
      "grad_norm": 0.875230516002503,
      "learning_rate": 4.169498113370245e-06,
      "loss": 0.6291,
      "step": 469
    },
    {
      "epoch": 4.017094017094017,
      "grad_norm": 1.272482921335239,
      "learning_rate": 4.166164361120036e-06,
      "loss": 0.3956,
      "step": 470
    },
    {
      "epoch": 4.0256410256410255,
      "grad_norm": 0.9556535023235093,
      "learning_rate": 4.162825269829165e-06,
      "loss": 0.5645,
      "step": 471
    },
    {
      "epoch": 4.034188034188034,
      "grad_norm": 0.9656405185189749,
      "learning_rate": 4.15948085019738e-06,
      "loss": 0.3914,
      "step": 472
    },
    {
      "epoch": 4.042735042735043,
      "grad_norm": 1.2321645936844894,
      "learning_rate": 4.156131112941509e-06,
      "loss": 0.588,
      "step": 473
    },
    {
      "epoch": 4.051282051282051,
      "grad_norm": 1.367549160162955,
      "learning_rate": 4.152776068795416e-06,
      "loss": 0.6519,
      "step": 474
    },
    {
      "epoch": 4.05982905982906,
      "grad_norm": 0.9481306468484307,
      "learning_rate": 4.149415728509971e-06,
      "loss": 0.5701,
      "step": 475
    },
    {
      "epoch": 4.068376068376068,
      "grad_norm": 1.1701525380439926,
      "learning_rate": 4.146050102853015e-06,
      "loss": 0.4844,
      "step": 476
    },
    {
      "epoch": 4.076923076923077,
      "grad_norm": 1.1247319279211934,
      "learning_rate": 4.1426792026093274e-06,
      "loss": 0.7878,
      "step": 477
    },
    {
      "epoch": 4.085470085470085,
      "grad_norm": 1.1237398995766223,
      "learning_rate": 4.139303038580586e-06,
      "loss": 0.4843,
      "step": 478
    },
    {
      "epoch": 4.094017094017094,
      "grad_norm": 1.0625870199426208,
      "learning_rate": 4.135921621585338e-06,
      "loss": 0.3677,
      "step": 479
    },
    {
      "epoch": 4.102564102564102,
      "grad_norm": 1.14727227991621,
      "learning_rate": 4.1325349624589625e-06,
      "loss": 0.6442,
      "step": 480
    },
    {
      "epoch": 4.111111111111111,
      "grad_norm": 1.0477961628751995,
      "learning_rate": 4.129143072053639e-06,
      "loss": 0.5369,
      "step": 481
    },
    {
      "epoch": 4.119658119658119,
      "grad_norm": 1.087010322607983,
      "learning_rate": 4.125745961238305e-06,
      "loss": 0.392,
      "step": 482
    },
    {
      "epoch": 4.128205128205128,
      "grad_norm": 1.2603627244731999,
      "learning_rate": 4.122343640898628e-06,
      "loss": 0.5996,
      "step": 483
    },
    {
      "epoch": 4.136752136752137,
      "grad_norm": 1.1308786469388739,
      "learning_rate": 4.118936121936973e-06,
      "loss": 0.6201,
      "step": 484
    },
    {
      "epoch": 4.145299145299146,
      "grad_norm": 0.9734248346069477,
      "learning_rate": 4.115523415272358e-06,
      "loss": 0.4135,
      "step": 485
    },
    {
      "epoch": 4.153846153846154,
      "grad_norm": 1.2509401315135145,
      "learning_rate": 4.112105531840427e-06,
      "loss": 0.7628,
      "step": 486
    },
    {
      "epoch": 4.162393162393163,
      "grad_norm": 1.2324367647560945,
      "learning_rate": 4.1086824825934126e-06,
      "loss": 0.4189,
      "step": 487
    },
    {
      "epoch": 4.170940170940171,
      "grad_norm": 1.5871825728125233,
      "learning_rate": 4.1052542785001e-06,
      "loss": 0.4659,
      "step": 488
    },
    {
      "epoch": 4.17948717948718,
      "grad_norm": 1.2427938238759526,
      "learning_rate": 4.101820930545792e-06,
      "loss": 0.5776,
      "step": 489
    },
    {
      "epoch": 4.188034188034188,
      "grad_norm": 1.1040081714838972,
      "learning_rate": 4.098382449732276e-06,
      "loss": 0.5746,
      "step": 490
    },
    {
      "epoch": 4.196581196581197,
      "grad_norm": 1.399147839889112,
      "learning_rate": 4.094938847077784e-06,
      "loss": 0.7351,
      "step": 491
    },
    {
      "epoch": 4.205128205128205,
      "grad_norm": 1.2039881642302797,
      "learning_rate": 4.091490133616965e-06,
      "loss": 0.3633,
      "step": 492
    },
    {
      "epoch": 4.213675213675214,
      "grad_norm": 1.363284686280309,
      "learning_rate": 4.08803632040084e-06,
      "loss": 0.4315,
      "step": 493
    },
    {
      "epoch": 4.222222222222222,
      "grad_norm": 1.0603492653742104,
      "learning_rate": 4.084577418496775e-06,
      "loss": 0.4569,
      "step": 494
    },
    {
      "epoch": 4.230769230769231,
      "grad_norm": 1.103163024491336,
      "learning_rate": 4.081113438988443e-06,
      "loss": 0.4104,
      "step": 495
    },
    {
      "epoch": 4.239316239316239,
      "grad_norm": 1.0528872661282709,
      "learning_rate": 4.077644392975785e-06,
      "loss": 0.3354,
      "step": 496
    },
    {
      "epoch": 4.247863247863248,
      "grad_norm": 1.2879686313297776,
      "learning_rate": 4.074170291574975e-06,
      "loss": 0.5563,
      "step": 497
    },
    {
      "epoch": 4.256410256410256,
      "grad_norm": 1.045397447019931,
      "learning_rate": 4.0706911459183915e-06,
      "loss": 0.5623,
      "step": 498
    },
    {
      "epoch": 4.264957264957265,
      "grad_norm": 0.9731163261614564,
      "learning_rate": 4.067206967154575e-06,
      "loss": 0.4018,
      "step": 499
    },
    {
      "epoch": 4.273504273504273,
      "grad_norm": 1.0969564649833976,
      "learning_rate": 4.063717766448194e-06,
      "loss": 0.4298,
      "step": 500
    },
    {
      "epoch": 4.282051282051282,
      "grad_norm": 1.0639228552178774,
      "learning_rate": 4.060223554980007e-06,
      "loss": 0.5563,
      "step": 501
    },
    {
      "epoch": 4.2905982905982905,
      "grad_norm": 1.2587006164595462,
      "learning_rate": 4.056724343946832e-06,
      "loss": 0.3347,
      "step": 502
    },
    {
      "epoch": 4.299145299145299,
      "grad_norm": 1.076742709469613,
      "learning_rate": 4.053220144561506e-06,
      "loss": 0.3666,
      "step": 503
    },
    {
      "epoch": 4.3076923076923075,
      "grad_norm": 1.1869023630845126,
      "learning_rate": 4.049710968052851e-06,
      "loss": 0.5493,
      "step": 504
    },
    {
      "epoch": 4.316239316239316,
      "grad_norm": 1.2762763414877694,
      "learning_rate": 4.046196825665638e-06,
      "loss": 0.4113,
      "step": 505
    },
    {
      "epoch": 4.3247863247863245,
      "grad_norm": 1.256286291317964,
      "learning_rate": 4.042677728660549e-06,
      "loss": 0.3814,
      "step": 506
    },
    {
      "epoch": 4.333333333333333,
      "grad_norm": 1.2060736250753987,
      "learning_rate": 4.039153688314146e-06,
      "loss": 0.5914,
      "step": 507
    },
    {
      "epoch": 4.3418803418803416,
      "grad_norm": 1.1176723905627919,
      "learning_rate": 4.035624715918827e-06,
      "loss": 0.6334,
      "step": 508
    },
    {
      "epoch": 4.35042735042735,
      "grad_norm": 1.5033666933742025,
      "learning_rate": 4.032090822782798e-06,
      "loss": 0.4644,
      "step": 509
    },
    {
      "epoch": 4.358974358974359,
      "grad_norm": 1.1687610354628746,
      "learning_rate": 4.028552020230031e-06,
      "loss": 0.4106,
      "step": 510
    },
    {
      "epoch": 4.367521367521368,
      "grad_norm": 1.2015668421384156,
      "learning_rate": 4.0250083196002285e-06,
      "loss": 0.5407,
      "step": 511
    },
    {
      "epoch": 4.3760683760683765,
      "grad_norm": 1.1147036921086226,
      "learning_rate": 4.021459732248792e-06,
      "loss": 0.358,
      "step": 512
    },
    {
      "epoch": 4.384615384615385,
      "grad_norm": 1.5367915145923337,
      "learning_rate": 4.017906269546778e-06,
      "loss": 0.4539,
      "step": 513
    },
    {
      "epoch": 4.3931623931623935,
      "grad_norm": 1.304473709441864,
      "learning_rate": 4.014347942880869e-06,
      "loss": 0.4259,
      "step": 514
    },
    {
      "epoch": 4.401709401709402,
      "grad_norm": 1.1833828587337736,
      "learning_rate": 4.0107847636533314e-06,
      "loss": 0.4551,
      "step": 515
    },
    {
      "epoch": 4.410256410256411,
      "grad_norm": 0.9701845087775417,
      "learning_rate": 4.0072167432819804e-06,
      "loss": 0.3402,
      "step": 516
    },
    {
      "epoch": 4.418803418803419,
      "grad_norm": 1.3429029717262404,
      "learning_rate": 4.003643893200148e-06,
      "loss": 0.5363,
      "step": 517
    },
    {
      "epoch": 4.427350427350428,
      "grad_norm": 1.4813898446999763,
      "learning_rate": 4.000066224856636e-06,
      "loss": 0.6563,
      "step": 518
    },
    {
      "epoch": 4.435897435897436,
      "grad_norm": 1.0620532686129334,
      "learning_rate": 3.996483749715694e-06,
      "loss": 0.5307,
      "step": 519
    },
    {
      "epoch": 4.444444444444445,
      "grad_norm": 1.1674492551898705,
      "learning_rate": 3.992896479256966e-06,
      "loss": 0.5044,
      "step": 520
    },
    {
      "epoch": 4.452991452991453,
      "grad_norm": 1.089978982595774,
      "learning_rate": 3.989304424975468e-06,
      "loss": 0.4327,
      "step": 521
    },
    {
      "epoch": 4.461538461538462,
      "grad_norm": 1.0065363497501418,
      "learning_rate": 3.985707598381544e-06,
      "loss": 0.55,
      "step": 522
    },
    {
      "epoch": 4.47008547008547,
      "grad_norm": 1.0697673878641412,
      "learning_rate": 3.9821060110008295e-06,
      "loss": 0.5825,
      "step": 523
    },
    {
      "epoch": 4.478632478632479,
      "grad_norm": 1.1226330146224837,
      "learning_rate": 3.978499674374214e-06,
      "loss": 0.5514,
      "step": 524
    },
    {
      "epoch": 4.487179487179487,
      "grad_norm": 1.178704468790716,
      "learning_rate": 3.974888600057808e-06,
      "loss": 0.4042,
      "step": 525
    },
    {
      "epoch": 4.495726495726496,
      "grad_norm": 1.1844014873988862,
      "learning_rate": 3.971272799622903e-06,
      "loss": 0.6905,
      "step": 526
    },
    {
      "epoch": 4.504273504273504,
      "grad_norm": 1.3063239169598142,
      "learning_rate": 3.967652284655933e-06,
      "loss": 0.464,
      "step": 527
    },
    {
      "epoch": 4.512820512820513,
      "grad_norm": 1.3805070176396315,
      "learning_rate": 3.964027066758442e-06,
      "loss": 0.3515,
      "step": 528
    },
    {
      "epoch": 4.521367521367521,
      "grad_norm": 1.0907984273825841,
      "learning_rate": 3.960397157547043e-06,
      "loss": 0.528,
      "step": 529
    },
    {
      "epoch": 4.52991452991453,
      "grad_norm": 1.2859054101894118,
      "learning_rate": 3.956762568653378e-06,
      "loss": 0.598,
      "step": 530
    },
    {
      "epoch": 4.538461538461538,
      "grad_norm": 1.2422004006544314,
      "learning_rate": 3.953123311724092e-06,
      "loss": 0.6026,
      "step": 531
    },
    {
      "epoch": 4.547008547008547,
      "grad_norm": 1.2436712062281594,
      "learning_rate": 3.9494793984207815e-06,
      "loss": 0.358,
      "step": 532
    },
    {
      "epoch": 4.555555555555555,
      "grad_norm": 1.117015372814124,
      "learning_rate": 3.945830840419966e-06,
      "loss": 0.5461,
      "step": 533
    },
    {
      "epoch": 4.564102564102564,
      "grad_norm": 1.0053532044910118,
      "learning_rate": 3.942177649413051e-06,
      "loss": 0.5622,
      "step": 534
    },
    {
      "epoch": 4.572649572649572,
      "grad_norm": 1.2133323539666498,
      "learning_rate": 3.938519837106284e-06,
      "loss": 0.3818,
      "step": 535
    },
    {
      "epoch": 4.581196581196581,
      "grad_norm": 1.3183491629414597,
      "learning_rate": 3.9348574152207245e-06,
      "loss": 0.42,
      "step": 536
    },
    {
      "epoch": 4.589743589743589,
      "grad_norm": 1.0855334417363243,
      "learning_rate": 3.931190395492198e-06,
      "loss": 0.4367,
      "step": 537
    },
    {
      "epoch": 4.598290598290598,
      "grad_norm": 1.1830080951781239,
      "learning_rate": 3.92751878967127e-06,
      "loss": 0.5654,
      "step": 538
    },
    {
      "epoch": 4.6068376068376065,
      "grad_norm": 1.0025162147085154,
      "learning_rate": 3.923842609523195e-06,
      "loss": 0.5085,
      "step": 539
    },
    {
      "epoch": 4.615384615384615,
      "grad_norm": 1.1938032588726277,
      "learning_rate": 3.92016186682789e-06,
      "loss": 0.3901,
      "step": 540
    },
    {
      "epoch": 4.6239316239316235,
      "grad_norm": 1.044455553238553,
      "learning_rate": 3.91647657337989e-06,
      "loss": 0.5414,
      "step": 541
    },
    {
      "epoch": 4.632478632478632,
      "grad_norm": 1.0604389385547883,
      "learning_rate": 3.9127867409883145e-06,
      "loss": 0.5697,
      "step": 542
    },
    {
      "epoch": 4.641025641025641,
      "grad_norm": 1.030741191950751,
      "learning_rate": 3.909092381476824e-06,
      "loss": 0.4111,
      "step": 543
    },
    {
      "epoch": 4.64957264957265,
      "grad_norm": 1.2599857607075067,
      "learning_rate": 3.905393506683589e-06,
      "loss": 0.7456,
      "step": 544
    },
    {
      "epoch": 4.6581196581196584,
      "grad_norm": 1.1979763813132545,
      "learning_rate": 3.901690128461248e-06,
      "loss": 0.4687,
      "step": 545
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 1.4424908787398834,
      "learning_rate": 3.897982258676867e-06,
      "loss": 0.6179,
      "step": 546
    },
    {
      "epoch": 4.6752136752136755,
      "grad_norm": 0.9608243421216695,
      "learning_rate": 3.894269909211911e-06,
      "loss": 0.5166,
      "step": 547
    },
    {
      "epoch": 4.683760683760684,
      "grad_norm": 1.2658167795968152,
      "learning_rate": 3.890553091962193e-06,
      "loss": 0.4927,
      "step": 548
    },
    {
      "epoch": 4.6923076923076925,
      "grad_norm": 1.4297839299427377,
      "learning_rate": 3.8868318188378475e-06,
      "loss": 0.3448,
      "step": 549
    },
    {
      "epoch": 4.700854700854701,
      "grad_norm": 1.1120064739717215,
      "learning_rate": 3.883106101763285e-06,
      "loss": 0.4685,
      "step": 550
    },
    {
      "epoch": 4.7094017094017095,
      "grad_norm": 1.093805953026953,
      "learning_rate": 3.879375952677156e-06,
      "loss": 0.4474,
      "step": 551
    },
    {
      "epoch": 4.717948717948718,
      "grad_norm": 1.0474731400789823,
      "learning_rate": 3.875641383532313e-06,
      "loss": 0.5464,
      "step": 552
    },
    {
      "epoch": 4.726495726495727,
      "grad_norm": 1.1262315965779182,
      "learning_rate": 3.871902406295775e-06,
      "loss": 0.4222,
      "step": 553
    },
    {
      "epoch": 4.735042735042735,
      "grad_norm": 0.964350644928307,
      "learning_rate": 3.868159032948681e-06,
      "loss": 0.4808,
      "step": 554
    },
    {
      "epoch": 4.743589743589744,
      "grad_norm": 1.0561833255928614,
      "learning_rate": 3.8644112754862614e-06,
      "loss": 0.6094,
      "step": 555
    },
    {
      "epoch": 4.752136752136752,
      "grad_norm": 1.0733675088449695,
      "learning_rate": 3.860659145917794e-06,
      "loss": 0.5114,
      "step": 556
    },
    {
      "epoch": 4.760683760683761,
      "grad_norm": 1.0653535243140473,
      "learning_rate": 3.856902656266563e-06,
      "loss": 0.4611,
      "step": 557
    },
    {
      "epoch": 4.769230769230769,
      "grad_norm": 1.1566280687891097,
      "learning_rate": 3.853141818569829e-06,
      "loss": 0.5141,
      "step": 558
    },
    {
      "epoch": 4.777777777777778,
      "grad_norm": 1.2272432059639735,
      "learning_rate": 3.849376644878783e-06,
      "loss": 0.4836,
      "step": 559
    },
    {
      "epoch": 4.786324786324786,
      "grad_norm": 1.0866113784148257,
      "learning_rate": 3.84560714725851e-06,
      "loss": 0.5022,
      "step": 560
    },
    {
      "epoch": 4.794871794871795,
      "grad_norm": 0.9904138308671785,
      "learning_rate": 3.841833337787951e-06,
      "loss": 0.5167,
      "step": 561
    },
    {
      "epoch": 4.803418803418803,
      "grad_norm": 1.2968094227448181,
      "learning_rate": 3.838055228559864e-06,
      "loss": 0.4508,
      "step": 562
    },
    {
      "epoch": 4.811965811965812,
      "grad_norm": 1.214225199676867,
      "learning_rate": 3.834272831680785e-06,
      "loss": 0.575,
      "step": 563
    },
    {
      "epoch": 4.82051282051282,
      "grad_norm": 1.2472021018422048,
      "learning_rate": 3.830486159270991e-06,
      "loss": 0.6268,
      "step": 564
    },
    {
      "epoch": 4.829059829059829,
      "grad_norm": 0.9172374646659589,
      "learning_rate": 3.826695223464455e-06,
      "loss": 0.6032,
      "step": 565
    },
    {
      "epoch": 4.837606837606837,
      "grad_norm": 1.0304247295278326,
      "learning_rate": 3.822900036408815e-06,
      "loss": 0.6203,
      "step": 566
    },
    {
      "epoch": 4.846153846153846,
      "grad_norm": 1.0666290948925694,
      "learning_rate": 3.819100610265332e-06,
      "loss": 0.4967,
      "step": 567
    },
    {
      "epoch": 4.854700854700854,
      "grad_norm": 1.2495026717724254,
      "learning_rate": 3.815296957208849e-06,
      "loss": 0.4664,
      "step": 568
    },
    {
      "epoch": 4.863247863247864,
      "grad_norm": 1.1007215063135107,
      "learning_rate": 3.811489089427756e-06,
      "loss": 0.51,
      "step": 569
    },
    {
      "epoch": 4.871794871794872,
      "grad_norm": 1.2148804714926256,
      "learning_rate": 3.8076770191239444e-06,
      "loss": 0.8572,
      "step": 570
    },
    {
      "epoch": 4.880341880341881,
      "grad_norm": 1.0075512142961376,
      "learning_rate": 3.8038607585127762e-06,
      "loss": 0.7746,
      "step": 571
    },
    {
      "epoch": 4.888888888888889,
      "grad_norm": 1.1675958255183994,
      "learning_rate": 3.8000403198230385e-06,
      "loss": 0.4149,
      "step": 572
    },
    {
      "epoch": 4.897435897435898,
      "grad_norm": 2.3479618762202166,
      "learning_rate": 3.7962157152969093e-06,
      "loss": 0.3843,
      "step": 573
    },
    {
      "epoch": 4.905982905982906,
      "grad_norm": 1.339128802411911,
      "learning_rate": 3.7923869571899115e-06,
      "loss": 0.4443,
      "step": 574
    },
    {
      "epoch": 4.914529914529915,
      "grad_norm": 1.2598653276333198,
      "learning_rate": 3.7885540577708806e-06,
      "loss": 0.599,
      "step": 575
    },
    {
      "epoch": 4.923076923076923,
      "grad_norm": 1.273835748560621,
      "learning_rate": 3.7847170293219223e-06,
      "loss": 0.5045,
      "step": 576
    },
    {
      "epoch": 4.931623931623932,
      "grad_norm": 1.1962056104312029,
      "learning_rate": 3.780875884138372e-06,
      "loss": 0.3395,
      "step": 577
    },
    {
      "epoch": 4.94017094017094,
      "grad_norm": 0.9688429291410264,
      "learning_rate": 3.7770306345287577e-06,
      "loss": 0.5456,
      "step": 578
    },
    {
      "epoch": 4.948717948717949,
      "grad_norm": 1.0556261135022196,
      "learning_rate": 3.7731812928147593e-06,
      "loss": 0.4656,
      "step": 579
    },
    {
      "epoch": 4.957264957264957,
      "grad_norm": 1.172036901000504,
      "learning_rate": 3.76932787133117e-06,
      "loss": 0.4111,
      "step": 580
    },
    {
      "epoch": 4.965811965811966,
      "grad_norm": 1.1936432934962213,
      "learning_rate": 3.7654703824258544e-06,
      "loss": 0.511,
      "step": 581
    },
    {
      "epoch": 4.9743589743589745,
      "grad_norm": 1.0927271152964717,
      "learning_rate": 3.7616088384597138e-06,
      "loss": 0.7687,
      "step": 582
    },
    {
      "epoch": 4.982905982905983,
      "grad_norm": 1.0498423497144054,
      "learning_rate": 3.757743251806639e-06,
      "loss": 0.522,
      "step": 583
    },
    {
      "epoch": 4.9914529914529915,
      "grad_norm": 1.0465049477381851,
      "learning_rate": 3.753873634853481e-06,
      "loss": 0.5185,
      "step": 584
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.1433429860697957,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.4767,
      "step": 585
    },
    {
      "epoch": 5.0085470085470085,
      "grad_norm": 1.1929331950145705,
      "learning_rate": 3.746122359658834e-06,
      "loss": 0.4947,
      "step": 586
    },
    {
      "epoch": 5.017094017094017,
      "grad_norm": 1.053420754339069,
      "learning_rate": 3.7422407262554567e-06,
      "loss": 0.7168,
      "step": 587
    },
    {
      "epoch": 5.0256410256410255,
      "grad_norm": 1.0144269826464443,
      "learning_rate": 3.738355112228134e-06,
      "loss": 0.3795,
      "step": 588
    },
    {
      "epoch": 5.034188034188034,
      "grad_norm": 1.1279663595509448,
      "learning_rate": 3.7344655300278887e-06,
      "loss": 0.6969,
      "step": 589
    },
    {
      "epoch": 5.042735042735043,
      "grad_norm": 1.2332439824894945,
      "learning_rate": 3.7305719921184626e-06,
      "loss": 0.4497,
      "step": 590
    },
    {
      "epoch": 5.051282051282051,
      "grad_norm": 1.2073548060349433,
      "learning_rate": 3.7266745109762668e-06,
      "loss": 0.3697,
      "step": 591
    },
    {
      "epoch": 5.05982905982906,
      "grad_norm": 1.3921748253309347,
      "learning_rate": 3.7227730990903556e-06,
      "loss": 0.6594,
      "step": 592
    },
    {
      "epoch": 5.068376068376068,
      "grad_norm": 1.6701256777936668,
      "learning_rate": 3.718867768962371e-06,
      "loss": 0.5206,
      "step": 593
    },
    {
      "epoch": 5.076923076923077,
      "grad_norm": 1.1440994803121367,
      "learning_rate": 3.714958533106515e-06,
      "loss": 0.5752,
      "step": 594
    },
    {
      "epoch": 5.085470085470085,
      "grad_norm": 1.224808376797521,
      "learning_rate": 3.711045404049507e-06,
      "loss": 0.5055,
      "step": 595
    },
    {
      "epoch": 5.094017094017094,
      "grad_norm": 1.7672289430629553,
      "learning_rate": 3.7071283943305367e-06,
      "loss": 0.6325,
      "step": 596
    },
    {
      "epoch": 5.102564102564102,
      "grad_norm": 1.3100332567694668,
      "learning_rate": 3.7032075165012323e-06,
      "loss": 0.4567,
      "step": 597
    },
    {
      "epoch": 5.111111111111111,
      "grad_norm": 1.094017044371432,
      "learning_rate": 3.699282783125616e-06,
      "loss": 0.622,
      "step": 598
    },
    {
      "epoch": 5.119658119658119,
      "grad_norm": 16.34482765121646,
      "learning_rate": 3.6953542067800647e-06,
      "loss": 0.5973,
      "step": 599
    },
    {
      "epoch": 5.128205128205128,
      "grad_norm": 2.1578147738458924,
      "learning_rate": 3.6914218000532697e-06,
      "loss": 0.3968,
      "step": 600
    },
    {
      "epoch": 5.136752136752137,
      "grad_norm": 1.440300424288331,
      "learning_rate": 3.6874855755461975e-06,
      "loss": 0.403,
      "step": 601
    },
    {
      "epoch": 5.145299145299146,
      "grad_norm": 1.4169316737597903,
      "learning_rate": 3.683545545872045e-06,
      "loss": 0.4174,
      "step": 602
    },
    {
      "epoch": 5.153846153846154,
      "grad_norm": 1.3735373305878908,
      "learning_rate": 3.679601723656205e-06,
      "loss": 0.534,
      "step": 603
    },
    {
      "epoch": 5.162393162393163,
      "grad_norm": 1.0510963346680822,
      "learning_rate": 3.675654121536225e-06,
      "loss": 0.3353,
      "step": 604
    },
    {
      "epoch": 5.170940170940171,
      "grad_norm": 1.1659301532257373,
      "learning_rate": 3.6717027521617593e-06,
      "loss": 0.6423,
      "step": 605
    },
    {
      "epoch": 5.17948717948718,
      "grad_norm": 1.1710337548444323,
      "learning_rate": 3.667747628194539e-06,
      "loss": 0.4896,
      "step": 606
    },
    {
      "epoch": 5.188034188034188,
      "grad_norm": 0.981800121939619,
      "learning_rate": 3.6637887623083235e-06,
      "loss": 0.4531,
      "step": 607
    },
    {
      "epoch": 5.196581196581197,
      "grad_norm": 1.3609555391016044,
      "learning_rate": 3.6598261671888623e-06,
      "loss": 0.3553,
      "step": 608
    },
    {
      "epoch": 5.205128205128205,
      "grad_norm": 1.3543886822633837,
      "learning_rate": 3.655859855533859e-06,
      "loss": 0.4541,
      "step": 609
    },
    {
      "epoch": 5.213675213675214,
      "grad_norm": 1.249408147562903,
      "learning_rate": 3.651889840052922e-06,
      "loss": 0.3994,
      "step": 610
    },
    {
      "epoch": 5.222222222222222,
      "grad_norm": 1.1992882246674452,
      "learning_rate": 3.6479161334675294e-06,
      "loss": 0.5442,
      "step": 611
    },
    {
      "epoch": 5.230769230769231,
      "grad_norm": 1.2907969275161841,
      "learning_rate": 3.643938748510989e-06,
      "loss": 0.4032,
      "step": 612
    },
    {
      "epoch": 5.239316239316239,
      "grad_norm": 1.2081170928909912,
      "learning_rate": 3.6399576979283914e-06,
      "loss": 0.7046,
      "step": 613
    },
    {
      "epoch": 5.247863247863248,
      "grad_norm": 1.0880564796064907,
      "learning_rate": 3.6359729944765785e-06,
      "loss": 0.5074,
      "step": 614
    },
    {
      "epoch": 5.256410256410256,
      "grad_norm": 1.242466798795238,
      "learning_rate": 3.631984650924094e-06,
      "loss": 0.5935,
      "step": 615
    },
    {
      "epoch": 5.264957264957265,
      "grad_norm": 1.0887798223946241,
      "learning_rate": 3.6279926800511455e-06,
      "loss": 0.7886,
      "step": 616
    },
    {
      "epoch": 5.273504273504273,
      "grad_norm": 1.2491679856445455,
      "learning_rate": 3.623997094649566e-06,
      "loss": 0.348,
      "step": 617
    },
    {
      "epoch": 5.282051282051282,
      "grad_norm": 1.294792724963306,
      "learning_rate": 3.6199979075227707e-06,
      "loss": 0.3407,
      "step": 618
    },
    {
      "epoch": 5.2905982905982905,
      "grad_norm": 1.3554021985070248,
      "learning_rate": 3.6159951314857145e-06,
      "loss": 0.3867,
      "step": 619
    },
    {
      "epoch": 5.299145299145299,
      "grad_norm": 0.9951487626604573,
      "learning_rate": 3.6119887793648535e-06,
      "loss": 0.495,
      "step": 620
    },
    {
      "epoch": 5.3076923076923075,
      "grad_norm": 1.5132542740857915,
      "learning_rate": 3.607978863998104e-06,
      "loss": 0.3828,
      "step": 621
    },
    {
      "epoch": 5.316239316239316,
      "grad_norm": 1.6796037288363033,
      "learning_rate": 3.6039653982347977e-06,
      "loss": 0.2866,
      "step": 622
    },
    {
      "epoch": 5.3247863247863245,
      "grad_norm": 1.7369544007805016,
      "learning_rate": 3.5999483949356458e-06,
      "loss": 0.3328,
      "step": 623
    },
    {
      "epoch": 5.333333333333333,
      "grad_norm": 1.03844187756572,
      "learning_rate": 3.595927866972694e-06,
      "loss": 0.3897,
      "step": 624
    },
    {
      "epoch": 5.3418803418803416,
      "grad_norm": 1.1397342002914956,
      "learning_rate": 3.5919038272292824e-06,
      "loss": 0.5871,
      "step": 625
    },
    {
      "epoch": 5.35042735042735,
      "grad_norm": 0.9852238046959051,
      "learning_rate": 3.587876288600004e-06,
      "loss": 0.3391,
      "step": 626
    },
    {
      "epoch": 5.358974358974359,
      "grad_norm": 1.2750514024039332,
      "learning_rate": 3.583845263990664e-06,
      "loss": 0.3513,
      "step": 627
    },
    {
      "epoch": 5.367521367521368,
      "grad_norm": 1.1066020720861434,
      "learning_rate": 3.5798107663182386e-06,
      "loss": 0.4639,
      "step": 628
    },
    {
      "epoch": 5.3760683760683765,
      "grad_norm": 1.1918671558636216,
      "learning_rate": 3.5757728085108318e-06,
      "loss": 0.2819,
      "step": 629
    },
    {
      "epoch": 5.384615384615385,
      "grad_norm": 1.3152243030901882,
      "learning_rate": 3.5717314035076355e-06,
      "loss": 0.5565,
      "step": 630
    },
    {
      "epoch": 5.3931623931623935,
      "grad_norm": 1.2710610311332853,
      "learning_rate": 3.5676865642588894e-06,
      "loss": 0.3711,
      "step": 631
    },
    {
      "epoch": 5.401709401709402,
      "grad_norm": 1.3852231938798323,
      "learning_rate": 3.563638303725835e-06,
      "loss": 0.5479,
      "step": 632
    },
    {
      "epoch": 5.410256410256411,
      "grad_norm": 1.2285067983502056,
      "learning_rate": 3.559586634880679e-06,
      "loss": 0.5268,
      "step": 633
    },
    {
      "epoch": 5.418803418803419,
      "grad_norm": 1.528384561793807,
      "learning_rate": 3.5555315707065496e-06,
      "loss": 0.4178,
      "step": 634
    },
    {
      "epoch": 5.427350427350428,
      "grad_norm": 1.1960031913100837,
      "learning_rate": 3.551473124197454e-06,
      "loss": 0.4534,
      "step": 635
    },
    {
      "epoch": 5.435897435897436,
      "grad_norm": 1.248321524655849,
      "learning_rate": 3.5474113083582382e-06,
      "loss": 0.7638,
      "step": 636
    },
    {
      "epoch": 5.444444444444445,
      "grad_norm": 1.0905064404938167,
      "learning_rate": 3.543346136204545e-06,
      "loss": 0.6237,
      "step": 637
    },
    {
      "epoch": 5.452991452991453,
      "grad_norm": 1.3017143288995983,
      "learning_rate": 3.539277620762772e-06,
      "loss": 0.7304,
      "step": 638
    },
    {
      "epoch": 5.461538461538462,
      "grad_norm": 1.119653015329565,
      "learning_rate": 3.53520577507003e-06,
      "loss": 0.5459,
      "step": 639
    },
    {
      "epoch": 5.47008547008547,
      "grad_norm": 1.346437171994515,
      "learning_rate": 3.5311306121741017e-06,
      "loss": 0.417,
      "step": 640
    },
    {
      "epoch": 5.478632478632479,
      "grad_norm": 1.0289127371308224,
      "learning_rate": 3.5270521451333984e-06,
      "loss": 0.4361,
      "step": 641
    },
    {
      "epoch": 5.487179487179487,
      "grad_norm": 1.2042095998787006,
      "learning_rate": 3.522970387016919e-06,
      "loss": 0.3922,
      "step": 642
    },
    {
      "epoch": 5.495726495726496,
      "grad_norm": 1.4121964704198753,
      "learning_rate": 3.5188853509042105e-06,
      "loss": 0.4761,
      "step": 643
    },
    {
      "epoch": 5.504273504273504,
      "grad_norm": 1.2150073180857026,
      "learning_rate": 3.5147970498853214e-06,
      "loss": 0.4089,
      "step": 644
    },
    {
      "epoch": 5.512820512820513,
      "grad_norm": 1.1270316353412924,
      "learning_rate": 3.5107054970607624e-06,
      "loss": 0.3747,
      "step": 645
    },
    {
      "epoch": 5.521367521367521,
      "grad_norm": 1.2041504969653847,
      "learning_rate": 3.5066107055414677e-06,
      "loss": 0.2461,
      "step": 646
    },
    {
      "epoch": 5.52991452991453,
      "grad_norm": 1.175505164637916,
      "learning_rate": 3.5025126884487447e-06,
      "loss": 0.5827,
      "step": 647
    },
    {
      "epoch": 5.538461538461538,
      "grad_norm": 1.3339817771254319,
      "learning_rate": 3.4984114589142388e-06,
      "loss": 0.3425,
      "step": 648
    },
    {
      "epoch": 5.547008547008547,
      "grad_norm": 1.2730063860154024,
      "learning_rate": 3.4943070300798913e-06,
      "loss": 0.2885,
      "step": 649
    },
    {
      "epoch": 5.555555555555555,
      "grad_norm": 1.1269118982496695,
      "learning_rate": 3.4901994150978926e-06,
      "loss": 0.7544,
      "step": 650
    },
    {
      "epoch": 5.564102564102564,
      "grad_norm": 1.4064894774351702,
      "learning_rate": 3.4860886271306433e-06,
      "loss": 0.4694,
      "step": 651
    },
    {
      "epoch": 5.572649572649572,
      "grad_norm": 1.0680338687097581,
      "learning_rate": 3.481974679350712e-06,
      "loss": 0.3943,
      "step": 652
    },
    {
      "epoch": 5.581196581196581,
      "grad_norm": 1.2287737197668491,
      "learning_rate": 3.4778575849407924e-06,
      "loss": 0.3581,
      "step": 653
    },
    {
      "epoch": 5.589743589743589,
      "grad_norm": 1.2852278699675668,
      "learning_rate": 3.473737357093662e-06,
      "loss": 0.4511,
      "step": 654
    },
    {
      "epoch": 5.598290598290598,
      "grad_norm": 1.3283726478419335,
      "learning_rate": 3.4696140090121377e-06,
      "loss": 0.3284,
      "step": 655
    },
    {
      "epoch": 5.6068376068376065,
      "grad_norm": 1.0510813178744065,
      "learning_rate": 3.465487553909035e-06,
      "loss": 0.3599,
      "step": 656
    },
    {
      "epoch": 5.615384615384615,
      "grad_norm": 1.0755691838628418,
      "learning_rate": 3.461358005007128e-06,
      "loss": 0.577,
      "step": 657
    },
    {
      "epoch": 5.6239316239316235,
      "grad_norm": 1.064075876776,
      "learning_rate": 3.4572253755390996e-06,
      "loss": 0.6382,
      "step": 658
    },
    {
      "epoch": 5.632478632478632,
      "grad_norm": 1.1490546293761381,
      "learning_rate": 3.4530896787475083e-06,
      "loss": 0.4093,
      "step": 659
    },
    {
      "epoch": 5.641025641025641,
      "grad_norm": 1.0908195852061844,
      "learning_rate": 3.4489509278847415e-06,
      "loss": 0.3019,
      "step": 660
    },
    {
      "epoch": 5.64957264957265,
      "grad_norm": 1.0838622648728164,
      "learning_rate": 3.44480913621297e-06,
      "loss": 0.5162,
      "step": 661
    },
    {
      "epoch": 5.6581196581196584,
      "grad_norm": 1.2952654131629489,
      "learning_rate": 3.44066431700411e-06,
      "loss": 0.3326,
      "step": 662
    },
    {
      "epoch": 5.666666666666667,
      "grad_norm": 1.11428294131442,
      "learning_rate": 3.436516483539781e-06,
      "loss": 0.4055,
      "step": 663
    },
    {
      "epoch": 5.6752136752136755,
      "grad_norm": 1.2330022201567412,
      "learning_rate": 3.432365649111257e-06,
      "loss": 0.4222,
      "step": 664
    },
    {
      "epoch": 5.683760683760684,
      "grad_norm": 1.1952087265548075,
      "learning_rate": 3.428211827019434e-06,
      "loss": 0.3838,
      "step": 665
    },
    {
      "epoch": 5.6923076923076925,
      "grad_norm": 1.1168628551342878,
      "learning_rate": 3.4240550305747776e-06,
      "loss": 0.2602,
      "step": 666
    },
    {
      "epoch": 5.700854700854701,
      "grad_norm": 2.630335521757739,
      "learning_rate": 3.4198952730972845e-06,
      "loss": 0.2775,
      "step": 667
    },
    {
      "epoch": 5.7094017094017095,
      "grad_norm": 1.2094615456741151,
      "learning_rate": 3.4157325679164416e-06,
      "loss": 0.5243,
      "step": 668
    },
    {
      "epoch": 5.717948717948718,
      "grad_norm": 1.091861892576226,
      "learning_rate": 3.4115669283711795e-06,
      "loss": 0.3371,
      "step": 669
    },
    {
      "epoch": 5.726495726495727,
      "grad_norm": 1.2671530096401382,
      "learning_rate": 3.407398367809832e-06,
      "loss": 0.553,
      "step": 670
    },
    {
      "epoch": 5.735042735042735,
      "grad_norm": 1.8341290350898378,
      "learning_rate": 3.403226899590096e-06,
      "loss": 0.4215,
      "step": 671
    },
    {
      "epoch": 5.743589743589744,
      "grad_norm": 1.0560810502003946,
      "learning_rate": 3.3990525370789793e-06,
      "loss": 0.5067,
      "step": 672
    },
    {
      "epoch": 5.752136752136752,
      "grad_norm": 1.1364367551725,
      "learning_rate": 3.3948752936527722e-06,
      "loss": 0.3745,
      "step": 673
    },
    {
      "epoch": 5.760683760683761,
      "grad_norm": 1.0851772890439473,
      "learning_rate": 3.3906951826969905e-06,
      "loss": 0.4404,
      "step": 674
    },
    {
      "epoch": 5.769230769230769,
      "grad_norm": 1.253804399448194,
      "learning_rate": 3.386512217606339e-06,
      "loss": 0.5596,
      "step": 675
    },
    {
      "epoch": 5.777777777777778,
      "grad_norm": 1.0118773013123081,
      "learning_rate": 3.3823264117846722e-06,
      "loss": 0.3556,
      "step": 676
    },
    {
      "epoch": 5.786324786324786,
      "grad_norm": 1.2600330912135416,
      "learning_rate": 3.378137778644945e-06,
      "loss": 0.4202,
      "step": 677
    },
    {
      "epoch": 5.794871794871795,
      "grad_norm": 1.1341711674144634,
      "learning_rate": 3.3739463316091696e-06,
      "loss": 0.3549,
      "step": 678
    },
    {
      "epoch": 5.803418803418803,
      "grad_norm": 1.3648356770575298,
      "learning_rate": 3.369752084108381e-06,
      "loss": 0.5829,
      "step": 679
    },
    {
      "epoch": 5.811965811965812,
      "grad_norm": 1.2817095099769975,
      "learning_rate": 3.3655550495825824e-06,
      "loss": 0.4679,
      "step": 680
    },
    {
      "epoch": 5.82051282051282,
      "grad_norm": 1.3209631982240306,
      "learning_rate": 3.3613552414807093e-06,
      "loss": 0.4057,
      "step": 681
    },
    {
      "epoch": 5.829059829059829,
      "grad_norm": 1.1394673881020136,
      "learning_rate": 3.3571526732605875e-06,
      "loss": 0.3996,
      "step": 682
    },
    {
      "epoch": 5.837606837606837,
      "grad_norm": 1.0155252392209944,
      "learning_rate": 3.352947358388884e-06,
      "loss": 0.4856,
      "step": 683
    },
    {
      "epoch": 5.846153846153846,
      "grad_norm": 1.2795057777037273,
      "learning_rate": 3.3487393103410683e-06,
      "loss": 0.5898,
      "step": 684
    },
    {
      "epoch": 5.854700854700854,
      "grad_norm": 1.2951968385232915,
      "learning_rate": 3.3445285426013683e-06,
      "loss": 0.405,
      "step": 685
    },
    {
      "epoch": 5.863247863247864,
      "grad_norm": 1.4918672670392705,
      "learning_rate": 3.3403150686627267e-06,
      "loss": 0.303,
      "step": 686
    },
    {
      "epoch": 5.871794871794872,
      "grad_norm": 2.006770881098824,
      "learning_rate": 3.336098902026758e-06,
      "loss": 0.4886,
      "step": 687
    },
    {
      "epoch": 5.880341880341881,
      "grad_norm": 1.2868325384326031,
      "learning_rate": 3.331880056203706e-06,
      "loss": 0.4736,
      "step": 688
    },
    {
      "epoch": 5.888888888888889,
      "grad_norm": 1.2436589514049399,
      "learning_rate": 3.3276585447123957e-06,
      "loss": 0.5197,
      "step": 689
    },
    {
      "epoch": 5.897435897435898,
      "grad_norm": 1.352912645734446,
      "learning_rate": 3.3234343810801995e-06,
      "loss": 0.5513,
      "step": 690
    },
    {
      "epoch": 5.905982905982906,
      "grad_norm": 1.1902349811855824,
      "learning_rate": 3.319207578842985e-06,
      "loss": 0.4307,
      "step": 691
    },
    {
      "epoch": 5.914529914529915,
      "grad_norm": 1.2876545508730952,
      "learning_rate": 3.314978151545076e-06,
      "loss": 0.6827,
      "step": 692
    },
    {
      "epoch": 5.923076923076923,
      "grad_norm": 1.4876272116617923,
      "learning_rate": 3.3107461127392072e-06,
      "loss": 0.4238,
      "step": 693
    },
    {
      "epoch": 5.931623931623932,
      "grad_norm": 1.0911236336199452,
      "learning_rate": 3.306511475986482e-06,
      "loss": 0.3803,
      "step": 694
    },
    {
      "epoch": 5.94017094017094,
      "grad_norm": 1.1717280095017706,
      "learning_rate": 3.3022742548563293e-06,
      "loss": 0.6268,
      "step": 695
    },
    {
      "epoch": 5.948717948717949,
      "grad_norm": 1.5095367793755137,
      "learning_rate": 3.2980344629264583e-06,
      "loss": 0.4046,
      "step": 696
    },
    {
      "epoch": 5.957264957264957,
      "grad_norm": 1.1688187493069797,
      "learning_rate": 3.293792113782816e-06,
      "loss": 0.3746,
      "step": 697
    },
    {
      "epoch": 5.965811965811966,
      "grad_norm": 2.38178908613328,
      "learning_rate": 3.289547221019546e-06,
      "loss": 0.386,
      "step": 698
    },
    {
      "epoch": 5.9743589743589745,
      "grad_norm": 1.0387739711325623,
      "learning_rate": 3.285299798238938e-06,
      "loss": 0.5492,
      "step": 699
    },
    {
      "epoch": 5.982905982905983,
      "grad_norm": 1.1945528947453434,
      "learning_rate": 3.281049859051394e-06,
      "loss": 0.2644,
      "step": 700
    },
    {
      "epoch": 5.9914529914529915,
      "grad_norm": 1.2000576561757432,
      "learning_rate": 3.276797417075377e-06,
      "loss": 0.4911,
      "step": 701
    },
    {
      "epoch": 6.0,
      "grad_norm": 1.6335354555713457,
      "learning_rate": 3.272542485937369e-06,
      "loss": 0.493,
      "step": 702
    },
    {
      "epoch": 6.0085470085470085,
      "grad_norm": 1.8696148210468555,
      "learning_rate": 3.26828507927183e-06,
      "loss": 0.3595,
      "step": 703
    },
    {
      "epoch": 6.017094017094017,
      "grad_norm": 1.394253213921132,
      "learning_rate": 3.264025210721153e-06,
      "loss": 0.3482,
      "step": 704
    },
    {
      "epoch": 6.0256410256410255,
      "grad_norm": 1.5215937520682763,
      "learning_rate": 3.2597628939356174e-06,
      "loss": 0.5041,
      "step": 705
    },
    {
      "epoch": 6.034188034188034,
      "grad_norm": 1.2837167314940414,
      "learning_rate": 3.25549814257335e-06,
      "loss": 0.5273,
      "step": 706
    },
    {
      "epoch": 6.042735042735043,
      "grad_norm": 1.7025099621293658,
      "learning_rate": 3.2512309703002776e-06,
      "loss": 0.5485,
      "step": 707
    },
    {
      "epoch": 6.051282051282051,
      "grad_norm": 2.0471758371089313,
      "learning_rate": 3.2469613907900847e-06,
      "loss": 0.3261,
      "step": 708
    },
    {
      "epoch": 6.05982905982906,
      "grad_norm": 1.918552218538202,
      "learning_rate": 3.2426894177241707e-06,
      "loss": 0.5351,
      "step": 709
    },
    {
      "epoch": 6.068376068376068,
      "grad_norm": 1.4591389372643695,
      "learning_rate": 3.2384150647916033e-06,
      "loss": 0.3315,
      "step": 710
    },
    {
      "epoch": 6.076923076923077,
      "grad_norm": 1.2241942485088706,
      "learning_rate": 3.2341383456890776e-06,
      "loss": 0.4083,
      "step": 711
    },
    {
      "epoch": 6.085470085470085,
      "grad_norm": 1.4233353451639101,
      "learning_rate": 3.229859274120869e-06,
      "loss": 0.3002,
      "step": 712
    },
    {
      "epoch": 6.094017094017094,
      "grad_norm": 1.1255702206320295,
      "learning_rate": 3.2255778637987935e-06,
      "loss": 0.2582,
      "step": 713
    },
    {
      "epoch": 6.102564102564102,
      "grad_norm": 1.5685617433721264,
      "learning_rate": 3.2212941284421595e-06,
      "loss": 0.2923,
      "step": 714
    },
    {
      "epoch": 6.111111111111111,
      "grad_norm": 1.4408254299138377,
      "learning_rate": 3.217008081777726e-06,
      "loss": 0.2306,
      "step": 715
    },
    {
      "epoch": 6.119658119658119,
      "grad_norm": 1.3513374270055847,
      "learning_rate": 3.2127197375396596e-06,
      "loss": 0.8027,
      "step": 716
    },
    {
      "epoch": 6.128205128205128,
      "grad_norm": 1.821456048145893,
      "learning_rate": 3.208429109469488e-06,
      "loss": 0.4388,
      "step": 717
    },
    {
      "epoch": 6.136752136752137,
      "grad_norm": 1.2713819489136216,
      "learning_rate": 3.204136211316057e-06,
      "loss": 0.5603,
      "step": 718
    },
    {
      "epoch": 6.145299145299146,
      "grad_norm": 1.5416645237683366,
      "learning_rate": 3.199841056835489e-06,
      "loss": 0.4057,
      "step": 719
    },
    {
      "epoch": 6.153846153846154,
      "grad_norm": 1.4766742725804678,
      "learning_rate": 3.195543659791132e-06,
      "loss": 0.4062,
      "step": 720
    },
    {
      "epoch": 6.162393162393163,
      "grad_norm": 1.0920488604517398,
      "learning_rate": 3.191244033953524e-06,
      "loss": 0.5131,
      "step": 721
    },
    {
      "epoch": 6.170940170940171,
      "grad_norm": 2.1075139352156635,
      "learning_rate": 3.1869421931003446e-06,
      "loss": 0.2847,
      "step": 722
    },
    {
      "epoch": 6.17948717948718,
      "grad_norm": 1.4061808008069692,
      "learning_rate": 3.182638151016369e-06,
      "loss": 0.5075,
      "step": 723
    },
    {
      "epoch": 6.188034188034188,
      "grad_norm": 0.9318584748785151,
      "learning_rate": 3.1783319214934274e-06,
      "loss": 0.4692,
      "step": 724
    },
    {
      "epoch": 6.196581196581197,
      "grad_norm": 1.5607675486820585,
      "learning_rate": 3.17402351833036e-06,
      "loss": 0.4791,
      "step": 725
    },
    {
      "epoch": 6.205128205128205,
      "grad_norm": 1.8849662864698848,
      "learning_rate": 3.1697129553329708e-06,
      "loss": 0.6563,
      "step": 726
    },
    {
      "epoch": 6.213675213675214,
      "grad_norm": 1.0331208261591494,
      "learning_rate": 3.1654002463139854e-06,
      "loss": 0.3876,
      "step": 727
    },
    {
      "epoch": 6.222222222222222,
      "grad_norm": 1.3672949819681985,
      "learning_rate": 3.1610854050930063e-06,
      "loss": 0.4545,
      "step": 728
    },
    {
      "epoch": 6.230769230769231,
      "grad_norm": 1.325503991861968,
      "learning_rate": 3.1567684454964674e-06,
      "loss": 0.4732,
      "step": 729
    },
    {
      "epoch": 6.239316239316239,
      "grad_norm": 0.9754927814726547,
      "learning_rate": 3.1524493813575936e-06,
      "loss": 0.5361,
      "step": 730
    },
    {
      "epoch": 6.247863247863248,
      "grad_norm": 1.475889572340759,
      "learning_rate": 3.1481282265163493e-06,
      "loss": 0.555,
      "step": 731
    },
    {
      "epoch": 6.256410256410256,
      "grad_norm": 1.2242919034236064,
      "learning_rate": 3.1438049948194006e-06,
      "loss": 0.3423,
      "step": 732
    },
    {
      "epoch": 6.264957264957265,
      "grad_norm": 1.1461098912240082,
      "learning_rate": 3.1394797001200705e-06,
      "loss": 0.4012,
      "step": 733
    },
    {
      "epoch": 6.273504273504273,
      "grad_norm": 1.24676230120588,
      "learning_rate": 3.1351523562782893e-06,
      "loss": 0.2398,
      "step": 734
    },
    {
      "epoch": 6.282051282051282,
      "grad_norm": 1.6064295658402483,
      "learning_rate": 3.1308229771605546e-06,
      "loss": 0.3138,
      "step": 735
    },
    {
      "epoch": 6.2905982905982905,
      "grad_norm": 1.1099122618300608,
      "learning_rate": 3.1264915766398872e-06,
      "loss": 0.4298,
      "step": 736
    },
    {
      "epoch": 6.299145299145299,
      "grad_norm": 1.1345723411356023,
      "learning_rate": 3.1221581685957837e-06,
      "loss": 0.3354,
      "step": 737
    },
    {
      "epoch": 6.3076923076923075,
      "grad_norm": 1.2491753534521146,
      "learning_rate": 3.117822766914174e-06,
      "loss": 0.6514,
      "step": 738
    },
    {
      "epoch": 6.316239316239316,
      "grad_norm": 1.3935328769910025,
      "learning_rate": 3.1134853854873774e-06,
      "loss": 0.384,
      "step": 739
    },
    {
      "epoch": 6.3247863247863245,
      "grad_norm": 1.025200062176103,
      "learning_rate": 3.109146038214055e-06,
      "loss": 0.4001,
      "step": 740
    },
    {
      "epoch": 6.333333333333333,
      "grad_norm": 1.2195206502277662,
      "learning_rate": 3.1048047389991693e-06,
      "loss": 0.5226,
      "step": 741
    },
    {
      "epoch": 6.3418803418803416,
      "grad_norm": 1.32123049941495,
      "learning_rate": 3.1004615017539375e-06,
      "loss": 0.3312,
      "step": 742
    },
    {
      "epoch": 6.35042735042735,
      "grad_norm": 1.2390374240886866,
      "learning_rate": 3.096116340395783e-06,
      "loss": 0.5162,
      "step": 743
    },
    {
      "epoch": 6.358974358974359,
      "grad_norm": 1.549536241283591,
      "learning_rate": 3.0917692688483023e-06,
      "loss": 0.3452,
      "step": 744
    },
    {
      "epoch": 6.367521367521368,
      "grad_norm": 1.163597656164462,
      "learning_rate": 3.0874203010412057e-06,
      "loss": 0.4066,
      "step": 745
    },
    {
      "epoch": 6.3760683760683765,
      "grad_norm": 1.4309744794804062,
      "learning_rate": 3.0830694509102835e-06,
      "loss": 0.3004,
      "step": 746
    },
    {
      "epoch": 6.384615384615385,
      "grad_norm": 1.9244526378821927,
      "learning_rate": 3.0787167323973584e-06,
      "loss": 0.4615,
      "step": 747
    },
    {
      "epoch": 6.3931623931623935,
      "grad_norm": 1.221426024391306,
      "learning_rate": 3.074362159450236e-06,
      "loss": 0.2793,
      "step": 748
    },
    {
      "epoch": 6.401709401709402,
      "grad_norm": 1.2043561131798144,
      "learning_rate": 3.070005746022669e-06,
      "loss": 0.2147,
      "step": 749
    },
    {
      "epoch": 6.410256410256411,
      "grad_norm": 1.6838636399693165,
      "learning_rate": 3.0656475060743065e-06,
      "loss": 0.593,
      "step": 750
    },
    {
      "epoch": 6.418803418803419,
      "grad_norm": 1.2949091730132467,
      "learning_rate": 3.061287453570646e-06,
      "loss": 0.5454,
      "step": 751
    },
    {
      "epoch": 6.427350427350428,
      "grad_norm": 1.4030184121773681,
      "learning_rate": 3.056925602483e-06,
      "loss": 0.3405,
      "step": 752
    },
    {
      "epoch": 6.435897435897436,
      "grad_norm": 1.2147418551070128,
      "learning_rate": 3.052561966788441e-06,
      "loss": 0.3789,
      "step": 753
    },
    {
      "epoch": 6.444444444444445,
      "grad_norm": 1.2739180484045751,
      "learning_rate": 3.0481965604697582e-06,
      "loss": 0.3262,
      "step": 754
    },
    {
      "epoch": 6.452991452991453,
      "grad_norm": 1.5444059820139582,
      "learning_rate": 3.043829397515419e-06,
      "loss": 0.3836,
      "step": 755
    },
    {
      "epoch": 6.461538461538462,
      "grad_norm": 1.2628409498676159,
      "learning_rate": 3.039460491919516e-06,
      "loss": 0.5492,
      "step": 756
    },
    {
      "epoch": 6.47008547008547,
      "grad_norm": 1.1975440242064967,
      "learning_rate": 3.0350898576817268e-06,
      "loss": 0.4043,
      "step": 757
    },
    {
      "epoch": 6.478632478632479,
      "grad_norm": 1.2299452398293043,
      "learning_rate": 3.03071750880727e-06,
      "loss": 0.5042,
      "step": 758
    },
    {
      "epoch": 6.487179487179487,
      "grad_norm": 1.296600850446519,
      "learning_rate": 3.0263434593068562e-06,
      "loss": 0.3901,
      "step": 759
    },
    {
      "epoch": 6.495726495726496,
      "grad_norm": 1.0714540115525397,
      "learning_rate": 3.021967723196647e-06,
      "loss": 0.5952,
      "step": 760
    },
    {
      "epoch": 6.504273504273504,
      "grad_norm": 1.2212295717383705,
      "learning_rate": 3.017590314498208e-06,
      "loss": 0.4616,
      "step": 761
    },
    {
      "epoch": 6.512820512820513,
      "grad_norm": 1.8417042430435893,
      "learning_rate": 3.0132112472384652e-06,
      "loss": 0.2779,
      "step": 762
    },
    {
      "epoch": 6.521367521367521,
      "grad_norm": 1.4681440287957088,
      "learning_rate": 3.0088305354496574e-06,
      "loss": 0.4951,
      "step": 763
    },
    {
      "epoch": 6.52991452991453,
      "grad_norm": 1.3911986079523737,
      "learning_rate": 3.004448193169294e-06,
      "loss": 0.4351,
      "step": 764
    },
    {
      "epoch": 6.538461538461538,
      "grad_norm": 1.131630270165073,
      "learning_rate": 3.0000642344401115e-06,
      "loss": 0.4227,
      "step": 765
    },
    {
      "epoch": 6.547008547008547,
      "grad_norm": 1.2607186615387505,
      "learning_rate": 2.9956786733100225e-06,
      "loss": 0.4701,
      "step": 766
    },
    {
      "epoch": 6.555555555555555,
      "grad_norm": 1.0732664543097763,
      "learning_rate": 2.9912915238320755e-06,
      "loss": 0.37,
      "step": 767
    },
    {
      "epoch": 6.564102564102564,
      "grad_norm": 1.0692993622926654,
      "learning_rate": 2.9869028000644102e-06,
      "loss": 0.4654,
      "step": 768
    },
    {
      "epoch": 6.572649572649572,
      "grad_norm": 1.4021715786112823,
      "learning_rate": 2.9825125160702096e-06,
      "loss": 0.6461,
      "step": 769
    },
    {
      "epoch": 6.581196581196581,
      "grad_norm": 1.4067036745056627,
      "learning_rate": 2.978120685917656e-06,
      "loss": 0.5253,
      "step": 770
    },
    {
      "epoch": 6.589743589743589,
      "grad_norm": 1.3331749163814144,
      "learning_rate": 2.9737273236798868e-06,
      "loss": 0.2276,
      "step": 771
    },
    {
      "epoch": 6.598290598290598,
      "grad_norm": 1.0606838977666249,
      "learning_rate": 2.9693324434349486e-06,
      "loss": 0.3774,
      "step": 772
    },
    {
      "epoch": 6.6068376068376065,
      "grad_norm": 1.3127138222755697,
      "learning_rate": 2.9649360592657526e-06,
      "loss": 0.3676,
      "step": 773
    },
    {
      "epoch": 6.615384615384615,
      "grad_norm": 1.2133475948912267,
      "learning_rate": 2.960538185260029e-06,
      "loss": 0.4022,
      "step": 774
    },
    {
      "epoch": 6.6239316239316235,
      "grad_norm": 1.0322313184125005,
      "learning_rate": 2.956138835510282e-06,
      "loss": 0.4228,
      "step": 775
    },
    {
      "epoch": 6.632478632478632,
      "grad_norm": 1.2342645552359022,
      "learning_rate": 2.9517380241137437e-06,
      "loss": 0.3714,
      "step": 776
    },
    {
      "epoch": 6.641025641025641,
      "grad_norm": 1.3392632273965621,
      "learning_rate": 2.9473357651723324e-06,
      "loss": 0.3647,
      "step": 777
    },
    {
      "epoch": 6.64957264957265,
      "grad_norm": 1.4710559108999355,
      "learning_rate": 2.942932072792602e-06,
      "loss": 0.5227,
      "step": 778
    },
    {
      "epoch": 6.6581196581196584,
      "grad_norm": 1.425011112204164,
      "learning_rate": 2.938526961085701e-06,
      "loss": 0.3658,
      "step": 779
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 1.3501743098448087,
      "learning_rate": 2.9341204441673267e-06,
      "loss": 0.3663,
      "step": 780
    },
    {
      "epoch": 6.6752136752136755,
      "grad_norm": 1.2064409829036324,
      "learning_rate": 2.929712536157677e-06,
      "loss": 0.4467,
      "step": 781
    },
    {
      "epoch": 6.683760683760684,
      "grad_norm": 1.1611425318187865,
      "learning_rate": 2.925303251181411e-06,
      "loss": 0.5549,
      "step": 782
    },
    {
      "epoch": 6.6923076923076925,
      "grad_norm": 1.6675008400391373,
      "learning_rate": 2.920892603367596e-06,
      "loss": 0.3228,
      "step": 783
    },
    {
      "epoch": 6.700854700854701,
      "grad_norm": 1.093008739070284,
      "learning_rate": 2.916480606849671e-06,
      "loss": 0.487,
      "step": 784
    },
    {
      "epoch": 6.7094017094017095,
      "grad_norm": 1.3507494915995428,
      "learning_rate": 2.9120672757653917e-06,
      "loss": 0.4002,
      "step": 785
    },
    {
      "epoch": 6.717948717948718,
      "grad_norm": 1.280608215049592,
      "learning_rate": 2.907652624256794e-06,
      "loss": 0.5651,
      "step": 786
    },
    {
      "epoch": 6.726495726495727,
      "grad_norm": 1.1277587047710411,
      "learning_rate": 2.903236666470143e-06,
      "loss": 0.5009,
      "step": 787
    },
    {
      "epoch": 6.735042735042735,
      "grad_norm": 1.4691185750909477,
      "learning_rate": 2.89881941655589e-06,
      "loss": 0.379,
      "step": 788
    },
    {
      "epoch": 6.743589743589744,
      "grad_norm": 1.3006869233227,
      "learning_rate": 2.8944008886686288e-06,
      "loss": 0.2883,
      "step": 789
    },
    {
      "epoch": 6.752136752136752,
      "grad_norm": 1.5404890921238719,
      "learning_rate": 2.889981096967045e-06,
      "loss": 0.5591,
      "step": 790
    },
    {
      "epoch": 6.760683760683761,
      "grad_norm": 1.5129641067947304,
      "learning_rate": 2.8855600556138757e-06,
      "loss": 0.4018,
      "step": 791
    },
    {
      "epoch": 6.769230769230769,
      "grad_norm": 1.7543516151257428,
      "learning_rate": 2.881137778775864e-06,
      "loss": 0.2992,
      "step": 792
    },
    {
      "epoch": 6.777777777777778,
      "grad_norm": 1.702284294452542,
      "learning_rate": 2.876714280623708e-06,
      "loss": 0.3827,
      "step": 793
    },
    {
      "epoch": 6.786324786324786,
      "grad_norm": 1.4614297879108673,
      "learning_rate": 2.872289575332023e-06,
      "loss": 0.3365,
      "step": 794
    },
    {
      "epoch": 6.794871794871795,
      "grad_norm": 1.3849915243782156,
      "learning_rate": 2.8678636770792907e-06,
      "loss": 0.4058,
      "step": 795
    },
    {
      "epoch": 6.803418803418803,
      "grad_norm": 1.26681353460394,
      "learning_rate": 2.863436600047815e-06,
      "loss": 0.3846,
      "step": 796
    },
    {
      "epoch": 6.811965811965812,
      "grad_norm": 1.6534025844078974,
      "learning_rate": 2.8590083584236792e-06,
      "loss": 0.5269,
      "step": 797
    },
    {
      "epoch": 6.82051282051282,
      "grad_norm": 1.360051665942373,
      "learning_rate": 2.854578966396697e-06,
      "loss": 0.5133,
      "step": 798
    },
    {
      "epoch": 6.829059829059829,
      "grad_norm": 1.0485488708326347,
      "learning_rate": 2.8501484381603685e-06,
      "loss": 0.2431,
      "step": 799
    },
    {
      "epoch": 6.837606837606837,
      "grad_norm": 1.4629730877013372,
      "learning_rate": 2.8457167879118332e-06,
      "loss": 0.6504,
      "step": 800
    },
    {
      "epoch": 6.846153846153846,
      "grad_norm": 1.2010217515771036,
      "learning_rate": 2.8412840298518295e-06,
      "loss": 0.4259,
      "step": 801
    },
    {
      "epoch": 6.854700854700854,
      "grad_norm": 1.5172571882549153,
      "learning_rate": 2.836850178184642e-06,
      "loss": 0.267,
      "step": 802
    },
    {
      "epoch": 6.863247863247864,
      "grad_norm": 1.274248209706548,
      "learning_rate": 2.8324152471180634e-06,
      "loss": 0.3201,
      "step": 803
    },
    {
      "epoch": 6.871794871794872,
      "grad_norm": 1.2852432803446472,
      "learning_rate": 2.8279792508633415e-06,
      "loss": 0.2461,
      "step": 804
    },
    {
      "epoch": 6.880341880341881,
      "grad_norm": 1.0075068351453549,
      "learning_rate": 2.8235422036351384e-06,
      "loss": 0.4763,
      "step": 805
    },
    {
      "epoch": 6.888888888888889,
      "grad_norm": 1.255947186502288,
      "learning_rate": 2.8191041196514874e-06,
      "loss": 0.2422,
      "step": 806
    },
    {
      "epoch": 6.897435897435898,
      "grad_norm": 1.226037306889786,
      "learning_rate": 2.8146650131337376e-06,
      "loss": 0.4444,
      "step": 807
    },
    {
      "epoch": 6.905982905982906,
      "grad_norm": 1.7899615704173477,
      "learning_rate": 2.81022489830652e-06,
      "loss": 0.4276,
      "step": 808
    },
    {
      "epoch": 6.914529914529915,
      "grad_norm": 0.9990455509140496,
      "learning_rate": 2.8057837893976958e-06,
      "loss": 0.4717,
      "step": 809
    },
    {
      "epoch": 6.923076923076923,
      "grad_norm": 1.255311068942597,
      "learning_rate": 2.8013417006383078e-06,
      "loss": 0.489,
      "step": 810
    },
    {
      "epoch": 6.931623931623932,
      "grad_norm": 0.914637503870123,
      "learning_rate": 2.7968986462625436e-06,
      "loss": 0.3367,
      "step": 811
    },
    {
      "epoch": 6.94017094017094,
      "grad_norm": 1.195207651209717,
      "learning_rate": 2.7924546405076837e-06,
      "loss": 0.4924,
      "step": 812
    },
    {
      "epoch": 6.948717948717949,
      "grad_norm": 1.2462780058107867,
      "learning_rate": 2.788009697614053e-06,
      "loss": 0.4046,
      "step": 813
    },
    {
      "epoch": 6.957264957264957,
      "grad_norm": 1.1795029432035888,
      "learning_rate": 2.7835638318249856e-06,
      "loss": 0.5524,
      "step": 814
    },
    {
      "epoch": 6.965811965811966,
      "grad_norm": 1.2531330155038323,
      "learning_rate": 2.7791170573867698e-06,
      "loss": 0.4246,
      "step": 815
    },
    {
      "epoch": 6.9743589743589745,
      "grad_norm": 1.3040915528226138,
      "learning_rate": 2.7746693885486044e-06,
      "loss": 0.4456,
      "step": 816
    },
    {
      "epoch": 6.982905982905983,
      "grad_norm": 1.293729203493905,
      "learning_rate": 2.770220839562556e-06,
      "loss": 0.4462,
      "step": 817
    },
    {
      "epoch": 6.9914529914529915,
      "grad_norm": 1.3294535147477113,
      "learning_rate": 2.765771424683513e-06,
      "loss": 0.2795,
      "step": 818
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.1599921647100595,
      "learning_rate": 2.761321158169134e-06,
      "loss": 0.4591,
      "step": 819
    },
    {
      "epoch": 7.0085470085470085,
      "grad_norm": 1.584598458934145,
      "learning_rate": 2.7568700542798112e-06,
      "loss": 0.4017,
      "step": 820
    },
    {
      "epoch": 7.017094017094017,
      "grad_norm": 1.3451576551405218,
      "learning_rate": 2.7524181272786153e-06,
      "loss": 0.6676,
      "step": 821
    },
    {
      "epoch": 7.0256410256410255,
      "grad_norm": 1.6154170245528818,
      "learning_rate": 2.747965391431261e-06,
      "loss": 0.3913,
      "step": 822
    },
    {
      "epoch": 7.034188034188034,
      "grad_norm": 1.1220733748458145,
      "learning_rate": 2.743511861006049e-06,
      "loss": 0.3535,
      "step": 823
    },
    {
      "epoch": 7.042735042735043,
      "grad_norm": 3.849959694375653,
      "learning_rate": 2.739057550273828e-06,
      "loss": 0.4753,
      "step": 824
    },
    {
      "epoch": 7.051282051282051,
      "grad_norm": 3.7643260373905805,
      "learning_rate": 2.7346024735079483e-06,
      "loss": 0.2384,
      "step": 825
    },
    {
      "epoch": 7.05982905982906,
      "grad_norm": 2.169051331573644,
      "learning_rate": 2.7301466449842147e-06,
      "loss": 0.3003,
      "step": 826
    },
    {
      "epoch": 7.068376068376068,
      "grad_norm": 1.1849827445410177,
      "learning_rate": 2.725690078980838e-06,
      "loss": 0.5765,
      "step": 827
    },
    {
      "epoch": 7.076923076923077,
      "grad_norm": 1.532957836247954,
      "learning_rate": 2.7212327897783963e-06,
      "loss": 0.4367,
      "step": 828
    },
    {
      "epoch": 7.085470085470085,
      "grad_norm": 1.9990747567784164,
      "learning_rate": 2.7167747916597825e-06,
      "loss": 0.1999,
      "step": 829
    },
    {
      "epoch": 7.094017094017094,
      "grad_norm": 1.0102970931842459,
      "learning_rate": 2.7123160989101623e-06,
      "loss": 0.3689,
      "step": 830
    },
    {
      "epoch": 7.102564102564102,
      "grad_norm": 1.456532466139522,
      "learning_rate": 2.7078567258169264e-06,
      "loss": 0.472,
      "step": 831
    },
    {
      "epoch": 7.111111111111111,
      "grad_norm": 1.4365042757644924,
      "learning_rate": 2.703396686669646e-06,
      "loss": 0.2607,
      "step": 832
    },
    {
      "epoch": 7.119658119658119,
      "grad_norm": 1.5344758159392708,
      "learning_rate": 2.698935995760027e-06,
      "loss": 0.2563,
      "step": 833
    },
    {
      "epoch": 7.128205128205128,
      "grad_norm": 1.8982223511921632,
      "learning_rate": 2.6944746673818623e-06,
      "loss": 0.2973,
      "step": 834
    },
    {
      "epoch": 7.136752136752137,
      "grad_norm": 1.3744560603995541,
      "learning_rate": 2.6900127158309903e-06,
      "loss": 0.4143,
      "step": 835
    },
    {
      "epoch": 7.145299145299146,
      "grad_norm": 1.3900882827182501,
      "learning_rate": 2.6855501554052433e-06,
      "loss": 0.4789,
      "step": 836
    },
    {
      "epoch": 7.153846153846154,
      "grad_norm": 1.9535660899006906,
      "learning_rate": 2.6810870004044065e-06,
      "loss": 0.3261,
      "step": 837
    },
    {
      "epoch": 7.162393162393163,
      "grad_norm": 1.2683908981688092,
      "learning_rate": 2.6766232651301694e-06,
      "loss": 0.3066,
      "step": 838
    },
    {
      "epoch": 7.170940170940171,
      "grad_norm": 1.1018165851573016,
      "learning_rate": 2.672158963886082e-06,
      "loss": 0.4697,
      "step": 839
    },
    {
      "epoch": 7.17948717948718,
      "grad_norm": 1.7661043091369681,
      "learning_rate": 2.667694110977506e-06,
      "loss": 0.3914,
      "step": 840
    },
    {
      "epoch": 7.188034188034188,
      "grad_norm": 1.2250139257227237,
      "learning_rate": 2.6632287207115735e-06,
      "loss": 0.4166,
      "step": 841
    },
    {
      "epoch": 7.196581196581197,
      "grad_norm": 1.143439941874872,
      "learning_rate": 2.6587628073971366e-06,
      "loss": 0.5522,
      "step": 842
    },
    {
      "epoch": 7.205128205128205,
      "grad_norm": 1.2894052072252156,
      "learning_rate": 2.654296385344724e-06,
      "loss": 0.3279,
      "step": 843
    },
    {
      "epoch": 7.213675213675214,
      "grad_norm": 1.4995290175071838,
      "learning_rate": 2.6498294688664937e-06,
      "loss": 0.4711,
      "step": 844
    },
    {
      "epoch": 7.222222222222222,
      "grad_norm": 1.273027740413982,
      "learning_rate": 2.6453620722761897e-06,
      "loss": 0.4302,
      "step": 845
    },
    {
      "epoch": 7.230769230769231,
      "grad_norm": 1.4543407022876726,
      "learning_rate": 2.6408942098890937e-06,
      "loss": 0.4384,
      "step": 846
    },
    {
      "epoch": 7.239316239316239,
      "grad_norm": 1.2122402658494311,
      "learning_rate": 2.6364258960219794e-06,
      "loss": 0.5011,
      "step": 847
    },
    {
      "epoch": 7.247863247863248,
      "grad_norm": 1.2195979451169634,
      "learning_rate": 2.631957144993068e-06,
      "loss": 0.3118,
      "step": 848
    },
    {
      "epoch": 7.256410256410256,
      "grad_norm": 1.0566516660523817,
      "learning_rate": 2.6274879711219816e-06,
      "loss": 0.4623,
      "step": 849
    },
    {
      "epoch": 7.264957264957265,
      "grad_norm": 1.0439581131472695,
      "learning_rate": 2.6230183887296955e-06,
      "loss": 0.5924,
      "step": 850
    },
    {
      "epoch": 7.273504273504273,
      "grad_norm": 1.14057396639862,
      "learning_rate": 2.6185484121384974e-06,
      "loss": 0.5444,
      "step": 851
    },
    {
      "epoch": 7.282051282051282,
      "grad_norm": 1.531318860119953,
      "learning_rate": 2.6140780556719354e-06,
      "loss": 0.2808,
      "step": 852
    },
    {
      "epoch": 7.2905982905982905,
      "grad_norm": 1.230993307344185,
      "learning_rate": 2.6096073336547757e-06,
      "loss": 0.2989,
      "step": 853
    },
    {
      "epoch": 7.299145299145299,
      "grad_norm": 1.2982927110560112,
      "learning_rate": 2.6051362604129553e-06,
      "loss": 0.3645,
      "step": 854
    },
    {
      "epoch": 7.3076923076923075,
      "grad_norm": 1.0710170660050595,
      "learning_rate": 2.6006648502735384e-06,
      "loss": 0.3189,
      "step": 855
    },
    {
      "epoch": 7.316239316239316,
      "grad_norm": 1.4129830299027597,
      "learning_rate": 2.5961931175646658e-06,
      "loss": 0.2819,
      "step": 856
    },
    {
      "epoch": 7.3247863247863245,
      "grad_norm": 1.1782095507296886,
      "learning_rate": 2.591721076615517e-06,
      "loss": 0.2548,
      "step": 857
    },
    {
      "epoch": 7.333333333333333,
      "grad_norm": 1.1081164425839658,
      "learning_rate": 2.587248741756253e-06,
      "loss": 0.413,
      "step": 858
    },
    {
      "epoch": 7.3418803418803416,
      "grad_norm": 1.0167105430488317,
      "learning_rate": 2.5827761273179795e-06,
      "loss": 0.4232,
      "step": 859
    },
    {
      "epoch": 7.35042735042735,
      "grad_norm": 1.9693966463338093,
      "learning_rate": 2.578303247632701e-06,
      "loss": 0.4598,
      "step": 860
    },
    {
      "epoch": 7.358974358974359,
      "grad_norm": 1.2184335947586225,
      "learning_rate": 2.5738301170332665e-06,
      "loss": 0.5022,
      "step": 861
    },
    {
      "epoch": 7.367521367521368,
      "grad_norm": 1.6767314445095651,
      "learning_rate": 2.5693567498533315e-06,
      "loss": 0.3089,
      "step": 862
    },
    {
      "epoch": 7.3760683760683765,
      "grad_norm": 1.1887353726846537,
      "learning_rate": 2.5648831604273117e-06,
      "loss": 0.3041,
      "step": 863
    },
    {
      "epoch": 7.384615384615385,
      "grad_norm": 1.257527921008963,
      "learning_rate": 2.560409363090331e-06,
      "loss": 0.218,
      "step": 864
    },
    {
      "epoch": 7.3931623931623935,
      "grad_norm": 1.5297790295923273,
      "learning_rate": 2.555935372178183e-06,
      "loss": 0.2525,
      "step": 865
    },
    {
      "epoch": 7.401709401709402,
      "grad_norm": 1.2253355789146119,
      "learning_rate": 2.5514612020272792e-06,
      "loss": 0.2754,
      "step": 866
    },
    {
      "epoch": 7.410256410256411,
      "grad_norm": 1.8586602173239468,
      "learning_rate": 2.546986866974606e-06,
      "loss": 0.3116,
      "step": 867
    },
    {
      "epoch": 7.418803418803419,
      "grad_norm": 1.3071314590457581,
      "learning_rate": 2.54251238135768e-06,
      "loss": 0.3596,
      "step": 868
    },
    {
      "epoch": 7.427350427350428,
      "grad_norm": 1.4114379208643046,
      "learning_rate": 2.5380377595144984e-06,
      "loss": 0.5204,
      "step": 869
    },
    {
      "epoch": 7.435897435897436,
      "grad_norm": 2.1954669593462475,
      "learning_rate": 2.533563015783494e-06,
      "loss": 0.4798,
      "step": 870
    },
    {
      "epoch": 7.444444444444445,
      "grad_norm": 1.0331522720681556,
      "learning_rate": 2.5290881645034932e-06,
      "loss": 0.4036,
      "step": 871
    },
    {
      "epoch": 7.452991452991453,
      "grad_norm": 1.512218877417084,
      "learning_rate": 2.524613220013664e-06,
      "loss": 0.2333,
      "step": 872
    },
    {
      "epoch": 7.461538461538462,
      "grad_norm": 1.0955389138492935,
      "learning_rate": 2.5201381966534748e-06,
      "loss": 0.6983,
      "step": 873
    },
    {
      "epoch": 7.47008547008547,
      "grad_norm": 1.4358219923803184,
      "learning_rate": 2.515663108762648e-06,
      "loss": 0.5395,
      "step": 874
    },
    {
      "epoch": 7.478632478632479,
      "grad_norm": 1.2250651733430429,
      "learning_rate": 2.511187970681109e-06,
      "loss": 0.3976,
      "step": 875
    },
    {
      "epoch": 7.487179487179487,
      "grad_norm": 1.2100214960869564,
      "learning_rate": 2.5067127967489464e-06,
      "loss": 0.3228,
      "step": 876
    },
    {
      "epoch": 7.495726495726496,
      "grad_norm": 1.03209917062225,
      "learning_rate": 2.5022376013063653e-06,
      "loss": 0.2953,
      "step": 877
    },
    {
      "epoch": 7.504273504273504,
      "grad_norm": 2.0442691089803264,
      "learning_rate": 2.497762398693636e-06,
      "loss": 0.3491,
      "step": 878
    },
    {
      "epoch": 7.512820512820513,
      "grad_norm": 1.1238804763641137,
      "learning_rate": 2.493287203251054e-06,
      "loss": 0.5455,
      "step": 879
    },
    {
      "epoch": 7.521367521367521,
      "grad_norm": 1.2347502797408516,
      "learning_rate": 2.4888120293188915e-06,
      "loss": 0.5476,
      "step": 880
    },
    {
      "epoch": 7.52991452991453,
      "grad_norm": 1.0667993198312395,
      "learning_rate": 2.484336891237353e-06,
      "loss": 0.4568,
      "step": 881
    },
    {
      "epoch": 7.538461538461538,
      "grad_norm": 1.1705030742580442,
      "learning_rate": 2.4798618033465256e-06,
      "loss": 0.3531,
      "step": 882
    },
    {
      "epoch": 7.547008547008547,
      "grad_norm": 1.4056506930020038,
      "learning_rate": 2.4753867799863365e-06,
      "loss": 0.262,
      "step": 883
    },
    {
      "epoch": 7.555555555555555,
      "grad_norm": 1.4336180176173927,
      "learning_rate": 2.470911835496508e-06,
      "loss": 0.3073,
      "step": 884
    },
    {
      "epoch": 7.564102564102564,
      "grad_norm": 1.2364038729002997,
      "learning_rate": 2.466436984216507e-06,
      "loss": 0.3548,
      "step": 885
    },
    {
      "epoch": 7.572649572649572,
      "grad_norm": 1.9618172986680815,
      "learning_rate": 2.4619622404855025e-06,
      "loss": 0.3784,
      "step": 886
    },
    {
      "epoch": 7.581196581196581,
      "grad_norm": 0.9993817480313798,
      "learning_rate": 2.4574876186423203e-06,
      "loss": 0.4865,
      "step": 887
    },
    {
      "epoch": 7.589743589743589,
      "grad_norm": 1.1318460373221935,
      "learning_rate": 2.4530131330253946e-06,
      "loss": 0.5475,
      "step": 888
    },
    {
      "epoch": 7.598290598290598,
      "grad_norm": 2.506285524005619,
      "learning_rate": 2.4485387979727216e-06,
      "loss": 0.3429,
      "step": 889
    },
    {
      "epoch": 7.6068376068376065,
      "grad_norm": 1.8244156537083576,
      "learning_rate": 2.4440646278218178e-06,
      "loss": 0.4488,
      "step": 890
    },
    {
      "epoch": 7.615384615384615,
      "grad_norm": 1.456426214018476,
      "learning_rate": 2.43959063690967e-06,
      "loss": 0.338,
      "step": 891
    },
    {
      "epoch": 7.6239316239316235,
      "grad_norm": 1.1028797644342125,
      "learning_rate": 2.435116839572689e-06,
      "loss": 0.4484,
      "step": 892
    },
    {
      "epoch": 7.632478632478632,
      "grad_norm": 1.4439295173359044,
      "learning_rate": 2.430643250146669e-06,
      "loss": 0.6095,
      "step": 893
    },
    {
      "epoch": 7.641025641025641,
      "grad_norm": 1.2732755672552745,
      "learning_rate": 2.426169882966735e-06,
      "loss": 0.5383,
      "step": 894
    },
    {
      "epoch": 7.64957264957265,
      "grad_norm": 1.0223416245596049,
      "learning_rate": 2.4216967523673e-06,
      "loss": 0.3725,
      "step": 895
    },
    {
      "epoch": 7.6581196581196584,
      "grad_norm": 1.1127930023622536,
      "learning_rate": 2.4172238726820205e-06,
      "loss": 0.1803,
      "step": 896
    },
    {
      "epoch": 7.666666666666667,
      "grad_norm": 1.187611997597461,
      "learning_rate": 2.4127512582437486e-06,
      "loss": 0.3982,
      "step": 897
    },
    {
      "epoch": 7.6752136752136755,
      "grad_norm": 1.3600435750463735,
      "learning_rate": 2.4082789233844837e-06,
      "loss": 0.4175,
      "step": 898
    },
    {
      "epoch": 7.683760683760684,
      "grad_norm": 1.3026323364977672,
      "learning_rate": 2.403806882435334e-06,
      "loss": 0.3755,
      "step": 899
    },
    {
      "epoch": 7.6923076923076925,
      "grad_norm": 1.1445423564052852,
      "learning_rate": 2.399335149726463e-06,
      "loss": 0.416,
      "step": 900
    },
    {
      "epoch": 7.700854700854701,
      "grad_norm": 1.453300314929189,
      "learning_rate": 2.394863739587045e-06,
      "loss": 0.2368,
      "step": 901
    },
    {
      "epoch": 7.7094017094017095,
      "grad_norm": 1.406547629276672,
      "learning_rate": 2.3903926663452255e-06,
      "loss": 0.3735,
      "step": 902
    },
    {
      "epoch": 7.717948717948718,
      "grad_norm": 1.4386212720763296,
      "learning_rate": 2.385921944328066e-06,
      "loss": 0.2898,
      "step": 903
    },
    {
      "epoch": 7.726495726495727,
      "grad_norm": 1.8255812573863117,
      "learning_rate": 2.3814515878615035e-06,
      "loss": 0.4494,
      "step": 904
    },
    {
      "epoch": 7.735042735042735,
      "grad_norm": 1.4340093814213342,
      "learning_rate": 2.376981611270305e-06,
      "loss": 0.4222,
      "step": 905
    },
    {
      "epoch": 7.743589743589744,
      "grad_norm": 1.3262164726354702,
      "learning_rate": 2.372512028878019e-06,
      "loss": 0.3349,
      "step": 906
    },
    {
      "epoch": 7.752136752136752,
      "grad_norm": 1.556934377276059,
      "learning_rate": 2.3680428550069327e-06,
      "loss": 0.2797,
      "step": 907
    },
    {
      "epoch": 7.760683760683761,
      "grad_norm": 1.1894029158833457,
      "learning_rate": 2.3635741039780214e-06,
      "loss": 0.3009,
      "step": 908
    },
    {
      "epoch": 7.769230769230769,
      "grad_norm": 1.1182938806964584,
      "learning_rate": 2.3591057901109063e-06,
      "loss": 0.4986,
      "step": 909
    },
    {
      "epoch": 7.777777777777778,
      "grad_norm": 1.0919187697035915,
      "learning_rate": 2.3546379277238107e-06,
      "loss": 0.4238,
      "step": 910
    },
    {
      "epoch": 7.786324786324786,
      "grad_norm": 1.4234512798748302,
      "learning_rate": 2.3501705311335067e-06,
      "loss": 0.3576,
      "step": 911
    },
    {
      "epoch": 7.794871794871795,
      "grad_norm": 1.345019755234573,
      "learning_rate": 2.3457036146552766e-06,
      "loss": 0.275,
      "step": 912
    },
    {
      "epoch": 7.803418803418803,
      "grad_norm": 1.2026943698155572,
      "learning_rate": 2.341237192602864e-06,
      "loss": 0.4876,
      "step": 913
    },
    {
      "epoch": 7.811965811965812,
      "grad_norm": 1.425767091915281,
      "learning_rate": 2.336771279288427e-06,
      "loss": 0.2407,
      "step": 914
    },
    {
      "epoch": 7.82051282051282,
      "grad_norm": 1.8197170307382808,
      "learning_rate": 2.332305889022494e-06,
      "loss": 0.2812,
      "step": 915
    },
    {
      "epoch": 7.829059829059829,
      "grad_norm": 2.0226814905047688,
      "learning_rate": 2.3278410361139198e-06,
      "loss": 0.4093,
      "step": 916
    },
    {
      "epoch": 7.837606837606837,
      "grad_norm": 1.5227436667459866,
      "learning_rate": 2.3233767348698314e-06,
      "loss": 0.3748,
      "step": 917
    },
    {
      "epoch": 7.846153846153846,
      "grad_norm": 0.9875096286316338,
      "learning_rate": 2.3189129995955944e-06,
      "loss": 0.3997,
      "step": 918
    },
    {
      "epoch": 7.854700854700854,
      "grad_norm": 1.5962327497251578,
      "learning_rate": 2.314449844594758e-06,
      "loss": 0.4961,
      "step": 919
    },
    {
      "epoch": 7.863247863247864,
      "grad_norm": 1.182953335931338,
      "learning_rate": 2.3099872841690105e-06,
      "loss": 0.4962,
      "step": 920
    },
    {
      "epoch": 7.871794871794872,
      "grad_norm": 1.638523718350465,
      "learning_rate": 2.305525332618138e-06,
      "loss": 0.4412,
      "step": 921
    },
    {
      "epoch": 7.880341880341881,
      "grad_norm": 1.4337420318321525,
      "learning_rate": 2.3010640042399748e-06,
      "loss": 0.43,
      "step": 922
    },
    {
      "epoch": 7.888888888888889,
      "grad_norm": 1.670604048799414,
      "learning_rate": 2.296603313330355e-06,
      "loss": 0.263,
      "step": 923
    },
    {
      "epoch": 7.897435897435898,
      "grad_norm": 1.1485379775019051,
      "learning_rate": 2.2921432741830744e-06,
      "loss": 0.4643,
      "step": 924
    },
    {
      "epoch": 7.905982905982906,
      "grad_norm": 2.867980041283883,
      "learning_rate": 2.2876839010898377e-06,
      "loss": 0.2376,
      "step": 925
    },
    {
      "epoch": 7.914529914529915,
      "grad_norm": 1.168055991443902,
      "learning_rate": 2.283225208340218e-06,
      "loss": 0.3508,
      "step": 926
    },
    {
      "epoch": 7.923076923076923,
      "grad_norm": 1.3338717397882187,
      "learning_rate": 2.2787672102216045e-06,
      "loss": 0.3976,
      "step": 927
    },
    {
      "epoch": 7.931623931623932,
      "grad_norm": 1.2019336080154466,
      "learning_rate": 2.2743099210191623e-06,
      "loss": 0.6231,
      "step": 928
    },
    {
      "epoch": 7.94017094017094,
      "grad_norm": 1.6402278408284279,
      "learning_rate": 2.2698533550157865e-06,
      "loss": 0.3036,
      "step": 929
    },
    {
      "epoch": 7.948717948717949,
      "grad_norm": 1.7450291076447808,
      "learning_rate": 2.265397526492052e-06,
      "loss": 0.3265,
      "step": 930
    },
    {
      "epoch": 7.957264957264957,
      "grad_norm": 1.2116360109877196,
      "learning_rate": 2.2609424497261723e-06,
      "loss": 0.3962,
      "step": 931
    },
    {
      "epoch": 7.965811965811966,
      "grad_norm": 1.504059791655357,
      "learning_rate": 2.2564881389939524e-06,
      "loss": 0.328,
      "step": 932
    },
    {
      "epoch": 7.9743589743589745,
      "grad_norm": 1.7027076220105453,
      "learning_rate": 2.25203460856874e-06,
      "loss": 0.2808,
      "step": 933
    },
    {
      "epoch": 7.982905982905983,
      "grad_norm": 1.3865934975804255,
      "learning_rate": 2.2475818727213843e-06,
      "loss": 0.2991,
      "step": 934
    },
    {
      "epoch": 7.9914529914529915,
      "grad_norm": 1.4673027031590777,
      "learning_rate": 2.24312994572019e-06,
      "loss": 0.3486,
      "step": 935
    },
    {
      "epoch": 8.0,
      "grad_norm": 1.049108221216376,
      "learning_rate": 2.238678841830867e-06,
      "loss": 0.4932,
      "step": 936
    },
    {
      "epoch": 8.008547008547009,
      "grad_norm": 1.3551983825674456,
      "learning_rate": 2.2342285753164876e-06,
      "loss": 0.2734,
      "step": 937
    },
    {
      "epoch": 8.017094017094017,
      "grad_norm": 1.4182004282586038,
      "learning_rate": 2.2297791604374443e-06,
      "loss": 0.3156,
      "step": 938
    },
    {
      "epoch": 8.025641025641026,
      "grad_norm": 1.0864965117814793,
      "learning_rate": 2.2253306114513964e-06,
      "loss": 0.2765,
      "step": 939
    },
    {
      "epoch": 8.034188034188034,
      "grad_norm": 1.2703882083327591,
      "learning_rate": 2.220882942613231e-06,
      "loss": 0.4824,
      "step": 940
    },
    {
      "epoch": 8.042735042735043,
      "grad_norm": 1.526912834087006,
      "learning_rate": 2.2164361681750148e-06,
      "loss": 0.3501,
      "step": 941
    },
    {
      "epoch": 8.051282051282051,
      "grad_norm": 2.4233778444054255,
      "learning_rate": 2.2119903023859475e-06,
      "loss": 0.2769,
      "step": 942
    },
    {
      "epoch": 8.05982905982906,
      "grad_norm": 1.4817064190089302,
      "learning_rate": 2.2075453594923175e-06,
      "loss": 0.564,
      "step": 943
    },
    {
      "epoch": 8.068376068376068,
      "grad_norm": 1.7198698064027493,
      "learning_rate": 2.2031013537374564e-06,
      "loss": 0.5413,
      "step": 944
    },
    {
      "epoch": 8.076923076923077,
      "grad_norm": 1.4476738121666382,
      "learning_rate": 2.1986582993616926e-06,
      "loss": 0.2316,
      "step": 945
    },
    {
      "epoch": 8.085470085470085,
      "grad_norm": 1.1074691461600432,
      "learning_rate": 2.194216210602305e-06,
      "loss": 0.3108,
      "step": 946
    },
    {
      "epoch": 8.094017094017094,
      "grad_norm": 1.834202942421801,
      "learning_rate": 2.1897751016934802e-06,
      "loss": 0.3146,
      "step": 947
    },
    {
      "epoch": 8.102564102564102,
      "grad_norm": 1.2978798214303746,
      "learning_rate": 2.1853349868662637e-06,
      "loss": 0.186,
      "step": 948
    },
    {
      "epoch": 8.11111111111111,
      "grad_norm": 1.3117945103185198,
      "learning_rate": 2.1808958803485134e-06,
      "loss": 0.3242,
      "step": 949
    },
    {
      "epoch": 8.11965811965812,
      "grad_norm": 1.6686535455884859,
      "learning_rate": 2.1764577963648616e-06,
      "loss": 0.4344,
      "step": 950
    },
    {
      "epoch": 8.128205128205128,
      "grad_norm": 1.3972878519691447,
      "learning_rate": 2.1720207491366598e-06,
      "loss": 0.3,
      "step": 951
    },
    {
      "epoch": 8.136752136752136,
      "grad_norm": 1.1625677465463464,
      "learning_rate": 2.167584752881937e-06,
      "loss": 0.2146,
      "step": 952
    },
    {
      "epoch": 8.145299145299145,
      "grad_norm": 1.601301969292571,
      "learning_rate": 2.163149821815358e-06,
      "loss": 0.564,
      "step": 953
    },
    {
      "epoch": 8.153846153846153,
      "grad_norm": 1.224853456198565,
      "learning_rate": 2.1587159701481718e-06,
      "loss": 0.5251,
      "step": 954
    },
    {
      "epoch": 8.162393162393162,
      "grad_norm": 1.357992600450483,
      "learning_rate": 2.154283212088168e-06,
      "loss": 0.4969,
      "step": 955
    },
    {
      "epoch": 8.17094017094017,
      "grad_norm": 1.426670407162958,
      "learning_rate": 2.1498515618396327e-06,
      "loss": 0.2752,
      "step": 956
    },
    {
      "epoch": 8.179487179487179,
      "grad_norm": 1.1704916649264339,
      "learning_rate": 2.145421033603304e-06,
      "loss": 0.1518,
      "step": 957
    },
    {
      "epoch": 8.188034188034187,
      "grad_norm": 1.8264402097576982,
      "learning_rate": 2.1409916415763216e-06,
      "loss": 0.2544,
      "step": 958
    },
    {
      "epoch": 8.196581196581196,
      "grad_norm": 1.4744665660554996,
      "learning_rate": 2.1365633999521852e-06,
      "loss": 0.302,
      "step": 959
    },
    {
      "epoch": 8.205128205128204,
      "grad_norm": 1.1587143039082641,
      "learning_rate": 2.1321363229207097e-06,
      "loss": 0.3356,
      "step": 960
    },
    {
      "epoch": 8.213675213675213,
      "grad_norm": 1.1394161124337572,
      "learning_rate": 2.127710424667978e-06,
      "loss": 0.3758,
      "step": 961
    },
    {
      "epoch": 8.222222222222221,
      "grad_norm": 1.61420409353586,
      "learning_rate": 2.1232857193762923e-06,
      "loss": 0.237,
      "step": 962
    },
    {
      "epoch": 8.23076923076923,
      "grad_norm": 1.3412922365080107,
      "learning_rate": 2.1188622212241366e-06,
      "loss": 0.2108,
      "step": 963
    },
    {
      "epoch": 8.239316239316238,
      "grad_norm": 1.220900622715855,
      "learning_rate": 2.114439944386125e-06,
      "loss": 0.4871,
      "step": 964
    },
    {
      "epoch": 8.247863247863247,
      "grad_norm": 1.1681317060272036,
      "learning_rate": 2.1100189030329557e-06,
      "loss": 0.4729,
      "step": 965
    },
    {
      "epoch": 8.256410256410255,
      "grad_norm": 2.148611373627357,
      "learning_rate": 2.105599111331372e-06,
      "loss": 0.2583,
      "step": 966
    },
    {
      "epoch": 8.264957264957266,
      "grad_norm": 1.4116678905851439,
      "learning_rate": 2.101180583444111e-06,
      "loss": 0.3968,
      "step": 967
    },
    {
      "epoch": 8.273504273504274,
      "grad_norm": 1.1484639753245225,
      "learning_rate": 2.0967633335298583e-06,
      "loss": 0.4482,
      "step": 968
    },
    {
      "epoch": 8.282051282051283,
      "grad_norm": 1.0727263766010298,
      "learning_rate": 2.0923473757432073e-06,
      "loss": 0.2581,
      "step": 969
    },
    {
      "epoch": 8.290598290598291,
      "grad_norm": 1.2886191668311129,
      "learning_rate": 2.0879327242346096e-06,
      "loss": 0.4631,
      "step": 970
    },
    {
      "epoch": 8.2991452991453,
      "grad_norm": 1.2055035364480693,
      "learning_rate": 2.0835193931503297e-06,
      "loss": 0.3323,
      "step": 971
    },
    {
      "epoch": 8.307692307692308,
      "grad_norm": 1.7152947572112152,
      "learning_rate": 2.079107396632404e-06,
      "loss": 0.2037,
      "step": 972
    },
    {
      "epoch": 8.316239316239317,
      "grad_norm": 1.1141021797533694,
      "learning_rate": 2.0746967488185903e-06,
      "loss": 0.5559,
      "step": 973
    },
    {
      "epoch": 8.324786324786325,
      "grad_norm": 1.1459993656130663,
      "learning_rate": 2.0702874638423233e-06,
      "loss": 0.3579,
      "step": 974
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 0.9871984116056022,
      "learning_rate": 2.0658795558326745e-06,
      "loss": 0.3504,
      "step": 975
    },
    {
      "epoch": 8.341880341880342,
      "grad_norm": 1.126408210461535,
      "learning_rate": 2.0614730389143004e-06,
      "loss": 0.2381,
      "step": 976
    },
    {
      "epoch": 8.350427350427351,
      "grad_norm": 1.5279937922363667,
      "learning_rate": 2.057067927207399e-06,
      "loss": 0.3458,
      "step": 977
    },
    {
      "epoch": 8.35897435897436,
      "grad_norm": 1.3740534569502523,
      "learning_rate": 2.052664234827668e-06,
      "loss": 0.454,
      "step": 978
    },
    {
      "epoch": 8.367521367521368,
      "grad_norm": 1.141519779977136,
      "learning_rate": 2.048261975886256e-06,
      "loss": 0.5811,
      "step": 979
    },
    {
      "epoch": 8.376068376068377,
      "grad_norm": 1.6726257371835274,
      "learning_rate": 2.0438611644897186e-06,
      "loss": 0.2511,
      "step": 980
    },
    {
      "epoch": 8.384615384615385,
      "grad_norm": 1.1118517949507838,
      "learning_rate": 2.0394618147399713e-06,
      "loss": 0.2559,
      "step": 981
    },
    {
      "epoch": 8.393162393162394,
      "grad_norm": 1.391439451118074,
      "learning_rate": 2.0350639407342474e-06,
      "loss": 0.2722,
      "step": 982
    },
    {
      "epoch": 8.401709401709402,
      "grad_norm": 1.481006180446899,
      "learning_rate": 2.030667556565052e-06,
      "loss": 0.2,
      "step": 983
    },
    {
      "epoch": 8.41025641025641,
      "grad_norm": 1.369017228519409,
      "learning_rate": 2.026272676320114e-06,
      "loss": 0.5728,
      "step": 984
    },
    {
      "epoch": 8.418803418803419,
      "grad_norm": 1.0477501912155225,
      "learning_rate": 2.021879314082344e-06,
      "loss": 0.267,
      "step": 985
    },
    {
      "epoch": 8.427350427350428,
      "grad_norm": 1.536820104698817,
      "learning_rate": 2.0174874839297912e-06,
      "loss": 0.3918,
      "step": 986
    },
    {
      "epoch": 8.435897435897436,
      "grad_norm": 1.270640833707206,
      "learning_rate": 2.01309719993559e-06,
      "loss": 0.4445,
      "step": 987
    },
    {
      "epoch": 8.444444444444445,
      "grad_norm": 1.3874319242758286,
      "learning_rate": 2.0087084761679245e-06,
      "loss": 0.364,
      "step": 988
    },
    {
      "epoch": 8.452991452991453,
      "grad_norm": 0.9453557375907675,
      "learning_rate": 2.0043213266899787e-06,
      "loss": 0.3354,
      "step": 989
    },
    {
      "epoch": 8.461538461538462,
      "grad_norm": 1.2460934799338634,
      "learning_rate": 1.9999357655598894e-06,
      "loss": 0.6679,
      "step": 990
    },
    {
      "epoch": 8.47008547008547,
      "grad_norm": 1.1321817356132304,
      "learning_rate": 1.995551806830706e-06,
      "loss": 0.3041,
      "step": 991
    },
    {
      "epoch": 8.478632478632479,
      "grad_norm": 1.535280683567388,
      "learning_rate": 1.9911694645503443e-06,
      "loss": 0.2659,
      "step": 992
    },
    {
      "epoch": 8.487179487179487,
      "grad_norm": 1.5817499556332901,
      "learning_rate": 1.986788752761536e-06,
      "loss": 0.4388,
      "step": 993
    },
    {
      "epoch": 8.495726495726496,
      "grad_norm": 1.63978371233733,
      "learning_rate": 1.9824096855017922e-06,
      "loss": 0.5096,
      "step": 994
    },
    {
      "epoch": 8.504273504273504,
      "grad_norm": 1.1861751601896062,
      "learning_rate": 1.978032276803354e-06,
      "loss": 0.4638,
      "step": 995
    },
    {
      "epoch": 8.512820512820513,
      "grad_norm": 1.2713512650546797,
      "learning_rate": 1.9736565406931446e-06,
      "loss": 0.3673,
      "step": 996
    },
    {
      "epoch": 8.521367521367521,
      "grad_norm": 1.1297283147346528,
      "learning_rate": 1.969282491192731e-06,
      "loss": 0.4807,
      "step": 997
    },
    {
      "epoch": 8.52991452991453,
      "grad_norm": 1.5692569320214733,
      "learning_rate": 1.9649101423182732e-06,
      "loss": 0.3172,
      "step": 998
    },
    {
      "epoch": 8.538461538461538,
      "grad_norm": 1.2481351983080389,
      "learning_rate": 1.960539508080485e-06,
      "loss": 0.4857,
      "step": 999
    },
    {
      "epoch": 8.547008547008547,
      "grad_norm": 1.1121259008864515,
      "learning_rate": 1.956170602484582e-06,
      "loss": 0.3605,
      "step": 1000
    },
    {
      "epoch": 8.555555555555555,
      "grad_norm": 1.0820395065466846,
      "learning_rate": 1.9518034395302413e-06,
      "loss": 0.5152,
      "step": 1001
    },
    {
      "epoch": 8.564102564102564,
      "grad_norm": 1.2170335969444013,
      "learning_rate": 1.94743803321156e-06,
      "loss": 0.1918,
      "step": 1002
    },
    {
      "epoch": 8.572649572649572,
      "grad_norm": 1.0768390804628318,
      "learning_rate": 1.9430743975170004e-06,
      "loss": 0.2212,
      "step": 1003
    },
    {
      "epoch": 8.581196581196581,
      "grad_norm": 1.1801978281526193,
      "learning_rate": 1.938712546429354e-06,
      "loss": 0.5013,
      "step": 1004
    },
    {
      "epoch": 8.58974358974359,
      "grad_norm": 1.2799936871315807,
      "learning_rate": 1.934352493925695e-06,
      "loss": 0.5044,
      "step": 1005
    },
    {
      "epoch": 8.598290598290598,
      "grad_norm": 1.1419776158740154,
      "learning_rate": 1.9299942539773316e-06,
      "loss": 0.2284,
      "step": 1006
    },
    {
      "epoch": 8.606837606837606,
      "grad_norm": 2.581226079905047,
      "learning_rate": 1.925637840549764e-06,
      "loss": 0.1372,
      "step": 1007
    },
    {
      "epoch": 8.615384615384615,
      "grad_norm": 1.4623611578834776,
      "learning_rate": 1.921283267602643e-06,
      "loss": 0.4478,
      "step": 1008
    },
    {
      "epoch": 8.623931623931623,
      "grad_norm": 2.0074150173839462,
      "learning_rate": 1.9169305490897173e-06,
      "loss": 0.3933,
      "step": 1009
    },
    {
      "epoch": 8.632478632478632,
      "grad_norm": 1.2589084565168156,
      "learning_rate": 1.9125796989587947e-06,
      "loss": 0.4317,
      "step": 1010
    },
    {
      "epoch": 8.64102564102564,
      "grad_norm": 1.1271801947834847,
      "learning_rate": 1.9082307311516985e-06,
      "loss": 0.4162,
      "step": 1011
    },
    {
      "epoch": 8.649572649572649,
      "grad_norm": 1.3583073699164292,
      "learning_rate": 1.9038836596042174e-06,
      "loss": 0.269,
      "step": 1012
    },
    {
      "epoch": 8.658119658119658,
      "grad_norm": 1.0956300687748248,
      "learning_rate": 1.8995384982460636e-06,
      "loss": 0.4542,
      "step": 1013
    },
    {
      "epoch": 8.666666666666666,
      "grad_norm": 1.411171017892716,
      "learning_rate": 1.895195261000831e-06,
      "loss": 0.2594,
      "step": 1014
    },
    {
      "epoch": 8.675213675213675,
      "grad_norm": 1.37127752697407,
      "learning_rate": 1.8908539617859455e-06,
      "loss": 0.2103,
      "step": 1015
    },
    {
      "epoch": 8.683760683760683,
      "grad_norm": 1.4011308842889736,
      "learning_rate": 1.8865146145126228e-06,
      "loss": 0.3916,
      "step": 1016
    },
    {
      "epoch": 8.692307692307692,
      "grad_norm": 1.784784106305036,
      "learning_rate": 1.8821772330858259e-06,
      "loss": 0.2464,
      "step": 1017
    },
    {
      "epoch": 8.7008547008547,
      "grad_norm": 1.4568451060436804,
      "learning_rate": 1.877841831404217e-06,
      "loss": 0.2571,
      "step": 1018
    },
    {
      "epoch": 8.709401709401709,
      "grad_norm": 1.2541641895523783,
      "learning_rate": 1.873508423360113e-06,
      "loss": 0.3484,
      "step": 1019
    },
    {
      "epoch": 8.717948717948717,
      "grad_norm": 1.790760709218908,
      "learning_rate": 1.8691770228394458e-06,
      "loss": 0.6167,
      "step": 1020
    },
    {
      "epoch": 8.726495726495726,
      "grad_norm": 1.341451765077332,
      "learning_rate": 1.8648476437217117e-06,
      "loss": 0.3222,
      "step": 1021
    },
    {
      "epoch": 8.735042735042736,
      "grad_norm": 1.0739656070864179,
      "learning_rate": 1.8605202998799299e-06,
      "loss": 0.3204,
      "step": 1022
    },
    {
      "epoch": 8.743589743589745,
      "grad_norm": 1.017114133042732,
      "learning_rate": 1.8561950051805994e-06,
      "loss": 0.4186,
      "step": 1023
    },
    {
      "epoch": 8.752136752136753,
      "grad_norm": 1.107854491656353,
      "learning_rate": 1.8518717734836522e-06,
      "loss": 0.4001,
      "step": 1024
    },
    {
      "epoch": 8.760683760683762,
      "grad_norm": 1.2546206655532879,
      "learning_rate": 1.8475506186424075e-06,
      "loss": 0.495,
      "step": 1025
    },
    {
      "epoch": 8.76923076923077,
      "grad_norm": 1.3154930449197013,
      "learning_rate": 1.8432315545035328e-06,
      "loss": 0.2821,
      "step": 1026
    },
    {
      "epoch": 8.777777777777779,
      "grad_norm": 1.4975866241382854,
      "learning_rate": 1.8389145949069953e-06,
      "loss": 0.3722,
      "step": 1027
    },
    {
      "epoch": 8.786324786324787,
      "grad_norm": 1.2831039302902338,
      "learning_rate": 1.8345997536860154e-06,
      "loss": 0.3783,
      "step": 1028
    },
    {
      "epoch": 8.794871794871796,
      "grad_norm": 1.1980975018106128,
      "learning_rate": 1.83028704466703e-06,
      "loss": 0.2603,
      "step": 1029
    },
    {
      "epoch": 8.803418803418804,
      "grad_norm": 2.209593515869201,
      "learning_rate": 1.8259764816696413e-06,
      "loss": 0.3781,
      "step": 1030
    },
    {
      "epoch": 8.811965811965813,
      "grad_norm": 1.2656408207095478,
      "learning_rate": 1.8216680785065734e-06,
      "loss": 0.4635,
      "step": 1031
    },
    {
      "epoch": 8.820512820512821,
      "grad_norm": 2.1089565512836748,
      "learning_rate": 1.8173618489836315e-06,
      "loss": 0.2241,
      "step": 1032
    },
    {
      "epoch": 8.82905982905983,
      "grad_norm": 1.2447903375211662,
      "learning_rate": 1.813057806899656e-06,
      "loss": 0.4539,
      "step": 1033
    },
    {
      "epoch": 8.837606837606838,
      "grad_norm": 1.0153965458845877,
      "learning_rate": 1.8087559660464766e-06,
      "loss": 0.4477,
      "step": 1034
    },
    {
      "epoch": 8.846153846153847,
      "grad_norm": 3.9606458120630346,
      "learning_rate": 1.8044563402088686e-06,
      "loss": 0.3163,
      "step": 1035
    },
    {
      "epoch": 8.854700854700855,
      "grad_norm": 1.7429909204407186,
      "learning_rate": 1.800158943164512e-06,
      "loss": 0.3911,
      "step": 1036
    },
    {
      "epoch": 8.863247863247864,
      "grad_norm": 1.509325256391988,
      "learning_rate": 1.7958637886839437e-06,
      "loss": 0.1908,
      "step": 1037
    },
    {
      "epoch": 8.871794871794872,
      "grad_norm": 1.3551384998854423,
      "learning_rate": 1.7915708905305124e-06,
      "loss": 0.4373,
      "step": 1038
    },
    {
      "epoch": 8.88034188034188,
      "grad_norm": 1.215439008720899,
      "learning_rate": 1.7872802624603408e-06,
      "loss": 0.4194,
      "step": 1039
    },
    {
      "epoch": 8.88888888888889,
      "grad_norm": 2.209864557912605,
      "learning_rate": 1.7829919182222752e-06,
      "loss": 0.3788,
      "step": 1040
    },
    {
      "epoch": 8.897435897435898,
      "grad_norm": 1.626332495031749,
      "learning_rate": 1.7787058715578415e-06,
      "loss": 0.4146,
      "step": 1041
    },
    {
      "epoch": 8.905982905982906,
      "grad_norm": 1.1918691292663193,
      "learning_rate": 1.7744221362012075e-06,
      "loss": 0.3042,
      "step": 1042
    },
    {
      "epoch": 8.914529914529915,
      "grad_norm": 1.2146958818698328,
      "learning_rate": 1.7701407258791323e-06,
      "loss": 0.4153,
      "step": 1043
    },
    {
      "epoch": 8.923076923076923,
      "grad_norm": 1.0513765409468407,
      "learning_rate": 1.7658616543109237e-06,
      "loss": 0.3588,
      "step": 1044
    },
    {
      "epoch": 8.931623931623932,
      "grad_norm": 1.3919842380301217,
      "learning_rate": 1.7615849352083975e-06,
      "loss": 0.4623,
      "step": 1045
    },
    {
      "epoch": 8.94017094017094,
      "grad_norm": 1.6990923788284649,
      "learning_rate": 1.7573105822758307e-06,
      "loss": 0.1482,
      "step": 1046
    },
    {
      "epoch": 8.948717948717949,
      "grad_norm": 1.5913449890668718,
      "learning_rate": 1.753038609209916e-06,
      "loss": 0.3842,
      "step": 1047
    },
    {
      "epoch": 8.957264957264957,
      "grad_norm": 1.3328131098581075,
      "learning_rate": 1.7487690296997234e-06,
      "loss": 0.3021,
      "step": 1048
    },
    {
      "epoch": 8.965811965811966,
      "grad_norm": 1.6596224898436513,
      "learning_rate": 1.7445018574266514e-06,
      "loss": 0.2747,
      "step": 1049
    },
    {
      "epoch": 8.974358974358974,
      "grad_norm": 1.4417679558155323,
      "learning_rate": 1.740237106064383e-06,
      "loss": 0.3087,
      "step": 1050
    },
    {
      "epoch": 8.982905982905983,
      "grad_norm": 1.2001346733735734,
      "learning_rate": 1.7359747892788476e-06,
      "loss": 0.4016,
      "step": 1051
    },
    {
      "epoch": 8.991452991452991,
      "grad_norm": 1.1721123567379175,
      "learning_rate": 1.7317149207281697e-06,
      "loss": 0.202,
      "step": 1052
    },
    {
      "epoch": 9.0,
      "grad_norm": 1.2206194854441548,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 0.5314,
      "step": 1053
    },
    {
      "epoch": 9.008547008547009,
      "grad_norm": 1.9473535963527984,
      "learning_rate": 1.723202582924624e-06,
      "loss": 0.3228,
      "step": 1054
    },
    {
      "epoch": 9.017094017094017,
      "grad_norm": 1.288253715894709,
      "learning_rate": 1.7189501409486061e-06,
      "loss": 0.3262,
      "step": 1055
    },
    {
      "epoch": 9.025641025641026,
      "grad_norm": 1.015888824030439,
      "learning_rate": 1.7147002017610626e-06,
      "loss": 0.3223,
      "step": 1056
    },
    {
      "epoch": 9.034188034188034,
      "grad_norm": 1.1877903780566137,
      "learning_rate": 1.7104527789804554e-06,
      "loss": 0.397,
      "step": 1057
    },
    {
      "epoch": 9.042735042735043,
      "grad_norm": 1.1021032536214928,
      "learning_rate": 1.7062078862171838e-06,
      "loss": 0.2936,
      "step": 1058
    },
    {
      "epoch": 9.051282051282051,
      "grad_norm": 0.9267012407122549,
      "learning_rate": 1.7019655370735428e-06,
      "loss": 0.3594,
      "step": 1059
    },
    {
      "epoch": 9.05982905982906,
      "grad_norm": 1.42199614285682,
      "learning_rate": 1.6977257451436714e-06,
      "loss": 0.5004,
      "step": 1060
    },
    {
      "epoch": 9.068376068376068,
      "grad_norm": 1.5329574926292904,
      "learning_rate": 1.6934885240135179e-06,
      "loss": 0.2812,
      "step": 1061
    },
    {
      "epoch": 9.076923076923077,
      "grad_norm": 1.9955114217965897,
      "learning_rate": 1.6892538872607936e-06,
      "loss": 0.2585,
      "step": 1062
    },
    {
      "epoch": 9.085470085470085,
      "grad_norm": 1.2354971874595881,
      "learning_rate": 1.6850218484549247e-06,
      "loss": 0.3017,
      "step": 1063
    },
    {
      "epoch": 9.094017094017094,
      "grad_norm": 1.2906710359133946,
      "learning_rate": 1.6807924211570151e-06,
      "loss": 0.5348,
      "step": 1064
    },
    {
      "epoch": 9.102564102564102,
      "grad_norm": 1.3827715724003555,
      "learning_rate": 1.6765656189198013e-06,
      "loss": 0.3824,
      "step": 1065
    },
    {
      "epoch": 9.11111111111111,
      "grad_norm": 1.7857256227411515,
      "learning_rate": 1.6723414552876052e-06,
      "loss": 0.1725,
      "step": 1066
    },
    {
      "epoch": 9.11965811965812,
      "grad_norm": 1.4869088752229858,
      "learning_rate": 1.6681199437962952e-06,
      "loss": 0.2514,
      "step": 1067
    },
    {
      "epoch": 9.128205128205128,
      "grad_norm": 1.2481499309084931,
      "learning_rate": 1.663901097973243e-06,
      "loss": 0.5908,
      "step": 1068
    },
    {
      "epoch": 9.136752136752136,
      "grad_norm": 1.1671070517620592,
      "learning_rate": 1.6596849313372737e-06,
      "loss": 0.3567,
      "step": 1069
    },
    {
      "epoch": 9.145299145299145,
      "grad_norm": 1.3165208006094509,
      "learning_rate": 1.6554714573986325e-06,
      "loss": 0.4676,
      "step": 1070
    },
    {
      "epoch": 9.153846153846153,
      "grad_norm": 1.5518295391328731,
      "learning_rate": 1.6512606896589323e-06,
      "loss": 0.5293,
      "step": 1071
    },
    {
      "epoch": 9.162393162393162,
      "grad_norm": 1.131322170174984,
      "learning_rate": 1.647052641611117e-06,
      "loss": 0.5835,
      "step": 1072
    },
    {
      "epoch": 9.17094017094017,
      "grad_norm": 1.3950148174594257,
      "learning_rate": 1.6428473267394135e-06,
      "loss": 0.3448,
      "step": 1073
    },
    {
      "epoch": 9.179487179487179,
      "grad_norm": 1.0946896683070888,
      "learning_rate": 1.6386447585192911e-06,
      "loss": 0.3427,
      "step": 1074
    },
    {
      "epoch": 9.188034188034187,
      "grad_norm": 1.3179053377437275,
      "learning_rate": 1.6344449504174193e-06,
      "loss": 0.361,
      "step": 1075
    },
    {
      "epoch": 9.196581196581196,
      "grad_norm": 1.294239250413899,
      "learning_rate": 1.63024791589162e-06,
      "loss": 0.2773,
      "step": 1076
    },
    {
      "epoch": 9.205128205128204,
      "grad_norm": 2.464763503764214,
      "learning_rate": 1.6260536683908302e-06,
      "loss": 0.3248,
      "step": 1077
    },
    {
      "epoch": 9.213675213675213,
      "grad_norm": 1.6874344831226797,
      "learning_rate": 1.621862221355056e-06,
      "loss": 0.3882,
      "step": 1078
    },
    {
      "epoch": 9.222222222222221,
      "grad_norm": 1.4390246409930711,
      "learning_rate": 1.6176735882153284e-06,
      "loss": 0.4268,
      "step": 1079
    },
    {
      "epoch": 9.23076923076923,
      "grad_norm": 1.198697926754215,
      "learning_rate": 1.613487782393661e-06,
      "loss": 0.3663,
      "step": 1080
    },
    {
      "epoch": 9.239316239316238,
      "grad_norm": 1.8995745824605574,
      "learning_rate": 1.6093048173030108e-06,
      "loss": 0.2581,
      "step": 1081
    },
    {
      "epoch": 9.247863247863247,
      "grad_norm": 1.0568489463788437,
      "learning_rate": 1.6051247063472286e-06,
      "loss": 0.3077,
      "step": 1082
    },
    {
      "epoch": 9.256410256410255,
      "grad_norm": 1.179620148126396,
      "learning_rate": 1.6009474629210202e-06,
      "loss": 0.3039,
      "step": 1083
    },
    {
      "epoch": 9.264957264957266,
      "grad_norm": 2.3385119254410247,
      "learning_rate": 1.5967731004099057e-06,
      "loss": 0.2822,
      "step": 1084
    },
    {
      "epoch": 9.273504273504274,
      "grad_norm": 1.4098002122577362,
      "learning_rate": 1.5926016321901688e-06,
      "loss": 0.2923,
      "step": 1085
    },
    {
      "epoch": 9.282051282051283,
      "grad_norm": 1.2847130932530688,
      "learning_rate": 1.5884330716288215e-06,
      "loss": 0.3617,
      "step": 1086
    },
    {
      "epoch": 9.290598290598291,
      "grad_norm": 1.1205572597044093,
      "learning_rate": 1.5842674320835598e-06,
      "loss": 0.1306,
      "step": 1087
    },
    {
      "epoch": 9.2991452991453,
      "grad_norm": 1.067879852287505,
      "learning_rate": 1.5801047269027164e-06,
      "loss": 0.2681,
      "step": 1088
    },
    {
      "epoch": 9.307692307692308,
      "grad_norm": 1.2697754842885247,
      "learning_rate": 1.5759449694252226e-06,
      "loss": 0.3835,
      "step": 1089
    },
    {
      "epoch": 9.316239316239317,
      "grad_norm": 1.115652242318011,
      "learning_rate": 1.571788172980566e-06,
      "loss": 0.3174,
      "step": 1090
    },
    {
      "epoch": 9.324786324786325,
      "grad_norm": 1.2589435696480922,
      "learning_rate": 1.567634350888743e-06,
      "loss": 0.2131,
      "step": 1091
    },
    {
      "epoch": 9.333333333333334,
      "grad_norm": 1.1703746042658,
      "learning_rate": 1.56348351646022e-06,
      "loss": 0.2466,
      "step": 1092
    },
    {
      "epoch": 9.341880341880342,
      "grad_norm": 1.8514476531000787,
      "learning_rate": 1.5593356829958906e-06,
      "loss": 0.4028,
      "step": 1093
    },
    {
      "epoch": 9.350427350427351,
      "grad_norm": 1.830700341859361,
      "learning_rate": 1.5551908637870316e-06,
      "loss": 0.424,
      "step": 1094
    },
    {
      "epoch": 9.35897435897436,
      "grad_norm": 1.2740235245585687,
      "learning_rate": 1.5510490721152594e-06,
      "loss": 0.3676,
      "step": 1095
    },
    {
      "epoch": 9.367521367521368,
      "grad_norm": 1.8244846681767757,
      "learning_rate": 1.5469103212524917e-06,
      "loss": 0.2519,
      "step": 1096
    },
    {
      "epoch": 9.376068376068377,
      "grad_norm": 1.1653789385736504,
      "learning_rate": 1.5427746244609015e-06,
      "loss": 0.4327,
      "step": 1097
    },
    {
      "epoch": 9.384615384615385,
      "grad_norm": 1.6539684322688117,
      "learning_rate": 1.5386419949928732e-06,
      "loss": 0.4085,
      "step": 1098
    },
    {
      "epoch": 9.393162393162394,
      "grad_norm": 1.1085990250805047,
      "learning_rate": 1.534512446090965e-06,
      "loss": 0.0987,
      "step": 1099
    },
    {
      "epoch": 9.401709401709402,
      "grad_norm": 1.1240053164248822,
      "learning_rate": 1.5303859909878632e-06,
      "loss": 0.2357,
      "step": 1100
    },
    {
      "epoch": 9.41025641025641,
      "grad_norm": 1.4361555001257007,
      "learning_rate": 1.5262626429063385e-06,
      "loss": 0.2324,
      "step": 1101
    },
    {
      "epoch": 9.418803418803419,
      "grad_norm": 1.4567991642591915,
      "learning_rate": 1.5221424150592078e-06,
      "loss": 0.1904,
      "step": 1102
    },
    {
      "epoch": 9.427350427350428,
      "grad_norm": 0.9078527832107429,
      "learning_rate": 1.518025320649289e-06,
      "loss": 0.3208,
      "step": 1103
    },
    {
      "epoch": 9.435897435897436,
      "grad_norm": 0.961997474880962,
      "learning_rate": 1.5139113728693575e-06,
      "loss": 0.2273,
      "step": 1104
    },
    {
      "epoch": 9.444444444444445,
      "grad_norm": 1.2170940431358712,
      "learning_rate": 1.509800584902108e-06,
      "loss": 0.1865,
      "step": 1105
    },
    {
      "epoch": 9.452991452991453,
      "grad_norm": 1.0896252626201706,
      "learning_rate": 1.5056929699201095e-06,
      "loss": 0.3603,
      "step": 1106
    },
    {
      "epoch": 9.461538461538462,
      "grad_norm": 1.2969179051429869,
      "learning_rate": 1.5015885410857617e-06,
      "loss": 0.2665,
      "step": 1107
    },
    {
      "epoch": 9.47008547008547,
      "grad_norm": 1.3077008732044582,
      "learning_rate": 1.4974873115512561e-06,
      "loss": 0.2858,
      "step": 1108
    },
    {
      "epoch": 9.478632478632479,
      "grad_norm": 1.5312364963489444,
      "learning_rate": 1.4933892944585331e-06,
      "loss": 0.3289,
      "step": 1109
    },
    {
      "epoch": 9.487179487179487,
      "grad_norm": 1.4001962940256754,
      "learning_rate": 1.489294502939238e-06,
      "loss": 0.1762,
      "step": 1110
    },
    {
      "epoch": 9.495726495726496,
      "grad_norm": 1.6856080795603618,
      "learning_rate": 1.4852029501146797e-06,
      "loss": 0.2352,
      "step": 1111
    },
    {
      "epoch": 9.504273504273504,
      "grad_norm": 1.7737389148061027,
      "learning_rate": 1.4811146490957903e-06,
      "loss": 0.4691,
      "step": 1112
    },
    {
      "epoch": 9.512820512820513,
      "grad_norm": 1.0477925330006121,
      "learning_rate": 1.477029612983082e-06,
      "loss": 0.3916,
      "step": 1113
    },
    {
      "epoch": 9.521367521367521,
      "grad_norm": 1.1845005270835856,
      "learning_rate": 1.4729478548666027e-06,
      "loss": 0.2385,
      "step": 1114
    },
    {
      "epoch": 9.52991452991453,
      "grad_norm": 1.1819763888213934,
      "learning_rate": 1.468869387825899e-06,
      "loss": 0.5128,
      "step": 1115
    },
    {
      "epoch": 9.538461538461538,
      "grad_norm": 1.0536988938300638,
      "learning_rate": 1.4647942249299708e-06,
      "loss": 0.2344,
      "step": 1116
    },
    {
      "epoch": 9.547008547008547,
      "grad_norm": 1.9011902118681425,
      "learning_rate": 1.4607223792372283e-06,
      "loss": 0.1493,
      "step": 1117
    },
    {
      "epoch": 9.555555555555555,
      "grad_norm": 1.3143466965397048,
      "learning_rate": 1.4566538637954556e-06,
      "loss": 0.6493,
      "step": 1118
    },
    {
      "epoch": 9.564102564102564,
      "grad_norm": 1.6921258588793364,
      "learning_rate": 1.452588691641763e-06,
      "loss": 0.5337,
      "step": 1119
    },
    {
      "epoch": 9.572649572649572,
      "grad_norm": 1.1271554002907007,
      "learning_rate": 1.4485268758025467e-06,
      "loss": 0.585,
      "step": 1120
    },
    {
      "epoch": 9.581196581196581,
      "grad_norm": 1.6575284887466888,
      "learning_rate": 1.4444684292934508e-06,
      "loss": 0.3152,
      "step": 1121
    },
    {
      "epoch": 9.58974358974359,
      "grad_norm": 1.1730847533193425,
      "learning_rate": 1.4404133651193214e-06,
      "loss": 0.475,
      "step": 1122
    },
    {
      "epoch": 9.598290598290598,
      "grad_norm": 1.1143229523313662,
      "learning_rate": 1.436361696274166e-06,
      "loss": 0.3746,
      "step": 1123
    },
    {
      "epoch": 9.606837606837606,
      "grad_norm": 1.4579201342685206,
      "learning_rate": 1.4323134357411114e-06,
      "loss": 0.2438,
      "step": 1124
    },
    {
      "epoch": 9.615384615384615,
      "grad_norm": 1.224526950260683,
      "learning_rate": 1.4282685964923643e-06,
      "loss": 0.2409,
      "step": 1125
    },
    {
      "epoch": 9.623931623931623,
      "grad_norm": 1.9183943687319494,
      "learning_rate": 1.4242271914891688e-06,
      "loss": 0.2619,
      "step": 1126
    },
    {
      "epoch": 9.632478632478632,
      "grad_norm": 1.448929760659896,
      "learning_rate": 1.4201892336817616e-06,
      "loss": 0.3426,
      "step": 1127
    },
    {
      "epoch": 9.64102564102564,
      "grad_norm": 1.2013122019818248,
      "learning_rate": 1.4161547360093364e-06,
      "loss": 0.4859,
      "step": 1128
    },
    {
      "epoch": 9.649572649572649,
      "grad_norm": 1.0596559301713948,
      "learning_rate": 1.4121237113999975e-06,
      "loss": 0.3119,
      "step": 1129
    },
    {
      "epoch": 9.658119658119658,
      "grad_norm": 1.4855292910426476,
      "learning_rate": 1.4080961727707185e-06,
      "loss": 0.2595,
      "step": 1130
    },
    {
      "epoch": 9.666666666666666,
      "grad_norm": 1.2478839118441194,
      "learning_rate": 1.4040721330273063e-06,
      "loss": 0.3786,
      "step": 1131
    },
    {
      "epoch": 9.675213675213675,
      "grad_norm": 3.9119278747334416,
      "learning_rate": 1.4000516050643549e-06,
      "loss": 0.2005,
      "step": 1132
    },
    {
      "epoch": 9.683760683760683,
      "grad_norm": 1.2618989022566294,
      "learning_rate": 1.3960346017652027e-06,
      "loss": 0.2733,
      "step": 1133
    },
    {
      "epoch": 9.692307692307692,
      "grad_norm": 1.1495751469091657,
      "learning_rate": 1.3920211360018971e-06,
      "loss": 0.3475,
      "step": 1134
    },
    {
      "epoch": 9.7008547008547,
      "grad_norm": 1.154458217155745,
      "learning_rate": 1.3880112206351476e-06,
      "loss": 0.4554,
      "step": 1135
    },
    {
      "epoch": 9.709401709401709,
      "grad_norm": 1.4267903377941606,
      "learning_rate": 1.3840048685142863e-06,
      "loss": 0.2523,
      "step": 1136
    },
    {
      "epoch": 9.717948717948717,
      "grad_norm": 1.2324200322804886,
      "learning_rate": 1.3800020924772295e-06,
      "loss": 0.3087,
      "step": 1137
    },
    {
      "epoch": 9.726495726495726,
      "grad_norm": 2.119358735727275,
      "learning_rate": 1.3760029053504346e-06,
      "loss": 0.2448,
      "step": 1138
    },
    {
      "epoch": 9.735042735042736,
      "grad_norm": 1.7034025319599941,
      "learning_rate": 1.372007319948855e-06,
      "loss": 0.4434,
      "step": 1139
    },
    {
      "epoch": 9.743589743589745,
      "grad_norm": 1.4386385418392187,
      "learning_rate": 1.3680153490759074e-06,
      "loss": 0.1699,
      "step": 1140
    },
    {
      "epoch": 9.752136752136753,
      "grad_norm": 1.82898779224831,
      "learning_rate": 1.3640270055234227e-06,
      "loss": 0.3602,
      "step": 1141
    },
    {
      "epoch": 9.760683760683762,
      "grad_norm": 1.5364042446197623,
      "learning_rate": 1.360042302071609e-06,
      "loss": 0.3554,
      "step": 1142
    },
    {
      "epoch": 9.76923076923077,
      "grad_norm": 1.2751685256780956,
      "learning_rate": 1.356061251489012e-06,
      "loss": 0.5297,
      "step": 1143
    },
    {
      "epoch": 9.777777777777779,
      "grad_norm": 1.7282120128773697,
      "learning_rate": 1.3520838665324704e-06,
      "loss": 0.2374,
      "step": 1144
    },
    {
      "epoch": 9.786324786324787,
      "grad_norm": 1.3425257816234517,
      "learning_rate": 1.3481101599470794e-06,
      "loss": 0.2757,
      "step": 1145
    },
    {
      "epoch": 9.794871794871796,
      "grad_norm": 1.2699459087777707,
      "learning_rate": 1.3441401444661416e-06,
      "loss": 0.3513,
      "step": 1146
    },
    {
      "epoch": 9.803418803418804,
      "grad_norm": 1.1393292998903113,
      "learning_rate": 1.3401738328111374e-06,
      "loss": 0.2073,
      "step": 1147
    },
    {
      "epoch": 9.811965811965813,
      "grad_norm": 1.1088535968070063,
      "learning_rate": 1.336211237691678e-06,
      "loss": 0.235,
      "step": 1148
    },
    {
      "epoch": 9.820512820512821,
      "grad_norm": 1.0270552038970258,
      "learning_rate": 1.3322523718054615e-06,
      "loss": 0.3529,
      "step": 1149
    },
    {
      "epoch": 9.82905982905983,
      "grad_norm": 1.5610839594093011,
      "learning_rate": 1.328297247838241e-06,
      "loss": 0.4928,
      "step": 1150
    },
    {
      "epoch": 9.837606837606838,
      "grad_norm": 1.292327740738013,
      "learning_rate": 1.3243458784637763e-06,
      "loss": 0.4766,
      "step": 1151
    },
    {
      "epoch": 9.846153846153847,
      "grad_norm": 1.7880250892242995,
      "learning_rate": 1.320398276343795e-06,
      "loss": 0.3798,
      "step": 1152
    },
    {
      "epoch": 9.854700854700855,
      "grad_norm": 1.154221527037192,
      "learning_rate": 1.3164544541279555e-06,
      "loss": 0.4733,
      "step": 1153
    },
    {
      "epoch": 9.863247863247864,
      "grad_norm": 1.4767581712130635,
      "learning_rate": 1.3125144244538038e-06,
      "loss": 0.3272,
      "step": 1154
    },
    {
      "epoch": 9.871794871794872,
      "grad_norm": 1.1375141309454042,
      "learning_rate": 1.3085781999467303e-06,
      "loss": 0.1772,
      "step": 1155
    },
    {
      "epoch": 9.88034188034188,
      "grad_norm": 1.357707685043935,
      "learning_rate": 1.304645793219936e-06,
      "loss": 0.4535,
      "step": 1156
    },
    {
      "epoch": 9.88888888888889,
      "grad_norm": 1.8407787280927563,
      "learning_rate": 1.3007172168743854e-06,
      "loss": 0.2293,
      "step": 1157
    },
    {
      "epoch": 9.897435897435898,
      "grad_norm": 1.2944703737109389,
      "learning_rate": 1.2967924834987687e-06,
      "loss": 0.449,
      "step": 1158
    },
    {
      "epoch": 9.905982905982906,
      "grad_norm": 1.361304325226804,
      "learning_rate": 1.2928716056694637e-06,
      "loss": 0.238,
      "step": 1159
    },
    {
      "epoch": 9.914529914529915,
      "grad_norm": 1.2097243748345254,
      "learning_rate": 1.288954595950494e-06,
      "loss": 0.4441,
      "step": 1160
    },
    {
      "epoch": 9.923076923076923,
      "grad_norm": 1.4844650371837154,
      "learning_rate": 1.285041466893485e-06,
      "loss": 0.317,
      "step": 1161
    },
    {
      "epoch": 9.931623931623932,
      "grad_norm": 1.173403562434299,
      "learning_rate": 1.2811322310376303e-06,
      "loss": 0.3963,
      "step": 1162
    },
    {
      "epoch": 9.94017094017094,
      "grad_norm": 1.2779391510526092,
      "learning_rate": 1.2772269009096456e-06,
      "loss": 0.5377,
      "step": 1163
    },
    {
      "epoch": 9.948717948717949,
      "grad_norm": 1.224475576365551,
      "learning_rate": 1.2733254890237334e-06,
      "loss": 0.2121,
      "step": 1164
    },
    {
      "epoch": 9.957264957264957,
      "grad_norm": 1.4870462629125007,
      "learning_rate": 1.2694280078815382e-06,
      "loss": 0.3333,
      "step": 1165
    },
    {
      "epoch": 9.965811965811966,
      "grad_norm": 1.123284137727133,
      "learning_rate": 1.2655344699721111e-06,
      "loss": 0.1703,
      "step": 1166
    },
    {
      "epoch": 9.974358974358974,
      "grad_norm": 1.4953977023064586,
      "learning_rate": 1.2616448877718672e-06,
      "loss": 0.2374,
      "step": 1167
    },
    {
      "epoch": 9.982905982905983,
      "grad_norm": 1.322352923347006,
      "learning_rate": 1.257759273744545e-06,
      "loss": 0.497,
      "step": 1168
    },
    {
      "epoch": 9.991452991452991,
      "grad_norm": 1.3765362036635373,
      "learning_rate": 1.253877640341166e-06,
      "loss": 0.4128,
      "step": 1169
    },
    {
      "epoch": 10.0,
      "grad_norm": 1.2816541186687818,
      "learning_rate": 1.2500000000000007e-06,
      "loss": 0.1786,
      "step": 1170
    },
    {
      "epoch": 10.008547008547009,
      "grad_norm": 1.045651825933841,
      "learning_rate": 1.2461263651465194e-06,
      "loss": 0.4029,
      "step": 1171
    },
    {
      "epoch": 10.017094017094017,
      "grad_norm": 1.2489254469634197,
      "learning_rate": 1.2422567481933604e-06,
      "loss": 0.2414,
      "step": 1172
    },
    {
      "epoch": 10.025641025641026,
      "grad_norm": 1.358095126324313,
      "learning_rate": 1.2383911615402873e-06,
      "loss": 0.4616,
      "step": 1173
    },
    {
      "epoch": 10.034188034188034,
      "grad_norm": 0.941362841502233,
      "learning_rate": 1.2345296175741466e-06,
      "loss": 0.4066,
      "step": 1174
    },
    {
      "epoch": 10.042735042735043,
      "grad_norm": 1.1007821459044609,
      "learning_rate": 1.2306721286688312e-06,
      "loss": 0.1387,
      "step": 1175
    },
    {
      "epoch": 10.051282051282051,
      "grad_norm": 1.1176796389778227,
      "learning_rate": 1.226818707185242e-06,
      "loss": 0.2619,
      "step": 1176
    },
    {
      "epoch": 10.05982905982906,
      "grad_norm": 1.7579873340296106,
      "learning_rate": 1.2229693654712433e-06,
      "loss": 0.4547,
      "step": 1177
    },
    {
      "epoch": 10.068376068376068,
      "grad_norm": 1.7828786292695857,
      "learning_rate": 1.2191241158616284e-06,
      "loss": 0.2923,
      "step": 1178
    },
    {
      "epoch": 10.076923076923077,
      "grad_norm": 1.5328882990850015,
      "learning_rate": 1.2152829706780786e-06,
      "loss": 0.2028,
      "step": 1179
    },
    {
      "epoch": 10.085470085470085,
      "grad_norm": 2.4020071504961598,
      "learning_rate": 1.2114459422291205e-06,
      "loss": 0.4144,
      "step": 1180
    },
    {
      "epoch": 10.094017094017094,
      "grad_norm": 1.5205253379430628,
      "learning_rate": 1.2076130428100894e-06,
      "loss": 0.2097,
      "step": 1181
    },
    {
      "epoch": 10.102564102564102,
      "grad_norm": 1.3015048316112432,
      "learning_rate": 1.203784284703091e-06,
      "loss": 0.3175,
      "step": 1182
    },
    {
      "epoch": 10.11111111111111,
      "grad_norm": 1.1770114406108907,
      "learning_rate": 1.1999596801769617e-06,
      "loss": 0.3426,
      "step": 1183
    },
    {
      "epoch": 10.11965811965812,
      "grad_norm": 1.5961214307061742,
      "learning_rate": 1.196139241487225e-06,
      "loss": 0.1844,
      "step": 1184
    },
    {
      "epoch": 10.128205128205128,
      "grad_norm": 1.096054730084331,
      "learning_rate": 1.1923229808760565e-06,
      "loss": 0.2849,
      "step": 1185
    },
    {
      "epoch": 10.136752136752136,
      "grad_norm": 1.5756734459360875,
      "learning_rate": 1.1885109105722454e-06,
      "loss": 0.1369,
      "step": 1186
    },
    {
      "epoch": 10.145299145299145,
      "grad_norm": 1.141051374841538,
      "learning_rate": 1.184703042791151e-06,
      "loss": 0.2324,
      "step": 1187
    },
    {
      "epoch": 10.153846153846153,
      "grad_norm": 1.4524342108279535,
      "learning_rate": 1.1808993897346679e-06,
      "loss": 0.2859,
      "step": 1188
    },
    {
      "epoch": 10.162393162393162,
      "grad_norm": 1.0352571140427305,
      "learning_rate": 1.1770999635911857e-06,
      "loss": 0.2178,
      "step": 1189
    },
    {
      "epoch": 10.17094017094017,
      "grad_norm": 1.8559620220081188,
      "learning_rate": 1.1733047765355466e-06,
      "loss": 0.2326,
      "step": 1190
    },
    {
      "epoch": 10.179487179487179,
      "grad_norm": 1.6108485889953668,
      "learning_rate": 1.1695138407290101e-06,
      "loss": 0.2345,
      "step": 1191
    },
    {
      "epoch": 10.188034188034187,
      "grad_norm": 1.445694921851245,
      "learning_rate": 1.1657271683192156e-06,
      "loss": 0.5413,
      "step": 1192
    },
    {
      "epoch": 10.196581196581196,
      "grad_norm": 1.2294376484774858,
      "learning_rate": 1.1619447714401367e-06,
      "loss": 0.3702,
      "step": 1193
    },
    {
      "epoch": 10.205128205128204,
      "grad_norm": 1.2929066334757782,
      "learning_rate": 1.1581666622120494e-06,
      "loss": 0.5578,
      "step": 1194
    },
    {
      "epoch": 10.213675213675213,
      "grad_norm": 1.1179736099867603,
      "learning_rate": 1.154392852741491e-06,
      "loss": 0.5457,
      "step": 1195
    },
    {
      "epoch": 10.222222222222221,
      "grad_norm": 1.2468241550484034,
      "learning_rate": 1.1506233551212186e-06,
      "loss": 0.366,
      "step": 1196
    },
    {
      "epoch": 10.23076923076923,
      "grad_norm": 0.9296405737354689,
      "learning_rate": 1.1468581814301718e-06,
      "loss": 0.1819,
      "step": 1197
    },
    {
      "epoch": 10.239316239316238,
      "grad_norm": 1.4238388630812857,
      "learning_rate": 1.1430973437334375e-06,
      "loss": 0.4314,
      "step": 1198
    },
    {
      "epoch": 10.247863247863247,
      "grad_norm": 0.9823335539626711,
      "learning_rate": 1.1393408540822073e-06,
      "loss": 0.2993,
      "step": 1199
    },
    {
      "epoch": 10.256410256410255,
      "grad_norm": 1.6141313708327418,
      "learning_rate": 1.1355887245137383e-06,
      "loss": 0.3024,
      "step": 1200
    },
    {
      "epoch": 10.264957264957266,
      "grad_norm": 1.3995125132746562,
      "learning_rate": 1.1318409670513194e-06,
      "loss": 0.4031,
      "step": 1201
    },
    {
      "epoch": 10.273504273504274,
      "grad_norm": 1.1327941072732368,
      "learning_rate": 1.1280975937042263e-06,
      "loss": 0.2056,
      "step": 1202
    },
    {
      "epoch": 10.282051282051283,
      "grad_norm": 1.3090370221081955,
      "learning_rate": 1.1243586164676873e-06,
      "loss": 0.1986,
      "step": 1203
    },
    {
      "epoch": 10.290598290598291,
      "grad_norm": 1.19361641233488,
      "learning_rate": 1.1206240473228447e-06,
      "loss": 0.2859,
      "step": 1204
    },
    {
      "epoch": 10.2991452991453,
      "grad_norm": 1.1504227602499402,
      "learning_rate": 1.1168938982367162e-06,
      "loss": 0.3263,
      "step": 1205
    },
    {
      "epoch": 10.307692307692308,
      "grad_norm": 2.3225905553046715,
      "learning_rate": 1.1131681811621529e-06,
      "loss": 0.387,
      "step": 1206
    },
    {
      "epoch": 10.316239316239317,
      "grad_norm": 1.338744538066502,
      "learning_rate": 1.1094469080378076e-06,
      "loss": 0.1859,
      "step": 1207
    },
    {
      "epoch": 10.324786324786325,
      "grad_norm": 1.352869738451569,
      "learning_rate": 1.1057300907880904e-06,
      "loss": 0.2305,
      "step": 1208
    },
    {
      "epoch": 10.333333333333334,
      "grad_norm": 1.5052743404655706,
      "learning_rate": 1.1020177413231334e-06,
      "loss": 0.154,
      "step": 1209
    },
    {
      "epoch": 10.341880341880342,
      "grad_norm": 1.2799078598685383,
      "learning_rate": 1.0983098715387528e-06,
      "loss": 0.1347,
      "step": 1210
    },
    {
      "epoch": 10.350427350427351,
      "grad_norm": 1.4931972984521482,
      "learning_rate": 1.0946064933164117e-06,
      "loss": 0.4827,
      "step": 1211
    },
    {
      "epoch": 10.35897435897436,
      "grad_norm": 1.8359920629352564,
      "learning_rate": 1.0909076185231762e-06,
      "loss": 0.3361,
      "step": 1212
    },
    {
      "epoch": 10.367521367521368,
      "grad_norm": 0.9290479814316553,
      "learning_rate": 1.0872132590116866e-06,
      "loss": 0.2929,
      "step": 1213
    },
    {
      "epoch": 10.376068376068377,
      "grad_norm": 1.2663449027008966,
      "learning_rate": 1.0835234266201109e-06,
      "loss": 0.2174,
      "step": 1214
    },
    {
      "epoch": 10.384615384615385,
      "grad_norm": 2.0664563187789775,
      "learning_rate": 1.079838133172111e-06,
      "loss": 0.1486,
      "step": 1215
    },
    {
      "epoch": 10.393162393162394,
      "grad_norm": 1.0989776984155346,
      "learning_rate": 1.0761573904768054e-06,
      "loss": 0.1891,
      "step": 1216
    },
    {
      "epoch": 10.401709401709402,
      "grad_norm": 1.0543354481576128,
      "learning_rate": 1.0724812103287304e-06,
      "loss": 0.2999,
      "step": 1217
    },
    {
      "epoch": 10.41025641025641,
      "grad_norm": 1.0538412063008713,
      "learning_rate": 1.0688096045078023e-06,
      "loss": 0.1605,
      "step": 1218
    },
    {
      "epoch": 10.418803418803419,
      "grad_norm": 1.5967998881825511,
      "learning_rate": 1.0651425847792767e-06,
      "loss": 0.3104,
      "step": 1219
    },
    {
      "epoch": 10.427350427350428,
      "grad_norm": 1.138119287272601,
      "learning_rate": 1.061480162893716e-06,
      "loss": 0.1738,
      "step": 1220
    },
    {
      "epoch": 10.435897435897436,
      "grad_norm": 1.2348448641155363,
      "learning_rate": 1.0578223505869494e-06,
      "loss": 0.2955,
      "step": 1221
    },
    {
      "epoch": 10.444444444444445,
      "grad_norm": 2.5194715080813403,
      "learning_rate": 1.0541691595800338e-06,
      "loss": 0.485,
      "step": 1222
    },
    {
      "epoch": 10.452991452991453,
      "grad_norm": 1.0000721895885394,
      "learning_rate": 1.0505206015792194e-06,
      "loss": 0.4439,
      "step": 1223
    },
    {
      "epoch": 10.461538461538462,
      "grad_norm": 1.0958363947374086,
      "learning_rate": 1.0468766882759094e-06,
      "loss": 0.3602,
      "step": 1224
    },
    {
      "epoch": 10.47008547008547,
      "grad_norm": 1.0825014890114124,
      "learning_rate": 1.043237431346622e-06,
      "loss": 0.2051,
      "step": 1225
    },
    {
      "epoch": 10.478632478632479,
      "grad_norm": 1.4035959711749553,
      "learning_rate": 1.0396028424529578e-06,
      "loss": 0.2622,
      "step": 1226
    },
    {
      "epoch": 10.487179487179487,
      "grad_norm": 1.3152449508712105,
      "learning_rate": 1.0359729332415582e-06,
      "loss": 0.2951,
      "step": 1227
    },
    {
      "epoch": 10.495726495726496,
      "grad_norm": 1.298812306086358,
      "learning_rate": 1.032347715344067e-06,
      "loss": 0.5657,
      "step": 1228
    },
    {
      "epoch": 10.504273504273504,
      "grad_norm": 1.2384576946208885,
      "learning_rate": 1.0287272003770982e-06,
      "loss": 0.1613,
      "step": 1229
    },
    {
      "epoch": 10.512820512820513,
      "grad_norm": 1.074312238661994,
      "learning_rate": 1.0251113999421936e-06,
      "loss": 0.2647,
      "step": 1230
    },
    {
      "epoch": 10.521367521367521,
      "grad_norm": 1.595259592896357,
      "learning_rate": 1.0215003256257874e-06,
      "loss": 0.4389,
      "step": 1231
    },
    {
      "epoch": 10.52991452991453,
      "grad_norm": 1.8489731376397716,
      "learning_rate": 1.0178939889991717e-06,
      "loss": 0.2919,
      "step": 1232
    },
    {
      "epoch": 10.538461538461538,
      "grad_norm": 1.4870289686970262,
      "learning_rate": 1.014292401618457e-06,
      "loss": 0.2796,
      "step": 1233
    },
    {
      "epoch": 10.547008547008547,
      "grad_norm": 1.1764260707798178,
      "learning_rate": 1.0106955750245323e-06,
      "loss": 0.3266,
      "step": 1234
    },
    {
      "epoch": 10.555555555555555,
      "grad_norm": 1.068302905807488,
      "learning_rate": 1.0071035207430352e-06,
      "loss": 0.2251,
      "step": 1235
    },
    {
      "epoch": 10.564102564102564,
      "grad_norm": 1.6533940675412566,
      "learning_rate": 1.0035162502843073e-06,
      "loss": 0.6334,
      "step": 1236
    },
    {
      "epoch": 10.572649572649572,
      "grad_norm": 1.2461535990621175,
      "learning_rate": 9.999337751433643e-07,
      "loss": 0.3432,
      "step": 1237
    },
    {
      "epoch": 10.581196581196581,
      "grad_norm": 0.928022739303377,
      "learning_rate": 9.963561067998531e-07,
      "loss": 0.4515,
      "step": 1238
    },
    {
      "epoch": 10.58974358974359,
      "grad_norm": 0.9670006631983897,
      "learning_rate": 9.927832567180193e-07,
      "loss": 0.3008,
      "step": 1239
    },
    {
      "epoch": 10.598290598290598,
      "grad_norm": 1.3374900402459684,
      "learning_rate": 9.892152363466692e-07,
      "loss": 0.2875,
      "step": 1240
    },
    {
      "epoch": 10.606837606837606,
      "grad_norm": 1.9904354279349568,
      "learning_rate": 9.856520571191316e-07,
      "loss": 0.3453,
      "step": 1241
    },
    {
      "epoch": 10.615384615384615,
      "grad_norm": 1.1713430132865819,
      "learning_rate": 9.820937304532221e-07,
      "loss": 0.347,
      "step": 1242
    },
    {
      "epoch": 10.623931623931623,
      "grad_norm": 1.33830343768472,
      "learning_rate": 9.78540267751209e-07,
      "loss": 0.2877,
      "step": 1243
    },
    {
      "epoch": 10.632478632478632,
      "grad_norm": 1.3015970073302343,
      "learning_rate": 9.749916803997717e-07,
      "loss": 0.3081,
      "step": 1244
    },
    {
      "epoch": 10.64102564102564,
      "grad_norm": 1.1242361345286183,
      "learning_rate": 9.714479797699695e-07,
      "loss": 0.2623,
      "step": 1245
    },
    {
      "epoch": 10.649572649572649,
      "grad_norm": 1.154568783965454,
      "learning_rate": 9.679091772172021e-07,
      "loss": 0.2735,
      "step": 1246
    },
    {
      "epoch": 10.658119658119658,
      "grad_norm": 1.3468667577457005,
      "learning_rate": 9.643752840811734e-07,
      "loss": 0.4489,
      "step": 1247
    },
    {
      "epoch": 10.666666666666666,
      "grad_norm": 1.120164463622952,
      "learning_rate": 9.608463116858544e-07,
      "loss": 0.4954,
      "step": 1248
    },
    {
      "epoch": 10.675213675213675,
      "grad_norm": 1.8882904303111139,
      "learning_rate": 9.573222713394513e-07,
      "loss": 0.3261,
      "step": 1249
    },
    {
      "epoch": 10.683760683760683,
      "grad_norm": 1.10929175594392,
      "learning_rate": 9.538031743343628e-07,
      "loss": 0.5187,
      "step": 1250
    },
    {
      "epoch": 10.692307692307692,
      "grad_norm": 1.1140665621309356,
      "learning_rate": 9.502890319471491e-07,
      "loss": 0.3435,
      "step": 1251
    },
    {
      "epoch": 10.7008547008547,
      "grad_norm": 1.8794392120766952,
      "learning_rate": 9.467798554384946e-07,
      "loss": 0.2568,
      "step": 1252
    },
    {
      "epoch": 10.709401709401709,
      "grad_norm": 1.221136395755458,
      "learning_rate": 9.432756560531691e-07,
      "loss": 0.2727,
      "step": 1253
    },
    {
      "epoch": 10.717948717948717,
      "grad_norm": 1.370772824557499,
      "learning_rate": 9.397764450199937e-07,
      "loss": 0.1289,
      "step": 1254
    },
    {
      "epoch": 10.726495726495726,
      "grad_norm": 1.0055054063061366,
      "learning_rate": 9.362822335518062e-07,
      "loss": 0.441,
      "step": 1255
    },
    {
      "epoch": 10.735042735042736,
      "grad_norm": 1.134449984636529,
      "learning_rate": 9.327930328454249e-07,
      "loss": 0.2073,
      "step": 1256
    },
    {
      "epoch": 10.743589743589745,
      "grad_norm": 1.3407780383754453,
      "learning_rate": 9.293088540816081e-07,
      "loss": 0.3687,
      "step": 1257
    },
    {
      "epoch": 10.752136752136753,
      "grad_norm": 1.5464482793782732,
      "learning_rate": 9.258297084250256e-07,
      "loss": 0.3144,
      "step": 1258
    },
    {
      "epoch": 10.760683760683762,
      "grad_norm": 1.0417626287781558,
      "learning_rate": 9.22355607024217e-07,
      "loss": 0.2077,
      "step": 1259
    },
    {
      "epoch": 10.76923076923077,
      "grad_norm": 1.6811935785110077,
      "learning_rate": 9.188865610115572e-07,
      "loss": 0.2955,
      "step": 1260
    },
    {
      "epoch": 10.777777777777779,
      "grad_norm": 2.27870136281711,
      "learning_rate": 9.154225815032242e-07,
      "loss": 0.308,
      "step": 1261
    },
    {
      "epoch": 10.786324786324787,
      "grad_norm": 1.1647120152217134,
      "learning_rate": 9.119636795991605e-07,
      "loss": 0.2365,
      "step": 1262
    },
    {
      "epoch": 10.794871794871796,
      "grad_norm": 2.012077640003533,
      "learning_rate": 9.085098663830366e-07,
      "loss": 0.3534,
      "step": 1263
    },
    {
      "epoch": 10.803418803418804,
      "grad_norm": 1.017212268548294,
      "learning_rate": 9.050611529222167e-07,
      "loss": 0.2761,
      "step": 1264
    },
    {
      "epoch": 10.811965811965813,
      "grad_norm": 1.102217264281939,
      "learning_rate": 9.01617550267726e-07,
      "loss": 0.4417,
      "step": 1265
    },
    {
      "epoch": 10.820512820512821,
      "grad_norm": 1.1395609377381528,
      "learning_rate": 8.98179069454209e-07,
      "loss": 0.2634,
      "step": 1266
    },
    {
      "epoch": 10.82905982905983,
      "grad_norm": 1.1768625766970986,
      "learning_rate": 8.947457214999006e-07,
      "loss": 0.3728,
      "step": 1267
    },
    {
      "epoch": 10.837606837606838,
      "grad_norm": 0.999996881687392,
      "learning_rate": 8.91317517406588e-07,
      "loss": 0.4249,
      "step": 1268
    },
    {
      "epoch": 10.846153846153847,
      "grad_norm": 1.2213184518043707,
      "learning_rate": 8.878944681595742e-07,
      "loss": 0.2689,
      "step": 1269
    },
    {
      "epoch": 10.854700854700855,
      "grad_norm": 1.2397285549777677,
      "learning_rate": 8.844765847276432e-07,
      "loss": 0.2904,
      "step": 1270
    },
    {
      "epoch": 10.863247863247864,
      "grad_norm": 1.1401456049615335,
      "learning_rate": 8.810638780630279e-07,
      "loss": 0.229,
      "step": 1271
    },
    {
      "epoch": 10.871794871794872,
      "grad_norm": 1.7363838065776036,
      "learning_rate": 8.776563591013729e-07,
      "loss": 0.1458,
      "step": 1272
    },
    {
      "epoch": 10.88034188034188,
      "grad_norm": 1.2492147539226588,
      "learning_rate": 8.742540387616966e-07,
      "loss": 0.541,
      "step": 1273
    },
    {
      "epoch": 10.88888888888889,
      "grad_norm": 1.054703332522137,
      "learning_rate": 8.708569279463622e-07,
      "loss": 0.5736,
      "step": 1274
    },
    {
      "epoch": 10.897435897435898,
      "grad_norm": 1.152302935892294,
      "learning_rate": 8.674650375410379e-07,
      "loss": 0.2423,
      "step": 1275
    },
    {
      "epoch": 10.905982905982906,
      "grad_norm": 1.2563902276802954,
      "learning_rate": 8.640783784146625e-07,
      "loss": 0.472,
      "step": 1276
    },
    {
      "epoch": 10.914529914529915,
      "grad_norm": 2.468428865936901,
      "learning_rate": 8.606969614194144e-07,
      "loss": 0.3641,
      "step": 1277
    },
    {
      "epoch": 10.923076923076923,
      "grad_norm": 1.219505533382249,
      "learning_rate": 8.573207973906736e-07,
      "loss": 0.3872,
      "step": 1278
    },
    {
      "epoch": 10.931623931623932,
      "grad_norm": 1.269203003869562,
      "learning_rate": 8.539498971469848e-07,
      "loss": 0.2392,
      "step": 1279
    },
    {
      "epoch": 10.94017094017094,
      "grad_norm": 0.9334967508461468,
      "learning_rate": 8.505842714900298e-07,
      "loss": 0.2905,
      "step": 1280
    },
    {
      "epoch": 10.948717948717949,
      "grad_norm": 1.5491194442069778,
      "learning_rate": 8.472239312045851e-07,
      "loss": 0.4097,
      "step": 1281
    },
    {
      "epoch": 10.957264957264957,
      "grad_norm": 1.175632296882277,
      "learning_rate": 8.438688870584913e-07,
      "loss": 0.3536,
      "step": 1282
    },
    {
      "epoch": 10.965811965811966,
      "grad_norm": 1.6321024132119002,
      "learning_rate": 8.405191498026197e-07,
      "loss": 0.3949,
      "step": 1283
    },
    {
      "epoch": 10.974358974358974,
      "grad_norm": 1.5361242486173952,
      "learning_rate": 8.371747301708358e-07,
      "loss": 0.3731,
      "step": 1284
    },
    {
      "epoch": 10.982905982905983,
      "grad_norm": 1.6109773712263131,
      "learning_rate": 8.338356388799637e-07,
      "loss": 0.3013,
      "step": 1285
    },
    {
      "epoch": 10.991452991452991,
      "grad_norm": 1.3208056491021436,
      "learning_rate": 8.305018866297562e-07,
      "loss": 0.312,
      "step": 1286
    },
    {
      "epoch": 11.0,
      "grad_norm": 1.2095249712528178,
      "learning_rate": 8.271734841028553e-07,
      "loss": 0.329,
      "step": 1287
    },
    {
      "epoch": 11.008547008547009,
      "grad_norm": 1.1339926584737854,
      "learning_rate": 8.238504419647602e-07,
      "loss": 0.3453,
      "step": 1288
    },
    {
      "epoch": 11.017094017094017,
      "grad_norm": 2.285728413611872,
      "learning_rate": 8.205327708637958e-07,
      "loss": 0.2377,
      "step": 1289
    },
    {
      "epoch": 11.025641025641026,
      "grad_norm": 0.8347020886437849,
      "learning_rate": 8.172204814310741e-07,
      "loss": 0.2597,
      "step": 1290
    },
    {
      "epoch": 11.034188034188034,
      "grad_norm": 0.9894332228301046,
      "learning_rate": 8.139135842804638e-07,
      "loss": 0.2353,
      "step": 1291
    },
    {
      "epoch": 11.042735042735043,
      "grad_norm": 1.1961555891311104,
      "learning_rate": 8.106120900085526e-07,
      "loss": 0.2439,
      "step": 1292
    },
    {
      "epoch": 11.051282051282051,
      "grad_norm": 2.4837226139453503,
      "learning_rate": 8.073160091946156e-07,
      "loss": 0.2131,
      "step": 1293
    },
    {
      "epoch": 11.05982905982906,
      "grad_norm": 1.2402400103690412,
      "learning_rate": 8.040253524005834e-07,
      "loss": 0.3762,
      "step": 1294
    },
    {
      "epoch": 11.068376068376068,
      "grad_norm": 1.0670179465841032,
      "learning_rate": 8.007401301710022e-07,
      "loss": 0.3589,
      "step": 1295
    },
    {
      "epoch": 11.076923076923077,
      "grad_norm": 1.2043916022680978,
      "learning_rate": 7.974603530330069e-07,
      "loss": 0.2431,
      "step": 1296
    },
    {
      "epoch": 11.085470085470085,
      "grad_norm": 1.0286938560514645,
      "learning_rate": 7.941860314962843e-07,
      "loss": 0.0952,
      "step": 1297
    },
    {
      "epoch": 11.094017094017094,
      "grad_norm": 1.0664360647879936,
      "learning_rate": 7.909171760530351e-07,
      "loss": 0.2457,
      "step": 1298
    },
    {
      "epoch": 11.102564102564102,
      "grad_norm": 3.129024390617159,
      "learning_rate": 7.876537971779493e-07,
      "loss": 0.1881,
      "step": 1299
    },
    {
      "epoch": 11.11111111111111,
      "grad_norm": 1.0282067050352575,
      "learning_rate": 7.843959053281663e-07,
      "loss": 0.4266,
      "step": 1300
    },
    {
      "epoch": 11.11965811965812,
      "grad_norm": 1.1975968400831338,
      "learning_rate": 7.811435109432417e-07,
      "loss": 0.4366,
      "step": 1301
    },
    {
      "epoch": 11.128205128205128,
      "grad_norm": 1.0744374162194799,
      "learning_rate": 7.778966244451169e-07,
      "loss": 0.2198,
      "step": 1302
    },
    {
      "epoch": 11.136752136752136,
      "grad_norm": 1.1029563296447737,
      "learning_rate": 7.746552562380829e-07,
      "loss": 0.3467,
      "step": 1303
    },
    {
      "epoch": 11.145299145299145,
      "grad_norm": 1.1960864725036724,
      "learning_rate": 7.714194167087466e-07,
      "loss": 0.3032,
      "step": 1304
    },
    {
      "epoch": 11.153846153846153,
      "grad_norm": 1.0760347397890864,
      "learning_rate": 7.681891162260016e-07,
      "loss": 0.1681,
      "step": 1305
    },
    {
      "epoch": 11.162393162393162,
      "grad_norm": 0.9750236130397525,
      "learning_rate": 7.649643651409916e-07,
      "loss": 0.3941,
      "step": 1306
    },
    {
      "epoch": 11.17094017094017,
      "grad_norm": 1.1613839311484029,
      "learning_rate": 7.617451737870754e-07,
      "loss": 0.4137,
      "step": 1307
    },
    {
      "epoch": 11.179487179487179,
      "grad_norm": 1.062582908339486,
      "learning_rate": 7.585315524797998e-07,
      "loss": 0.28,
      "step": 1308
    },
    {
      "epoch": 11.188034188034187,
      "grad_norm": 0.8296624704516735,
      "learning_rate": 7.553235115168598e-07,
      "loss": 0.1661,
      "step": 1309
    },
    {
      "epoch": 11.196581196581196,
      "grad_norm": 1.0142078743901737,
      "learning_rate": 7.521210611780715e-07,
      "loss": 0.1358,
      "step": 1310
    },
    {
      "epoch": 11.205128205128204,
      "grad_norm": 1.0029814830645243,
      "learning_rate": 7.489242117253342e-07,
      "loss": 0.2322,
      "step": 1311
    },
    {
      "epoch": 11.213675213675213,
      "grad_norm": 1.4087404893921196,
      "learning_rate": 7.457329734026012e-07,
      "loss": 0.2598,
      "step": 1312
    },
    {
      "epoch": 11.222222222222221,
      "grad_norm": 1.2902537318460452,
      "learning_rate": 7.425473564358457e-07,
      "loss": 0.2689,
      "step": 1313
    },
    {
      "epoch": 11.23076923076923,
      "grad_norm": 1.3368847754229851,
      "learning_rate": 7.393673710330271e-07,
      "loss": 0.1781,
      "step": 1314
    },
    {
      "epoch": 11.239316239316238,
      "grad_norm": 1.2385657214763104,
      "learning_rate": 7.361930273840581e-07,
      "loss": 0.6036,
      "step": 1315
    },
    {
      "epoch": 11.247863247863247,
      "grad_norm": 1.1567242628436323,
      "learning_rate": 7.330243356607758e-07,
      "loss": 0.4502,
      "step": 1316
    },
    {
      "epoch": 11.256410256410255,
      "grad_norm": 1.0159771072872956,
      "learning_rate": 7.298613060169035e-07,
      "loss": 0.3977,
      "step": 1317
    },
    {
      "epoch": 11.264957264957266,
      "grad_norm": 1.3932241780307368,
      "learning_rate": 7.267039485880225e-07,
      "loss": 0.3772,
      "step": 1318
    },
    {
      "epoch": 11.273504273504274,
      "grad_norm": 2.6569957477431574,
      "learning_rate": 7.235522734915393e-07,
      "loss": 0.1337,
      "step": 1319
    },
    {
      "epoch": 11.282051282051283,
      "grad_norm": 1.0130803300138838,
      "learning_rate": 7.204062908266491e-07,
      "loss": 0.2043,
      "step": 1320
    },
    {
      "epoch": 11.290598290598291,
      "grad_norm": 1.2815538738053691,
      "learning_rate": 7.172660106743073e-07,
      "loss": 0.5477,
      "step": 1321
    },
    {
      "epoch": 11.2991452991453,
      "grad_norm": 1.1054111222974445,
      "learning_rate": 7.14131443097198e-07,
      "loss": 0.3482,
      "step": 1322
    },
    {
      "epoch": 11.307692307692308,
      "grad_norm": 0.9503250393747996,
      "learning_rate": 7.110025981396976e-07,
      "loss": 0.1972,
      "step": 1323
    },
    {
      "epoch": 11.316239316239317,
      "grad_norm": 1.3510519482608403,
      "learning_rate": 7.078794858278462e-07,
      "loss": 0.2791,
      "step": 1324
    },
    {
      "epoch": 11.324786324786325,
      "grad_norm": 1.1784971548342345,
      "learning_rate": 7.047621161693152e-07,
      "loss": 0.3825,
      "step": 1325
    },
    {
      "epoch": 11.333333333333334,
      "grad_norm": 1.0519903179740981,
      "learning_rate": 7.016504991533727e-07,
      "loss": 0.1166,
      "step": 1326
    },
    {
      "epoch": 11.341880341880342,
      "grad_norm": 1.230631119598931,
      "learning_rate": 6.985446447508526e-07,
      "loss": 0.3721,
      "step": 1327
    },
    {
      "epoch": 11.350427350427351,
      "grad_norm": 1.5015192550894227,
      "learning_rate": 6.954445629141246e-07,
      "loss": 0.4144,
      "step": 1328
    },
    {
      "epoch": 11.35897435897436,
      "grad_norm": 1.0329732984020763,
      "learning_rate": 6.923502635770618e-07,
      "loss": 0.3516,
      "step": 1329
    },
    {
      "epoch": 11.367521367521368,
      "grad_norm": 1.0689751035375088,
      "learning_rate": 6.892617566550044e-07,
      "loss": 0.3322,
      "step": 1330
    },
    {
      "epoch": 11.376068376068377,
      "grad_norm": 1.0234832818727786,
      "learning_rate": 6.861790520447356e-07,
      "loss": 0.1614,
      "step": 1331
    },
    {
      "epoch": 11.384615384615385,
      "grad_norm": 1.3705378171282174,
      "learning_rate": 6.831021596244425e-07,
      "loss": 0.1148,
      "step": 1332
    },
    {
      "epoch": 11.393162393162394,
      "grad_norm": 2.687284547442591,
      "learning_rate": 6.800310892536884e-07,
      "loss": 0.3965,
      "step": 1333
    },
    {
      "epoch": 11.401709401709402,
      "grad_norm": 1.122865008233458,
      "learning_rate": 6.769658507733815e-07,
      "loss": 0.2215,
      "step": 1334
    },
    {
      "epoch": 11.41025641025641,
      "grad_norm": 1.4255571677021555,
      "learning_rate": 6.739064540057425e-07,
      "loss": 0.3261,
      "step": 1335
    },
    {
      "epoch": 11.418803418803419,
      "grad_norm": 1.000000239576537,
      "learning_rate": 6.708529087542717e-07,
      "loss": 0.3153,
      "step": 1336
    },
    {
      "epoch": 11.427350427350428,
      "grad_norm": 1.6496826359295964,
      "learning_rate": 6.678052248037184e-07,
      "loss": 0.3163,
      "step": 1337
    },
    {
      "epoch": 11.435897435897436,
      "grad_norm": 1.2866215785516808,
      "learning_rate": 6.64763411920053e-07,
      "loss": 0.3916,
      "step": 1338
    },
    {
      "epoch": 11.444444444444445,
      "grad_norm": 1.0576686463394254,
      "learning_rate": 6.617274798504286e-07,
      "loss": 0.1983,
      "step": 1339
    },
    {
      "epoch": 11.452991452991453,
      "grad_norm": 1.1094778246746804,
      "learning_rate": 6.586974383231573e-07,
      "loss": 0.1522,
      "step": 1340
    },
    {
      "epoch": 11.461538461538462,
      "grad_norm": 1.0465398985429073,
      "learning_rate": 6.556732970476748e-07,
      "loss": 0.3405,
      "step": 1341
    },
    {
      "epoch": 11.47008547008547,
      "grad_norm": 1.0863827719627053,
      "learning_rate": 6.526550657145089e-07,
      "loss": 0.1914,
      "step": 1342
    },
    {
      "epoch": 11.478632478632479,
      "grad_norm": 1.2352159374155984,
      "learning_rate": 6.496427539952499e-07,
      "loss": 0.2707,
      "step": 1343
    },
    {
      "epoch": 11.487179487179487,
      "grad_norm": 1.1636422042646013,
      "learning_rate": 6.4663637154252e-07,
      "loss": 0.3073,
      "step": 1344
    },
    {
      "epoch": 11.495726495726496,
      "grad_norm": 0.9400787003787303,
      "learning_rate": 6.436359279899426e-07,
      "loss": 0.2746,
      "step": 1345
    },
    {
      "epoch": 11.504273504273504,
      "grad_norm": 1.7599302281828313,
      "learning_rate": 6.406414329521079e-07,
      "loss": 0.3627,
      "step": 1346
    },
    {
      "epoch": 11.512820512820513,
      "grad_norm": 1.0452282974681053,
      "learning_rate": 6.376528960245476e-07,
      "loss": 0.3472,
      "step": 1347
    },
    {
      "epoch": 11.521367521367521,
      "grad_norm": 1.1671960949053146,
      "learning_rate": 6.346703267836998e-07,
      "loss": 0.4321,
      "step": 1348
    },
    {
      "epoch": 11.52991452991453,
      "grad_norm": 1.0530089555475837,
      "learning_rate": 6.316937347868787e-07,
      "loss": 0.1937,
      "step": 1349
    },
    {
      "epoch": 11.538461538461538,
      "grad_norm": 1.519317404025633,
      "learning_rate": 6.28723129572247e-07,
      "loss": 0.4331,
      "step": 1350
    },
    {
      "epoch": 11.547008547008547,
      "grad_norm": 1.0524691505080326,
      "learning_rate": 6.257585206587843e-07,
      "loss": 0.2391,
      "step": 1351
    },
    {
      "epoch": 11.555555555555555,
      "grad_norm": 2.2094469058936594,
      "learning_rate": 6.227999175462521e-07,
      "loss": 0.5341,
      "step": 1352
    },
    {
      "epoch": 11.564102564102564,
      "grad_norm": 1.0170031770981227,
      "learning_rate": 6.198473297151705e-07,
      "loss": 0.2465,
      "step": 1353
    },
    {
      "epoch": 11.572649572649572,
      "grad_norm": 1.2489717726242522,
      "learning_rate": 6.169007666267824e-07,
      "loss": 0.4308,
      "step": 1354
    },
    {
      "epoch": 11.581196581196581,
      "grad_norm": 1.4302549005618361,
      "learning_rate": 6.139602377230247e-07,
      "loss": 0.3328,
      "step": 1355
    },
    {
      "epoch": 11.58974358974359,
      "grad_norm": 1.0017349912737998,
      "learning_rate": 6.110257524264998e-07,
      "loss": 0.1254,
      "step": 1356
    },
    {
      "epoch": 11.598290598290598,
      "grad_norm": 1.0916907629282087,
      "learning_rate": 6.080973201404444e-07,
      "loss": 0.5605,
      "step": 1357
    },
    {
      "epoch": 11.606837606837606,
      "grad_norm": 1.0348888975048909,
      "learning_rate": 6.051749502486967e-07,
      "loss": 0.2208,
      "step": 1358
    },
    {
      "epoch": 11.615384615384615,
      "grad_norm": 1.3165290208428866,
      "learning_rate": 6.022586521156714e-07,
      "loss": 0.2646,
      "step": 1359
    },
    {
      "epoch": 11.623931623931623,
      "grad_norm": 1.2837933636074768,
      "learning_rate": 5.993484350863246e-07,
      "loss": 0.164,
      "step": 1360
    },
    {
      "epoch": 11.632478632478632,
      "grad_norm": 2.412451254753726,
      "learning_rate": 5.964443084861265e-07,
      "loss": 0.0772,
      "step": 1361
    },
    {
      "epoch": 11.64102564102564,
      "grad_norm": 1.7479905420082966,
      "learning_rate": 5.935462816210325e-07,
      "loss": 0.3332,
      "step": 1362
    },
    {
      "epoch": 11.649572649572649,
      "grad_norm": 0.9679775296832205,
      "learning_rate": 5.906543637774512e-07,
      "loss": 0.3549,
      "step": 1363
    },
    {
      "epoch": 11.658119658119658,
      "grad_norm": 1.3334212644868122,
      "learning_rate": 5.877685642222163e-07,
      "loss": 0.4342,
      "step": 1364
    },
    {
      "epoch": 11.666666666666666,
      "grad_norm": 1.6850702917590108,
      "learning_rate": 5.848888922025553e-07,
      "loss": 0.3346,
      "step": 1365
    },
    {
      "epoch": 11.675213675213675,
      "grad_norm": 1.1306411787231243,
      "learning_rate": 5.820153569460596e-07,
      "loss": 0.1642,
      "step": 1366
    },
    {
      "epoch": 11.683760683760683,
      "grad_norm": 1.8273706494458912,
      "learning_rate": 5.791479676606587e-07,
      "loss": 0.2132,
      "step": 1367
    },
    {
      "epoch": 11.692307692307692,
      "grad_norm": 1.2185234414343529,
      "learning_rate": 5.762867335345851e-07,
      "loss": 0.454,
      "step": 1368
    },
    {
      "epoch": 11.7008547008547,
      "grad_norm": 1.315369472839878,
      "learning_rate": 5.734316637363505e-07,
      "loss": 0.3286,
      "step": 1369
    },
    {
      "epoch": 11.709401709401709,
      "grad_norm": 1.2250823833793434,
      "learning_rate": 5.705827674147125e-07,
      "loss": 0.2436,
      "step": 1370
    },
    {
      "epoch": 11.717948717948717,
      "grad_norm": 1.1492279019896923,
      "learning_rate": 5.67740053698646e-07,
      "loss": 0.4002,
      "step": 1371
    },
    {
      "epoch": 11.726495726495726,
      "grad_norm": 1.35392039503661,
      "learning_rate": 5.649035316973142e-07,
      "loss": 0.4827,
      "step": 1372
    },
    {
      "epoch": 11.735042735042736,
      "grad_norm": 1.0492032813030379,
      "learning_rate": 5.620732105000415e-07,
      "loss": 0.1662,
      "step": 1373
    },
    {
      "epoch": 11.743589743589745,
      "grad_norm": 1.2987614026152967,
      "learning_rate": 5.5924909917628e-07,
      "loss": 0.3225,
      "step": 1374
    },
    {
      "epoch": 11.752136752136753,
      "grad_norm": 1.1037691937130272,
      "learning_rate": 5.564312067755856e-07,
      "loss": 0.2383,
      "step": 1375
    },
    {
      "epoch": 11.760683760683762,
      "grad_norm": 1.4303796759619123,
      "learning_rate": 5.536195423275839e-07,
      "loss": 0.3351,
      "step": 1376
    },
    {
      "epoch": 11.76923076923077,
      "grad_norm": 1.0452517599489266,
      "learning_rate": 5.508141148419443e-07,
      "loss": 0.1719,
      "step": 1377
    },
    {
      "epoch": 11.777777777777779,
      "grad_norm": 1.2734071534409364,
      "learning_rate": 5.48014933308352e-07,
      "loss": 0.3308,
      "step": 1378
    },
    {
      "epoch": 11.786324786324787,
      "grad_norm": 1.3858295600616295,
      "learning_rate": 5.45222006696477e-07,
      "loss": 0.3838,
      "step": 1379
    },
    {
      "epoch": 11.794871794871796,
      "grad_norm": 1.0090102173530728,
      "learning_rate": 5.424353439559446e-07,
      "loss": 0.3294,
      "step": 1380
    },
    {
      "epoch": 11.803418803418804,
      "grad_norm": 1.5702211951848428,
      "learning_rate": 5.396549540163106e-07,
      "loss": 0.3281,
      "step": 1381
    },
    {
      "epoch": 11.811965811965813,
      "grad_norm": 2.0057111464447672,
      "learning_rate": 5.36880845787028e-07,
      "loss": 0.2921,
      "step": 1382
    },
    {
      "epoch": 11.820512820512821,
      "grad_norm": 1.21502886222698,
      "learning_rate": 5.341130281574233e-07,
      "loss": 0.4285,
      "step": 1383
    },
    {
      "epoch": 11.82905982905983,
      "grad_norm": 1.1066519508861632,
      "learning_rate": 5.313515099966627e-07,
      "loss": 0.3674,
      "step": 1384
    },
    {
      "epoch": 11.837606837606838,
      "grad_norm": 1.4189752940939921,
      "learning_rate": 5.28596300153728e-07,
      "loss": 0.3384,
      "step": 1385
    },
    {
      "epoch": 11.846153846153847,
      "grad_norm": 1.0497925251189915,
      "learning_rate": 5.258474074573878e-07,
      "loss": 0.3128,
      "step": 1386
    },
    {
      "epoch": 11.854700854700855,
      "grad_norm": 1.5489841745380557,
      "learning_rate": 5.231048407161657e-07,
      "loss": 0.4025,
      "step": 1387
    },
    {
      "epoch": 11.863247863247864,
      "grad_norm": 0.9941263366209754,
      "learning_rate": 5.203686087183149e-07,
      "loss": 0.2513,
      "step": 1388
    },
    {
      "epoch": 11.871794871794872,
      "grad_norm": 1.4119603032969352,
      "learning_rate": 5.176387202317915e-07,
      "loss": 0.4545,
      "step": 1389
    },
    {
      "epoch": 11.88034188034188,
      "grad_norm": 1.0132661355105295,
      "learning_rate": 5.149151840042224e-07,
      "loss": 0.4799,
      "step": 1390
    },
    {
      "epoch": 11.88888888888889,
      "grad_norm": 1.1066098785113443,
      "learning_rate": 5.121980087628802e-07,
      "loss": 0.2781,
      "step": 1391
    },
    {
      "epoch": 11.897435897435898,
      "grad_norm": 0.9564853712646907,
      "learning_rate": 5.094872032146562e-07,
      "loss": 0.2008,
      "step": 1392
    },
    {
      "epoch": 11.905982905982906,
      "grad_norm": 1.4412517815970023,
      "learning_rate": 5.06782776046027e-07,
      "loss": 0.3013,
      "step": 1393
    },
    {
      "epoch": 11.914529914529915,
      "grad_norm": 1.0156654862669703,
      "learning_rate": 5.040847359230327e-07,
      "loss": 0.2307,
      "step": 1394
    },
    {
      "epoch": 11.923076923076923,
      "grad_norm": 1.0769449871390542,
      "learning_rate": 5.013930914912477e-07,
      "loss": 0.3282,
      "step": 1395
    },
    {
      "epoch": 11.931623931623932,
      "grad_norm": 1.096688158237559,
      "learning_rate": 4.98707851375749e-07,
      "loss": 0.2744,
      "step": 1396
    },
    {
      "epoch": 11.94017094017094,
      "grad_norm": 1.4568030911492071,
      "learning_rate": 4.96029024181095e-07,
      "loss": 0.3547,
      "step": 1397
    },
    {
      "epoch": 11.948717948717949,
      "grad_norm": 1.101802823212953,
      "learning_rate": 4.933566184912931e-07,
      "loss": 0.2269,
      "step": 1398
    },
    {
      "epoch": 11.957264957264957,
      "grad_norm": 1.3169879884728675,
      "learning_rate": 4.906906428697736e-07,
      "loss": 0.3043,
      "step": 1399
    },
    {
      "epoch": 11.965811965811966,
      "grad_norm": 1.5033493354032195,
      "learning_rate": 4.880311058593617e-07,
      "loss": 0.5169,
      "step": 1400
    },
    {
      "epoch": 11.974358974358974,
      "grad_norm": 1.8372923383003767,
      "learning_rate": 4.853780159822521e-07,
      "loss": 0.4537,
      "step": 1401
    },
    {
      "epoch": 11.982905982905983,
      "grad_norm": 1.1792625592862642,
      "learning_rate": 4.827313817399809e-07,
      "loss": 0.2065,
      "step": 1402
    },
    {
      "epoch": 11.991452991452991,
      "grad_norm": 1.126875070334918,
      "learning_rate": 4.800912116133955e-07,
      "loss": 0.2456,
      "step": 1403
    },
    {
      "epoch": 12.0,
      "grad_norm": 1.5697513878589628,
      "learning_rate": 4.774575140626317e-07,
      "loss": 0.4124,
      "step": 1404
    },
    {
      "epoch": 12.008547008547009,
      "grad_norm": 0.8438163695497493,
      "learning_rate": 4.748302975270838e-07,
      "loss": 0.3199,
      "step": 1405
    },
    {
      "epoch": 12.017094017094017,
      "grad_norm": 1.5079877456318747,
      "learning_rate": 4.7220957042537793e-07,
      "loss": 0.4273,
      "step": 1406
    },
    {
      "epoch": 12.025641025641026,
      "grad_norm": 1.0229727290063586,
      "learning_rate": 4.6959534115534665e-07,
      "loss": 0.1734,
      "step": 1407
    },
    {
      "epoch": 12.034188034188034,
      "grad_norm": 0.97072370113068,
      "learning_rate": 4.669876180940014e-07,
      "loss": 0.2477,
      "step": 1408
    },
    {
      "epoch": 12.042735042735043,
      "grad_norm": 1.211064910341796,
      "learning_rate": 4.6438640959750285e-07,
      "loss": 0.2913,
      "step": 1409
    },
    {
      "epoch": 12.051282051282051,
      "grad_norm": 2.8135483449892176,
      "learning_rate": 4.617917240011394e-07,
      "loss": 0.2096,
      "step": 1410
    },
    {
      "epoch": 12.05982905982906,
      "grad_norm": 1.0958182096094111,
      "learning_rate": 4.592035696192948e-07,
      "loss": 0.1916,
      "step": 1411
    },
    {
      "epoch": 12.068376068376068,
      "grad_norm": 1.2127130625533227,
      "learning_rate": 4.566219547454251e-07,
      "loss": 0.2594,
      "step": 1412
    },
    {
      "epoch": 12.076923076923077,
      "grad_norm": 1.1936809604241543,
      "learning_rate": 4.5404688765203236e-07,
      "loss": 0.2925,
      "step": 1413
    },
    {
      "epoch": 12.085470085470085,
      "grad_norm": 1.1130311861207984,
      "learning_rate": 4.514783765906369e-07,
      "loss": 0.2031,
      "step": 1414
    },
    {
      "epoch": 12.094017094017094,
      "grad_norm": 1.3007322916470179,
      "learning_rate": 4.489164297917492e-07,
      "loss": 0.4502,
      "step": 1415
    },
    {
      "epoch": 12.102564102564102,
      "grad_norm": 1.0994068000291886,
      "learning_rate": 4.463610554648459e-07,
      "loss": 0.4473,
      "step": 1416
    },
    {
      "epoch": 12.11111111111111,
      "grad_norm": 1.1115069843985332,
      "learning_rate": 4.438122617983442e-07,
      "loss": 0.1048,
      "step": 1417
    },
    {
      "epoch": 12.11965811965812,
      "grad_norm": 1.0846637850713525,
      "learning_rate": 4.4127005695957374e-07,
      "loss": 0.2828,
      "step": 1418
    },
    {
      "epoch": 12.128205128205128,
      "grad_norm": 0.8866260859036948,
      "learning_rate": 4.3873444909474985e-07,
      "loss": 0.216,
      "step": 1419
    },
    {
      "epoch": 12.136752136752136,
      "grad_norm": 1.092900767346248,
      "learning_rate": 4.3620544632895e-07,
      "loss": 0.274,
      "step": 1420
    },
    {
      "epoch": 12.145299145299145,
      "grad_norm": 1.0884795291953004,
      "learning_rate": 4.336830567660855e-07,
      "loss": 0.4049,
      "step": 1421
    },
    {
      "epoch": 12.153846153846153,
      "grad_norm": 1.2438059455775952,
      "learning_rate": 4.311672884888757e-07,
      "loss": 0.4181,
      "step": 1422
    },
    {
      "epoch": 12.162393162393162,
      "grad_norm": 1.2209213090493276,
      "learning_rate": 4.286581495588249e-07,
      "loss": 0.3342,
      "step": 1423
    },
    {
      "epoch": 12.17094017094017,
      "grad_norm": 1.0078786941883342,
      "learning_rate": 4.2615564801619325e-07,
      "loss": 0.4002,
      "step": 1424
    },
    {
      "epoch": 12.179487179487179,
      "grad_norm": 1.3416635010873221,
      "learning_rate": 4.2365979187997094e-07,
      "loss": 0.2419,
      "step": 1425
    },
    {
      "epoch": 12.188034188034187,
      "grad_norm": 1.1175759363671287,
      "learning_rate": 4.21170589147856e-07,
      "loss": 0.2284,
      "step": 1426
    },
    {
      "epoch": 12.196581196581196,
      "grad_norm": 1.2925695930187564,
      "learning_rate": 4.1868804779622437e-07,
      "loss": 0.3257,
      "step": 1427
    },
    {
      "epoch": 12.205128205128204,
      "grad_norm": 0.9971032549939878,
      "learning_rate": 4.1621217578010686e-07,
      "loss": 0.2485,
      "step": 1428
    },
    {
      "epoch": 12.213675213675213,
      "grad_norm": 1.46223278685261,
      "learning_rate": 4.137429810331639e-07,
      "loss": 0.4119,
      "step": 1429
    },
    {
      "epoch": 12.222222222222221,
      "grad_norm": 1.0768684889712687,
      "learning_rate": 4.1128047146765936e-07,
      "loss": 0.2551,
      "step": 1430
    },
    {
      "epoch": 12.23076923076923,
      "grad_norm": 0.9499969600745545,
      "learning_rate": 4.088246549744332e-07,
      "loss": 0.3378,
      "step": 1431
    },
    {
      "epoch": 12.239316239316238,
      "grad_norm": 1.06970202901976,
      "learning_rate": 4.063755394228811e-07,
      "loss": 0.3059,
      "step": 1432
    },
    {
      "epoch": 12.247863247863247,
      "grad_norm": 1.1829683368454302,
      "learning_rate": 4.039331326609239e-07,
      "loss": 0.217,
      "step": 1433
    },
    {
      "epoch": 12.256410256410255,
      "grad_norm": 1.4564822793866774,
      "learning_rate": 4.0149744251498537e-07,
      "loss": 0.2916,
      "step": 1434
    },
    {
      "epoch": 12.264957264957266,
      "grad_norm": 1.3856240797241763,
      "learning_rate": 3.990684767899677e-07,
      "loss": 0.297,
      "step": 1435
    },
    {
      "epoch": 12.273504273504274,
      "grad_norm": 1.256964556701651,
      "learning_rate": 3.9664624326922447e-07,
      "loss": 0.4278,
      "step": 1436
    },
    {
      "epoch": 12.282051282051283,
      "grad_norm": 1.265967197096582,
      "learning_rate": 3.9423074971453785e-07,
      "loss": 0.3981,
      "step": 1437
    },
    {
      "epoch": 12.290598290598291,
      "grad_norm": 1.3605620355633015,
      "learning_rate": 3.918220038660908e-07,
      "loss": 0.2225,
      "step": 1438
    },
    {
      "epoch": 12.2991452991453,
      "grad_norm": 0.8642733786395712,
      "learning_rate": 3.8942001344244416e-07,
      "loss": 0.2764,
      "step": 1439
    },
    {
      "epoch": 12.307692307692308,
      "grad_norm": 1.0067964022511735,
      "learning_rate": 3.8702478614051353e-07,
      "loss": 0.2164,
      "step": 1440
    },
    {
      "epoch": 12.316239316239317,
      "grad_norm": 1.0214820970789222,
      "learning_rate": 3.846363296355404e-07,
      "loss": 0.3077,
      "step": 1441
    },
    {
      "epoch": 12.324786324786325,
      "grad_norm": 0.8436702793433775,
      "learning_rate": 3.822546515810724e-07,
      "loss": 0.308,
      "step": 1442
    },
    {
      "epoch": 12.333333333333334,
      "grad_norm": 0.9345090665422185,
      "learning_rate": 3.798797596089351e-07,
      "loss": 0.1793,
      "step": 1443
    },
    {
      "epoch": 12.341880341880342,
      "grad_norm": 0.8187825789284637,
      "learning_rate": 3.7751166132920877e-07,
      "loss": 0.1257,
      "step": 1444
    },
    {
      "epoch": 12.350427350427351,
      "grad_norm": 0.970866368219796,
      "learning_rate": 3.751503643302035e-07,
      "loss": 0.1998,
      "step": 1445
    },
    {
      "epoch": 12.35897435897436,
      "grad_norm": 1.5138688300904741,
      "learning_rate": 3.727958761784375e-07,
      "loss": 0.421,
      "step": 1446
    },
    {
      "epoch": 12.367521367521368,
      "grad_norm": 1.0829254665533683,
      "learning_rate": 3.7044820441860806e-07,
      "loss": 0.4511,
      "step": 1447
    },
    {
      "epoch": 12.376068376068377,
      "grad_norm": 1.1777412706647559,
      "learning_rate": 3.681073565735718e-07,
      "loss": 0.3406,
      "step": 1448
    },
    {
      "epoch": 12.384615384615385,
      "grad_norm": 1.3337291401337419,
      "learning_rate": 3.6577334014432003e-07,
      "loss": 0.4862,
      "step": 1449
    },
    {
      "epoch": 12.393162393162394,
      "grad_norm": 0.9920625204158359,
      "learning_rate": 3.634461626099495e-07,
      "loss": 0.3188,
      "step": 1450
    },
    {
      "epoch": 12.401709401709402,
      "grad_norm": 1.1732324209380567,
      "learning_rate": 3.611258314276461e-07,
      "loss": 0.2023,
      "step": 1451
    },
    {
      "epoch": 12.41025641025641,
      "grad_norm": 1.4368434603949196,
      "learning_rate": 3.5881235403265713e-07,
      "loss": 0.4928,
      "step": 1452
    },
    {
      "epoch": 12.418803418803419,
      "grad_norm": 1.209880026192477,
      "learning_rate": 3.56505737838265e-07,
      "loss": 0.5794,
      "step": 1453
    },
    {
      "epoch": 12.427350427350428,
      "grad_norm": 1.0765303844783172,
      "learning_rate": 3.5420599023576946e-07,
      "loss": 0.3588,
      "step": 1454
    },
    {
      "epoch": 12.435897435897436,
      "grad_norm": 1.1952856792258182,
      "learning_rate": 3.51913118594458e-07,
      "loss": 0.3529,
      "step": 1455
    },
    {
      "epoch": 12.444444444444445,
      "grad_norm": 1.4189895503204382,
      "learning_rate": 3.4962713026158697e-07,
      "loss": 0.3579,
      "step": 1456
    },
    {
      "epoch": 12.452991452991453,
      "grad_norm": 1.0904847944667027,
      "learning_rate": 3.473480325623535e-07,
      "loss": 0.3049,
      "step": 1457
    },
    {
      "epoch": 12.461538461538462,
      "grad_norm": 1.0804400329952155,
      "learning_rate": 3.450758327998768e-07,
      "loss": 0.3236,
      "step": 1458
    },
    {
      "epoch": 12.47008547008547,
      "grad_norm": 0.9096956137546562,
      "learning_rate": 3.428105382551716e-07,
      "loss": 0.3332,
      "step": 1459
    },
    {
      "epoch": 12.478632478632479,
      "grad_norm": 0.9262886907290493,
      "learning_rate": 3.405521561871247e-07,
      "loss": 0.164,
      "step": 1460
    },
    {
      "epoch": 12.487179487179487,
      "grad_norm": 1.0500627883818996,
      "learning_rate": 3.3830069383247343e-07,
      "loss": 0.5438,
      "step": 1461
    },
    {
      "epoch": 12.495726495726496,
      "grad_norm": 1.2508699587963135,
      "learning_rate": 3.3605615840578224e-07,
      "loss": 0.2713,
      "step": 1462
    },
    {
      "epoch": 12.504273504273504,
      "grad_norm": 1.2153830905771117,
      "learning_rate": 3.3381855709941733e-07,
      "loss": 0.1354,
      "step": 1463
    },
    {
      "epoch": 12.512820512820513,
      "grad_norm": 1.1888683890499405,
      "learning_rate": 3.315878970835268e-07,
      "loss": 0.2409,
      "step": 1464
    },
    {
      "epoch": 12.521367521367521,
      "grad_norm": 1.1059794294375975,
      "learning_rate": 3.293641855060162e-07,
      "loss": 0.3336,
      "step": 1465
    },
    {
      "epoch": 12.52991452991453,
      "grad_norm": 1.1840113168496267,
      "learning_rate": 3.2714742949252447e-07,
      "loss": 0.2742,
      "step": 1466
    },
    {
      "epoch": 12.538461538461538,
      "grad_norm": 2.1592550264073345,
      "learning_rate": 3.249376361464021e-07,
      "loss": 0.2109,
      "step": 1467
    },
    {
      "epoch": 12.547008547008547,
      "grad_norm": 1.0812187409812728,
      "learning_rate": 3.227348125486904e-07,
      "loss": 0.2329,
      "step": 1468
    },
    {
      "epoch": 12.555555555555555,
      "grad_norm": 1.1007960406636101,
      "learning_rate": 3.2053896575809426e-07,
      "loss": 0.3259,
      "step": 1469
    },
    {
      "epoch": 12.564102564102564,
      "grad_norm": 1.1818858763469258,
      "learning_rate": 3.1835010281096426e-07,
      "loss": 0.2348,
      "step": 1470
    },
    {
      "epoch": 12.572649572649572,
      "grad_norm": 1.036774434563237,
      "learning_rate": 3.1616823072127157e-07,
      "loss": 0.358,
      "step": 1471
    },
    {
      "epoch": 12.581196581196581,
      "grad_norm": 0.9144896563328214,
      "learning_rate": 3.1399335648058555e-07,
      "loss": 0.2064,
      "step": 1472
    },
    {
      "epoch": 12.58974358974359,
      "grad_norm": 1.1880151496219533,
      "learning_rate": 3.118254870580506e-07,
      "loss": 0.0903,
      "step": 1473
    },
    {
      "epoch": 12.598290598290598,
      "grad_norm": 1.086336466044215,
      "learning_rate": 3.096646294003675e-07,
      "loss": 0.339,
      "step": 1474
    },
    {
      "epoch": 12.606837606837606,
      "grad_norm": 1.446415527228014,
      "learning_rate": 3.075107904317667e-07,
      "loss": 0.5688,
      "step": 1475
    },
    {
      "epoch": 12.615384615384615,
      "grad_norm": 0.8813039270395069,
      "learning_rate": 3.0536397705398845e-07,
      "loss": 0.2948,
      "step": 1476
    },
    {
      "epoch": 12.623931623931623,
      "grad_norm": 0.9608868381854022,
      "learning_rate": 3.032241961462612e-07,
      "loss": 0.3666,
      "step": 1477
    },
    {
      "epoch": 12.632478632478632,
      "grad_norm": 1.0169318443426791,
      "learning_rate": 3.010914545652771e-07,
      "loss": 0.205,
      "step": 1478
    },
    {
      "epoch": 12.64102564102564,
      "grad_norm": 1.2527583493254357,
      "learning_rate": 2.9896575914517166e-07,
      "loss": 0.3132,
      "step": 1479
    },
    {
      "epoch": 12.649572649572649,
      "grad_norm": 1.4280662690242643,
      "learning_rate": 2.9684711669750313e-07,
      "loss": 0.1883,
      "step": 1480
    },
    {
      "epoch": 12.658119658119658,
      "grad_norm": 1.1301566337681153,
      "learning_rate": 2.9473553401122875e-07,
      "loss": 0.1991,
      "step": 1481
    },
    {
      "epoch": 12.666666666666666,
      "grad_norm": 1.0433471063867608,
      "learning_rate": 2.9263101785268253e-07,
      "loss": 0.2524,
      "step": 1482
    },
    {
      "epoch": 12.675213675213675,
      "grad_norm": 1.7274104052810784,
      "learning_rate": 2.9053357496555635e-07,
      "loss": 0.1876,
      "step": 1483
    },
    {
      "epoch": 12.683760683760683,
      "grad_norm": 1.036859974163086,
      "learning_rate": 2.8844321207087465e-07,
      "loss": 0.1416,
      "step": 1484
    },
    {
      "epoch": 12.692307692307692,
      "grad_norm": 0.7738557295444807,
      "learning_rate": 2.8635993586697555e-07,
      "loss": 0.1175,
      "step": 1485
    },
    {
      "epoch": 12.7008547008547,
      "grad_norm": 1.187284383819613,
      "learning_rate": 2.84283753029489e-07,
      "loss": 0.2627,
      "step": 1486
    },
    {
      "epoch": 12.709401709401709,
      "grad_norm": 0.8173656024155544,
      "learning_rate": 2.822146702113157e-07,
      "loss": 0.1562,
      "step": 1487
    },
    {
      "epoch": 12.717948717948717,
      "grad_norm": 1.1095747046438942,
      "learning_rate": 2.8015269404260333e-07,
      "loss": 0.1976,
      "step": 1488
    },
    {
      "epoch": 12.726495726495726,
      "grad_norm": 0.9916720254317867,
      "learning_rate": 2.780978311307278e-07,
      "loss": 0.1927,
      "step": 1489
    },
    {
      "epoch": 12.735042735042736,
      "grad_norm": 0.8982994853794152,
      "learning_rate": 2.7605008806027205e-07,
      "loss": 0.1149,
      "step": 1490
    },
    {
      "epoch": 12.743589743589745,
      "grad_norm": 0.8876994162052723,
      "learning_rate": 2.7400947139300443e-07,
      "loss": 0.3522,
      "step": 1491
    },
    {
      "epoch": 12.752136752136753,
      "grad_norm": 1.1108135115334203,
      "learning_rate": 2.7197598766785544e-07,
      "loss": 0.2567,
      "step": 1492
    },
    {
      "epoch": 12.760683760683762,
      "grad_norm": 0.9849565773663954,
      "learning_rate": 2.6994964340090163e-07,
      "loss": 0.2471,
      "step": 1493
    },
    {
      "epoch": 12.76923076923077,
      "grad_norm": 1.3189832219649773,
      "learning_rate": 2.679304450853401e-07,
      "loss": 0.2143,
      "step": 1494
    },
    {
      "epoch": 12.777777777777779,
      "grad_norm": 1.2483986620984422,
      "learning_rate": 2.6591839919146963e-07,
      "loss": 0.6708,
      "step": 1495
    },
    {
      "epoch": 12.786324786324787,
      "grad_norm": 1.1032130060771996,
      "learning_rate": 2.6391351216667053e-07,
      "loss": 0.373,
      "step": 1496
    },
    {
      "epoch": 12.794871794871796,
      "grad_norm": 1.2634362960620278,
      "learning_rate": 2.6191579043538333e-07,
      "loss": 0.1604,
      "step": 1497
    },
    {
      "epoch": 12.803418803418804,
      "grad_norm": 1.0765362403824454,
      "learning_rate": 2.599252403990873e-07,
      "loss": 0.3473,
      "step": 1498
    },
    {
      "epoch": 12.811965811965813,
      "grad_norm": 1.1233098805300556,
      "learning_rate": 2.5794186843628247e-07,
      "loss": 0.2566,
      "step": 1499
    },
    {
      "epoch": 12.820512820512821,
      "grad_norm": 1.0012569569825553,
      "learning_rate": 2.5596568090246546e-07,
      "loss": 0.2917,
      "step": 1500
    },
    {
      "epoch": 12.82905982905983,
      "grad_norm": 2.5769967962871263,
      "learning_rate": 2.539966841301117e-07,
      "loss": 0.2105,
      "step": 1501
    },
    {
      "epoch": 12.837606837606838,
      "grad_norm": 1.0551006567181496,
      "learning_rate": 2.5203488442865574e-07,
      "loss": 0.5103,
      "step": 1502
    },
    {
      "epoch": 12.846153846153847,
      "grad_norm": 1.0739659574771914,
      "learning_rate": 2.5008028808446995e-07,
      "loss": 0.3336,
      "step": 1503
    },
    {
      "epoch": 12.854700854700855,
      "grad_norm": 1.0004107986619046,
      "learning_rate": 2.481329013608419e-07,
      "loss": 0.189,
      "step": 1504
    },
    {
      "epoch": 12.863247863247864,
      "grad_norm": 1.198361498583922,
      "learning_rate": 2.4619273049796e-07,
      "loss": 0.2824,
      "step": 1505
    },
    {
      "epoch": 12.871794871794872,
      "grad_norm": 1.0439673382608252,
      "learning_rate": 2.4425978171288807e-07,
      "loss": 0.3656,
      "step": 1506
    },
    {
      "epoch": 12.88034188034188,
      "grad_norm": 2.5685003689848434,
      "learning_rate": 2.4233406119954756e-07,
      "loss": 0.1698,
      "step": 1507
    },
    {
      "epoch": 12.88888888888889,
      "grad_norm": 1.0793633165726377,
      "learning_rate": 2.404155751286988e-07,
      "loss": 0.2515,
      "step": 1508
    },
    {
      "epoch": 12.897435897435898,
      "grad_norm": 2.2961950625371492,
      "learning_rate": 2.385043296479195e-07,
      "loss": 0.1088,
      "step": 1509
    },
    {
      "epoch": 12.905982905982906,
      "grad_norm": 1.111817950970189,
      "learning_rate": 2.3660033088158646e-07,
      "loss": 0.3174,
      "step": 1510
    },
    {
      "epoch": 12.914529914529915,
      "grad_norm": 1.0582422873750743,
      "learning_rate": 2.3470358493085433e-07,
      "loss": 0.545,
      "step": 1511
    },
    {
      "epoch": 12.923076923076923,
      "grad_norm": 1.0443314605511564,
      "learning_rate": 2.3281409787363652e-07,
      "loss": 0.2725,
      "step": 1512
    },
    {
      "epoch": 12.931623931623932,
      "grad_norm": 2.47403877787119,
      "learning_rate": 2.309318757645873e-07,
      "loss": 0.1496,
      "step": 1513
    },
    {
      "epoch": 12.94017094017094,
      "grad_norm": 1.3943374589312603,
      "learning_rate": 2.2905692463508045e-07,
      "loss": 0.2497,
      "step": 1514
    },
    {
      "epoch": 12.948717948717949,
      "grad_norm": 0.9048792274362765,
      "learning_rate": 2.271892504931905e-07,
      "loss": 0.1958,
      "step": 1515
    },
    {
      "epoch": 12.957264957264957,
      "grad_norm": 1.3903380713051325,
      "learning_rate": 2.253288593236755e-07,
      "loss": 0.5352,
      "step": 1516
    },
    {
      "epoch": 12.965811965811966,
      "grad_norm": 1.312412248496653,
      "learning_rate": 2.234757570879534e-07,
      "loss": 0.4187,
      "step": 1517
    },
    {
      "epoch": 12.974358974358974,
      "grad_norm": 1.0728502906390496,
      "learning_rate": 2.2162994972408647e-07,
      "loss": 0.4241,
      "step": 1518
    },
    {
      "epoch": 12.982905982905983,
      "grad_norm": 1.4832237419691372,
      "learning_rate": 2.1979144314676239e-07,
      "loss": 0.3298,
      "step": 1519
    },
    {
      "epoch": 12.991452991452991,
      "grad_norm": 1.8596141275548963,
      "learning_rate": 2.17960243247273e-07,
      "loss": 0.2932,
      "step": 1520
    },
    {
      "epoch": 13.0,
      "grad_norm": 1.756329510002311,
      "learning_rate": 2.1613635589349756e-07,
      "loss": 0.3875,
      "step": 1521
    },
    {
      "epoch": 13.008547008547009,
      "grad_norm": 1.1505738852548575,
      "learning_rate": 2.1431978692988298e-07,
      "loss": 0.2159,
      "step": 1522
    },
    {
      "epoch": 13.017094017094017,
      "grad_norm": 1.0857319212781678,
      "learning_rate": 2.1251054217742484e-07,
      "loss": 0.2768,
      "step": 1523
    },
    {
      "epoch": 13.025641025641026,
      "grad_norm": 1.0267844636312484,
      "learning_rate": 2.107086274336484e-07,
      "loss": 0.3875,
      "step": 1524
    },
    {
      "epoch": 13.034188034188034,
      "grad_norm": 0.9343948903201362,
      "learning_rate": 2.0891404847259267e-07,
      "loss": 0.3016,
      "step": 1525
    },
    {
      "epoch": 13.042735042735043,
      "grad_norm": 1.155357237625711,
      "learning_rate": 2.0712681104478742e-07,
      "loss": 0.2082,
      "step": 1526
    },
    {
      "epoch": 13.051282051282051,
      "grad_norm": 1.1515917888558194,
      "learning_rate": 2.0534692087724017e-07,
      "loss": 0.4446,
      "step": 1527
    },
    {
      "epoch": 13.05982905982906,
      "grad_norm": 0.8190061036835171,
      "learning_rate": 2.0357438367341248e-07,
      "loss": 0.1272,
      "step": 1528
    },
    {
      "epoch": 13.068376068376068,
      "grad_norm": 0.7712922541814253,
      "learning_rate": 2.0180920511320562e-07,
      "loss": 0.0751,
      "step": 1529
    },
    {
      "epoch": 13.076923076923077,
      "grad_norm": 1.0294870961933071,
      "learning_rate": 2.0005139085293945e-07,
      "loss": 0.2942,
      "step": 1530
    },
    {
      "epoch": 13.085470085470085,
      "grad_norm": 0.8439613811790312,
      "learning_rate": 1.983009465253377e-07,
      "loss": 0.2488,
      "step": 1531
    },
    {
      "epoch": 13.094017094017094,
      "grad_norm": 1.1553682818767042,
      "learning_rate": 1.9655787773950691e-07,
      "loss": 0.4796,
      "step": 1532
    },
    {
      "epoch": 13.102564102564102,
      "grad_norm": 1.2396451404876954,
      "learning_rate": 1.9482219008091885e-07,
      "loss": 0.3161,
      "step": 1533
    },
    {
      "epoch": 13.11111111111111,
      "grad_norm": 0.8384235193749258,
      "learning_rate": 1.9309388911139427e-07,
      "loss": 0.2178,
      "step": 1534
    },
    {
      "epoch": 13.11965811965812,
      "grad_norm": 0.9678932445180488,
      "learning_rate": 1.9137298036908392e-07,
      "loss": 0.1476,
      "step": 1535
    },
    {
      "epoch": 13.128205128205128,
      "grad_norm": 1.14547754453679,
      "learning_rate": 1.896594693684503e-07,
      "loss": 0.3058,
      "step": 1536
    },
    {
      "epoch": 13.136752136752136,
      "grad_norm": 1.037919981668251,
      "learning_rate": 1.8795336160025157e-07,
      "loss": 0.2297,
      "step": 1537
    },
    {
      "epoch": 13.145299145299145,
      "grad_norm": 1.45584546731332,
      "learning_rate": 1.8625466253152314e-07,
      "loss": 0.6431,
      "step": 1538
    },
    {
      "epoch": 13.153846153846153,
      "grad_norm": 0.9426394319446211,
      "learning_rate": 1.8456337760555915e-07,
      "loss": 0.1778,
      "step": 1539
    },
    {
      "epoch": 13.162393162393162,
      "grad_norm": 1.3071195563971694,
      "learning_rate": 1.8287951224189555e-07,
      "loss": 0.3046,
      "step": 1540
    },
    {
      "epoch": 13.17094017094017,
      "grad_norm": 1.0013067940381606,
      "learning_rate": 1.8120307183629533e-07,
      "loss": 0.1752,
      "step": 1541
    },
    {
      "epoch": 13.179487179487179,
      "grad_norm": 1.009937942806977,
      "learning_rate": 1.7953406176072636e-07,
      "loss": 0.1444,
      "step": 1542
    },
    {
      "epoch": 13.188034188034187,
      "grad_norm": 0.907238022289693,
      "learning_rate": 1.778724873633486e-07,
      "loss": 0.2081,
      "step": 1543
    },
    {
      "epoch": 13.196581196581196,
      "grad_norm": 1.1561662315881491,
      "learning_rate": 1.7621835396849528e-07,
      "loss": 0.3784,
      "step": 1544
    },
    {
      "epoch": 13.205128205128204,
      "grad_norm": 1.0786060827716064,
      "learning_rate": 1.745716668766545e-07,
      "loss": 0.2069,
      "step": 1545
    },
    {
      "epoch": 13.213675213675213,
      "grad_norm": 1.143617253857377,
      "learning_rate": 1.7293243136445398e-07,
      "loss": 0.2093,
      "step": 1546
    },
    {
      "epoch": 13.222222222222221,
      "grad_norm": 0.9032057647054741,
      "learning_rate": 1.713006526846439e-07,
      "loss": 0.2647,
      "step": 1547
    },
    {
      "epoch": 13.23076923076923,
      "grad_norm": 1.1492208541704318,
      "learning_rate": 1.6967633606608082e-07,
      "loss": 0.2843,
      "step": 1548
    },
    {
      "epoch": 13.239316239316238,
      "grad_norm": 1.1662669216870896,
      "learning_rate": 1.6805948671370726e-07,
      "loss": 0.519,
      "step": 1549
    },
    {
      "epoch": 13.247863247863247,
      "grad_norm": 1.1127259115281096,
      "learning_rate": 1.6645010980854082e-07,
      "loss": 0.3714,
      "step": 1550
    },
    {
      "epoch": 13.256410256410255,
      "grad_norm": 0.78442378098213,
      "learning_rate": 1.6484821050765209e-07,
      "loss": 0.0919,
      "step": 1551
    },
    {
      "epoch": 13.264957264957266,
      "grad_norm": 1.1438475548638625,
      "learning_rate": 1.6325379394415168e-07,
      "loss": 0.4341,
      "step": 1552
    },
    {
      "epoch": 13.273504273504274,
      "grad_norm": 0.9868013923057376,
      "learning_rate": 1.6166686522717217e-07,
      "loss": 0.3411,
      "step": 1553
    },
    {
      "epoch": 13.282051282051283,
      "grad_norm": 0.906498367822162,
      "learning_rate": 1.600874294418528e-07,
      "loss": 0.1169,
      "step": 1554
    },
    {
      "epoch": 13.290598290598291,
      "grad_norm": 1.330656388274806,
      "learning_rate": 1.5851549164932118e-07,
      "loss": 0.3343,
      "step": 1555
    },
    {
      "epoch": 13.2991452991453,
      "grad_norm": 1.0982661982531574,
      "learning_rate": 1.569510568866803e-07,
      "loss": 0.1931,
      "step": 1556
    },
    {
      "epoch": 13.307692307692308,
      "grad_norm": 1.007238900807903,
      "learning_rate": 1.5539413016698923e-07,
      "loss": 0.3144,
      "step": 1557
    },
    {
      "epoch": 13.316239316239317,
      "grad_norm": 0.974750212176006,
      "learning_rate": 1.5384471647924781e-07,
      "loss": 0.435,
      "step": 1558
    },
    {
      "epoch": 13.324786324786325,
      "grad_norm": 1.181406302290183,
      "learning_rate": 1.5230282078838255e-07,
      "loss": 0.2805,
      "step": 1559
    },
    {
      "epoch": 13.333333333333334,
      "grad_norm": 1.0014146267263002,
      "learning_rate": 1.507684480352292e-07,
      "loss": 0.2635,
      "step": 1560
    },
    {
      "epoch": 13.341880341880342,
      "grad_norm": 1.2320491733226004,
      "learning_rate": 1.4924160313651598e-07,
      "loss": 0.3606,
      "step": 1561
    },
    {
      "epoch": 13.350427350427351,
      "grad_norm": 0.9911464162208959,
      "learning_rate": 1.4772229098485053e-07,
      "loss": 0.2405,
      "step": 1562
    },
    {
      "epoch": 13.35897435897436,
      "grad_norm": 1.2551361681677689,
      "learning_rate": 1.46210516448701e-07,
      "loss": 0.4468,
      "step": 1563
    },
    {
      "epoch": 13.367521367521368,
      "grad_norm": 1.0520156172205075,
      "learning_rate": 1.447062843723826e-07,
      "loss": 0.2223,
      "step": 1564
    },
    {
      "epoch": 13.376068376068377,
      "grad_norm": 0.9628315407802782,
      "learning_rate": 1.432095995760424e-07,
      "loss": 0.2264,
      "step": 1565
    },
    {
      "epoch": 13.384615384615385,
      "grad_norm": 1.2836910046347718,
      "learning_rate": 1.417204668556421e-07,
      "loss": 0.5595,
      "step": 1566
    },
    {
      "epoch": 13.393162393162394,
      "grad_norm": 1.1889920690605311,
      "learning_rate": 1.402388909829447e-07,
      "loss": 0.2119,
      "step": 1567
    },
    {
      "epoch": 13.401709401709402,
      "grad_norm": 1.0303893972646192,
      "learning_rate": 1.387648767054961e-07,
      "loss": 0.2094,
      "step": 1568
    },
    {
      "epoch": 13.41025641025641,
      "grad_norm": 0.8281095566416786,
      "learning_rate": 1.3729842874661365e-07,
      "loss": 0.3305,
      "step": 1569
    },
    {
      "epoch": 13.418803418803419,
      "grad_norm": 1.094637661606762,
      "learning_rate": 1.35839551805369e-07,
      "loss": 0.3879,
      "step": 1570
    },
    {
      "epoch": 13.427350427350428,
      "grad_norm": 1.1817341481124783,
      "learning_rate": 1.3438825055657246e-07,
      "loss": 0.2283,
      "step": 1571
    },
    {
      "epoch": 13.435897435897436,
      "grad_norm": 1.167606047139076,
      "learning_rate": 1.3294452965076031e-07,
      "loss": 0.3251,
      "step": 1572
    },
    {
      "epoch": 13.444444444444445,
      "grad_norm": 1.0632336244379381,
      "learning_rate": 1.31508393714177e-07,
      "loss": 0.3327,
      "step": 1573
    },
    {
      "epoch": 13.452991452991453,
      "grad_norm": 1.1356919410815811,
      "learning_rate": 1.3007984734876217e-07,
      "loss": 0.1932,
      "step": 1574
    },
    {
      "epoch": 13.461538461538462,
      "grad_norm": 1.1019667814822336,
      "learning_rate": 1.286588951321363e-07,
      "loss": 0.4479,
      "step": 1575
    },
    {
      "epoch": 13.47008547008547,
      "grad_norm": 1.2456887117109419,
      "learning_rate": 1.272455416175844e-07,
      "loss": 0.2,
      "step": 1576
    },
    {
      "epoch": 13.478632478632479,
      "grad_norm": 0.8768840020349393,
      "learning_rate": 1.258397913340423e-07,
      "loss": 0.1898,
      "step": 1577
    },
    {
      "epoch": 13.487179487179487,
      "grad_norm": 0.8610819031956608,
      "learning_rate": 1.2444164878608307e-07,
      "loss": 0.1477,
      "step": 1578
    },
    {
      "epoch": 13.495726495726496,
      "grad_norm": 1.0823294789754605,
      "learning_rate": 1.2305111845390043e-07,
      "loss": 0.1623,
      "step": 1579
    },
    {
      "epoch": 13.504273504273504,
      "grad_norm": 0.9631874294889519,
      "learning_rate": 1.2166820479329572e-07,
      "loss": 0.3338,
      "step": 1580
    },
    {
      "epoch": 13.512820512820513,
      "grad_norm": 1.0530613205307584,
      "learning_rate": 1.2029291223566413e-07,
      "loss": 0.1973,
      "step": 1581
    },
    {
      "epoch": 13.521367521367521,
      "grad_norm": 0.8717117935351345,
      "learning_rate": 1.1892524518797993e-07,
      "loss": 0.2069,
      "step": 1582
    },
    {
      "epoch": 13.52991452991453,
      "grad_norm": 0.9608587841521585,
      "learning_rate": 1.1756520803278188e-07,
      "loss": 0.2495,
      "step": 1583
    },
    {
      "epoch": 13.538461538461538,
      "grad_norm": 1.5519958028368797,
      "learning_rate": 1.1621280512815941e-07,
      "loss": 0.171,
      "step": 1584
    },
    {
      "epoch": 13.547008547008547,
      "grad_norm": 0.9131328583735898,
      "learning_rate": 1.1486804080773878e-07,
      "loss": 0.2453,
      "step": 1585
    },
    {
      "epoch": 13.555555555555555,
      "grad_norm": 0.99728593250585,
      "learning_rate": 1.1353091938067024e-07,
      "loss": 0.2087,
      "step": 1586
    },
    {
      "epoch": 13.564102564102564,
      "grad_norm": 1.0702964515597055,
      "learning_rate": 1.1220144513161197e-07,
      "loss": 0.3104,
      "step": 1587
    },
    {
      "epoch": 13.572649572649572,
      "grad_norm": 1.077157969082805,
      "learning_rate": 1.1087962232071814e-07,
      "loss": 0.3807,
      "step": 1588
    },
    {
      "epoch": 13.581196581196581,
      "grad_norm": 1.1826954134832977,
      "learning_rate": 1.0956545518362532e-07,
      "loss": 0.2355,
      "step": 1589
    },
    {
      "epoch": 13.58974358974359,
      "grad_norm": 1.156905738771152,
      "learning_rate": 1.0825894793143721e-07,
      "loss": 0.1391,
      "step": 1590
    },
    {
      "epoch": 13.598290598290598,
      "grad_norm": 1.3506966457826057,
      "learning_rate": 1.0696010475071267e-07,
      "loss": 0.2835,
      "step": 1591
    },
    {
      "epoch": 13.606837606837606,
      "grad_norm": 1.3963385583754653,
      "learning_rate": 1.0566892980345245e-07,
      "loss": 0.4713,
      "step": 1592
    },
    {
      "epoch": 13.615384615384615,
      "grad_norm": 1.1187293180307003,
      "learning_rate": 1.0438542722708444e-07,
      "loss": 0.4176,
      "step": 1593
    },
    {
      "epoch": 13.623931623931623,
      "grad_norm": 1.2856600898305055,
      "learning_rate": 1.0310960113445179e-07,
      "loss": 0.4702,
      "step": 1594
    },
    {
      "epoch": 13.632478632478632,
      "grad_norm": 1.1958490213395467,
      "learning_rate": 1.0184145561379949e-07,
      "loss": 0.1315,
      "step": 1595
    },
    {
      "epoch": 13.64102564102564,
      "grad_norm": 0.7705316255345382,
      "learning_rate": 1.0058099472876004e-07,
      "loss": 0.1864,
      "step": 1596
    },
    {
      "epoch": 13.649572649572649,
      "grad_norm": 0.8840288370368232,
      "learning_rate": 9.932822251834173e-08,
      "loss": 0.338,
      "step": 1597
    },
    {
      "epoch": 13.658119658119658,
      "grad_norm": 1.0348235453397847,
      "learning_rate": 9.808314299691591e-08,
      "loss": 0.3332,
      "step": 1598
    },
    {
      "epoch": 13.666666666666666,
      "grad_norm": 1.1828653933998128,
      "learning_rate": 9.684576015420277e-08,
      "loss": 0.3737,
      "step": 1599
    },
    {
      "epoch": 13.675213675213675,
      "grad_norm": 1.0091289878163927,
      "learning_rate": 9.561607795526007e-08,
      "loss": 0.2419,
      "step": 1600
    },
    {
      "epoch": 13.683760683760683,
      "grad_norm": 1.0828857407343973,
      "learning_rate": 9.439410034046942e-08,
      "loss": 0.2528,
      "step": 1601
    },
    {
      "epoch": 13.692307692307692,
      "grad_norm": 1.2959615159059084,
      "learning_rate": 9.317983122552332e-08,
      "loss": 0.1915,
      "step": 1602
    },
    {
      "epoch": 13.7008547008547,
      "grad_norm": 1.379162456698023,
      "learning_rate": 9.197327450141402e-08,
      "loss": 0.3602,
      "step": 1603
    },
    {
      "epoch": 13.709401709401709,
      "grad_norm": 1.1832886065543922,
      "learning_rate": 9.077443403441994e-08,
      "loss": 0.3007,
      "step": 1604
    },
    {
      "epoch": 13.717948717948717,
      "grad_norm": 1.159394233790722,
      "learning_rate": 8.958331366609424e-08,
      "loss": 0.2803,
      "step": 1605
    },
    {
      "epoch": 13.726495726495726,
      "grad_norm": 0.9755941982498277,
      "learning_rate": 8.839991721325047e-08,
      "loss": 0.2278,
      "step": 1606
    },
    {
      "epoch": 13.735042735042736,
      "grad_norm": 1.1557656098603584,
      "learning_rate": 8.72242484679528e-08,
      "loss": 0.3131,
      "step": 1607
    },
    {
      "epoch": 13.743589743589745,
      "grad_norm": 1.1722877958343039,
      "learning_rate": 8.605631119750297e-08,
      "loss": 0.3569,
      "step": 1608
    },
    {
      "epoch": 13.752136752136753,
      "grad_norm": 0.9489527958140329,
      "learning_rate": 8.489610914442697e-08,
      "loss": 0.3719,
      "step": 1609
    },
    {
      "epoch": 13.760683760683762,
      "grad_norm": 0.9781463695410046,
      "learning_rate": 8.374364602646512e-08,
      "loss": 0.2721,
      "step": 1610
    },
    {
      "epoch": 13.76923076923077,
      "grad_norm": 1.2276231195845517,
      "learning_rate": 8.259892553655946e-08,
      "loss": 0.2396,
      "step": 1611
    },
    {
      "epoch": 13.777777777777779,
      "grad_norm": 0.8125741331372663,
      "learning_rate": 8.146195134284052e-08,
      "loss": 0.282,
      "step": 1612
    },
    {
      "epoch": 13.786324786324787,
      "grad_norm": 1.0716815459854674,
      "learning_rate": 8.033272708861673e-08,
      "loss": 0.3969,
      "step": 1613
    },
    {
      "epoch": 13.794871794871796,
      "grad_norm": 0.9978778449841479,
      "learning_rate": 7.921125639236416e-08,
      "loss": 0.2373,
      "step": 1614
    },
    {
      "epoch": 13.803418803418804,
      "grad_norm": 0.921081013180687,
      "learning_rate": 7.809754284771181e-08,
      "loss": 0.225,
      "step": 1615
    },
    {
      "epoch": 13.811965811965813,
      "grad_norm": 1.3841256382567415,
      "learning_rate": 7.699159002343248e-08,
      "loss": 0.2923,
      "step": 1616
    },
    {
      "epoch": 13.820512820512821,
      "grad_norm": 1.0531740575034831,
      "learning_rate": 7.589340146343077e-08,
      "loss": 0.1851,
      "step": 1617
    },
    {
      "epoch": 13.82905982905983,
      "grad_norm": 1.052402359782648,
      "learning_rate": 7.48029806867312e-08,
      "loss": 0.2215,
      "step": 1618
    },
    {
      "epoch": 13.837606837606838,
      "grad_norm": 0.9474502677015596,
      "learning_rate": 7.372033118746708e-08,
      "loss": 0.3537,
      "step": 1619
    },
    {
      "epoch": 13.846153846153847,
      "grad_norm": 1.0433066188917253,
      "learning_rate": 7.264545643486997e-08,
      "loss": 0.192,
      "step": 1620
    },
    {
      "epoch": 13.854700854700855,
      "grad_norm": 1.0503434380249135,
      "learning_rate": 7.157835987325807e-08,
      "loss": 0.4817,
      "step": 1621
    },
    {
      "epoch": 13.863247863247864,
      "grad_norm": 1.0208969742480898,
      "learning_rate": 7.051904492202472e-08,
      "loss": 0.1586,
      "step": 1622
    },
    {
      "epoch": 13.871794871794872,
      "grad_norm": 1.0691421470631344,
      "learning_rate": 6.946751497562909e-08,
      "loss": 0.3521,
      "step": 1623
    },
    {
      "epoch": 13.88034188034188,
      "grad_norm": 1.2245483605550895,
      "learning_rate": 6.842377340358252e-08,
      "loss": 0.5142,
      "step": 1624
    },
    {
      "epoch": 13.88888888888889,
      "grad_norm": 1.0065600482948194,
      "learning_rate": 6.738782355044048e-08,
      "loss": 0.3602,
      "step": 1625
    },
    {
      "epoch": 13.897435897435898,
      "grad_norm": 1.0654700012366067,
      "learning_rate": 6.635966873579063e-08,
      "loss": 0.3004,
      "step": 1626
    },
    {
      "epoch": 13.905982905982906,
      "grad_norm": 1.3042405801995822,
      "learning_rate": 6.5339312254242e-08,
      "loss": 0.3589,
      "step": 1627
    },
    {
      "epoch": 13.914529914529915,
      "grad_norm": 1.1138127249535994,
      "learning_rate": 6.432675737541499e-08,
      "loss": 0.3743,
      "step": 1628
    },
    {
      "epoch": 13.923076923076923,
      "grad_norm": 0.809005507517847,
      "learning_rate": 6.332200734393057e-08,
      "loss": 0.1839,
      "step": 1629
    },
    {
      "epoch": 13.931623931623932,
      "grad_norm": 1.0368280042421736,
      "learning_rate": 6.232506537939942e-08,
      "loss": 0.1791,
      "step": 1630
    },
    {
      "epoch": 13.94017094017094,
      "grad_norm": 1.0270604550679647,
      "learning_rate": 6.13359346764128e-08,
      "loss": 0.4125,
      "step": 1631
    },
    {
      "epoch": 13.948717948717949,
      "grad_norm": 1.1754201805715523,
      "learning_rate": 6.035461840453116e-08,
      "loss": 0.4687,
      "step": 1632
    },
    {
      "epoch": 13.957264957264957,
      "grad_norm": 0.8604966197339347,
      "learning_rate": 5.938111970827526e-08,
      "loss": 0.2878,
      "step": 1633
    },
    {
      "epoch": 13.965811965811966,
      "grad_norm": 1.1086668737564982,
      "learning_rate": 5.841544170711422e-08,
      "loss": 0.1041,
      "step": 1634
    },
    {
      "epoch": 13.974358974358974,
      "grad_norm": 0.9009406977063691,
      "learning_rate": 5.745758749545749e-08,
      "loss": 0.1912,
      "step": 1635
    },
    {
      "epoch": 13.982905982905983,
      "grad_norm": 1.1802740851626485,
      "learning_rate": 5.650756014264347e-08,
      "loss": 0.4936,
      "step": 1636
    },
    {
      "epoch": 13.991452991452991,
      "grad_norm": 1.2300056190706103,
      "learning_rate": 5.556536269293006e-08,
      "loss": 0.4424,
      "step": 1637
    },
    {
      "epoch": 14.0,
      "grad_norm": 1.0137995981094534,
      "learning_rate": 5.463099816548578e-08,
      "loss": 0.3284,
      "step": 1638
    },
    {
      "epoch": 14.008547008547009,
      "grad_norm": 0.9709594304196405,
      "learning_rate": 5.3704469554379527e-08,
      "loss": 0.2297,
      "step": 1639
    },
    {
      "epoch": 14.017094017094017,
      "grad_norm": 1.1299871999082094,
      "learning_rate": 5.278577982857025e-08,
      "loss": 0.3918,
      "step": 1640
    },
    {
      "epoch": 14.025641025641026,
      "grad_norm": 1.1449136156565218,
      "learning_rate": 5.1874931931897854e-08,
      "loss": 0.3966,
      "step": 1641
    },
    {
      "epoch": 14.034188034188034,
      "grad_norm": 0.9844915462506715,
      "learning_rate": 5.097192878307455e-08,
      "loss": 0.2968,
      "step": 1642
    },
    {
      "epoch": 14.042735042735043,
      "grad_norm": 1.0431617240132687,
      "learning_rate": 5.0076773275675174e-08,
      "loss": 0.4419,
      "step": 1643
    },
    {
      "epoch": 14.051282051282051,
      "grad_norm": 1.0470663059857832,
      "learning_rate": 4.91894682781266e-08,
      "loss": 0.2256,
      "step": 1644
    },
    {
      "epoch": 14.05982905982906,
      "grad_norm": 1.0482134797568037,
      "learning_rate": 4.831001663370083e-08,
      "loss": 0.4594,
      "step": 1645
    },
    {
      "epoch": 14.068376068376068,
      "grad_norm": 0.9581719946948042,
      "learning_rate": 4.743842116050334e-08,
      "loss": 0.2997,
      "step": 1646
    },
    {
      "epoch": 14.076923076923077,
      "grad_norm": 1.1271714744759658,
      "learning_rate": 4.657468465146642e-08,
      "loss": 0.3799,
      "step": 1647
    },
    {
      "epoch": 14.085470085470085,
      "grad_norm": 0.9991717412883093,
      "learning_rate": 4.571880987433886e-08,
      "loss": 0.3411,
      "step": 1648
    },
    {
      "epoch": 14.094017094017094,
      "grad_norm": 0.9440052310542084,
      "learning_rate": 4.487079957167767e-08,
      "loss": 0.2588,
      "step": 1649
    },
    {
      "epoch": 14.102564102564102,
      "grad_norm": 1.1730488791362796,
      "learning_rate": 4.40306564608381e-08,
      "loss": 0.264,
      "step": 1650
    },
    {
      "epoch": 14.11111111111111,
      "grad_norm": 1.3216740626779884,
      "learning_rate": 4.319838323396691e-08,
      "loss": 0.1883,
      "step": 1651
    },
    {
      "epoch": 14.11965811965812,
      "grad_norm": 0.7823448055638057,
      "learning_rate": 4.237398255799191e-08,
      "loss": 0.1273,
      "step": 1652
    },
    {
      "epoch": 14.128205128205128,
      "grad_norm": 0.9551378725158102,
      "learning_rate": 4.155745707461467e-08,
      "loss": 0.3024,
      "step": 1653
    },
    {
      "epoch": 14.136752136752136,
      "grad_norm": 1.0106178858369212,
      "learning_rate": 4.0748809400301403e-08,
      "loss": 0.3491,
      "step": 1654
    },
    {
      "epoch": 14.145299145299145,
      "grad_norm": 1.3301723397017644,
      "learning_rate": 3.994804212627462e-08,
      "loss": 0.4849,
      "step": 1655
    },
    {
      "epoch": 14.153846153846153,
      "grad_norm": 1.0907574351446336,
      "learning_rate": 3.9155157818505654e-08,
      "loss": 0.2302,
      "step": 1656
    },
    {
      "epoch": 14.162393162393162,
      "grad_norm": 1.0599072945861983,
      "learning_rate": 3.8370159017704636e-08,
      "loss": 0.4176,
      "step": 1657
    },
    {
      "epoch": 14.17094017094017,
      "grad_norm": 1.1921847870816986,
      "learning_rate": 3.759304823931359e-08,
      "loss": 0.4857,
      "step": 1658
    },
    {
      "epoch": 14.179487179487179,
      "grad_norm": 1.0941612458721797,
      "learning_rate": 3.682382797349976e-08,
      "loss": 0.4745,
      "step": 1659
    },
    {
      "epoch": 14.188034188034187,
      "grad_norm": 1.1544133425051988,
      "learning_rate": 3.6062500685143943e-08,
      "loss": 0.2755,
      "step": 1660
    },
    {
      "epoch": 14.196581196581196,
      "grad_norm": 1.2723721479857397,
      "learning_rate": 3.5309068813836056e-08,
      "loss": 0.6096,
      "step": 1661
    },
    {
      "epoch": 14.205128205128204,
      "grad_norm": 1.047213471583824,
      "learning_rate": 3.4563534773866256e-08,
      "loss": 0.4001,
      "step": 1662
    },
    {
      "epoch": 14.213675213675213,
      "grad_norm": 1.1591614287101293,
      "learning_rate": 3.382590095421606e-08,
      "loss": 0.2664,
      "step": 1663
    },
    {
      "epoch": 14.222222222222221,
      "grad_norm": 0.9914669953998568,
      "learning_rate": 3.309616971855195e-08,
      "loss": 0.1926,
      "step": 1664
    },
    {
      "epoch": 14.23076923076923,
      "grad_norm": 1.0844370003359844,
      "learning_rate": 3.237434340521789e-08,
      "loss": 0.2233,
      "step": 1665
    },
    {
      "epoch": 14.239316239316238,
      "grad_norm": 1.0158802297687415,
      "learning_rate": 3.166042432722671e-08,
      "loss": 0.2643,
      "step": 1666
    },
    {
      "epoch": 14.247863247863247,
      "grad_norm": 1.2925442972530232,
      "learning_rate": 3.095441477225347e-08,
      "loss": 0.2349,
      "step": 1667
    },
    {
      "epoch": 14.256410256410255,
      "grad_norm": 0.9924388816355043,
      "learning_rate": 3.025631700262877e-08,
      "loss": 0.2553,
      "step": 1668
    },
    {
      "epoch": 14.264957264957266,
      "grad_norm": 1.069284498726041,
      "learning_rate": 2.9566133255329864e-08,
      "loss": 0.3725,
      "step": 1669
    },
    {
      "epoch": 14.273504273504274,
      "grad_norm": 1.1929109155842428,
      "learning_rate": 2.888386574197488e-08,
      "loss": 0.2351,
      "step": 1670
    },
    {
      "epoch": 14.282051282051283,
      "grad_norm": 0.9617405737468387,
      "learning_rate": 2.8209516648814996e-08,
      "loss": 0.0816,
      "step": 1671
    },
    {
      "epoch": 14.290598290598291,
      "grad_norm": 0.9147539896413376,
      "learning_rate": 2.7543088136727792e-08,
      "loss": 0.3169,
      "step": 1672
    },
    {
      "epoch": 14.2991452991453,
      "grad_norm": 0.795512505038877,
      "learning_rate": 2.688458234121033e-08,
      "loss": 0.0757,
      "step": 1673
    },
    {
      "epoch": 14.307692307692308,
      "grad_norm": 0.9502257124356851,
      "learning_rate": 2.6234001372372196e-08,
      "loss": 0.3862,
      "step": 1674
    },
    {
      "epoch": 14.316239316239317,
      "grad_norm": 1.274230365165183,
      "learning_rate": 2.5591347314928572e-08,
      "loss": 0.2647,
      "step": 1675
    },
    {
      "epoch": 14.324786324786325,
      "grad_norm": 0.967532783276075,
      "learning_rate": 2.495662222819356e-08,
      "loss": 0.3305,
      "step": 1676
    },
    {
      "epoch": 14.333333333333334,
      "grad_norm": 1.3224798768697412,
      "learning_rate": 2.4329828146074096e-08,
      "loss": 0.3332,
      "step": 1677
    },
    {
      "epoch": 14.341880341880342,
      "grad_norm": 1.0727985919335998,
      "learning_rate": 2.3710967077063275e-08,
      "loss": 0.3429,
      "step": 1678
    },
    {
      "epoch": 14.350427350427351,
      "grad_norm": 1.2004949487299452,
      "learning_rate": 2.310004100423313e-08,
      "loss": 0.287,
      "step": 1679
    },
    {
      "epoch": 14.35897435897436,
      "grad_norm": 1.0812921492853662,
      "learning_rate": 2.2497051885228825e-08,
      "loss": 0.3371,
      "step": 1680
    },
    {
      "epoch": 14.367521367521368,
      "grad_norm": 0.9348664846892308,
      "learning_rate": 2.190200165226336e-08,
      "loss": 0.1468,
      "step": 1681
    },
    {
      "epoch": 14.376068376068377,
      "grad_norm": 0.979520377810362,
      "learning_rate": 2.131489221210953e-08,
      "loss": 0.2964,
      "step": 1682
    },
    {
      "epoch": 14.384615384615385,
      "grad_norm": 0.9953585036774258,
      "learning_rate": 2.0735725446094924e-08,
      "loss": 0.2317,
      "step": 1683
    },
    {
      "epoch": 14.393162393162394,
      "grad_norm": 1.0211981474940792,
      "learning_rate": 2.016450321009611e-08,
      "loss": 0.2613,
      "step": 1684
    },
    {
      "epoch": 14.401709401709402,
      "grad_norm": 0.9977623186417789,
      "learning_rate": 1.9601227334531958e-08,
      "loss": 0.2357,
      "step": 1685
    },
    {
      "epoch": 14.41025641025641,
      "grad_norm": 1.0577263179561622,
      "learning_rate": 1.904589962435782e-08,
      "loss": 0.328,
      "step": 1686
    },
    {
      "epoch": 14.418803418803419,
      "grad_norm": 0.9932052880875452,
      "learning_rate": 1.8498521859060814e-08,
      "loss": 0.2766,
      "step": 1687
    },
    {
      "epoch": 14.427350427350428,
      "grad_norm": 1.2058132888141184,
      "learning_rate": 1.795909579265259e-08,
      "loss": 0.4555,
      "step": 1688
    },
    {
      "epoch": 14.435897435897436,
      "grad_norm": 0.9202271237956986,
      "learning_rate": 1.7427623153664364e-08,
      "loss": 0.2043,
      "step": 1689
    },
    {
      "epoch": 14.444444444444445,
      "grad_norm": 1.0762171371983105,
      "learning_rate": 1.6904105645142443e-08,
      "loss": 0.39,
      "step": 1690
    },
    {
      "epoch": 14.452991452991453,
      "grad_norm": 1.1389216220816722,
      "learning_rate": 1.638854494464104e-08,
      "loss": 0.2034,
      "step": 1691
    },
    {
      "epoch": 14.461538461538462,
      "grad_norm": 0.8740065264182487,
      "learning_rate": 1.5880942704217528e-08,
      "loss": 0.2881,
      "step": 1692
    },
    {
      "epoch": 14.47008547008547,
      "grad_norm": 0.7538828934098073,
      "learning_rate": 1.5381300550427748e-08,
      "loss": 0.1935,
      "step": 1693
    },
    {
      "epoch": 14.478632478632479,
      "grad_norm": 0.9410330802083953,
      "learning_rate": 1.4889620084319878e-08,
      "loss": 0.3392,
      "step": 1694
    },
    {
      "epoch": 14.487179487179487,
      "grad_norm": 1.0121163042552828,
      "learning_rate": 1.4405902881430289e-08,
      "loss": 0.1335,
      "step": 1695
    },
    {
      "epoch": 14.495726495726496,
      "grad_norm": 1.072963591656099,
      "learning_rate": 1.393015049177715e-08,
      "loss": 0.2879,
      "step": 1696
    },
    {
      "epoch": 14.504273504273504,
      "grad_norm": 0.9404008286999908,
      "learning_rate": 1.3462364439857379e-08,
      "loss": 0.3634,
      "step": 1697
    },
    {
      "epoch": 14.512820512820513,
      "grad_norm": 1.075575665244736,
      "learning_rate": 1.3002546224639146e-08,
      "loss": 0.3437,
      "step": 1698
    },
    {
      "epoch": 14.521367521367521,
      "grad_norm": 0.9955741016463133,
      "learning_rate": 1.2550697319560211e-08,
      "loss": 0.521,
      "step": 1699
    },
    {
      "epoch": 14.52991452991453,
      "grad_norm": 0.9076329314518284,
      "learning_rate": 1.2106819172520434e-08,
      "loss": 0.3091,
      "step": 1700
    },
    {
      "epoch": 14.538461538461538,
      "grad_norm": 0.9706053653755294,
      "learning_rate": 1.1670913205878431e-08,
      "loss": 0.2192,
      "step": 1701
    },
    {
      "epoch": 14.547008547008547,
      "grad_norm": 0.9722292014471331,
      "learning_rate": 1.1242980816447147e-08,
      "loss": 0.1934,
      "step": 1702
    },
    {
      "epoch": 14.555555555555555,
      "grad_norm": 1.2950023777523025,
      "learning_rate": 1.0823023375489128e-08,
      "loss": 0.2387,
      "step": 1703
    },
    {
      "epoch": 14.564102564102564,
      "grad_norm": 0.9826157475331373,
      "learning_rate": 1.0411042228711254e-08,
      "loss": 0.1015,
      "step": 1704
    },
    {
      "epoch": 14.572649572649572,
      "grad_norm": 1.202253302780247,
      "learning_rate": 1.0007038696262517e-08,
      "loss": 0.4506,
      "step": 1705
    },
    {
      "epoch": 14.581196581196581,
      "grad_norm": 1.0946650709225376,
      "learning_rate": 9.611014072727354e-09,
      "loss": 0.246,
      "step": 1706
    },
    {
      "epoch": 14.58974358974359,
      "grad_norm": 1.0774018997500412,
      "learning_rate": 9.222969627123435e-09,
      "loss": 0.4581,
      "step": 1707
    },
    {
      "epoch": 14.598290598290598,
      "grad_norm": 1.3051886054710957,
      "learning_rate": 8.842906602896661e-09,
      "loss": 0.2653,
      "step": 1708
    },
    {
      "epoch": 14.606837606837606,
      "grad_norm": 0.9800272459228091,
      "learning_rate": 8.470826217917006e-09,
      "loss": 0.1108,
      "step": 1709
    },
    {
      "epoch": 14.615384615384615,
      "grad_norm": 0.9277483890590144,
      "learning_rate": 8.106729664475178e-09,
      "loss": 0.4336,
      "step": 1710
    },
    {
      "epoch": 14.623931623931623,
      "grad_norm": 1.0716060943444674,
      "learning_rate": 7.750618109278464e-09,
      "loss": 0.2783,
      "step": 1711
    },
    {
      "epoch": 14.632478632478632,
      "grad_norm": 0.9851690922805892,
      "learning_rate": 7.402492693447671e-09,
      "loss": 0.1646,
      "step": 1712
    },
    {
      "epoch": 14.64102564102564,
      "grad_norm": 1.491798773298588,
      "learning_rate": 7.062354532512416e-09,
      "loss": 0.5989,
      "step": 1713
    },
    {
      "epoch": 14.649572649572649,
      "grad_norm": 0.9900391674662283,
      "learning_rate": 6.730204716407507e-09,
      "loss": 0.3693,
      "step": 1714
    },
    {
      "epoch": 14.658119658119658,
      "grad_norm": 1.0409210475517114,
      "learning_rate": 6.406044309471004e-09,
      "loss": 0.1636,
      "step": 1715
    },
    {
      "epoch": 14.666666666666666,
      "grad_norm": 1.1496312917251512,
      "learning_rate": 6.089874350439507e-09,
      "loss": 0.3444,
      "step": 1716
    },
    {
      "epoch": 14.675213675213675,
      "grad_norm": 1.2119259909711284,
      "learning_rate": 5.781695852444258e-09,
      "loss": 0.4824,
      "step": 1717
    },
    {
      "epoch": 14.683760683760683,
      "grad_norm": 1.098314603031555,
      "learning_rate": 5.481509803009766e-09,
      "loss": 0.2783,
      "step": 1718
    },
    {
      "epoch": 14.692307692307692,
      "grad_norm": 0.7245471807790067,
      "learning_rate": 5.189317164049634e-09,
      "loss": 0.0919,
      "step": 1719
    },
    {
      "epoch": 14.7008547008547,
      "grad_norm": 1.09639322805763,
      "learning_rate": 4.905118871862402e-09,
      "loss": 0.3483,
      "step": 1720
    },
    {
      "epoch": 14.709401709401709,
      "grad_norm": 1.0581128651122704,
      "learning_rate": 4.62891583713071e-09,
      "loss": 0.09,
      "step": 1721
    },
    {
      "epoch": 14.717948717948717,
      "grad_norm": 1.0395913771941103,
      "learning_rate": 4.3607089449165806e-09,
      "loss": 0.3622,
      "step": 1722
    },
    {
      "epoch": 14.726495726495726,
      "grad_norm": 1.1486440562166906,
      "learning_rate": 4.100499054659757e-09,
      "loss": 0.2662,
      "step": 1723
    },
    {
      "epoch": 14.735042735042736,
      "grad_norm": 0.8951556891047434,
      "learning_rate": 3.848287000174089e-09,
      "loss": 0.3152,
      "step": 1724
    },
    {
      "epoch": 14.743589743589745,
      "grad_norm": 1.139371446672755,
      "learning_rate": 3.6040735896455957e-09,
      "loss": 0.273,
      "step": 1725
    },
    {
      "epoch": 14.752136752136753,
      "grad_norm": 0.8938981908344934,
      "learning_rate": 3.367859605628854e-09,
      "loss": 0.2162,
      "step": 1726
    },
    {
      "epoch": 14.760683760683762,
      "grad_norm": 1.1693798769445003,
      "learning_rate": 3.139645805046165e-09,
      "loss": 0.2869,
      "step": 1727
    },
    {
      "epoch": 14.76923076923077,
      "grad_norm": 0.9333554324995236,
      "learning_rate": 2.919432919183396e-09,
      "loss": 0.1761,
      "step": 1728
    },
    {
      "epoch": 14.777777777777779,
      "grad_norm": 1.24121704933367,
      "learning_rate": 2.7072216536885855e-09,
      "loss": 0.3746,
      "step": 1729
    },
    {
      "epoch": 14.786324786324787,
      "grad_norm": 1.1976434166923222,
      "learning_rate": 2.5030126885694505e-09,
      "loss": 0.4664,
      "step": 1730
    },
    {
      "epoch": 14.794871794871796,
      "grad_norm": 0.7888451308473411,
      "learning_rate": 2.3068066781908873e-09,
      "loss": 0.1526,
      "step": 1731
    },
    {
      "epoch": 14.803418803418804,
      "grad_norm": 1.0609440995043917,
      "learning_rate": 2.118604251273859e-09,
      "loss": 0.1695,
      "step": 1732
    },
    {
      "epoch": 14.811965811965813,
      "grad_norm": 1.1284468963364922,
      "learning_rate": 1.9384060108923463e-09,
      "loss": 0.4076,
      "step": 1733
    },
    {
      "epoch": 14.820512820512821,
      "grad_norm": 0.9200801478575613,
      "learning_rate": 1.766212534471401e-09,
      "loss": 0.1048,
      "step": 1734
    },
    {
      "epoch": 14.82905982905983,
      "grad_norm": 1.0748364878920569,
      "learning_rate": 1.6020243737865926e-09,
      "loss": 0.4537,
      "step": 1735
    },
    {
      "epoch": 14.837606837606838,
      "grad_norm": 1.111423242662358,
      "learning_rate": 1.4458420549606777e-09,
      "loss": 0.2354,
      "step": 1736
    },
    {
      "epoch": 14.846153846153847,
      "grad_norm": 0.9085030432997364,
      "learning_rate": 1.297666078462767e-09,
      "loss": 0.4356,
      "step": 1737
    },
    {
      "epoch": 14.854700854700855,
      "grad_norm": 0.7275980986093856,
      "learning_rate": 1.1574969191061047e-09,
      "loss": 0.1204,
      "step": 1738
    },
    {
      "epoch": 14.863247863247864,
      "grad_norm": 0.9569122217056513,
      "learning_rate": 1.0253350260480688e-09,
      "loss": 0.1361,
      "step": 1739
    },
    {
      "epoch": 14.871794871794872,
      "grad_norm": 1.1601832719859042,
      "learning_rate": 9.011808227865626e-10,
      "loss": 0.384,
      "step": 1740
    },
    {
      "epoch": 14.88034188034188,
      "grad_norm": 0.884213582013597,
      "learning_rate": 7.850347071597376e-10,
      "loss": 0.2421,
      "step": 1741
    },
    {
      "epoch": 14.88888888888889,
      "grad_norm": 0.8832450863042082,
      "learning_rate": 6.768970513457151e-10,
      "loss": 0.1231,
      "step": 1742
    },
    {
      "epoch": 14.897435897435898,
      "grad_norm": 1.3622457616085866,
      "learning_rate": 5.767682018595344e-10,
      "loss": 0.3513,
      "step": 1743
    },
    {
      "epoch": 14.905982905982906,
      "grad_norm": 1.055014911090151,
      "learning_rate": 4.846484795528739e-10,
      "loss": 0.2323,
      "step": 1744
    },
    {
      "epoch": 14.914529914529915,
      "grad_norm": 0.9330647451830272,
      "learning_rate": 4.0053817961321903e-10,
      "loss": 0.1243,
      "step": 1745
    },
    {
      "epoch": 14.923076923076923,
      "grad_norm": 0.9676723236179351,
      "learning_rate": 3.2443757156330746e-10,
      "loss": 0.2195,
      "step": 1746
    },
    {
      "epoch": 14.931623931623932,
      "grad_norm": 1.0828377737529897,
      "learning_rate": 2.563468992586304e-10,
      "loss": 0.3462,
      "step": 1747
    },
    {
      "epoch": 14.94017094017094,
      "grad_norm": 0.8574676515336324,
      "learning_rate": 1.9626638088854344e-10,
      "loss": 0.1641,
      "step": 1748
    },
    {
      "epoch": 14.948717948717949,
      "grad_norm": 0.9772627888768189,
      "learning_rate": 1.4419620897432318e-10,
      "loss": 0.1304,
      "step": 1749
    },
    {
      "epoch": 14.957264957264957,
      "grad_norm": 0.8578581155444094,
      "learning_rate": 1.0013655036916758e-10,
      "loss": 0.2403,
      "step": 1750
    },
    {
      "epoch": 14.965811965811966,
      "grad_norm": 0.9304603448605594,
      "learning_rate": 6.408754625736313e-11,
      "loss": 0.0621,
      "step": 1751
    },
    {
      "epoch": 14.974358974358974,
      "grad_norm": 1.1238292040246003,
      "learning_rate": 3.604931215400731e-11,
      "loss": 0.3074,
      "step": 1752
    },
    {
      "epoch": 14.982905982905983,
      "grad_norm": 1.1079431347000677,
      "learning_rate": 1.6021937904731054e-11,
      "loss": 0.3654,
      "step": 1753
    },
    {
      "epoch": 14.991452991452991,
      "grad_norm": 1.0356429914866991,
      "learning_rate": 4.005487684866083e-12,
      "loss": 0.2706,
      "step": 1754
    },
    {
      "epoch": 15.0,
      "grad_norm": 1.20860366920909,
      "learning_rate": 0.0,
      "loss": 0.3097,
      "step": 1755
    },
    {
      "epoch": 15.0,
      "step": 1755,
      "total_flos": 74086762168320.0,
      "train_loss": 0.44032181025060835,
      "train_runtime": 4826.2525,
      "train_samples_per_second": 2.539,
      "train_steps_per_second": 0.364
    }
  ],
  "logging_steps": 1,
  "max_steps": 1755,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 74086762168320.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}