{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 11184,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.4905116461224677,
      "learning_rate": 1.1904761904761906e-07,
      "loss": 0.9843,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2693075805631095,
      "learning_rate": 2.3809523809523811e-07,
      "loss": 1.0091,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5831426297391533,
      "learning_rate": 3.5714285714285716e-07,
      "loss": 1.0207,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.70115280643876,
      "learning_rate": 4.7619047619047623e-07,
      "loss": 1.0528,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5896350012719145,
      "learning_rate": 5.952380952380953e-07,
      "loss": 0.9976,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.678044135204474,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.9859,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4753765370838066,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.0166,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8571617127762683,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.9656,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.225002256736644,
      "learning_rate": 1.0714285714285714e-06,
      "loss": 0.9945,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.05074743400638,
      "learning_rate": 1.1904761904761906e-06,
      "loss": 0.9454,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.933169236750314,
      "learning_rate": 1.3095238095238096e-06,
      "loss": 0.9807,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9296810111917027,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 1.0305,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.507442396090275,
      "learning_rate": 1.5476190476190479e-06,
      "loss": 0.9231,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1634451962972794,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.9076,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2048145648848256,
      "learning_rate": 1.7857142857142859e-06,
      "loss": 0.9458,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1749341125616883,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.9422,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1552379316015964,
      "learning_rate": 2.023809523809524e-06,
      "loss": 0.9676,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9511412393481435,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.8984,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1810622414758247,
      "learning_rate": 2.261904761904762e-06,
      "loss": 0.965,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9072795673181119,
      "learning_rate": 2.380952380952381e-06,
      "loss": 0.8563,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8798943350980488,
      "learning_rate": 2.5e-06,
      "loss": 0.8715,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8454573628718905,
      "learning_rate": 2.6190476190476192e-06,
      "loss": 0.9177,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8598971710141431,
      "learning_rate": 2.7380952380952387e-06,
      "loss": 0.845,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8920658704848323,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.898,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7549670548703341,
      "learning_rate": 2.9761904761904763e-06,
      "loss": 0.8803,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8274854514850731,
      "learning_rate": 3.0952380952380957e-06,
      "loss": 0.9007,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8018993367708901,
      "learning_rate": 3.2142857142857147e-06,
      "loss": 0.8992,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8285002412423733,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.8875,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7889926553340196,
      "learning_rate": 3.4523809523809528e-06,
      "loss": 0.8671,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6716775454451835,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.8774,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6840674635883848,
      "learning_rate": 3.690476190476191e-06,
      "loss": 0.8464,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8040303420139149,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.837,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8814860363398729,
      "learning_rate": 3.928571428571429e-06,
      "loss": 0.8938,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9995961190327338,
      "learning_rate": 4.047619047619048e-06,
      "loss": 0.8735,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7890926449168132,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.8596,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7271316175341241,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.8595,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7201610664240254,
      "learning_rate": 4.404761904761905e-06,
      "loss": 0.8388,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8373331876517197,
      "learning_rate": 4.523809523809524e-06,
      "loss": 0.8452,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6854820900462705,
      "learning_rate": 4.642857142857144e-06,
      "loss": 0.8317,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5451080405172236,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.8081,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7808355493682025,
      "learning_rate": 4.880952380952381e-06,
      "loss": 0.8621,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7470932896010775,
      "learning_rate": 5e-06,
      "loss": 0.8401,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5922034298188192,
      "learning_rate": 5.119047619047619e-06,
      "loss": 0.8075,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6920602362891849,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.8371,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7955612134313848,
      "learning_rate": 5.357142857142857e-06,
      "loss": 0.8723,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.78700270631042,
      "learning_rate": 5.476190476190477e-06,
      "loss": 0.7942,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6570356872442532,
      "learning_rate": 5.595238095238096e-06,
      "loss": 0.8044,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6661626816426142,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.8036,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8707617348406472,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.7969,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7974090359202415,
      "learning_rate": 5.9523809523809525e-06,
      "loss": 0.8468,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7011181835924537,
      "learning_rate": 6.071428571428571e-06,
      "loss": 0.8058,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6086652974307393,
      "learning_rate": 6.1904761904761914e-06,
      "loss": 0.7712,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7119316612675569,
      "learning_rate": 6.30952380952381e-06,
      "loss": 0.7781,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6834105979065176,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 0.8216,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8171580949673858,
      "learning_rate": 6.547619047619048e-06,
      "loss": 0.8408,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.694803958631982,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.7847,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6810075325733252,
      "learning_rate": 6.785714285714287e-06,
      "loss": 0.7927,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8087248463402614,
      "learning_rate": 6.9047619047619055e-06,
      "loss": 0.8524,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6046904198366219,
      "learning_rate": 7.023809523809524e-06,
      "loss": 0.7973,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6986163081410239,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.8133,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7104394397045953,
      "learning_rate": 7.261904761904762e-06,
      "loss": 0.7919,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6759181027760283,
      "learning_rate": 7.380952380952382e-06,
      "loss": 0.8118,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6996270970019214,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.7887,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9526944850755016,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.2165,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7440376212880004,
      "learning_rate": 7.738095238095238e-06,
      "loss": 0.8315,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6880719844392995,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.8063,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7404343735132985,
      "learning_rate": 7.976190476190477e-06,
      "loss": 0.7538,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6309644450083883,
      "learning_rate": 8.095238095238097e-06,
      "loss": 0.7778,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7857153141356396,
      "learning_rate": 8.214285714285714e-06,
      "loss": 0.7998,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6476995456973781,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.7764,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6276745993958004,
      "learning_rate": 8.452380952380953e-06,
      "loss": 0.7424,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7744017979124047,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.7722,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.697746967134904,
      "learning_rate": 8.690476190476192e-06,
      "loss": 0.7896,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.794972422302961,
      "learning_rate": 8.80952380952381e-06,
      "loss": 0.7723,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6844038366258636,
      "learning_rate": 8.92857142857143e-06,
      "loss": 0.7636,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6732225265860294,
      "learning_rate": 9.047619047619049e-06,
      "loss": 0.7475,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6949450100119404,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.7436,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6227005261964506,
      "learning_rate": 9.285714285714288e-06,
      "loss": 0.7363,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8212741825433421,
      "learning_rate": 9.404761904761905e-06,
      "loss": 0.7979,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7634905153202906,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.822,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7596076496074272,
      "learning_rate": 9.642857142857144e-06,
      "loss": 0.7613,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7327674145799108,
      "learning_rate": 9.761904761904762e-06,
      "loss": 0.7556,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.733817028718878,
      "learning_rate": 9.880952380952381e-06,
      "loss": 0.7786,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8172182525934336,
      "learning_rate": 1e-05,
      "loss": 0.765,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6331972796814931,
      "learning_rate": 1.011904761904762e-05,
      "loss": 0.7493,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8213481104774921,
      "learning_rate": 1.0238095238095238e-05,
      "loss": 0.7703,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5667277258448238,
      "learning_rate": 1.0357142857142859e-05,
      "loss": 0.2024,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7181668047542872,
      "learning_rate": 1.0476190476190477e-05,
      "loss": 0.7727,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6718828398346983,
      "learning_rate": 1.0595238095238096e-05,
      "loss": 0.752,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.968711444179766,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 0.787,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6787864478355967,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 0.8095,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.704495221535615,
      "learning_rate": 1.0952380952380955e-05,
      "loss": 0.7647,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6541191371570237,
      "learning_rate": 1.1071428571428572e-05,
      "loss": 0.7642,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6803300206598802,
      "learning_rate": 1.1190476190476192e-05,
      "loss": 0.7374,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6869960716974999,
      "learning_rate": 1.130952380952381e-05,
      "loss": 0.7995,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7070626406635038,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.7689,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6083646402774259,
      "learning_rate": 1.1547619047619047e-05,
      "loss": 0.7795,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6674460378696793,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.7663,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5913906747542741,
      "learning_rate": 1.1785714285714287e-05,
      "loss": 0.7413,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6990784441711254,
      "learning_rate": 1.1904761904761905e-05,
      "loss": 0.7866,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5862444389491024,
      "learning_rate": 1.2023809523809525e-05,
      "loss": 0.7447,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5007416413045386,
      "learning_rate": 1.2142857142857142e-05,
      "loss": 0.7347,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6660939284217758,
      "learning_rate": 1.2261904761904763e-05,
      "loss": 0.7841,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5625751981094747,
      "learning_rate": 1.2380952380952383e-05,
      "loss": 0.7689,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7103829185710463,
      "learning_rate": 1.25e-05,
      "loss": 0.7684,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6494710363763432,
      "learning_rate": 1.261904761904762e-05,
      "loss": 0.7988,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6183548736294103,
      "learning_rate": 1.2738095238095238e-05,
      "loss": 0.7485,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6088295949429023,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 0.7688,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6190366038488115,
      "learning_rate": 1.2976190476190478e-05,
      "loss": 0.7415,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6466413436635802,
      "learning_rate": 1.3095238095238096e-05,
      "loss": 0.7557,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.649199747794475,
      "learning_rate": 1.3214285714285716e-05,
      "loss": 0.7503,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6307815292843437,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.7482,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7146719635576578,
      "learning_rate": 1.3452380952380954e-05,
      "loss": 0.7643,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6673506041514341,
      "learning_rate": 1.3571428571428574e-05,
      "loss": 0.8159,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6379063200343927,
      "learning_rate": 1.3690476190476192e-05,
      "loss": 0.7253,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6191191565349611,
      "learning_rate": 1.3809523809523811e-05,
      "loss": 0.7405,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6884317017525743,
      "learning_rate": 1.3928571428571429e-05,
      "loss": 0.7448,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6751741281565017,
      "learning_rate": 1.4047619047619048e-05,
      "loss": 0.7768,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6345768114338896,
      "learning_rate": 1.416666666666667e-05,
      "loss": 0.7654,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6129373456099391,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.7475,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6111121687272495,
      "learning_rate": 1.4404761904761907e-05,
      "loss": 0.7574,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5970763369165566,
      "learning_rate": 1.4523809523809524e-05,
      "loss": 0.7685,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7040182170851765,
      "learning_rate": 1.4642857142857144e-05,
      "loss": 0.7491,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7174626264727028,
      "learning_rate": 1.4761904761904763e-05,
      "loss": 0.772,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.673974594364649,
      "learning_rate": 1.4880952380952383e-05,
      "loss": 0.785,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6370557113271064,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7812,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5488037753992542,
      "learning_rate": 1.511904761904762e-05,
      "loss": 0.7333,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6138246500420873,
      "learning_rate": 1.523809523809524e-05,
      "loss": 0.7633,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6307139093394719,
      "learning_rate": 1.535714285714286e-05,
      "loss": 0.7538,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6081278143376319,
      "learning_rate": 1.5476190476190476e-05,
      "loss": 0.7237,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6247257544752407,
      "learning_rate": 1.5595238095238098e-05,
      "loss": 0.7447,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6145393866795815,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.7229,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6800047588743964,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 0.7621,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6476462020432572,
      "learning_rate": 1.5952380952380954e-05,
      "loss": 0.722,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6802692100284877,
      "learning_rate": 1.6071428571428572e-05,
      "loss": 0.7831,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5884147441951202,
      "learning_rate": 1.6190476190476193e-05,
      "loss": 0.7894,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9085266684212221,
      "learning_rate": 1.630952380952381e-05,
      "loss": 0.6967,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6998459177688401,
      "learning_rate": 1.642857142857143e-05,
      "loss": 0.2276,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.162238833267921,
      "learning_rate": 1.6547619047619046e-05,
      "loss": 0.7829,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6732613200808061,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.7279,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.100000508218701,
      "learning_rate": 1.678571428571429e-05,
      "loss": 0.7865,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6286068092068148,
      "learning_rate": 1.6904761904761906e-05,
      "loss": 0.7501,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6809020074813019,
      "learning_rate": 1.7023809523809524e-05,
      "loss": 0.7492,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8611185301764633,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.7902,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6523902461451291,
      "learning_rate": 1.7261904761904763e-05,
      "loss": 0.7628,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.603332150068893,
      "learning_rate": 1.7380952380952384e-05,
      "loss": 0.7969,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6279734661188838,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.784,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5557672906407286,
      "learning_rate": 1.761904761904762e-05,
      "loss": 0.7269,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6260911647240376,
      "learning_rate": 1.7738095238095237e-05,
      "loss": 0.7765,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5892475611201051,
      "learning_rate": 1.785714285714286e-05,
      "loss": 0.7318,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5723577602938895,
      "learning_rate": 1.797619047619048e-05,
      "loss": 0.7351,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5630697637918489,
      "learning_rate": 1.8095238095238097e-05,
      "loss": 0.7782,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.635294620118572,
      "learning_rate": 1.8214285714285715e-05,
      "loss": 0.7915,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5444394283210853,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.7589,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5852988336213298,
      "learning_rate": 1.8452380952380954e-05,
      "loss": 0.7419,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5489018730640458,
      "learning_rate": 1.8571428571428575e-05,
      "loss": 0.7616,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.57411167963052,
      "learning_rate": 1.8690476190476193e-05,
      "loss": 0.7812,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5362165165416939,
      "learning_rate": 1.880952380952381e-05,
      "loss": 0.7546,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5898238778327553,
      "learning_rate": 1.892857142857143e-05,
      "loss": 0.7186,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2369921770542007,
      "learning_rate": 1.904761904761905e-05,
      "loss": 0.7179,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.704481712322241,
      "learning_rate": 1.916666666666667e-05,
      "loss": 0.7425,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5780462540023076,
      "learning_rate": 1.928571428571429e-05,
      "loss": 0.7661,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5929495083043826,
      "learning_rate": 1.9404761904761906e-05,
      "loss": 0.7127,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6491819100667442,
      "learning_rate": 1.9523809523809524e-05,
      "loss": 0.7557,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1972806129733917,
      "learning_rate": 1.9642857142857145e-05,
      "loss": 0.2278,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3372841611675992,
      "learning_rate": 1.9761904761904763e-05,
      "loss": 0.7735,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6201975123106436,
      "learning_rate": 1.9880952380952384e-05,
      "loss": 0.705,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.696255291449838,
      "learning_rate": 2e-05,
      "loss": 0.7778,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7794274172821811,
      "learning_rate": 2.011904761904762e-05,
      "loss": 0.7618,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6143963346575989,
      "learning_rate": 2.023809523809524e-05,
      "loss": 0.7784,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6501079922513006,
      "learning_rate": 2.0357142857142858e-05,
      "loss": 0.7922,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5730166479685945,
      "learning_rate": 2.0476190476190476e-05,
      "loss": 0.7422,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6848557430432394,
      "learning_rate": 2.0595238095238094e-05,
      "loss": 0.7519,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6654190845449479,
      "learning_rate": 2.0714285714285718e-05,
      "loss": 0.7791,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6170766171768842,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.7425,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5576812597474525,
      "learning_rate": 2.0952380952380954e-05,
      "loss": 0.7826,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5786056538650225,
      "learning_rate": 2.1071428571428575e-05,
      "loss": 0.7262,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6014713961058313,
      "learning_rate": 2.1190476190476193e-05,
      "loss": 0.7616,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6731286347537808,
      "learning_rate": 2.130952380952381e-05,
      "loss": 0.8216,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6115864705338812,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.7619,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4942771766242318,
      "learning_rate": 2.154761904761905e-05,
      "loss": 0.7059,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5873926004778744,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 0.7282,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6009377263761441,
      "learning_rate": 2.1785714285714285e-05,
      "loss": 0.7495,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5882532701418745,
      "learning_rate": 2.190476190476191e-05,
      "loss": 0.746,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6702582281007632,
      "learning_rate": 2.2023809523809527e-05,
      "loss": 0.7867,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5033526083270368,
      "learning_rate": 2.2142857142857145e-05,
      "loss": 0.7363,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.604520819206373,
      "learning_rate": 2.2261904761904766e-05,
      "loss": 0.7568,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5892612052935836,
      "learning_rate": 2.2380952380952384e-05,
      "loss": 0.7711,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7066735591552987,
      "learning_rate": 2.25e-05,
      "loss": 0.7533,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6040257373716639,
      "learning_rate": 2.261904761904762e-05,
      "loss": 0.7558,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5307546424299912,
      "learning_rate": 2.273809523809524e-05,
      "loss": 0.727,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5690551656480883,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 0.7425,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5332280581541026,
      "learning_rate": 2.2976190476190476e-05,
      "loss": 0.7516,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5660635904840328,
      "learning_rate": 2.3095238095238094e-05,
      "loss": 0.7633,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6267232718510237,
      "learning_rate": 2.3214285714285718e-05,
      "loss": 0.7862,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5389776744583166,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.7533,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5377812700166781,
      "learning_rate": 2.3452380952380957e-05,
      "loss": 0.7699,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5779223936322543,
      "learning_rate": 2.3571428571428575e-05,
      "loss": 0.7617,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5501015354122613,
      "learning_rate": 2.3690476190476192e-05,
      "loss": 0.7442,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6269641458173392,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.7951,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6325252520162454,
      "learning_rate": 2.392857142857143e-05,
      "loss": 0.7972,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5838401543889118,
      "learning_rate": 2.404761904761905e-05,
      "loss": 0.7593,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5642739179445266,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 0.7373,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5853934531065161,
      "learning_rate": 2.4285714285714285e-05,
      "loss": 0.7527,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6530058306848152,
      "learning_rate": 2.440476190476191e-05,
      "loss": 0.7337,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5417794190525068,
      "learning_rate": 2.4523809523809527e-05,
      "loss": 0.7413,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5977332430144849,
      "learning_rate": 2.4642857142857145e-05,
      "loss": 0.7422,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6008019924306953,
      "learning_rate": 2.4761904761904766e-05,
      "loss": 0.7695,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8943714984094318,
      "learning_rate": 2.4880952380952383e-05,
      "loss": 0.7707,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.493443014486216,
      "learning_rate": 2.5e-05,
      "loss": 0.723,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6108826841826349,
      "learning_rate": 2.5119047619047622e-05,
      "loss": 0.781,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6447251278095719,
      "learning_rate": 2.523809523809524e-05,
      "loss": 0.7488,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6253431023840471,
      "learning_rate": 2.5357142857142858e-05,
      "loss": 0.7533,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.754396958248776,
      "learning_rate": 2.5476190476190476e-05,
      "loss": 0.7965,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6042957940575493,
      "learning_rate": 2.5595238095238097e-05,
      "loss": 0.7244,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5378718281939072,
      "learning_rate": 2.5714285714285718e-05,
      "loss": 0.7354,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6533853085783494,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 0.7863,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5365712691702396,
      "learning_rate": 2.5952380952380957e-05,
      "loss": 0.7519,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5623445470469434,
      "learning_rate": 2.6071428571428574e-05,
      "loss": 0.7393,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5898841498240613,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 0.7512,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5484603819379246,
      "learning_rate": 2.630952380952381e-05,
      "loss": 0.7631,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5583894126798397,
      "learning_rate": 2.642857142857143e-05,
      "loss": 0.7692,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5159607625147283,
      "learning_rate": 2.654761904761905e-05,
      "loss": 0.7434,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5847632521351462,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.7082,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5750053490741818,
      "learning_rate": 2.6785714285714284e-05,
      "loss": 0.7265,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5353359314891155,
      "learning_rate": 2.690476190476191e-05,
      "loss": 0.7249,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5577619120216777,
      "learning_rate": 2.7023809523809527e-05,
      "loss": 0.7602,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4688256463436997,
      "learning_rate": 2.7142857142857148e-05,
      "loss": 0.7263,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4471158745558642,
      "learning_rate": 2.7261904761904765e-05,
      "loss": 0.7149,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.538200562556444,
      "learning_rate": 2.7380952380952383e-05,
      "loss": 0.7586,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48405514422357504,
      "learning_rate": 2.75e-05,
      "loss": 0.677,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5436079640571804,
      "learning_rate": 2.7619047619047622e-05,
      "loss": 0.7744,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5091861090720476,
      "learning_rate": 2.773809523809524e-05,
      "loss": 0.7462,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5662504525172081,
      "learning_rate": 2.7857142857142858e-05,
      "loss": 0.7537,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5318413251905137,
      "learning_rate": 2.7976190476190475e-05,
      "loss": 0.7406,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6037528352666494,
      "learning_rate": 2.8095238095238096e-05,
      "loss": 0.7595,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5647506160279632,
      "learning_rate": 2.8214285714285718e-05,
      "loss": 0.7836,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5696693617365941,
      "learning_rate": 2.833333333333334e-05,
      "loss": 0.773,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6239866991190165,
      "learning_rate": 2.8452380952380957e-05,
      "loss": 0.748,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5281120646672359,
      "learning_rate": 2.8571428571428574e-05,
      "loss": 0.7723,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5372501559044877,
      "learning_rate": 2.8690476190476192e-05,
      "loss": 0.726,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5459452965522313,
      "learning_rate": 2.8809523809523813e-05,
      "loss": 0.7366,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4716900838385467,
      "learning_rate": 2.892857142857143e-05,
      "loss": 0.7334,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.614616869905398,
      "learning_rate": 2.904761904761905e-05,
      "loss": 0.7618,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5777574016853546,
      "learning_rate": 2.9166666666666666e-05,
      "loss": 0.7475,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5816797570702905,
      "learning_rate": 2.9285714285714288e-05,
      "loss": 0.7575,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5358560625085461,
      "learning_rate": 2.940476190476191e-05,
      "loss": 0.7632,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5618907486525954,
      "learning_rate": 2.9523809523809526e-05,
      "loss": 0.7484,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5720517294524703,
      "learning_rate": 2.9642857142857148e-05,
      "loss": 0.7593,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6249943186636568,
      "learning_rate": 2.9761904761904765e-05,
      "loss": 0.789,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5669581628844208,
      "learning_rate": 2.9880952380952383e-05,
      "loss": 0.7577,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5313122777051179,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.7197,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5186342060850352,
      "learning_rate": 3.0119047619047622e-05,
      "loss": 0.7205,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6137832258944671,
      "learning_rate": 3.023809523809524e-05,
      "loss": 0.7294,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5704395694216968,
      "learning_rate": 3.0357142857142857e-05,
      "loss": 0.7268,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5699440025595037,
      "learning_rate": 3.047619047619048e-05,
      "loss": 0.7263,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5978320404650848,
      "learning_rate": 3.059523809523809e-05,
      "loss": 0.7389,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6730323332819415,
      "learning_rate": 3.071428571428572e-05,
      "loss": 0.8013,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5325682693287718,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 0.7774,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6278381442080764,
      "learning_rate": 3.095238095238095e-05,
      "loss": 0.7947,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.45352073460393916,
      "learning_rate": 3.107142857142858e-05,
      "loss": 0.6819,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1248728663840308,
      "learning_rate": 3.1190476190476195e-05,
      "loss": 0.3111,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9866435487257205,
      "learning_rate": 3.130952380952381e-05,
      "loss": 0.7683,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6039965225036162,
      "learning_rate": 3.142857142857143e-05,
      "loss": 0.8023,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.661366251516469,
      "learning_rate": 3.154761904761905e-05,
      "loss": 0.7449,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6083977989381968,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 0.7608,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6052846980109551,
      "learning_rate": 3.1785714285714284e-05,
      "loss": 0.7404,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6167092562335443,
      "learning_rate": 3.190476190476191e-05,
      "loss": 0.7468,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6187380609261965,
      "learning_rate": 3.2023809523809526e-05,
      "loss": 0.7982,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.46257200376992896,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 0.7356,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5394434358707887,
      "learning_rate": 3.226190476190477e-05,
      "loss": 0.7372,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48690352022987715,
      "learning_rate": 3.2380952380952386e-05,
      "loss": 0.7467,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.539872648150301,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.7382,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5910986523216055,
      "learning_rate": 3.261904761904762e-05,
      "loss": 0.7476,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6046865797006944,
      "learning_rate": 3.273809523809524e-05,
      "loss": 0.7904,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5320764414226274,
      "learning_rate": 3.285714285714286e-05,
      "loss": 0.7587,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8721170794118804,
      "learning_rate": 3.2976190476190475e-05,
      "loss": 0.2929,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.590426838074977,
      "learning_rate": 3.309523809523809e-05,
      "loss": 0.736,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4873693002234687,
      "learning_rate": 3.321428571428572e-05,
      "loss": 0.7242,
      "step": 279
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5285252295484378,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.7331,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5099158617230611,
      "learning_rate": 3.345238095238096e-05,
      "loss": 0.7325,
      "step": 281
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5880058766798582,
      "learning_rate": 3.357142857142858e-05,
      "loss": 0.7933,
      "step": 282
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5591985897014409,
      "learning_rate": 3.3690476190476195e-05,
      "loss": 0.2753,
      "step": 283
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5298586743098739,
      "learning_rate": 3.380952380952381e-05,
      "loss": 0.7546,
      "step": 284
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48110503801740934,
      "learning_rate": 3.392857142857143e-05,
      "loss": 0.7353,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5488875412704809,
      "learning_rate": 3.404761904761905e-05,
      "loss": 0.7385,
      "step": 286
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.512715018574185,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 0.7282,
      "step": 287
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5429248979849669,
      "learning_rate": 3.4285714285714284e-05,
      "loss": 0.7087,
      "step": 288
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5786668482375372,
      "learning_rate": 3.440476190476191e-05,
      "loss": 0.7857,
      "step": 289
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6976414802787998,
      "learning_rate": 3.4523809523809526e-05,
      "loss": 0.7335,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5529079495911252,
      "learning_rate": 3.4642857142857144e-05,
      "loss": 0.7819,
      "step": 291
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5190347495900116,
      "learning_rate": 3.476190476190477e-05,
      "loss": 0.7582,
      "step": 292
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47941946927353946,
      "learning_rate": 3.4880952380952386e-05,
      "loss": 0.7659,
      "step": 293
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.514875752015368,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 0.7602,
      "step": 294
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47887994685301455,
      "learning_rate": 3.511904761904762e-05,
      "loss": 0.74,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5427943576446719,
      "learning_rate": 3.523809523809524e-05,
      "loss": 0.78,
      "step": 296
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5404405772102042,
      "learning_rate": 3.535714285714286e-05,
      "loss": 0.7787,
      "step": 297
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5449657252666567,
      "learning_rate": 3.5476190476190475e-05,
      "loss": 0.7597,
      "step": 298
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.55371239257055,
      "learning_rate": 3.55952380952381e-05,
      "loss": 0.7553,
      "step": 299
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.541821897883467,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.7472,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5286462351859228,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 0.7685,
      "step": 301
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4689369173237389,
      "learning_rate": 3.595238095238096e-05,
      "loss": 0.7318,
      "step": 302
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5114938509959724,
      "learning_rate": 3.607142857142858e-05,
      "loss": 0.744,
      "step": 303
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5177429629813635,
      "learning_rate": 3.6190476190476195e-05,
      "loss": 0.7209,
      "step": 304
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5455228681867046,
      "learning_rate": 3.630952380952381e-05,
      "loss": 0.8061,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5109330763354822,
      "learning_rate": 3.642857142857143e-05,
      "loss": 0.7387,
      "step": 306
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5031842524114989,
      "learning_rate": 3.654761904761905e-05,
      "loss": 0.7504,
      "step": 307
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6633267373132385,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.2759,
      "step": 308
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5036374938741508,
      "learning_rate": 3.6785714285714283e-05,
      "loss": 0.7144,
      "step": 309
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6381461807489422,
      "learning_rate": 3.690476190476191e-05,
      "loss": 0.7527,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5366557747275517,
      "learning_rate": 3.7023809523809526e-05,
      "loss": 0.7726,
      "step": 311
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48616082517557,
      "learning_rate": 3.714285714285715e-05,
      "loss": 0.715,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4883369873004085,
      "learning_rate": 3.726190476190477e-05,
      "loss": 0.7591,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5046015433872738,
      "learning_rate": 3.7380952380952386e-05,
      "loss": 0.7325,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48896729091483904,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.7612,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49403584135312845,
      "learning_rate": 3.761904761904762e-05,
      "loss": 0.7274,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49061680419264636,
      "learning_rate": 3.773809523809524e-05,
      "loss": 0.7084,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6041387651907814,
      "learning_rate": 3.785714285714286e-05,
      "loss": 0.7129,
      "step": 318
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44909254806698473,
      "learning_rate": 3.7976190476190474e-05,
      "loss": 0.7178,
      "step": 319
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5173217599832592,
      "learning_rate": 3.80952380952381e-05,
      "loss": 0.7415,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5025531531822874,
      "learning_rate": 3.821428571428572e-05,
      "loss": 0.7451,
      "step": 321
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5706585439232325,
      "learning_rate": 3.833333333333334e-05,
      "loss": 0.7679,
      "step": 322
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5266940135321423,
      "learning_rate": 3.845238095238096e-05,
      "loss": 0.7346,
      "step": 323
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4766115438367989,
      "learning_rate": 3.857142857142858e-05,
      "loss": 0.7321,
      "step": 324
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5134132228045268,
      "learning_rate": 3.8690476190476195e-05,
      "loss": 0.7299,
      "step": 325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5374792986383184,
      "learning_rate": 3.880952380952381e-05,
      "loss": 0.7516,
      "step": 326
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5808694110709398,
      "learning_rate": 3.892857142857143e-05,
      "loss": 0.765,
      "step": 327
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5064888428571029,
      "learning_rate": 3.904761904761905e-05,
      "loss": 0.7549,
      "step": 328
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5541164703342127,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 0.7414,
      "step": 329
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.555305891813835,
      "learning_rate": 3.928571428571429e-05,
      "loss": 0.7682,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5368569645441458,
      "learning_rate": 3.940476190476191e-05,
      "loss": 0.7531,
      "step": 331
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47715680363079566,
      "learning_rate": 3.9523809523809526e-05,
      "loss": 0.7401,
      "step": 332
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46507309721392986,
      "learning_rate": 3.964285714285715e-05,
      "loss": 0.7572,
      "step": 333
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5904079494622209,
      "learning_rate": 3.976190476190477e-05,
      "loss": 0.7521,
      "step": 334
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6373644133003155,
      "learning_rate": 3.9880952380952386e-05,
      "loss": 0.7685,
      "step": 335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4968900762450253,
      "learning_rate": 4e-05,
      "loss": 0.7472,
      "step": 336
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5176366793131504,
      "learning_rate": 3.999999916131208e-05,
      "loss": 0.7417,
      "step": 337
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6298844368670057,
      "learning_rate": 3.999999664524838e-05,
      "loss": 0.7076,
      "step": 338
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5786244797339942,
      "learning_rate": 3.999999245180912e-05,
      "loss": 0.783,
      "step": 339
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5744779578190186,
      "learning_rate": 3.999998658099464e-05,
      "loss": 0.7738,
      "step": 340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49567813259348054,
      "learning_rate": 3.999997903280544e-05,
      "loss": 0.7207,
      "step": 341
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4472200088156858,
      "learning_rate": 3.999996980724216e-05,
      "loss": 0.7424,
      "step": 342
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5539319545609397,
      "learning_rate": 3.999995890430556e-05,
      "loss": 0.7669,
      "step": 343
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5114555121265989,
      "learning_rate": 3.999994632399656e-05,
      "loss": 0.7657,
      "step": 344
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5267269146476349,
      "learning_rate": 3.999993206631622e-05,
      "loss": 0.7459,
      "step": 345
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5651283092588717,
      "learning_rate": 3.9999916131265726e-05,
      "loss": 0.7884,
      "step": 346
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5604442618354375,
      "learning_rate": 3.9999898518846424e-05,
      "loss": 0.7798,
      "step": 347
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5795730131598683,
      "learning_rate": 3.9999879229059785e-05,
      "loss": 0.7787,
      "step": 348
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5375644420070966,
      "learning_rate": 3.999985826190743e-05,
      "loss": 0.7498,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5932741705018203,
      "learning_rate": 3.999983561739112e-05,
      "loss": 0.7226,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5351412970472065,
      "learning_rate": 3.999981129551274e-05,
      "loss": 0.7693,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6112962619691243,
      "learning_rate": 3.999978529627435e-05,
      "loss": 0.7832,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5881158686613928,
      "learning_rate": 3.9999757619678115e-05,
      "loss": 0.7242,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5085713258416195,
      "learning_rate": 3.9999728265726364e-05,
      "loss": 0.7344,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5237409369920965,
      "learning_rate": 3.999969723442155e-05,
      "loss": 0.7913,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5403542434396562,
      "learning_rate": 3.99996645257663e-05,
      "loss": 0.7513,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5341749468639716,
      "learning_rate": 3.999963013976333e-05,
      "loss": 0.746,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44930380573564654,
      "learning_rate": 3.999959407641553e-05,
      "loss": 0.7843,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4674721902838872,
      "learning_rate": 3.999955633572593e-05,
      "loss": 0.789,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5086012736046716,
      "learning_rate": 3.999951691769769e-05,
      "loss": 0.7328,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5304395171647385,
      "learning_rate": 3.999947582233413e-05,
      "loss": 0.792,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4562254407817604,
      "learning_rate": 3.999943304963868e-05,
      "loss": 0.7371,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5029487968186257,
      "learning_rate": 3.999938859961493e-05,
      "loss": 0.7613,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44564779501682844,
      "learning_rate": 3.999934247226662e-05,
      "loss": 0.7532,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5046851552885737,
      "learning_rate": 3.99992946675976e-05,
      "loss": 0.7554,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4786037361061595,
      "learning_rate": 3.99992451856119e-05,
      "loss": 0.7547,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4613692942755683,
      "learning_rate": 3.999919402631365e-05,
      "loss": 0.7251,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5048887502124114,
      "learning_rate": 3.999914118970715e-05,
      "loss": 0.7166,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47276577761958083,
      "learning_rate": 3.999908667579684e-05,
      "loss": 0.7262,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5083165727914501,
      "learning_rate": 3.9999030484587274e-05,
      "loss": 0.7729,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.51834681663405,
      "learning_rate": 3.999897261608318e-05,
      "loss": 0.7934,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5046084591812421,
      "learning_rate": 3.99989130702894e-05,
      "loss": 0.7695,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5090437342934677,
      "learning_rate": 3.999885184721094e-05,
      "loss": 0.7376,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46553597788248546,
      "learning_rate": 3.999878894685292e-05,
      "loss": 0.7322,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.445824336800517,
      "learning_rate": 3.999872436922063e-05,
      "loss": 0.7089,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45864185215685876,
      "learning_rate": 3.9998658114319475e-05,
      "loss": 0.7578,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5340513947434615,
      "learning_rate": 3.9998590182155024e-05,
      "loss": 0.7835,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49134365115754175,
      "learning_rate": 3.999852057273296e-05,
      "loss": 0.7266,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5858836038526174,
      "learning_rate": 3.999844928605913e-05,
      "loss": 0.7507,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5091038275099742,
      "learning_rate": 3.9998376322139506e-05,
      "loss": 0.7275,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4916399216814001,
      "learning_rate": 3.999830168098021e-05,
      "loss": 0.7307,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5027498818974724,
      "learning_rate": 3.999822536258752e-05,
      "loss": 0.7358,
      "step": 382
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5036775800123834,
      "learning_rate": 3.99981473669678e-05,
      "loss": 0.7642,
      "step": 383
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49765872350449053,
      "learning_rate": 3.9998067694127626e-05,
      "loss": 0.7627,
      "step": 384
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47712158249621345,
      "learning_rate": 3.999798634407366e-05,
      "loss": 0.7187,
      "step": 385
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5600977879482325,
      "learning_rate": 3.999790331681273e-05,
      "loss": 0.7578,
      "step": 386
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48429850142375996,
      "learning_rate": 3.9997818612351804e-05,
      "loss": 0.8121,
      "step": 387
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5217066481607935,
      "learning_rate": 3.999773223069798e-05,
      "loss": 0.7631,
      "step": 388
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8830414498924135,
      "learning_rate": 3.9997644171858504e-05,
      "loss": 0.3171,
      "step": 389
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.59529966586921,
      "learning_rate": 3.999755443584077e-05,
      "loss": 0.7887,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48061629728346705,
      "learning_rate": 3.999746302265228e-05,
      "loss": 0.7332,
      "step": 391
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4838294359164395,
      "learning_rate": 3.999736993230073e-05,
      "loss": 0.7436,
      "step": 392
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5025434867930174,
      "learning_rate": 3.9997275164793915e-05,
      "loss": 0.753,
      "step": 393
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5583171950968815,
      "learning_rate": 3.9997178720139785e-05,
      "loss": 0.7412,
      "step": 394
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5300989896390733,
      "learning_rate": 3.9997080598346414e-05,
      "loss": 0.7559,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.096818118333641,
      "learning_rate": 3.9996980799422054e-05,
      "loss": 0.3227,
      "step": 396
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.536849902980891,
      "learning_rate": 3.999687932337506e-05,
      "loss": 0.7664,
      "step": 397
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4561979314858516,
      "learning_rate": 3.9996776170213945e-05,
      "loss": 0.7602,
      "step": 398
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5582312712607754,
      "learning_rate": 3.9996671339947366e-05,
      "loss": 0.7394,
      "step": 399
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5244478461489286,
      "learning_rate": 3.9996564832584114e-05,
      "loss": 0.7527,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5064984370542686,
      "learning_rate": 3.999645664813311e-05,
      "loss": 0.7166,
      "step": 401
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5156938833560681,
      "learning_rate": 3.999634678660344e-05,
      "loss": 0.7533,
      "step": 402
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4400955769660254,
      "learning_rate": 3.9996235248004326e-05,
      "loss": 0.7464,
      "step": 403
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48000568925662107,
      "learning_rate": 3.999612203234511e-05,
      "loss": 0.7883,
      "step": 404
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44952444028392663,
      "learning_rate": 3.999600713963527e-05,
      "loss": 0.7489,
      "step": 405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5242543411866685,
      "learning_rate": 3.999589056988447e-05,
      "loss": 0.7633,
      "step": 406
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5667996261520698,
      "learning_rate": 3.9995772323102484e-05,
      "loss": 0.7959,
      "step": 407
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48203301877848487,
      "learning_rate": 3.9995652399299215e-05,
      "loss": 0.7296,
      "step": 408
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4545478613022514,
      "learning_rate": 3.999553079848473e-05,
      "loss": 0.7019,
      "step": 409
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47264357991648576,
      "learning_rate": 3.999540752066923e-05,
      "loss": 0.7522,
      "step": 410
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5008123336820316,
      "learning_rate": 3.9995282565863034e-05,
      "loss": 0.7473,
      "step": 411
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.439231221460751,
      "learning_rate": 3.999515593407665e-05,
      "loss": 0.716,
      "step": 412
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4783080451393385,
      "learning_rate": 3.9995027625320684e-05,
      "loss": 0.7876,
      "step": 413
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.516739168782736,
      "learning_rate": 3.99948976396059e-05,
      "loss": 0.7645,
      "step": 414
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45035984717574457,
      "learning_rate": 3.999476597694319e-05,
      "loss": 0.76,
      "step": 415
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4635509822404213,
      "learning_rate": 3.999463263734361e-05,
      "loss": 0.7327,
      "step": 416
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5060335213537814,
      "learning_rate": 3.999449762081835e-05,
      "loss": 0.7213,
      "step": 417
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5996771278299274,
      "learning_rate": 3.9994360927378706e-05,
      "loss": 0.8216,
      "step": 418
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45900474068914343,
      "learning_rate": 3.9994222557036165e-05,
      "loss": 0.7149,
      "step": 419
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.512883747536977,
      "learning_rate": 3.9994082509802325e-05,
      "loss": 0.7651,
      "step": 420
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.398703982673895,
      "learning_rate": 3.999394078568893e-05,
      "loss": 0.7378,
      "step": 421
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4832897632801979,
      "learning_rate": 3.9993797384707865e-05,
      "loss": 0.7623,
      "step": 422
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4436409620995274,
      "learning_rate": 3.9993652306871165e-05,
      "loss": 0.7157,
      "step": 423
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4388782893070347,
      "learning_rate": 3.999350555219099e-05,
      "loss": 0.7174,
      "step": 424
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4754526047443349,
      "learning_rate": 3.9993357120679655e-05,
      "loss": 0.7705,
      "step": 425
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.512859659971535,
      "learning_rate": 3.999320701234959e-05,
      "loss": 0.6982,
      "step": 426
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5293117001400206,
      "learning_rate": 3.9993055227213415e-05,
      "loss": 0.7186,
      "step": 427
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43994158070834694,
      "learning_rate": 3.9992901765283836e-05,
      "loss": 0.7403,
      "step": 428
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5449888267601656,
      "learning_rate": 3.999274662657374e-05,
      "loss": 0.7622,
      "step": 429
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4430247666754252,
      "learning_rate": 3.9992589811096115e-05,
      "loss": 0.7721,
      "step": 430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4210616831115738,
      "learning_rate": 3.999243131886413e-05,
      "loss": 0.7649,
      "step": 431
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.469340637925769,
      "learning_rate": 3.9992271149891086e-05,
      "loss": 0.7321,
      "step": 432
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46267902095437097,
      "learning_rate": 3.999210930419039e-05,
      "loss": 0.7189,
      "step": 433
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4903578892542254,
      "learning_rate": 3.999194578177565e-05,
      "loss": 0.7353,
      "step": 434
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4827565970603046,
      "learning_rate": 3.9991780582660545e-05,
      "loss": 0.7559,
      "step": 435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47610171282967856,
      "learning_rate": 3.9991613706858956e-05,
      "loss": 0.7071,
      "step": 436
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4487581114437705,
      "learning_rate": 3.999144515438486e-05,
      "loss": 0.7431,
      "step": 437
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5082305730352467,
      "learning_rate": 3.999127492525241e-05,
      "loss": 0.742,
      "step": 438
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44748407530407974,
      "learning_rate": 3.9991103019475875e-05,
      "loss": 0.7451,
      "step": 439
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4520307134875093,
      "learning_rate": 3.9990929437069675e-05,
      "loss": 0.7391,
      "step": 440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4602918791426955,
      "learning_rate": 3.9990754178048356e-05,
      "loss": 0.7639,
      "step": 441
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5034639038452932,
      "learning_rate": 3.999057724242664e-05,
      "loss": 0.708,
      "step": 442
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5590919976419049,
      "learning_rate": 3.999039863021934e-05,
      "loss": 0.7995,
      "step": 443
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45776533897643884,
      "learning_rate": 3.999021834144146e-05,
      "loss": 0.7643,
      "step": 444
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5441542098655722,
      "learning_rate": 3.99900363761081e-05,
      "loss": 0.8004,
      "step": 445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5307082583975898,
      "learning_rate": 3.998985273423454e-05,
      "loss": 0.7791,
      "step": 446
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45433379228188253,
      "learning_rate": 3.9989667415836166e-05,
      "loss": 0.6911,
      "step": 447
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5823396476105079,
      "learning_rate": 3.9989480420928536e-05,
      "loss": 0.7694,
      "step": 448
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42123915823831604,
      "learning_rate": 3.9989291749527314e-05,
      "loss": 0.7328,
      "step": 449
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4393332419703939,
      "learning_rate": 3.998910140164834e-05,
      "loss": 0.7422,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49573223324596516,
      "learning_rate": 3.998890937730757e-05,
      "loss": 0.7745,
      "step": 451
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46940777856630533,
      "learning_rate": 3.9988715676521115e-05,
      "loss": 0.76,
      "step": 452
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5231887109587556,
      "learning_rate": 3.998852029930521e-05,
      "loss": 0.7945,
      "step": 453
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47267615155781123,
      "learning_rate": 3.9988323245676256e-05,
      "loss": 0.7359,
      "step": 454
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44678692713887525,
      "learning_rate": 3.9988124515650765e-05,
      "loss": 0.7639,
      "step": 455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5141630621618691,
      "learning_rate": 3.9987924109245406e-05,
      "loss": 0.7397,
      "step": 456
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5086125931204177,
      "learning_rate": 3.998772202647699e-05,
      "loss": 0.7349,
      "step": 457
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4495688457863861,
      "learning_rate": 3.998751826736248e-05,
      "loss": 0.7518,
      "step": 458
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49826365444110116,
      "learning_rate": 3.998731283191895e-05,
      "loss": 0.7722,
      "step": 459
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4564400926275752,
      "learning_rate": 3.998710572016362e-05,
      "loss": 0.7206,
      "step": 460
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49699504395915767,
      "learning_rate": 3.998689693211388e-05,
      "loss": 0.7591,
      "step": 461
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48986649133481924,
      "learning_rate": 3.998668646778723e-05,
      "loss": 0.7491,
      "step": 462
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49558850451005604,
      "learning_rate": 3.998647432720132e-05,
      "loss": 0.8231,
      "step": 463
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45502091624905966,
      "learning_rate": 3.998626051037396e-05,
      "loss": 0.7662,
      "step": 464
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44650067542817434,
      "learning_rate": 3.998604501732305e-05,
      "loss": 0.7466,
      "step": 465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4929656108861082,
      "learning_rate": 3.998582784806669e-05,
      "loss": 0.7581,
      "step": 466
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46383041962875815,
      "learning_rate": 3.998560900262309e-05,
      "loss": 0.6819,
      "step": 467
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4355744278557774,
      "learning_rate": 3.998538848101059e-05,
      "loss": 0.7298,
      "step": 468
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4832994808210268,
      "learning_rate": 3.99851662832477e-05,
      "loss": 0.7435,
      "step": 469
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.40768232495083195,
      "learning_rate": 3.998494240935305e-05,
      "loss": 0.6971,
      "step": 470
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0722646887039862,
      "learning_rate": 3.998471685934541e-05,
      "loss": 0.4103,
      "step": 471
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4985045668648099,
      "learning_rate": 3.998448963324371e-05,
      "loss": 0.7193,
      "step": 472
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6023527477002727,
      "learning_rate": 3.9984260731066995e-05,
      "loss": 0.7466,
      "step": 473
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5195189296066353,
      "learning_rate": 3.998403015283447e-05,
      "loss": 0.7018,
      "step": 474
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47652131717138335,
      "learning_rate": 3.998379789856547e-05,
      "loss": 0.718,
      "step": 475
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6967900782631196,
      "learning_rate": 3.998356396827947e-05,
      "loss": 0.8459,
      "step": 476
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5355319809687373,
      "learning_rate": 3.99833283619961e-05,
      "loss": 0.7474,
      "step": 477
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5390852496059754,
      "learning_rate": 3.998309107973511e-05,
      "loss": 0.7049,
      "step": 478
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47767370189750263,
      "learning_rate": 3.9982852121516405e-05,
      "loss": 0.734,
      "step": 479
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5247411809486707,
      "learning_rate": 3.9982611487360026e-05,
      "loss": 0.7688,
      "step": 480
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5328429910161419,
      "learning_rate": 3.998236917728616e-05,
      "loss": 0.7906,
      "step": 481
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4620575067424612,
      "learning_rate": 3.998212519131512e-05,
      "loss": 0.7332,
      "step": 482
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43272946967347065,
      "learning_rate": 3.9981879529467366e-05,
      "loss": 0.7323,
      "step": 483
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4271410966692234,
      "learning_rate": 3.998163219176352e-05,
      "loss": 0.7285,
      "step": 484
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4681156979811844,
      "learning_rate": 3.998138317822431e-05,
      "loss": 0.7187,
      "step": 485
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47507588204874146,
      "learning_rate": 3.998113248887062e-05,
      "loss": 0.7655,
      "step": 486
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4887411781357356,
      "learning_rate": 3.998088012372348e-05,
      "loss": 0.7361,
      "step": 487
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4775697840147466,
      "learning_rate": 3.998062608280405e-05,
      "loss": 0.8162,
      "step": 488
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4965796719248768,
      "learning_rate": 3.998037036613365e-05,
      "loss": 0.735,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4835640570889444,
      "learning_rate": 3.9980112973733714e-05,
      "loss": 0.7383,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4453489881147271,
      "learning_rate": 3.997985390562583e-05,
      "loss": 0.7196,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.107316908812484,
      "learning_rate": 3.997959316183173e-05,
      "loss": 0.3611,
      "step": 492
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4352578732454092,
      "learning_rate": 3.9979330742373286e-05,
      "loss": 0.6868,
      "step": 493
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4940248029419971,
      "learning_rate": 3.99790666472725e-05,
      "loss": 0.7666,
      "step": 494
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5254546606776338,
      "learning_rate": 3.997880087655152e-05,
      "loss": 0.7177,
      "step": 495
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5018861148713596,
      "learning_rate": 3.997853343023263e-05,
      "loss": 0.739,
      "step": 496
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4304884708442096,
      "learning_rate": 3.9978264308338287e-05,
      "loss": 0.7393,
      "step": 497
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4572357452122689,
      "learning_rate": 3.9977993510891036e-05,
      "loss": 0.7331,
      "step": 498
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5044335949760187,
      "learning_rate": 3.9977721037913594e-05,
      "loss": 0.7414,
      "step": 499
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5123186486525863,
      "learning_rate": 3.997744688942882e-05,
      "loss": 0.7616,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4534059795401516,
      "learning_rate": 3.99771710654597e-05,
      "loss": 0.7762,
      "step": 501
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43795159186243776,
      "learning_rate": 3.997689356602937e-05,
      "loss": 0.6904,
      "step": 502
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3971445030129901,
      "learning_rate": 3.99766143911611e-05,
      "loss": 0.7654,
      "step": 503
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4552227504788532,
      "learning_rate": 3.997633354087831e-05,
      "loss": 0.7578,
      "step": 504
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4888772772878807,
      "learning_rate": 3.9976051015204544e-05,
      "loss": 0.7592,
      "step": 505
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4612493258491078,
      "learning_rate": 3.9975766814163506e-05,
      "loss": 0.7743,
      "step": 506
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44152385716251613,
      "learning_rate": 3.997548093777904e-05,
      "loss": 0.7208,
      "step": 507
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5451473709891111,
      "learning_rate": 3.997519338607511e-05,
      "loss": 0.8038,
      "step": 508
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48631770119828854,
      "learning_rate": 3.997490415907582e-05,
      "loss": 0.7647,
      "step": 509
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5266907463963029,
      "learning_rate": 3.997461325680545e-05,
      "loss": 0.7541,
      "step": 510
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4765410195125885,
      "learning_rate": 3.9974320679288395e-05,
      "loss": 0.7075,
      "step": 511
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4570460535764937,
      "learning_rate": 3.997402642654919e-05,
      "loss": 0.7694,
      "step": 512
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42334751128783654,
      "learning_rate": 3.99737304986125e-05,
      "loss": 0.7367,
      "step": 513
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4791266121294545,
      "learning_rate": 3.997343289550316e-05,
      "loss": 0.7467,
      "step": 514
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4534823044259912,
      "learning_rate": 3.9973133617246125e-05,
      "loss": 0.7509,
      "step": 515
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4713440118335267,
      "learning_rate": 3.9972832663866496e-05,
      "loss": 0.7924,
      "step": 516
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4693735811441656,
      "learning_rate": 3.997253003538952e-05,
      "loss": 0.7601,
      "step": 517
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4435354530459694,
      "learning_rate": 3.997222573184056e-05,
      "loss": 0.7589,
      "step": 518
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4990684725783971,
      "learning_rate": 3.997191975324515e-05,
      "loss": 0.7947,
      "step": 519
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4266501950300463,
      "learning_rate": 3.9971612099628945e-05,
      "loss": 0.7401,
      "step": 520
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47714766679303333,
      "learning_rate": 3.997130277101776e-05,
      "loss": 0.7768,
      "step": 521
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5531781807999284,
      "learning_rate": 3.997099176743753e-05,
      "loss": 0.7741,
      "step": 522
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4623897034711402,
      "learning_rate": 3.997067908891434e-05,
      "loss": 0.72,
      "step": 523
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48730169384974636,
      "learning_rate": 3.9970364735474406e-05,
      "loss": 0.7656,
      "step": 524
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45280689957287745,
      "learning_rate": 3.9970048707144104e-05,
      "loss": 0.7463,
      "step": 525
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4666943684560168,
      "learning_rate": 3.996973100394994e-05,
      "loss": 0.7537,
      "step": 526
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47149122956203127,
      "learning_rate": 3.996941162591855e-05,
      "loss": 0.7662,
      "step": 527
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49039690852414985,
      "learning_rate": 3.9969090573076725e-05,
      "loss": 0.7763,
      "step": 528
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4705230899084127,
      "learning_rate": 3.996876784545139e-05,
      "loss": 0.7509,
      "step": 529
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4613724098163056,
      "learning_rate": 3.996844344306961e-05,
      "loss": 0.7621,
      "step": 530
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4073899833232935,
      "learning_rate": 3.9968117365958594e-05,
      "loss": 0.7194,
      "step": 531
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4797157754887573,
      "learning_rate": 3.9967789614145694e-05,
      "loss": 0.7842,
      "step": 532
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3673329775351421,
      "learning_rate": 3.996746018765839e-05,
      "loss": 0.7076,
      "step": 533
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43941442096248573,
      "learning_rate": 3.996712908652431e-05,
      "loss": 0.7345,
      "step": 534
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40735507183739295,
      "learning_rate": 3.996679631077124e-05,
      "loss": 0.7331,
      "step": 535
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4507962477697629,
      "learning_rate": 3.996646186042707e-05,
      "loss": 0.7392,
      "step": 536
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47582459589284204,
      "learning_rate": 3.996612573551986e-05,
      "loss": 0.7882,
      "step": 537
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4472199465269131,
      "learning_rate": 3.9965787936077795e-05,
      "loss": 0.7752,
      "step": 538
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44776253933323285,
      "learning_rate": 3.996544846212921e-05,
      "loss": 0.7393,
      "step": 539
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5367692661233083,
      "learning_rate": 3.996510731370257e-05,
      "loss": 0.7964,
      "step": 540
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.464378376155837,
      "learning_rate": 3.99647644908265e-05,
      "loss": 0.7533,
      "step": 541
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4555973718391219,
      "learning_rate": 3.996441999352973e-05,
      "loss": 0.7393,
      "step": 542
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4692269205053077,
      "learning_rate": 3.996407382184118e-05,
      "loss": 0.7327,
      "step": 543
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4544648649970294,
      "learning_rate": 3.996372597578987e-05,
      "loss": 0.7567,
      "step": 544
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5023650016870457,
      "learning_rate": 3.996337645540496e-05,
      "loss": 0.7479,
      "step": 545
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4370496048358207,
      "learning_rate": 3.9963025260715784e-05,
      "loss": 0.715,
      "step": 546
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45660557533958673,
      "learning_rate": 3.996267239175179e-05,
      "loss": 0.733,
      "step": 547
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43201987493759425,
      "learning_rate": 3.9962317848542565e-05,
      "loss": 0.7111,
      "step": 548
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.464314062492322,
      "learning_rate": 3.9961961631117853e-05,
      "loss": 0.7483,
      "step": 549
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4918158748652747,
      "learning_rate": 3.996160373950753e-05,
      "loss": 0.7622,
      "step": 550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4396151796761339,
      "learning_rate": 3.99612441737416e-05,
      "loss": 0.7321,
      "step": 551
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4265622884170004,
      "learning_rate": 3.9960882933850236e-05,
      "loss": 0.7347,
      "step": 552
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47823477199839715,
      "learning_rate": 3.996052001986373e-05,
      "loss": 0.7651,
      "step": 553
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46762040567745305,
      "learning_rate": 3.996015543181251e-05,
      "loss": 0.7929,
      "step": 554
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4931202835047349,
      "learning_rate": 3.995978916972716e-05,
      "loss": 0.733,
      "step": 555
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4976909168072441,
      "learning_rate": 3.9959421233638405e-05,
      "loss": 0.7716,
      "step": 556
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45718688166464033,
      "learning_rate": 3.9959051623577085e-05,
      "loss": 0.6898,
      "step": 557
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44443651874263973,
      "learning_rate": 3.995868033957422e-05,
      "loss": 0.7514,
      "step": 558
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5117042081822992,
      "learning_rate": 3.995830738166093e-05,
      "loss": 0.7671,
      "step": 559
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5698222258187892,
      "learning_rate": 3.995793274986851e-05,
      "loss": 0.8088,
      "step": 560
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3801865445766651,
      "learning_rate": 3.995755644422837e-05,
      "loss": 0.6714,
      "step": 561
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46930399867210215,
      "learning_rate": 3.995717846477207e-05,
      "loss": 0.7281,
      "step": 562
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.36829837066403276,
      "learning_rate": 3.995679881153132e-05,
      "loss": 0.7297,
      "step": 563
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3998551654970105,
      "learning_rate": 3.9956417484537956e-05,
      "loss": 0.7034,
      "step": 564
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41641072964678466,
      "learning_rate": 3.9956034483823954e-05,
      "loss": 0.7423,
      "step": 565
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4239368167408748,
      "learning_rate": 3.995564980942145e-05,
      "loss": 0.738,
      "step": 566
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3716217610327783,
      "learning_rate": 3.995526346136269e-05,
      "loss": 0.7064,
      "step": 567
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4667257593078147,
      "learning_rate": 3.995487543968008e-05,
      "loss": 0.7644,
      "step": 568
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4753403023263858,
      "learning_rate": 3.995448574440617e-05,
      "loss": 0.7117,
      "step": 569
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42844798943217355,
      "learning_rate": 3.995409437557364e-05,
      "loss": 0.7816,
      "step": 570
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44860524433248267,
      "learning_rate": 3.9953701333215314e-05,
      "loss": 0.7616,
      "step": 571
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.404626438140634,
      "learning_rate": 3.995330661736416e-05,
      "loss": 0.7023,
      "step": 572
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3962824720483544,
      "learning_rate": 3.995291022805326e-05,
      "loss": 0.7366,
      "step": 573
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4894751680508126,
      "learning_rate": 3.995251216531589e-05,
      "loss": 0.7931,
      "step": 574
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4263506602856468,
      "learning_rate": 3.995211242918542e-05,
      "loss": 0.7725,
      "step": 575
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4314395482001416,
      "learning_rate": 3.9951711019695375e-05,
      "loss": 0.7232,
      "step": 576
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.443696311981766,
      "learning_rate": 3.995130793687942e-05,
      "loss": 0.7192,
      "step": 577
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39600655315605565,
      "learning_rate": 3.9950903180771365e-05,
      "loss": 0.7285,
      "step": 578
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4875819743026769,
      "learning_rate": 3.995049675140516e-05,
      "loss": 0.6929,
      "step": 579
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4780791040982828,
      "learning_rate": 3.995008864881488e-05,
      "loss": 0.7354,
      "step": 580
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47529392056444514,
      "learning_rate": 3.9949678873034766e-05,
      "loss": 0.7395,
      "step": 581
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4388577199648977,
      "learning_rate": 3.9949267424099175e-05,
      "loss": 0.7179,
      "step": 582
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45123248549812983,
      "learning_rate": 3.994885430204261e-05,
      "loss": 0.7799,
      "step": 583
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46664085514825127,
      "learning_rate": 3.994843950689974e-05,
      "loss": 0.7711,
      "step": 584
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4733140635878419,
      "learning_rate": 3.994802303870533e-05,
      "loss": 0.7391,
      "step": 585
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.466379382584718,
      "learning_rate": 3.994760489749432e-05,
      "loss": 0.766,
      "step": 586
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3703076704643255,
      "learning_rate": 3.994718508330178e-05,
      "loss": 0.7144,
      "step": 587
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4148198955977376,
      "learning_rate": 3.994676359616292e-05,
      "loss": 0.7337,
      "step": 588
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4082328292220436,
      "learning_rate": 3.9946340436113084e-05,
      "loss": 0.7445,
      "step": 589
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4144377998509284,
      "learning_rate": 3.994591560318776e-05,
      "loss": 0.7355,
      "step": 590
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5354669867635111,
      "learning_rate": 3.994548909742259e-05,
      "loss": 0.8047,
      "step": 591
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44754476250124814,
      "learning_rate": 3.994506091885333e-05,
      "loss": 0.7515,
      "step": 592
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47062657773487515,
      "learning_rate": 3.994463106751591e-05,
      "loss": 0.7219,
      "step": 593
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42867642234269004,
      "learning_rate": 3.994419954344635e-05,
      "loss": 0.7139,
      "step": 594
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4777379078600945,
      "learning_rate": 3.9943766346680884e-05,
      "loss": 0.7598,
      "step": 595
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46761932908886567,
      "learning_rate": 3.994333147725581e-05,
      "loss": 0.7677,
      "step": 596
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4371014327716791,
      "learning_rate": 3.994289493520761e-05,
      "loss": 0.727,
      "step": 597
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48557677968989477,
      "learning_rate": 3.994245672057289e-05,
      "loss": 0.7928,
      "step": 598
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40570453048562394,
      "learning_rate": 3.994201683338842e-05,
      "loss": 0.7165,
      "step": 599
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4219021501971988,
      "learning_rate": 3.994157527369107e-05,
      "loss": 0.7402,
      "step": 600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4510539953764998,
      "learning_rate": 3.994113204151789e-05,
      "loss": 0.7212,
      "step": 601
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47279300182602335,
      "learning_rate": 3.994068713690606e-05,
      "loss": 0.7577,
      "step": 602
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42454192608846536,
      "learning_rate": 3.9940240559892875e-05,
      "loss": 0.6935,
      "step": 603
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4382533945531273,
      "learning_rate": 3.993979231051579e-05,
      "loss": 0.7268,
      "step": 604
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4347253610670376,
      "learning_rate": 3.993934238881241e-05,
      "loss": 0.7473,
      "step": 605
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4113518736725938,
      "learning_rate": 3.9938890794820465e-05,
      "loss": 0.6988,
      "step": 606
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5009054085420184,
      "learning_rate": 3.9938437528577824e-05,
      "loss": 0.7863,
      "step": 607
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4276602842681522,
      "learning_rate": 3.993798259012252e-05,
      "loss": 0.7411,
      "step": 608
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4293617941027217,
      "learning_rate": 3.993752597949268e-05,
      "loss": 0.7277,
      "step": 609
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4220858848245686,
      "learning_rate": 3.993706769672662e-05,
      "loss": 0.7393,
      "step": 610
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43555138828846285,
      "learning_rate": 3.9936607741862774e-05,
      "loss": 0.6922,
      "step": 611
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40605579242306133,
      "learning_rate": 3.993614611493971e-05,
      "loss": 0.6964,
      "step": 612
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42367399278219686,
      "learning_rate": 3.993568281599615e-05,
      "loss": 0.7238,
      "step": 613
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4465613035996987,
      "learning_rate": 3.993521784507095e-05,
      "loss": 0.768,
      "step": 614
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4978294171231172,
      "learning_rate": 3.9934751202203104e-05,
      "loss": 0.7651,
      "step": 615
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.36332998253770427,
      "learning_rate": 3.993428288743175e-05,
      "loss": 0.7004,
      "step": 616
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4551766438587616,
      "learning_rate": 3.993381290079616e-05,
      "loss": 0.7187,
      "step": 617
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5172086700294333,
      "learning_rate": 3.993334124233576e-05,
      "loss": 0.7853,
      "step": 618
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5329919713681869,
      "learning_rate": 3.99328679120901e-05,
      "loss": 0.7839,
      "step": 619
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43524730173796067,
      "learning_rate": 3.993239291009888e-05,
      "loss": 0.7399,
      "step": 620
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5183038189535283,
      "learning_rate": 3.993191623640195e-05,
      "loss": 0.7857,
      "step": 621
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4339590871940094,
      "learning_rate": 3.9931437891039266e-05,
      "loss": 0.6927,
      "step": 622
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4815237340543706,
      "learning_rate": 3.9930957874050964e-05,
      "loss": 0.7151,
      "step": 623
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44600272955287806,
      "learning_rate": 3.9930476185477286e-05,
      "loss": 0.7306,
      "step": 624
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4581028950103957,
      "learning_rate": 3.992999282535865e-05,
      "loss": 0.6917,
      "step": 625
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40086011225800455,
      "learning_rate": 3.992950779373558e-05,
      "loss": 0.7229,
      "step": 626
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4490453684938336,
      "learning_rate": 3.992902109064876e-05,
      "loss": 0.7126,
      "step": 627
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41370252141439856,
      "learning_rate": 3.992853271613901e-05,
      "loss": 0.6751,
      "step": 628
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4404344648534978,
      "learning_rate": 3.9928042670247294e-05,
      "loss": 0.7594,
      "step": 629
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46025657095332717,
      "learning_rate": 3.99275509530147e-05,
      "loss": 0.7567,
      "step": 630
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4669378504080285,
      "learning_rate": 3.992705756448248e-05,
      "loss": 0.7287,
      "step": 631
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4486010290487258,
      "learning_rate": 3.9926562504692e-05,
      "loss": 0.7086,
      "step": 632
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47000381291179194,
      "learning_rate": 3.992606577368479e-05,
      "loss": 0.7613,
      "step": 633
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40864259298912603,
      "learning_rate": 3.992556737150251e-05,
      "loss": 0.7578,
      "step": 634
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4486309100910012,
      "learning_rate": 3.992506729818696e-05,
      "loss": 0.7038,
      "step": 635
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4112653270335197,
      "learning_rate": 3.9924565553780076e-05,
      "loss": 0.7142,
      "step": 636
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4183358933124458,
      "learning_rate": 3.9924062138323945e-05,
      "loss": 0.7061,
      "step": 637
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41295105373993335,
      "learning_rate": 3.992355705186078e-05,
      "loss": 0.7743,
      "step": 638
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9264745418905802,
      "learning_rate": 3.992305029443295e-05,
      "loss": 0.4348,
      "step": 639
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4749407588307009,
      "learning_rate": 3.992254186608295e-05,
      "loss": 0.7004,
      "step": 640
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4108418559661517,
      "learning_rate": 3.992203176685343e-05,
      "loss": 0.7138,
      "step": 641
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5431779969063293,
      "learning_rate": 3.992151999678716e-05,
      "loss": 0.7687,
      "step": 642
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4328719403501901,
      "learning_rate": 3.9921006555927065e-05,
      "loss": 0.67,
      "step": 643
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4281254179463862,
      "learning_rate": 3.9920491444316215e-05,
      "loss": 0.7225,
      "step": 644
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.422381193581763,
      "learning_rate": 3.99199746619978e-05,
      "loss": 0.7009,
      "step": 645
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49278383005739645,
      "learning_rate": 3.9919456209015174e-05,
      "loss": 0.6994,
      "step": 646
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48739254192009934,
      "learning_rate": 3.991893608541181e-05,
      "loss": 0.7737,
      "step": 647
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4336480124570362,
      "learning_rate": 3.9918414291231326e-05,
      "loss": 0.7635,
      "step": 648
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4868525543182586,
      "learning_rate": 3.9917890826517496e-05,
      "loss": 0.7089,
      "step": 649
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5444614073323037,
      "learning_rate": 3.9917365691314216e-05,
      "loss": 0.7911,
      "step": 650
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5719819393565672,
      "learning_rate": 3.991683888566553e-05,
      "loss": 0.7298,
      "step": 651
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4816207865800524,
      "learning_rate": 3.991631040961562e-05,
      "loss": 0.7678,
      "step": 652
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4308352199981906,
      "learning_rate": 3.991578026320881e-05,
      "loss": 0.7212,
      "step": 653
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4869195633888084,
      "learning_rate": 3.9915248446489564e-05,
      "loss": 0.7837,
      "step": 654
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5254416096263118,
      "learning_rate": 3.9914714959502474e-05,
      "loss": 0.7524,
      "step": 655
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4390853785723994,
      "learning_rate": 3.99141798022923e-05,
      "loss": 0.7479,
      "step": 656
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4957765773343556,
      "learning_rate": 3.9913642974903906e-05,
      "loss": 0.7386,
      "step": 657
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45044579403902296,
      "learning_rate": 3.9913104477382335e-05,
      "loss": 0.776,
      "step": 658
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4435171305346514,
      "learning_rate": 3.991256430977274e-05,
      "loss": 0.7545,
      "step": 659
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4327798741493769,
      "learning_rate": 3.991202247212042e-05,
      "loss": 0.7318,
      "step": 660
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42705472159424046,
      "learning_rate": 3.991147896447082e-05,
      "loss": 0.7009,
      "step": 661
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5219932585179906,
      "learning_rate": 3.991093378686953e-05,
      "loss": 0.7862,
      "step": 662
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4116486964185547,
      "learning_rate": 3.9910386939362274e-05,
      "loss": 0.7257,
      "step": 663
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4324786672437277,
      "learning_rate": 3.990983842199491e-05,
      "loss": 0.7398,
      "step": 664
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4659448724563981,
      "learning_rate": 3.990928823481344e-05,
      "loss": 0.7498,
      "step": 665
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45220300013226095,
      "learning_rate": 3.9908736377864005e-05,
      "loss": 0.7059,
      "step": 666
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41671349033072885,
      "learning_rate": 3.9908182851192905e-05,
      "loss": 0.7386,
      "step": 667
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42779271851422995,
      "learning_rate": 3.9907627654846536e-05,
      "loss": 0.7249,
      "step": 668
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4317757082164683,
      "learning_rate": 3.990707078887149e-05,
      "loss": 0.7232,
      "step": 669
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41469394796727627,
      "learning_rate": 3.990651225331446e-05,
      "loss": 0.7332,
      "step": 670
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43945332715774427,
      "learning_rate": 3.9905952048222286e-05,
      "loss": 0.7293,
      "step": 671
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4420001998103126,
      "learning_rate": 3.990539017364195e-05,
      "loss": 0.7782,
      "step": 672
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43286908889413706,
      "learning_rate": 3.990482662962058e-05,
      "loss": 0.7191,
      "step": 673
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48542037566255963,
      "learning_rate": 3.9904261416205444e-05,
      "loss": 0.7683,
      "step": 674
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4594589208614641,
      "learning_rate": 3.990369453344394e-05,
      "loss": 0.7214,
      "step": 675
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44793872818867214,
      "learning_rate": 3.990312598138361e-05,
      "loss": 0.7624,
      "step": 676
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46594918325298285,
      "learning_rate": 3.9902555760072144e-05,
      "loss": 0.7527,
      "step": 677
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4813847004909254,
      "learning_rate": 3.990198386955737e-05,
      "loss": 0.7753,
      "step": 678
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.466286641974618,
      "learning_rate": 3.990141030988723e-05,
      "loss": 0.5708,
      "step": 679
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46303562871424975,
      "learning_rate": 3.990083508110985e-05,
      "loss": 0.7171,
      "step": 680
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8965992026015187,
      "learning_rate": 3.990025818327346e-05,
      "loss": 0.5146,
      "step": 681
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6056950021425844,
      "learning_rate": 3.989967961642646e-05,
      "loss": 0.812,
      "step": 682
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5489574509938849,
      "learning_rate": 3.9899099380617354e-05,
      "loss": 0.7935,
      "step": 683
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5116281620044508,
      "learning_rate": 3.989851747589482e-05,
      "loss": 0.7174,
      "step": 684
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48678902042966177,
      "learning_rate": 3.989793390230765e-05,
      "loss": 0.7416,
      "step": 685
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48293532621281127,
      "learning_rate": 3.98973486599048e-05,
      "loss": 0.7693,
      "step": 686
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5687915770710246,
      "learning_rate": 3.989676174873535e-05,
      "loss": 0.8248,
      "step": 687
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5072407542598744,
      "learning_rate": 3.9896173168848515e-05,
      "loss": 0.7166,
      "step": 688
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.436325401385422,
      "learning_rate": 3.989558292029367e-05,
      "loss": 0.4841,
      "step": 689
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5187344795837392,
      "learning_rate": 3.989499100312031e-05,
      "loss": 0.754,
      "step": 690
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4572875432778379,
      "learning_rate": 3.989439741737808e-05,
      "loss": 0.767,
      "step": 691
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6123448825678597,
      "learning_rate": 3.989380216311677e-05,
      "loss": 0.7619,
      "step": 692
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4847269587736593,
      "learning_rate": 3.98932052403863e-05,
      "loss": 0.8033,
      "step": 693
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43835995643309583,
      "learning_rate": 3.989260664923672e-05,
      "loss": 0.7375,
      "step": 694
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5288002053555223,
      "learning_rate": 3.989200638971825e-05,
      "loss": 0.7456,
      "step": 695
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4589683641273149,
      "learning_rate": 3.989140446188123e-05,
      "loss": 0.7442,
      "step": 696
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5316235250887035,
      "learning_rate": 3.989080086577614e-05,
      "loss": 0.7683,
      "step": 697
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45702916983419,
      "learning_rate": 3.9890195601453606e-05,
      "loss": 0.7545,
      "step": 698
    },
    {
      "epoch": 0.06,
      "grad_norm": 39.049237588179906,
      "learning_rate": 3.988958866896439e-05,
      "loss": 1.6505,
      "step": 699
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45991836978343964,
      "learning_rate": 3.988898006835938e-05,
      "loss": 0.721,
      "step": 700
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3900194180602027,
      "learning_rate": 3.9888369799689644e-05,
      "loss": 0.6928,
      "step": 701
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6144160542798972,
      "learning_rate": 3.988775786300635e-05,
      "loss": 0.8123,
      "step": 702
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4113959562161541,
      "learning_rate": 3.9887144258360824e-05,
      "loss": 0.766,
      "step": 703
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5195681160791645,
      "learning_rate": 3.988652898580452e-05,
      "loss": 0.7597,
      "step": 704
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4043814744824493,
      "learning_rate": 3.988591204538905e-05,
      "loss": 0.7105,
      "step": 705
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5230232907227962,
      "learning_rate": 3.9885293437166155e-05,
      "loss": 0.7266,
      "step": 706
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40849545839730095,
      "learning_rate": 3.988467316118771e-05,
      "loss": 0.7367,
      "step": 707
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41297999993749196,
      "learning_rate": 3.9884051217505734e-05,
      "loss": 0.773,
      "step": 708
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.472366314630076,
      "learning_rate": 3.988342760617241e-05,
      "loss": 0.7329,
      "step": 709
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4117877263861336,
      "learning_rate": 3.988280232724003e-05,
      "loss": 0.7318,
      "step": 710
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44108664734741354,
      "learning_rate": 3.9882175380761016e-05,
      "loss": 0.7693,
      "step": 711
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49403344279303785,
      "learning_rate": 3.988154676678797e-05,
      "loss": 0.7893,
      "step": 712
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45964008049988214,
      "learning_rate": 3.9880916485373606e-05,
      "loss": 0.7837,
      "step": 713
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4550020127555738,
      "learning_rate": 3.9880284536570796e-05,
      "loss": 0.7428,
      "step": 714
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4302102680465257,
      "learning_rate": 3.9879650920432527e-05,
      "loss": 0.7265,
      "step": 715
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48629001390965515,
      "learning_rate": 3.987901563701194e-05,
      "loss": 0.7729,
      "step": 716
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4515856093397961,
      "learning_rate": 3.987837868636232e-05,
      "loss": 0.7187,
      "step": 717
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4440680761421225,
      "learning_rate": 3.9877740068537084e-05,
      "loss": 0.7009,
      "step": 718
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4808904602374052,
      "learning_rate": 3.9877099783589795e-05,
      "loss": 0.7674,
      "step": 719
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43417762116408204,
      "learning_rate": 3.987645783157416e-05,
      "loss": 0.758,
      "step": 720
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4386391274664717,
      "learning_rate": 3.9875814212544006e-05,
      "loss": 0.7403,
      "step": 721
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42499590678439764,
      "learning_rate": 3.9875168926553315e-05,
      "loss": 0.7371,
      "step": 722
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47443513380197755,
      "learning_rate": 3.987452197365622e-05,
      "loss": 0.7611,
      "step": 723
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4829129968521857,
      "learning_rate": 3.987387335390696e-05,
      "loss": 0.7162,
      "step": 724
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5333103004423845,
      "learning_rate": 3.987322306735994e-05,
      "loss": 0.7579,
      "step": 725
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5093571547103453,
      "learning_rate": 3.987257111406971e-05,
      "loss": 0.7972,
      "step": 726
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43860131668565844,
      "learning_rate": 3.987191749409094e-05,
      "loss": 0.7167,
      "step": 727
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4573430705784016,
      "learning_rate": 3.987126220747845e-05,
      "loss": 0.7262,
      "step": 728
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.42894711541394903,
      "learning_rate": 3.987060525428719e-05,
      "loss": 0.717,
      "step": 729
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.944315097614231,
      "learning_rate": 3.986994663457227e-05,
      "loss": 1.5123,
      "step": 730
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1470628247812038,
      "learning_rate": 3.986928634838893e-05,
      "loss": 0.7343,
      "step": 731
    },
    {
      "epoch": 0.07,
      "grad_norm": 28.243608068448264,
      "learning_rate": 3.986862439579253e-05,
      "loss": 0.9085,
      "step": 732
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3461715241533585,
      "learning_rate": 3.98679607768386e-05,
      "loss": 0.762,
      "step": 733
    },
    {
      "epoch": 0.07,
      "grad_norm": 102.40195862627498,
      "learning_rate": 3.9867295491582784e-05,
      "loss": 2.3109,
      "step": 734
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.491500958624565,
      "learning_rate": 3.9866628540080905e-05,
      "loss": 1.2742,
      "step": 735
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.535355907439943,
      "learning_rate": 3.986595992238887e-05,
      "loss": 0.997,
      "step": 736
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.2938701987897385,
      "learning_rate": 3.986528963856278e-05,
      "loss": 0.9315,
      "step": 737
    },
    {
      "epoch": 0.07,
      "grad_norm": 26.612198278020767,
      "learning_rate": 3.986461768865884e-05,
      "loss": 1.0648,
      "step": 738
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.30560916747687,
      "learning_rate": 3.9863944072733394e-05,
      "loss": 2.1963,
      "step": 739
    },
    {
      "epoch": 0.07,
      "grad_norm": 21.682261994016102,
      "learning_rate": 3.986326879084295e-05,
      "loss": 2.4565,
      "step": 740
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.519835798180415,
      "learning_rate": 3.986259184304414e-05,
      "loss": 1.8651,
      "step": 741
    },
    {
      "epoch": 0.07,
      "grad_norm": 38.68374417107596,
      "learning_rate": 3.986191322939375e-05,
      "loss": 2.3358,
      "step": 742
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.40392700841701,
      "learning_rate": 3.986123294994868e-05,
      "loss": 1.4752,
      "step": 743
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.742101240402859,
      "learning_rate": 3.9860551004765985e-05,
      "loss": 1.2859,
      "step": 744
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.69373187816688,
      "learning_rate": 3.9859867393902864e-05,
      "loss": 1.6474,
      "step": 745
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.296709708564427,
      "learning_rate": 3.985918211741665e-05,
      "loss": 1.5982,
      "step": 746
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.103750232533572,
      "learning_rate": 3.985849517536481e-05,
      "loss": 1.1549,
      "step": 747
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1861212061471735,
      "learning_rate": 3.985780656780498e-05,
      "loss": 1.0904,
      "step": 748
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1288353465194056,
      "learning_rate": 3.9857116294794874e-05,
      "loss": 1.0699,
      "step": 749
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5950287730406025,
      "learning_rate": 3.9856424356392425e-05,
      "loss": 1.0034,
      "step": 750
    },
    {
      "epoch": 0.07,
      "grad_norm": 80.57026117108826,
      "learning_rate": 3.9855730752655636e-05,
      "loss": 4.2776,
      "step": 751
    },
    {
      "epoch": 0.07,
      "grad_norm": 68.17373122290626,
      "learning_rate": 3.985503548364269e-05,
      "loss": 1.4619,
      "step": 752
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.949484961640497,
      "learning_rate": 3.9854338549411895e-05,
      "loss": 1.4326,
      "step": 753
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.894056252789349,
      "learning_rate": 3.9853639950021705e-05,
      "loss": 1.0492,
      "step": 754
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.249308677859596,
      "learning_rate": 3.9852939685530706e-05,
      "loss": 1.0501,
      "step": 755
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.538504991745647,
      "learning_rate": 3.985223775599763e-05,
      "loss": 1.1029,
      "step": 756
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0618908375504519,
      "learning_rate": 3.985153416148136e-05,
      "loss": 0.941,
      "step": 757
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7225063098240772,
      "learning_rate": 3.985082890204089e-05,
      "loss": 0.9402,
      "step": 758
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4027662999523147,
      "learning_rate": 3.9850121977735366e-05,
      "loss": 0.9892,
      "step": 759
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.447220737338147,
      "learning_rate": 3.984941338862409e-05,
      "loss": 0.9116,
      "step": 760
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2840673276979593,
      "learning_rate": 3.984870313476649e-05,
      "loss": 0.922,
      "step": 761
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7528263316159265,
      "learning_rate": 3.984799121622212e-05,
      "loss": 0.9384,
      "step": 762
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.824944018130355,
      "learning_rate": 3.9847277633050704e-05,
      "loss": 0.9909,
      "step": 763
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.675340451273989,
      "learning_rate": 3.984656238531208e-05,
      "loss": 0.9006,
      "step": 764
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.484059359234763,
      "learning_rate": 3.984584547306624e-05,
      "loss": 0.8344,
      "step": 765
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8707001643624318,
      "learning_rate": 3.98451268963733e-05,
      "loss": 0.8754,
      "step": 766
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1187369444133126,
      "learning_rate": 3.9844406655293545e-05,
      "loss": 0.9097,
      "step": 767
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0823130102198766,
      "learning_rate": 3.9843684749887364e-05,
      "loss": 0.8649,
      "step": 768
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.753855258326931,
      "learning_rate": 3.984296118021531e-05,
      "loss": 0.8783,
      "step": 769
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6593005544647018,
      "learning_rate": 3.984223594633806e-05,
      "loss": 0.8572,
      "step": 770
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.014515702784008,
      "learning_rate": 3.984150904831645e-05,
      "loss": 0.832,
      "step": 771
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6207126082687671,
      "learning_rate": 3.984078048621144e-05,
      "loss": 0.8029,
      "step": 772
    },
    {
      "epoch": 0.07,
      "grad_norm": 102.29164527811568,
      "learning_rate": 3.9840050260084124e-05,
      "loss": 4.4715,
      "step": 773
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.08593522956233,
      "learning_rate": 3.983931836999576e-05,
      "loss": 1.2905,
      "step": 774
    },
    {
      "epoch": 0.07,
      "grad_norm": 9.421392235891995,
      "learning_rate": 3.983858481600773e-05,
      "loss": 0.8937,
      "step": 775
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5036424538582829,
      "learning_rate": 3.983784959818154e-05,
      "loss": 0.9305,
      "step": 776
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.290645770917935,
      "learning_rate": 3.983711271657886e-05,
      "loss": 0.9092,
      "step": 777
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8042684178890913,
      "learning_rate": 3.983637417126151e-05,
      "loss": 0.8976,
      "step": 778
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.05723955127239,
      "learning_rate": 3.9835633962291396e-05,
      "loss": 0.9531,
      "step": 779
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7954176321697404,
      "learning_rate": 3.983489208973063e-05,
      "loss": 0.9028,
      "step": 780
    },
    {
      "epoch": 0.07,
      "grad_norm": 90.93976515137715,
      "learning_rate": 3.983414855364141e-05,
      "loss": 5.1701,
      "step": 781
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.549221420891714,
      "learning_rate": 3.983340335408611e-05,
      "loss": 0.8965,
      "step": 782
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.4061892053390315,
      "learning_rate": 3.983265649112722e-05,
      "loss": 0.9335,
      "step": 783
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1723612012235782,
      "learning_rate": 3.9831907964827386e-05,
      "loss": 0.9994,
      "step": 784
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0813312478353,
      "learning_rate": 3.983115777524938e-05,
      "loss": 0.8672,
      "step": 785
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9169422056745102,
      "learning_rate": 3.983040592245612e-05,
      "loss": 0.8545,
      "step": 786
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8238480218541222,
      "learning_rate": 3.9829652406510666e-05,
      "loss": 0.8446,
      "step": 787
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7037776793539492,
      "learning_rate": 3.982889722747621e-05,
      "loss": 0.8495,
      "step": 788
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.593780838157691,
      "learning_rate": 3.982814038541609e-05,
      "loss": 0.859,
      "step": 789
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6892163204788312,
      "learning_rate": 3.982738188039379e-05,
      "loss": 0.8537,
      "step": 790
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6732484036998018,
      "learning_rate": 3.982662171247291e-05,
      "loss": 0.8853,
      "step": 791
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6008729028364662,
      "learning_rate": 3.9825859881717214e-05,
      "loss": 0.8788,
      "step": 792
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5381316085866348,
      "learning_rate": 3.9825096388190594e-05,
      "loss": 0.7712,
      "step": 793
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5022041638907259,
      "learning_rate": 3.982433123195708e-05,
      "loss": 0.8616,
      "step": 794
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5298110035701256,
      "learning_rate": 3.982356441308085e-05,
      "loss": 0.8674,
      "step": 795
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.570702935722321,
      "learning_rate": 3.982279593162621e-05,
      "loss": 0.8598,
      "step": 796
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.537066461101011,
      "learning_rate": 3.982202578765762e-05,
      "loss": 0.845,
      "step": 797
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5571914775657044,
      "learning_rate": 3.982125398123966e-05,
      "loss": 0.813,
      "step": 798
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48328222877586047,
      "learning_rate": 3.982048051243707e-05,
      "loss": 0.784,
      "step": 799
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5980460806047941,
      "learning_rate": 3.981970538131472e-05,
      "loss": 0.8264,
      "step": 800
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4615364537989429,
      "learning_rate": 3.981892858793761e-05,
      "loss": 0.8012,
      "step": 801
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5383643308211881,
      "learning_rate": 3.98181501323709e-05,
      "loss": 0.8219,
      "step": 802
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5543228113357235,
      "learning_rate": 3.981737001467987e-05,
      "loss": 0.8566,
      "step": 803
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5479534061646634,
      "learning_rate": 3.981658823492995e-05,
      "loss": 0.7865,
      "step": 804
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46553888638141355,
      "learning_rate": 3.981580479318671e-05,
      "loss": 0.7529,
      "step": 805
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4580148048774779,
      "learning_rate": 3.981501968951585e-05,
      "loss": 0.8179,
      "step": 806
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47949385582622084,
      "learning_rate": 3.981423292398321e-05,
      "loss": 0.8051,
      "step": 807
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5377900062354212,
      "learning_rate": 3.98134444966548e-05,
      "loss": 0.8283,
      "step": 808
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5148540132870172,
      "learning_rate": 3.981265440759672e-05,
      "loss": 0.8198,
      "step": 809
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47719646928104636,
      "learning_rate": 3.981186265687525e-05,
      "loss": 0.7887,
      "step": 810
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5039449597911119,
      "learning_rate": 3.981106924455678e-05,
      "loss": 0.8302,
      "step": 811
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.536047229928186,
      "learning_rate": 3.981027417070785e-05,
      "loss": 0.7967,
      "step": 812
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3958499468948143,
      "learning_rate": 3.980947743539516e-05,
      "loss": 0.7535,
      "step": 813
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48816488794138135,
      "learning_rate": 3.9808679038685526e-05,
      "loss": 0.7994,
      "step": 814
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5477991662983397,
      "learning_rate": 3.98078789806459e-05,
      "loss": 0.8099,
      "step": 815
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48347100146493405,
      "learning_rate": 3.9807077261343374e-05,
      "loss": 0.7994,
      "step": 816
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40858104056645367,
      "learning_rate": 3.980627388084521e-05,
      "loss": 0.7818,
      "step": 817
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5085876416345199,
      "learning_rate": 3.980546883921878e-05,
      "loss": 0.8007,
      "step": 818
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4581424655820648,
      "learning_rate": 3.980466213653159e-05,
      "loss": 0.8269,
      "step": 819
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4347421516010553,
      "learning_rate": 3.9803853772851304e-05,
      "loss": 0.8106,
      "step": 820
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5190234886223117,
      "learning_rate": 3.980304374824572e-05,
      "loss": 0.7673,
      "step": 821
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4908114025172036,
      "learning_rate": 3.980223206278278e-05,
      "loss": 0.778,
      "step": 822
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4439057537296685,
      "learning_rate": 3.9801418716530545e-05,
      "loss": 0.7562,
      "step": 823
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.459081293992922,
      "learning_rate": 3.980060370955723e-05,
      "loss": 0.7495,
      "step": 824
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46057565861568006,
      "learning_rate": 3.97997870419312e-05,
      "loss": 0.812,
      "step": 825
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49225043291813353,
      "learning_rate": 3.979896871372095e-05,
      "loss": 0.8107,
      "step": 826
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46365432052862376,
      "learning_rate": 3.97981487249951e-05,
      "loss": 0.8229,
      "step": 827
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43872466361792267,
      "learning_rate": 3.9797327075822425e-05,
      "loss": 0.7847,
      "step": 828
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.391479917920385,
      "learning_rate": 3.9796503766271845e-05,
      "loss": 0.7425,
      "step": 829
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4843142166176709,
      "learning_rate": 3.97956787964124e-05,
      "loss": 0.7875,
      "step": 830
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4392288394497174,
      "learning_rate": 3.979485216631327e-05,
      "loss": 0.776,
      "step": 831
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4816467820064895,
      "learning_rate": 3.97940238760438e-05,
      "loss": 0.8004,
      "step": 832
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4604567568145682,
      "learning_rate": 3.9793193925673456e-05,
      "loss": 0.8152,
      "step": 833
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4334831029372247,
      "learning_rate": 3.979236231527185e-05,
      "loss": 0.7613,
      "step": 834
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.41006696691977,
      "learning_rate": 3.9791529044908704e-05,
      "loss": 0.772,
      "step": 835
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49036422959326087,
      "learning_rate": 3.979069411465393e-05,
      "loss": 0.7362,
      "step": 836
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46131670649167056,
      "learning_rate": 3.978985752457753e-05,
      "loss": 0.784,
      "step": 837
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.458896949201363,
      "learning_rate": 3.9789019274749685e-05,
      "loss": 0.7377,
      "step": 838
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5156640937437811,
      "learning_rate": 3.97881793652407e-05,
      "loss": 0.7742,
      "step": 839
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4655959191573969,
      "learning_rate": 3.978733779612101e-05,
      "loss": 0.7381,
      "step": 840
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4837945984329557,
      "learning_rate": 3.9786494567461184e-05,
      "loss": 0.7822,
      "step": 841
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.45833771389454336,
      "learning_rate": 3.978564967933196e-05,
      "loss": 0.7593,
      "step": 842
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4770285024725713,
      "learning_rate": 3.9784803131804195e-05,
      "loss": 0.7712,
      "step": 843
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.483180010099671,
      "learning_rate": 3.9783954924948884e-05,
      "loss": 0.7454,
      "step": 844
    },
    {
      "epoch": 0.08,
      "grad_norm": 244.81894066646248,
      "learning_rate": 3.978310505883716e-05,
      "loss": 3.839,
      "step": 845
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.9143443597407575,
      "learning_rate": 3.978225353354032e-05,
      "loss": 0.824,
      "step": 846
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.690681568380832,
      "learning_rate": 3.978140034912976e-05,
      "loss": 0.9536,
      "step": 847
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.284335874132491,
      "learning_rate": 3.978054550567704e-05,
      "loss": 0.8542,
      "step": 848
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.225279538874514,
      "learning_rate": 3.977968900325386e-05,
      "loss": 0.841,
      "step": 849
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.049605334262809,
      "learning_rate": 3.977883084193205e-05,
      "loss": 0.8352,
      "step": 850
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9672058808866641,
      "learning_rate": 3.9777971021783585e-05,
      "loss": 0.8246,
      "step": 851
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4862644282991992,
      "learning_rate": 3.9777109542880574e-05,
      "loss": 0.8272,
      "step": 852
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.885597860216329,
      "learning_rate": 3.977624640529527e-05,
      "loss": 0.7929,
      "step": 853
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9592928196806143,
      "learning_rate": 3.977538160910007e-05,
      "loss": 0.84,
      "step": 854
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.940369781430907,
      "learning_rate": 3.977451515436749e-05,
      "loss": 0.8019,
      "step": 855
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9353410694484238,
      "learning_rate": 3.977364704117021e-05,
      "loss": 0.7626,
      "step": 856
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5946345240249756,
      "learning_rate": 3.977277726958103e-05,
      "loss": 0.8087,
      "step": 857
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6522582449316415,
      "learning_rate": 3.97719058396729e-05,
      "loss": 0.7784,
      "step": 858
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6980989532485041,
      "learning_rate": 3.9771032751518914e-05,
      "loss": 0.7923,
      "step": 859
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6213732456705754,
      "learning_rate": 3.977015800519228e-05,
      "loss": 0.7871,
      "step": 860
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5674042619417339,
      "learning_rate": 3.976928160076637e-05,
      "loss": 0.7844,
      "step": 861
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5857849926681215,
      "learning_rate": 3.976840353831469e-05,
      "loss": 0.7759,
      "step": 862
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49605661909361454,
      "learning_rate": 3.976752381791088e-05,
      "loss": 0.7643,
      "step": 863
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4613799302287111,
      "learning_rate": 3.976664243962872e-05,
      "loss": 0.7338,
      "step": 864
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6819436736189345,
      "learning_rate": 3.976575940354213e-05,
      "loss": 0.7848,
      "step": 865
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5277203136111713,
      "learning_rate": 3.976487470972517e-05,
      "loss": 0.7884,
      "step": 866
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5142723794580124,
      "learning_rate": 3.976398835825204e-05,
      "loss": 0.7337,
      "step": 867
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6422344825853924,
      "learning_rate": 3.976310034919707e-05,
      "loss": 0.8251,
      "step": 868
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5617495311217854,
      "learning_rate": 3.9762210682634746e-05,
      "loss": 0.7792,
      "step": 869
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6178589516537082,
      "learning_rate": 3.976131935863968e-05,
      "loss": 0.8214,
      "step": 870
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.516013540297012,
      "learning_rate": 3.976042637728662e-05,
      "loss": 0.7736,
      "step": 871
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5700640215521318,
      "learning_rate": 3.975953173865047e-05,
      "loss": 0.76,
      "step": 872
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5113940127299399,
      "learning_rate": 3.975863544280625e-05,
      "loss": 0.7556,
      "step": 873
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4869856895456197,
      "learning_rate": 3.975773748982914e-05,
      "loss": 0.7654,
      "step": 874
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4886133641152854,
      "learning_rate": 3.975683787979445e-05,
      "loss": 0.774,
      "step": 875
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5172492870833474,
      "learning_rate": 3.9755936612777626e-05,
      "loss": 0.7385,
      "step": 876
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5317379141933964,
      "learning_rate": 3.975503368885425e-05,
      "loss": 0.7747,
      "step": 877
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5097392479464279,
      "learning_rate": 3.975412910810007e-05,
      "loss": 0.7464,
      "step": 878
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5319686759725818,
      "learning_rate": 3.975322287059094e-05,
      "loss": 0.7774,
      "step": 879
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4956824084845786,
      "learning_rate": 3.975231497640285e-05,
      "loss": 0.7512,
      "step": 880
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41762275357874606,
      "learning_rate": 3.9751405425611966e-05,
      "loss": 0.7535,
      "step": 881
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5524734095323984,
      "learning_rate": 3.975049421829456e-05,
      "loss": 0.8001,
      "step": 882
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5248900031973595,
      "learning_rate": 3.974958135452706e-05,
      "loss": 0.7981,
      "step": 883
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.515867293793552,
      "learning_rate": 3.9748666834386015e-05,
      "loss": 0.762,
      "step": 884
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4812774592283238,
      "learning_rate": 3.9747750657948145e-05,
      "loss": 0.7284,
      "step": 885
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4030300295185488,
      "learning_rate": 3.974683282529027e-05,
      "loss": 0.7444,
      "step": 886
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.43716269410597386,
      "learning_rate": 3.974591333648937e-05,
      "loss": 0.7287,
      "step": 887
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46012716654014957,
      "learning_rate": 3.9744992191622574e-05,
      "loss": 0.7448,
      "step": 888
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.45316924463724456,
      "learning_rate": 3.974406939076712e-05,
      "loss": 0.7476,
      "step": 889
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.40145509116609723,
      "learning_rate": 3.974314493400042e-05,
      "loss": 0.7127,
      "step": 890
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5049656691225627,
      "learning_rate": 3.974221882139999e-05,
      "loss": 0.7282,
      "step": 891
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42926399960870726,
      "learning_rate": 3.974129105304352e-05,
      "loss": 0.7182,
      "step": 892
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.459563987825906,
      "learning_rate": 3.97403616290088e-05,
      "loss": 0.7483,
      "step": 893
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4523531292172176,
      "learning_rate": 3.9739430549373796e-05,
      "loss": 0.757,
      "step": 894
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5348204366687123,
      "learning_rate": 3.9738497814216584e-05,
      "loss": 0.8188,
      "step": 895
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4539531842580808,
      "learning_rate": 3.973756342361541e-05,
      "loss": 0.7675,
      "step": 896
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5024922886145092,
      "learning_rate": 3.973662737764862e-05,
      "loss": 0.804,
      "step": 897
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46089550563788956,
      "learning_rate": 3.973568967639473e-05,
      "loss": 0.7786,
      "step": 898
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5063672990817168,
      "learning_rate": 3.973475031993238e-05,
      "loss": 0.7462,
      "step": 899
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4085728020506605,
      "learning_rate": 3.973380930834036e-05,
      "loss": 0.6991,
      "step": 900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48681728248990946,
      "learning_rate": 3.9732866641697586e-05,
      "loss": 0.7612,
      "step": 901
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4975601709156168,
      "learning_rate": 3.973192232008311e-05,
      "loss": 0.7497,
      "step": 902
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48623986496732563,
      "learning_rate": 3.9730976343576144e-05,
      "loss": 0.7297,
      "step": 903
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5052403974750618,
      "learning_rate": 3.973002871225602e-05,
      "loss": 0.7869,
      "step": 904
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.162353533963621,
      "learning_rate": 3.972907942620221e-05,
      "loss": 0.7698,
      "step": 905
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44966727648670773,
      "learning_rate": 3.972812848549435e-05,
      "loss": 0.7283,
      "step": 906
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4624674200625444,
      "learning_rate": 3.9727175890212166e-05,
      "loss": 0.7572,
      "step": 907
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5551588247268922,
      "learning_rate": 3.972622164043557e-05,
      "loss": 0.7297,
      "step": 908
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5609355747804353,
      "learning_rate": 3.972526573624459e-05,
      "loss": 0.7305,
      "step": 909
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46127298689775315,
      "learning_rate": 3.9724308177719395e-05,
      "loss": 0.7194,
      "step": 910
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4683186801733709,
      "learning_rate": 3.972334896494029e-05,
      "loss": 0.742,
      "step": 911
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5684251252305937,
      "learning_rate": 3.972238809798773e-05,
      "loss": 0.7253,
      "step": 912
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.396607873881449,
      "learning_rate": 3.972142557694229e-05,
      "loss": 0.7127,
      "step": 913
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48920501770478575,
      "learning_rate": 3.9720461401884716e-05,
      "loss": 0.7911,
      "step": 914
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49085115577444116,
      "learning_rate": 3.971949557289586e-05,
      "loss": 0.7673,
      "step": 915
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5116725974774421,
      "learning_rate": 3.971852809005672e-05,
      "loss": 0.7979,
      "step": 916
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4807966400639766,
      "learning_rate": 3.971755895344844e-05,
      "loss": 0.7356,
      "step": 917
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.401358813857337,
      "learning_rate": 3.971658816315231e-05,
      "loss": 0.7482,
      "step": 918
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.420379083093671,
      "learning_rate": 3.971561571924974e-05,
      "loss": 0.7244,
      "step": 919
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4263328569417266,
      "learning_rate": 3.971464162182229e-05,
      "loss": 0.7074,
      "step": 920
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4493900341511835,
      "learning_rate": 3.9713665870951657e-05,
      "loss": 0.7196,
      "step": 921
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48945343563842386,
      "learning_rate": 3.971268846671968e-05,
      "loss": 0.7518,
      "step": 922
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4563854537989886,
      "learning_rate": 3.971170940920832e-05,
      "loss": 0.7327,
      "step": 923
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5396058393246441,
      "learning_rate": 3.97107286984997e-05,
      "loss": 0.7879,
      "step": 924
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4551385328729665,
      "learning_rate": 3.970974633467607e-05,
      "loss": 0.7806,
      "step": 925
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49991026461820004,
      "learning_rate": 3.970876231781982e-05,
      "loss": 0.7604,
      "step": 926
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4689958731207649,
      "learning_rate": 3.970777664801347e-05,
      "loss": 0.7667,
      "step": 927
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46888323170843355,
      "learning_rate": 3.97067893253397e-05,
      "loss": 0.7458,
      "step": 928
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5305980594426118,
      "learning_rate": 3.9705800349881306e-05,
      "loss": 0.7722,
      "step": 929
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42305766567466896,
      "learning_rate": 3.970480972172124e-05,
      "loss": 0.7546,
      "step": 930
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3878875624242735,
      "learning_rate": 3.9703817440942575e-05,
      "loss": 0.6696,
      "step": 931
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49016092538590017,
      "learning_rate": 3.970282350762854e-05,
      "loss": 0.6882,
      "step": 932
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46566699388525723,
      "learning_rate": 3.970182792186249e-05,
      "loss": 0.7487,
      "step": 933
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4242417752457773,
      "learning_rate": 3.970083068372792e-05,
      "loss": 0.7055,
      "step": 934
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3907637210825002,
      "learning_rate": 3.969983179330848e-05,
      "loss": 0.6959,
      "step": 935
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4494633213218372,
      "learning_rate": 3.9698831250687944e-05,
      "loss": 0.7287,
      "step": 936
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42096556085232123,
      "learning_rate": 3.9697829055950216e-05,
      "loss": 0.7164,
      "step": 937
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48603572976532655,
      "learning_rate": 3.969682520917936e-05,
      "loss": 0.7397,
      "step": 938
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4725289889584046,
      "learning_rate": 3.969581971045955e-05,
      "loss": 0.7906,
      "step": 939
    },
    {
      "epoch": 0.08,
      "grad_norm": 187.06275546033825,
      "learning_rate": 3.9694812559875135e-05,
      "loss": 7.0476,
      "step": 940
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2126569912292264,
      "learning_rate": 3.9693803757510575e-05,
      "loss": 0.7261,
      "step": 941
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4720390981437714,
      "learning_rate": 3.9692793303450476e-05,
      "loss": 0.788,
      "step": 942
    },
    {
      "epoch": 0.08,
      "grad_norm": 36.01953253082479,
      "learning_rate": 3.9691781197779584e-05,
      "loss": 0.9763,
      "step": 943
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.704246184560814,
      "learning_rate": 3.9690767440582784e-05,
      "loss": 1.0244,
      "step": 944
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.787091623425756,
      "learning_rate": 3.9689752031945105e-05,
      "loss": 0.9068,
      "step": 945
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6831118951275603,
      "learning_rate": 3.9688734971951704e-05,
      "loss": 0.8926,
      "step": 946
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5753783686201177,
      "learning_rate": 3.968771626068787e-05,
      "loss": 0.8846,
      "step": 947
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6647587794905796,
      "learning_rate": 3.968669589823905e-05,
      "loss": 0.9183,
      "step": 948
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.601781753829527,
      "learning_rate": 3.968567388469083e-05,
      "loss": 0.8918,
      "step": 949
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.907653750626235,
      "learning_rate": 3.9684650220128914e-05,
      "loss": 0.861,
      "step": 950
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0566348827843681,
      "learning_rate": 3.968362490463915e-05,
      "loss": 0.8117,
      "step": 951
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7089076658490596,
      "learning_rate": 3.968259793830754e-05,
      "loss": 0.9155,
      "step": 952
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.975930912557807,
      "learning_rate": 3.9681569321220215e-05,
      "loss": 0.7951,
      "step": 953
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6601704601319249,
      "learning_rate": 3.968053905346343e-05,
      "loss": 0.7895,
      "step": 954
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7461406650500564,
      "learning_rate": 3.967950713512361e-05,
      "loss": 0.8149,
      "step": 955
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8502168912280219,
      "learning_rate": 3.96784735662873e-05,
      "loss": 0.8176,
      "step": 956
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7663131221463622,
      "learning_rate": 3.967743834704117e-05,
      "loss": 0.8311,
      "step": 957
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7694525651283932,
      "learning_rate": 3.9676401477472053e-05,
      "loss": 0.8604,
      "step": 958
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.663605419585027,
      "learning_rate": 3.967536295766691e-05,
      "loss": 0.8053,
      "step": 959
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6956029066440678,
      "learning_rate": 3.9674322787712825e-05,
      "loss": 0.8249,
      "step": 960
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5920218598119088,
      "learning_rate": 3.967328096769706e-05,
      "loss": 0.8134,
      "step": 961
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5831667478867999,
      "learning_rate": 3.967223749770697e-05,
      "loss": 0.7648,
      "step": 962
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7440131504849704,
      "learning_rate": 3.967119237783009e-05,
      "loss": 0.8388,
      "step": 963
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5973935619322341,
      "learning_rate": 3.967014560815405e-05,
      "loss": 0.7602,
      "step": 964
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5360661564888763,
      "learning_rate": 3.966909718876666e-05,
      "loss": 0.7737,
      "step": 965
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4712843055646287,
      "learning_rate": 3.966804711975585e-05,
      "loss": 0.7828,
      "step": 966
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46724904949118645,
      "learning_rate": 3.966699540120967e-05,
      "loss": 0.7611,
      "step": 967
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4345576951545582,
      "learning_rate": 3.966594203321634e-05,
      "loss": 0.7613,
      "step": 968
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44868729417747993,
      "learning_rate": 3.966488701586421e-05,
      "loss": 0.7654,
      "step": 969
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44591712868983807,
      "learning_rate": 3.966383034924175e-05,
      "loss": 0.7363,
      "step": 970
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49535657836689595,
      "learning_rate": 3.966277203343758e-05,
      "loss": 0.816,
      "step": 971
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4766674526179854,
      "learning_rate": 3.966171206854047e-05,
      "loss": 0.7942,
      "step": 972
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4138961582297848,
      "learning_rate": 3.966065045463932e-05,
      "loss": 0.7206,
      "step": 973
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4539617180724506,
      "learning_rate": 3.9659587191823154e-05,
      "loss": 0.7499,
      "step": 974
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.41943199267959236,
      "learning_rate": 3.965852228018116e-05,
      "loss": 0.779,
      "step": 975
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46156603365706833,
      "learning_rate": 3.965745571980264e-05,
      "loss": 0.7706,
      "step": 976
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.41641994914802016,
      "learning_rate": 3.965638751077705e-05,
      "loss": 0.7209,
      "step": 977
    },
    {
      "epoch": 0.09,
      "grad_norm": 105.63902544408259,
      "learning_rate": 3.965531765319397e-05,
      "loss": 8.2051,
      "step": 978
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.926859852285284,
      "learning_rate": 3.9654246147143145e-05,
      "loss": 0.7715,
      "step": 979
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1439101395078977,
      "learning_rate": 3.9653172992714434e-05,
      "loss": 0.8168,
      "step": 980
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0115293885990888,
      "learning_rate": 3.9652098189997825e-05,
      "loss": 0.7631,
      "step": 981
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.996149428212821,
      "learning_rate": 3.965102173908349e-05,
      "loss": 0.8782,
      "step": 982
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5057763959352457,
      "learning_rate": 3.9649943640061686e-05,
      "loss": 0.7841,
      "step": 983
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.546720449587582,
      "learning_rate": 3.964886389302284e-05,
      "loss": 0.825,
      "step": 984
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8538963250772162,
      "learning_rate": 3.9647782498057515e-05,
      "loss": 0.8422,
      "step": 985
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3246234813069895,
      "learning_rate": 3.9646699455256396e-05,
      "loss": 0.7781,
      "step": 986
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4048605676075139,
      "learning_rate": 3.964561476471033e-05,
      "loss": 0.7768,
      "step": 987
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6918914619459189,
      "learning_rate": 3.964452842651026e-05,
      "loss": 0.7821,
      "step": 988
    },
    {
      "epoch": 0.09,
      "grad_norm": 73.03199910128643,
      "learning_rate": 3.964344044074733e-05,
      "loss": 8.7854,
      "step": 989
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.074168846377257,
      "learning_rate": 3.964235080751277e-05,
      "loss": 0.8298,
      "step": 990
    },
    {
      "epoch": 0.09,
      "grad_norm": 22.014869113299735,
      "learning_rate": 3.9641259526897974e-05,
      "loss": 1.0942,
      "step": 991
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.198791935851004,
      "learning_rate": 3.9640166598994454e-05,
      "loss": 0.8415,
      "step": 992
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.543468583247521,
      "learning_rate": 3.963907202389389e-05,
      "loss": 0.936,
      "step": 993
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6085248296144012,
      "learning_rate": 3.963797580168806e-05,
      "loss": 0.8493,
      "step": 994
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.830534295883685,
      "learning_rate": 3.963687793246893e-05,
      "loss": 0.8774,
      "step": 995
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8771430712017086,
      "learning_rate": 3.9635778416328556e-05,
      "loss": 0.7806,
      "step": 996
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8648638861030026,
      "learning_rate": 3.963467725335916e-05,
      "loss": 0.8081,
      "step": 997
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7515413761289608,
      "learning_rate": 3.96335744436531e-05,
      "loss": 0.7739,
      "step": 998
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8581966552788584,
      "learning_rate": 3.963246998730286e-05,
      "loss": 0.806,
      "step": 999
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6096224124147751,
      "learning_rate": 3.963136388440107e-05,
      "loss": 0.7429,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6278751818015775,
      "learning_rate": 3.96302561350405e-05,
      "loss": 0.7874,
      "step": 1001
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5782261032934457,
      "learning_rate": 3.9629146739314055e-05,
      "loss": 0.8067,
      "step": 1002
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6757566110194757,
      "learning_rate": 3.962803569731478e-05,
      "loss": 0.8032,
      "step": 1003
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5403908348657669,
      "learning_rate": 3.962692300913586e-05,
      "loss": 0.74,
      "step": 1004
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6324070226459847,
      "learning_rate": 3.962580867487061e-05,
      "loss": 0.8161,
      "step": 1005
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6437670043090575,
      "learning_rate": 3.962469269461249e-05,
      "loss": 0.757,
      "step": 1006
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6534995409060194,
      "learning_rate": 3.962357506845509e-05,
      "loss": 0.7758,
      "step": 1007
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5622136375511128,
      "learning_rate": 3.9622455796492144e-05,
      "loss": 0.7899,
      "step": 1008
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6049500997341049,
      "learning_rate": 3.962133487881754e-05,
      "loss": 0.7614,
      "step": 1009
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6956865939284989,
      "learning_rate": 3.9620212315525265e-05,
      "loss": 0.7817,
      "step": 1010
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5636711606257613,
      "learning_rate": 3.961908810670948e-05,
      "loss": 0.768,
      "step": 1011
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.161116408579692,
      "learning_rate": 3.9617962252464474e-05,
      "loss": 0.8116,
      "step": 1012
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5296242189867049,
      "learning_rate": 3.961683475288466e-05,
      "loss": 0.7617,
      "step": 1013
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4984800607767209,
      "learning_rate": 3.961570560806461e-05,
      "loss": 0.7722,
      "step": 1014
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5453670021097238,
      "learning_rate": 3.961457481809902e-05,
      "loss": 0.7607,
      "step": 1015
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4343559573801026,
      "learning_rate": 3.961344238308274e-05,
      "loss": 0.7376,
      "step": 1016
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5558810762480867,
      "learning_rate": 3.9612308303110715e-05,
      "loss": 0.7543,
      "step": 1017
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4525386660607209,
      "learning_rate": 3.961117257827809e-05,
      "loss": 0.7621,
      "step": 1018
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.502863165222504,
      "learning_rate": 3.96100352086801e-05,
      "loss": 0.7794,
      "step": 1019
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6538711678519408,
      "learning_rate": 3.960889619441214e-05,
      "loss": 0.8007,
      "step": 1020
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5238096759329232,
      "learning_rate": 3.9607755535569745e-05,
      "loss": 0.787,
      "step": 1021
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4859843892589709,
      "learning_rate": 3.9606613232248564e-05,
      "loss": 0.7745,
      "step": 1022
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4249286727912984,
      "learning_rate": 3.9605469284544416e-05,
      "loss": 0.8031,
      "step": 1023
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4641807444699075,
      "learning_rate": 3.960432369255323e-05,
      "loss": 0.7848,
      "step": 1024
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47867278210583775,
      "learning_rate": 3.96031764563711e-05,
      "loss": 0.7475,
      "step": 1025
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49140713363692545,
      "learning_rate": 3.960202757609423e-05,
      "loss": 0.7973,
      "step": 1026
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.559506452717162,
      "learning_rate": 3.960087705181899e-05,
      "loss": 0.8004,
      "step": 1027
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5102760593704562,
      "learning_rate": 3.959972488364185e-05,
      "loss": 0.7741,
      "step": 1028
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43885971697077664,
      "learning_rate": 3.959857107165946e-05,
      "loss": 0.7587,
      "step": 1029
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48926404028848586,
      "learning_rate": 3.959741561596858e-05,
      "loss": 0.7572,
      "step": 1030
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5475415059515697,
      "learning_rate": 3.959625851666613e-05,
      "loss": 0.7949,
      "step": 1031
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47488282893047556,
      "learning_rate": 3.959509977384914e-05,
      "loss": 0.746,
      "step": 1032
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42610683837729263,
      "learning_rate": 3.959393938761479e-05,
      "loss": 0.7201,
      "step": 1033
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4312534748484394,
      "learning_rate": 3.959277735806041e-05,
      "loss": 0.7031,
      "step": 1034
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.38960688905125257,
      "learning_rate": 3.959161368528346e-05,
      "loss": 0.7202,
      "step": 1035
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43475965032209807,
      "learning_rate": 3.959044836938153e-05,
      "loss": 0.7181,
      "step": 1036
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4045088202238398,
      "learning_rate": 3.958928141045235e-05,
      "loss": 0.7395,
      "step": 1037
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49410982055320235,
      "learning_rate": 3.95881128085938e-05,
      "loss": 0.7608,
      "step": 1038
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4558241795615959,
      "learning_rate": 3.9586942563903884e-05,
      "loss": 0.749,
      "step": 1039
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43277928408376204,
      "learning_rate": 3.958577067648075e-05,
      "loss": 0.7436,
      "step": 1040
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4657961219416923,
      "learning_rate": 3.958459714642268e-05,
      "loss": 0.7178,
      "step": 1041
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.38705315305097543,
      "learning_rate": 3.95834219738281e-05,
      "loss": 0.7259,
      "step": 1042
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4530062770179624,
      "learning_rate": 3.9582245158795576e-05,
      "loss": 0.7498,
      "step": 1043
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44081170948231047,
      "learning_rate": 3.9581066701423796e-05,
      "loss": 0.7251,
      "step": 1044
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3912548505154261,
      "learning_rate": 3.95798866018116e-05,
      "loss": 0.7136,
      "step": 1045
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46914012007461087,
      "learning_rate": 3.957870486005796e-05,
      "loss": 0.7882,
      "step": 1046
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44206118506977166,
      "learning_rate": 3.957752147626199e-05,
      "loss": 0.7724,
      "step": 1047
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3852289566466024,
      "learning_rate": 3.957633645052294e-05,
      "loss": 0.7287,
      "step": 1048
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4509795662235156,
      "learning_rate": 3.9575149782940194e-05,
      "loss": 0.7645,
      "step": 1049
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4123550047171538,
      "learning_rate": 3.957396147361327e-05,
      "loss": 0.7181,
      "step": 1050
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44228949683700236,
      "learning_rate": 3.957277152264184e-05,
      "loss": 0.7459,
      "step": 1051
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44785410687250155,
      "learning_rate": 3.95715799301257e-05,
      "loss": 0.7893,
      "step": 1052
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.428849913469544,
      "learning_rate": 3.957038669616479e-05,
      "loss": 0.7925,
      "step": 1053
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.409872633804563,
      "learning_rate": 3.956919182085918e-05,
      "loss": 0.7439,
      "step": 1054
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45514322393072565,
      "learning_rate": 3.9567995304309086e-05,
      "loss": 0.7285,
      "step": 1055
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4079360084078462,
      "learning_rate": 3.956679714661486e-05,
      "loss": 0.723,
      "step": 1056
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44198254343341115,
      "learning_rate": 3.9565597347876985e-05,
      "loss": 0.7432,
      "step": 1057
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.40403502893849247,
      "learning_rate": 3.956439590819609e-05,
      "loss": 0.7608,
      "step": 1058
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4239482217750803,
      "learning_rate": 3.956319282767294e-05,
      "loss": 0.7741,
      "step": 1059
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4087507475023897,
      "learning_rate": 3.956198810640843e-05,
      "loss": 0.7597,
      "step": 1060
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42128962179526624,
      "learning_rate": 3.956078174450361e-05,
      "loss": 0.7347,
      "step": 1061
    },
    {
      "epoch": 0.09,
      "grad_norm": 76.49165224253706,
      "learning_rate": 3.955957374205964e-05,
      "loss": 6.2389,
      "step": 1062
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7833152685265331,
      "learning_rate": 3.955836409917785e-05,
      "loss": 0.7561,
      "step": 1063
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.9504991595621233,
      "learning_rate": 3.9557152815959676e-05,
      "loss": 0.7424,
      "step": 1064
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.607915932817685,
      "learning_rate": 3.955593989250672e-05,
      "loss": 0.7759,
      "step": 1065
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.759374044169199,
      "learning_rate": 3.95547253289207e-05,
      "loss": 0.7999,
      "step": 1066
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7864766593874519,
      "learning_rate": 3.9553509125303486e-05,
      "loss": 0.7493,
      "step": 1067
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9100370644244401,
      "learning_rate": 3.955229128175708e-05,
      "loss": 0.7492,
      "step": 1068
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9975745066356574,
      "learning_rate": 3.9551071798383605e-05,
      "loss": 0.7336,
      "step": 1069
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.2178611897917717,
      "learning_rate": 3.954985067528536e-05,
      "loss": 0.7869,
      "step": 1070
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6145038808378588,
      "learning_rate": 3.9548627912564746e-05,
      "loss": 0.7953,
      "step": 1071
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.726742240329883,
      "learning_rate": 3.9547403510324315e-05,
      "loss": 0.7376,
      "step": 1072
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0581689832403016,
      "learning_rate": 3.954617746866676e-05,
      "loss": 0.7446,
      "step": 1073
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5630975589605307,
      "learning_rate": 3.954494978769491e-05,
      "loss": 0.7907,
      "step": 1074
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6278157996351217,
      "learning_rate": 3.9543720467511724e-05,
      "loss": 0.7566,
      "step": 1075
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47730837390464553,
      "learning_rate": 3.95424895082203e-05,
      "loss": 0.7121,
      "step": 1076
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48329693827158704,
      "learning_rate": 3.954125690992389e-05,
      "loss": 0.6686,
      "step": 1077
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47087331153099443,
      "learning_rate": 3.9540022672725857e-05,
      "loss": 0.7415,
      "step": 1078
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47468070187477485,
      "learning_rate": 3.953878679672973e-05,
      "loss": 0.7615,
      "step": 1079
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5851109674744317,
      "learning_rate": 3.953754928203914e-05,
      "loss": 0.8017,
      "step": 1080
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45754347184425204,
      "learning_rate": 3.9536310128757895e-05,
      "loss": 0.7228,
      "step": 1081
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5936850686608534,
      "learning_rate": 3.9535069336989905e-05,
      "loss": 0.7888,
      "step": 1082
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.472563938267618,
      "learning_rate": 3.953382690683925e-05,
      "loss": 0.7427,
      "step": 1083
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46240036173280447,
      "learning_rate": 3.9532582838410124e-05,
      "loss": 0.72,
      "step": 1084
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41545002463909203,
      "learning_rate": 3.953133713180686e-05,
      "loss": 0.7328,
      "step": 1085
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5147141486986895,
      "learning_rate": 3.953008978713394e-05,
      "loss": 0.7063,
      "step": 1086
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40042294941883705,
      "learning_rate": 3.952884080449598e-05,
      "loss": 0.696,
      "step": 1087
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5111184828951182,
      "learning_rate": 3.952759018399772e-05,
      "loss": 0.7571,
      "step": 1088
    },
    {
      "epoch": 0.1,
      "grad_norm": 129.76266358383307,
      "learning_rate": 3.9526337925744056e-05,
      "loss": 7.5771,
      "step": 1089
    },
    {
      "epoch": 0.1,
      "grad_norm": 211.65853267331076,
      "learning_rate": 3.952508402984001e-05,
      "loss": 12.6679,
      "step": 1090
    },
    {
      "epoch": 0.1,
      "grad_norm": 55.6364380734918,
      "learning_rate": 3.952382849639075e-05,
      "loss": 2.5921,
      "step": 1091
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.247729134683355,
      "learning_rate": 3.9522571325501576e-05,
      "loss": 0.9055,
      "step": 1092
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.196757371986058,
      "learning_rate": 3.952131251727792e-05,
      "loss": 1.0092,
      "step": 1093
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.625927925988165,
      "learning_rate": 3.9520052071825355e-05,
      "loss": 0.9312,
      "step": 1094
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5264149891949264,
      "learning_rate": 3.9518789989249605e-05,
      "loss": 0.9992,
      "step": 1095
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.55072158170491,
      "learning_rate": 3.95175262696565e-05,
      "loss": 0.9237,
      "step": 1096
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2067661108960115,
      "learning_rate": 3.9516260913152045e-05,
      "loss": 0.9741,
      "step": 1097
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8729368370444308,
      "learning_rate": 3.951499391984236e-05,
      "loss": 0.9255,
      "step": 1098
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3801396159945536,
      "learning_rate": 3.95137252898337e-05,
      "loss": 0.9066,
      "step": 1099
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9564636236985447,
      "learning_rate": 3.951245502323246e-05,
      "loss": 0.8154,
      "step": 1100
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0517804805134328,
      "learning_rate": 3.951118312014519e-05,
      "loss": 0.8586,
      "step": 1101
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2742156082256195,
      "learning_rate": 3.9509909580678555e-05,
      "loss": 0.9038,
      "step": 1102
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9918485015374686,
      "learning_rate": 3.9508634404939363e-05,
      "loss": 0.8022,
      "step": 1103
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9505414509573836,
      "learning_rate": 3.950735759303456e-05,
      "loss": 0.8777,
      "step": 1104
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6717900032801587,
      "learning_rate": 3.9506079145071245e-05,
      "loss": 0.7982,
      "step": 1105
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9473579145051456,
      "learning_rate": 3.9504799061156625e-05,
      "loss": 0.8857,
      "step": 1106
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.736539911237458,
      "learning_rate": 3.950351734139806e-05,
      "loss": 0.8255,
      "step": 1107
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8427410506541544,
      "learning_rate": 3.950223398590306e-05,
      "loss": 0.8137,
      "step": 1108
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7896731555784607,
      "learning_rate": 3.950094899477923e-05,
      "loss": 0.819,
      "step": 1109
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6504350673471173,
      "learning_rate": 3.9499662368134376e-05,
      "loss": 0.8144,
      "step": 1110
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6781449453188586,
      "learning_rate": 3.949837410607638e-05,
      "loss": 0.8718,
      "step": 1111
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7934113251722616,
      "learning_rate": 3.94970842087133e-05,
      "loss": 0.8889,
      "step": 1112
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5964705916177785,
      "learning_rate": 3.9495792676153314e-05,
      "loss": 0.7864,
      "step": 1113
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6344619578870333,
      "learning_rate": 3.9494499508504734e-05,
      "loss": 0.8137,
      "step": 1114
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5790584354865754,
      "learning_rate": 3.949320470587603e-05,
      "loss": 0.8289,
      "step": 1115
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5209951303017388,
      "learning_rate": 3.949190826837579e-05,
      "loss": 0.8033,
      "step": 1116
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.597415657023988,
      "learning_rate": 3.949061019611274e-05,
      "loss": 0.7946,
      "step": 1117
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6239662290096958,
      "learning_rate": 3.9489310489195756e-05,
      "loss": 0.8378,
      "step": 1118
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5823508559532594,
      "learning_rate": 3.948800914773383e-05,
      "loss": 0.8471,
      "step": 1119
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5120526487683353,
      "learning_rate": 3.948670617183612e-05,
      "loss": 0.7956,
      "step": 1120
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5455173787758001,
      "learning_rate": 3.9485401561611896e-05,
      "loss": 0.8086,
      "step": 1121
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5069958473163028,
      "learning_rate": 3.948409531717057e-05,
      "loss": 0.748,
      "step": 1122
    },
    {
      "epoch": 0.1,
      "grad_norm": 62.037569715218716,
      "learning_rate": 3.9482787438621706e-05,
      "loss": 9.4115,
      "step": 1123
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6748045914494565,
      "learning_rate": 3.9481477926074994e-05,
      "loss": 0.7407,
      "step": 1124
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3723875841831579,
      "learning_rate": 3.948016677964025e-05,
      "loss": 0.8336,
      "step": 1125
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9371982927870585,
      "learning_rate": 3.9478853999427446e-05,
      "loss": 0.7748,
      "step": 1126
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5744114098249682,
      "learning_rate": 3.9477539585546676e-05,
      "loss": 0.8178,
      "step": 1127
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0511807446481085,
      "learning_rate": 3.947622353810819e-05,
      "loss": 0.7825,
      "step": 1128
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6328431826751678,
      "learning_rate": 3.947490585722236e-05,
      "loss": 0.7296,
      "step": 1129
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7688716460956517,
      "learning_rate": 3.947358654299969e-05,
      "loss": 0.7553,
      "step": 1130
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6926194121420979,
      "learning_rate": 3.947226559555083e-05,
      "loss": 0.7756,
      "step": 1131
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6828894292530936,
      "learning_rate": 3.9470943014986574e-05,
      "loss": 0.8095,
      "step": 1132
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.655093353669352,
      "learning_rate": 3.9469618801417846e-05,
      "loss": 0.7643,
      "step": 1133
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7152128029131267,
      "learning_rate": 3.94682929549557e-05,
      "loss": 0.8281,
      "step": 1134
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6974444259319373,
      "learning_rate": 3.946696547571134e-05,
      "loss": 0.8016,
      "step": 1135
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5744840475446685,
      "learning_rate": 3.946563636379609e-05,
      "loss": 0.7686,
      "step": 1136
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5982572679567171,
      "learning_rate": 3.9464305619321436e-05,
      "loss": 0.8007,
      "step": 1137
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7450134348957443,
      "learning_rate": 3.946297324239896e-05,
      "loss": 0.8013,
      "step": 1138
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5607228961568611,
      "learning_rate": 3.9461639233140433e-05,
      "loss": 0.7829,
      "step": 1139
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4927100950314772,
      "learning_rate": 3.946030359165773e-05,
      "loss": 0.7217,
      "step": 1140
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7023974525549211,
      "learning_rate": 3.945896631806287e-05,
      "loss": 0.778,
      "step": 1141
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5764885719028876,
      "learning_rate": 3.945762741246799e-05,
      "loss": 0.7836,
      "step": 1142
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5201274219824239,
      "learning_rate": 3.945628687498541e-05,
      "loss": 0.7773,
      "step": 1143
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5033130217329636,
      "learning_rate": 3.945494470572755e-05,
      "loss": 0.7765,
      "step": 1144
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4635460771040858,
      "learning_rate": 3.945360090480697e-05,
      "loss": 0.7116,
      "step": 1145
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5313333836033228,
      "learning_rate": 3.9452255472336375e-05,
      "loss": 0.7275,
      "step": 1146
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8445307533313972,
      "learning_rate": 3.94509084084286e-05,
      "loss": 0.7823,
      "step": 1147
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.481360090777458,
      "learning_rate": 3.9449559713196636e-05,
      "loss": 0.7773,
      "step": 1148
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46191073096390955,
      "learning_rate": 3.944820938675358e-05,
      "loss": 0.7461,
      "step": 1149
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4233877158684677,
      "learning_rate": 3.9446857429212695e-05,
      "loss": 0.7504,
      "step": 1150
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5258266552213272,
      "learning_rate": 3.9445503840687365e-05,
      "loss": 0.7419,
      "step": 1151
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6204548998441423,
      "learning_rate": 3.9444148621291107e-05,
      "loss": 0.8041,
      "step": 1152
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45246914423128615,
      "learning_rate": 3.944279177113759e-05,
      "loss": 0.7641,
      "step": 1153
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4801496435964734,
      "learning_rate": 3.944143329034061e-05,
      "loss": 0.7696,
      "step": 1154
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4532160478981047,
      "learning_rate": 3.944007317901409e-05,
      "loss": 0.7459,
      "step": 1155
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4062222325010684,
      "learning_rate": 3.943871143727212e-05,
      "loss": 0.7005,
      "step": 1156
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45273136161152877,
      "learning_rate": 3.9437348065228897e-05,
      "loss": 0.7916,
      "step": 1157
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46919938718232823,
      "learning_rate": 3.943598306299876e-05,
      "loss": 0.7382,
      "step": 1158
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4997337884326267,
      "learning_rate": 3.94346164306962e-05,
      "loss": 0.7733,
      "step": 1159
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40104206807122084,
      "learning_rate": 3.943324816843584e-05,
      "loss": 0.7058,
      "step": 1160
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4342486795457491,
      "learning_rate": 3.9431878276332414e-05,
      "loss": 0.7551,
      "step": 1161
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.487964125596612,
      "learning_rate": 3.943050675450082e-05,
      "loss": 0.7273,
      "step": 1162
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5094323831528451,
      "learning_rate": 3.942913360305611e-05,
      "loss": 0.7402,
      "step": 1163
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44575953926232015,
      "learning_rate": 3.942775882211341e-05,
      "loss": 0.7243,
      "step": 1164
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4264053541484483,
      "learning_rate": 3.942638241178805e-05,
      "loss": 0.7756,
      "step": 1165
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44825669588571965,
      "learning_rate": 3.942500437219545e-05,
      "loss": 0.7592,
      "step": 1166
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4762661517988632,
      "learning_rate": 3.9423624703451205e-05,
      "loss": 0.7239,
      "step": 1167
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41468318389280484,
      "learning_rate": 3.942224340567101e-05,
      "loss": 0.7342,
      "step": 1168
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4746437614949873,
      "learning_rate": 3.9420860478970716e-05,
      "loss": 0.7644,
      "step": 1169
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47936775798505876,
      "learning_rate": 3.94194759234663e-05,
      "loss": 0.767,
      "step": 1170
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5009254492930449,
      "learning_rate": 3.9418089739273895e-05,
      "loss": 0.7294,
      "step": 1171
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44239639322996865,
      "learning_rate": 3.941670192650976e-05,
      "loss": 0.7781,
      "step": 1172
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43284723736920716,
      "learning_rate": 3.941531248529028e-05,
      "loss": 0.7713,
      "step": 1173
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5366474366400434,
      "learning_rate": 3.941392141573199e-05,
      "loss": 0.7198,
      "step": 1174
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5090449582626522,
      "learning_rate": 3.9412528717951556e-05,
      "loss": 0.7681,
      "step": 1175
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4141136514599923,
      "learning_rate": 3.941113439206579e-05,
      "loss": 0.7033,
      "step": 1176
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45280863501506613,
      "learning_rate": 3.940973843819162e-05,
      "loss": 0.72,
      "step": 1177
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46488600135851843,
      "learning_rate": 3.940834085644612e-05,
      "loss": 0.7932,
      "step": 1178
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4225395801013948,
      "learning_rate": 3.940694164694653e-05,
      "loss": 0.7271,
      "step": 1179
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47212782893690136,
      "learning_rate": 3.940554080981016e-05,
      "loss": 0.7419,
      "step": 1180
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4418074611157787,
      "learning_rate": 3.940413834515454e-05,
      "loss": 0.7471,
      "step": 1181
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45192874132134603,
      "learning_rate": 3.940273425309726e-05,
      "loss": 0.7396,
      "step": 1182
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48500340218509924,
      "learning_rate": 3.940132853375609e-05,
      "loss": 0.7609,
      "step": 1183
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4193699571976979,
      "learning_rate": 3.9399921187248925e-05,
      "loss": 0.7302,
      "step": 1184
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46463043050743047,
      "learning_rate": 3.939851221369381e-05,
      "loss": 0.7387,
      "step": 1185
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43242275019783694,
      "learning_rate": 3.939710161320889e-05,
      "loss": 0.7599,
      "step": 1186
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43952964814493845,
      "learning_rate": 3.939568938591249e-05,
      "loss": 0.7336,
      "step": 1187
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5043670844168008,
      "learning_rate": 3.939427553192304e-05,
      "loss": 0.7736,
      "step": 1188
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49692736961606293,
      "learning_rate": 3.9392860051359135e-05,
      "loss": 0.7789,
      "step": 1189
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3974828189278222,
      "learning_rate": 3.939144294433947e-05,
      "loss": 0.7162,
      "step": 1190
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4323810855110649,
      "learning_rate": 3.93900242109829e-05,
      "loss": 0.7174,
      "step": 1191
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4275219594519422,
      "learning_rate": 3.938860385140842e-05,
      "loss": 0.7564,
      "step": 1192
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49548792945404674,
      "learning_rate": 3.938718186573516e-05,
      "loss": 0.7377,
      "step": 1193
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44511018298133587,
      "learning_rate": 3.938575825408236e-05,
      "loss": 0.6903,
      "step": 1194
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45238191203947714,
      "learning_rate": 3.9384333016569433e-05,
      "loss": 0.7894,
      "step": 1195
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49420943815313917,
      "learning_rate": 3.93829061533159e-05,
      "loss": 0.8029,
      "step": 1196
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48592492921255676,
      "learning_rate": 3.9381477664441436e-05,
      "loss": 0.7946,
      "step": 1197
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4677728779274136,
      "learning_rate": 3.938004755006586e-05,
      "loss": 0.7367,
      "step": 1198
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4489273921980025,
      "learning_rate": 3.9378615810309085e-05,
      "loss": 0.7406,
      "step": 1199
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45440134093921464,
      "learning_rate": 3.9377182445291214e-05,
      "loss": 0.7542,
      "step": 1200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45826963483054023,
      "learning_rate": 3.937574745513245e-05,
      "loss": 0.7458,
      "step": 1201
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6098208134908127,
      "learning_rate": 3.937431083995315e-05,
      "loss": 0.7887,
      "step": 1202
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5085078461611056,
      "learning_rate": 3.93728725998738e-05,
      "loss": 0.7224,
      "step": 1203
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48548933335637035,
      "learning_rate": 3.9371432735015017e-05,
      "loss": 0.7406,
      "step": 1204
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40690008662998595,
      "learning_rate": 3.9369991245497564e-05,
      "loss": 0.7354,
      "step": 1205
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5373443621048082,
      "learning_rate": 3.936854813144234e-05,
      "loss": 0.7764,
      "step": 1206
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4222385714285651,
      "learning_rate": 3.9367103392970385e-05,
      "loss": 0.6909,
      "step": 1207
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48862882708045086,
      "learning_rate": 3.936565703020285e-05,
      "loss": 0.8038,
      "step": 1208
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4171706578766399,
      "learning_rate": 3.936420904326104e-05,
      "loss": 0.712,
      "step": 1209
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4771140882791814,
      "learning_rate": 3.9362759432266415e-05,
      "loss": 0.7272,
      "step": 1210
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.418361847069613,
      "learning_rate": 3.936130819734054e-05,
      "loss": 0.7076,
      "step": 1211
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48149582408705427,
      "learning_rate": 3.935985533860512e-05,
      "loss": 0.8091,
      "step": 1212
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42105624491867394,
      "learning_rate": 3.935840085618202e-05,
      "loss": 0.7616,
      "step": 1213
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.446673187004786,
      "learning_rate": 3.935694475019322e-05,
      "loss": 0.7591,
      "step": 1214
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4338921860988,
      "learning_rate": 3.9355487020760836e-05,
      "loss": 0.7496,
      "step": 1215
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40502122811029706,
      "learning_rate": 3.935402766800714e-05,
      "loss": 0.7132,
      "step": 1216
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4348847448151212,
      "learning_rate": 3.935256669205451e-05,
      "loss": 0.7266,
      "step": 1217
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4072282054125831,
      "learning_rate": 3.935110409302548e-05,
      "loss": 0.7028,
      "step": 1218
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47099997624169937,
      "learning_rate": 3.934963987104273e-05,
      "loss": 0.76,
      "step": 1219
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48816182759055865,
      "learning_rate": 3.9348174026229045e-05,
      "loss": 0.7612,
      "step": 1220
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4073813667499251,
      "learning_rate": 3.9346706558707374e-05,
      "loss": 0.7533,
      "step": 1221
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3855439593025972,
      "learning_rate": 3.934523746860078e-05,
      "loss": 0.7081,
      "step": 1222
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.474947360371556,
      "learning_rate": 3.93437667560325e-05,
      "loss": 0.742,
      "step": 1223
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43089198439225224,
      "learning_rate": 3.934229442112585e-05,
      "loss": 0.7085,
      "step": 1224
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4359173955992016,
      "learning_rate": 3.934082046400433e-05,
      "loss": 0.7745,
      "step": 1225
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40886781015932633,
      "learning_rate": 3.933934488479156e-05,
      "loss": 0.7096,
      "step": 1226
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44392932304267496,
      "learning_rate": 3.933786768361129e-05,
      "loss": 0.7053,
      "step": 1227
    },
    {
      "epoch": 0.11,
      "grad_norm": 79.6363336827904,
      "learning_rate": 3.9336388860587404e-05,
      "loss": 5.7073,
      "step": 1228
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.826600643711508,
      "learning_rate": 3.9334908415843944e-05,
      "loss": 1.1126,
      "step": 1229
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.570860335720692,
      "learning_rate": 3.933342634950506e-05,
      "loss": 0.7127,
      "step": 1230
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.846749035853758,
      "learning_rate": 3.933194266169507e-05,
      "loss": 0.7999,
      "step": 1231
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8129389418036703,
      "learning_rate": 3.933045735253838e-05,
      "loss": 0.7933,
      "step": 1232
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.629872679098973,
      "learning_rate": 3.932897042215959e-05,
      "loss": 0.8812,
      "step": 1233
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.873687957485471,
      "learning_rate": 3.932748187068338e-05,
      "loss": 0.7979,
      "step": 1234
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.696894583599638,
      "learning_rate": 3.932599169823461e-05,
      "loss": 0.7828,
      "step": 1235
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7871045675540579,
      "learning_rate": 3.932449990493827e-05,
      "loss": 0.7439,
      "step": 1236
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8016184956487349,
      "learning_rate": 3.9323006490919445e-05,
      "loss": 0.7627,
      "step": 1237
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9178813607943582,
      "learning_rate": 3.9321511456303414e-05,
      "loss": 0.8455,
      "step": 1238
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7861165636757224,
      "learning_rate": 3.932001480121555e-05,
      "loss": 0.8118,
      "step": 1239
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5783000646194437,
      "learning_rate": 3.931851652578137e-05,
      "loss": 0.775,
      "step": 1240
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.571664982421585,
      "learning_rate": 3.9317016630126546e-05,
      "loss": 0.7099,
      "step": 1241
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7160353601892996,
      "learning_rate": 3.931551511437686e-05,
      "loss": 0.8061,
      "step": 1242
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5562316469806352,
      "learning_rate": 3.9314011978658256e-05,
      "loss": 0.787,
      "step": 1243
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5120938568581694,
      "learning_rate": 3.931250722309679e-05,
      "loss": 0.785,
      "step": 1244
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5858135344505428,
      "learning_rate": 3.931100084781866e-05,
      "loss": 0.7411,
      "step": 1245
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5019641239078362,
      "learning_rate": 3.9309492852950216e-05,
      "loss": 0.7448,
      "step": 1246
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47577583295217246,
      "learning_rate": 3.930798323861793e-05,
      "loss": 0.7379,
      "step": 1247
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46520828555953336,
      "learning_rate": 3.9306472004948404e-05,
      "loss": 0.7246,
      "step": 1248
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4811341393677244,
      "learning_rate": 3.930495915206839e-05,
      "loss": 0.75,
      "step": 1249
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4848457563625481,
      "learning_rate": 3.9303444680104764e-05,
      "loss": 0.7366,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5031587361421899,
      "learning_rate": 3.9301928589184545e-05,
      "loss": 0.7274,
      "step": 1251
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5172761001315915,
      "learning_rate": 3.9300410879434886e-05,
      "loss": 0.7765,
      "step": 1252
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5580352216231027,
      "learning_rate": 3.929889155098308e-05,
      "loss": 0.7573,
      "step": 1253
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5576722485005807,
      "learning_rate": 3.929737060395655e-05,
      "loss": 0.806,
      "step": 1254
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5737271843630097,
      "learning_rate": 3.929584803848284e-05,
      "loss": 0.81,
      "step": 1255
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4797944993032185,
      "learning_rate": 3.929432385468966e-05,
      "loss": 0.7241,
      "step": 1256
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4001112147084526,
      "learning_rate": 3.929279805270485e-05,
      "loss": 0.7054,
      "step": 1257
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45854306921902144,
      "learning_rate": 3.929127063265636e-05,
      "loss": 0.772,
      "step": 1258
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4538793937301648,
      "learning_rate": 3.9289741594672304e-05,
      "loss": 0.7516,
      "step": 1259
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4685132934967583,
      "learning_rate": 3.928821093888091e-05,
      "loss": 0.7548,
      "step": 1260
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47437675111341066,
      "learning_rate": 3.928667866541056e-05,
      "loss": 0.7608,
      "step": 1261
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45854993188106913,
      "learning_rate": 3.928514477438977e-05,
      "loss": 0.7174,
      "step": 1262
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.489834384495078,
      "learning_rate": 3.928360926594717e-05,
      "loss": 0.7657,
      "step": 1263
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42224678802247423,
      "learning_rate": 3.928207214021156e-05,
      "loss": 0.7137,
      "step": 1264
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4123342825269307,
      "learning_rate": 3.928053339731184e-05,
      "loss": 0.6981,
      "step": 1265
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48676782882125413,
      "learning_rate": 3.9278993037377066e-05,
      "loss": 0.7591,
      "step": 1266
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.39858404460154073,
      "learning_rate": 3.927745106053643e-05,
      "loss": 0.6999,
      "step": 1267
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45755597584082647,
      "learning_rate": 3.9275907466919255e-05,
      "loss": 0.7469,
      "step": 1268
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3948656312533755,
      "learning_rate": 3.927436225665501e-05,
      "loss": 0.6526,
      "step": 1269
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4159509540264599,
      "learning_rate": 3.9272815429873275e-05,
      "loss": 0.7415,
      "step": 1270
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4868374718308464,
      "learning_rate": 3.927126698670378e-05,
      "loss": 0.7672,
      "step": 1271
    },
    {
      "epoch": 0.11,
      "grad_norm": 121.99364314866082,
      "learning_rate": 3.92697169272764e-05,
      "loss": 7.6662,
      "step": 1272
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6835126376191463,
      "learning_rate": 3.926816525172113e-05,
      "loss": 0.7406,
      "step": 1273
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8999280487580353,
      "learning_rate": 3.9266611960168114e-05,
      "loss": 0.7036,
      "step": 1274
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8144135522684053,
      "learning_rate": 3.926505705274762e-05,
      "loss": 0.7524,
      "step": 1275
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.906396578334179,
      "learning_rate": 3.926350052959006e-05,
      "loss": 0.856,
      "step": 1276
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.891123019372691,
      "learning_rate": 3.926194239082597e-05,
      "loss": 0.8271,
      "step": 1277
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.213288310971589,
      "learning_rate": 3.926038263658603e-05,
      "loss": 0.796,
      "step": 1278
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2328102472207125,
      "learning_rate": 3.925882126700107e-05,
      "loss": 0.7926,
      "step": 1279
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1602980777378231,
      "learning_rate": 3.925725828220202e-05,
      "loss": 0.7473,
      "step": 1280
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.860151548051606,
      "learning_rate": 3.925569368231998e-05,
      "loss": 0.7638,
      "step": 1281
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6930227167063545,
      "learning_rate": 3.9254127467486165e-05,
      "loss": 0.739,
      "step": 1282
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9870157508663251,
      "learning_rate": 3.925255963783194e-05,
      "loss": 0.7365,
      "step": 1283
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8266948313160167,
      "learning_rate": 3.925099019348878e-05,
      "loss": 0.7174,
      "step": 1284
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.845262375348477,
      "learning_rate": 3.924941913458832e-05,
      "loss": 0.8101,
      "step": 1285
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9254979281853579,
      "learning_rate": 3.924784646126233e-05,
      "loss": 0.7935,
      "step": 1286
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6096667459464864,
      "learning_rate": 3.9246272173642704e-05,
      "loss": 0.7702,
      "step": 1287
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6461520790392359,
      "learning_rate": 3.924469627186147e-05,
      "loss": 0.7801,
      "step": 1288
    },
    {
      "epoch": 0.12,
      "grad_norm": 96.41785324967816,
      "learning_rate": 3.924311875605081e-05,
      "loss": 11.8963,
      "step": 1289
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.882129787406697,
      "learning_rate": 3.9241539626343006e-05,
      "loss": 0.7904,
      "step": 1290
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.770365847919543,
      "learning_rate": 3.923995888287052e-05,
      "loss": 1.0646,
      "step": 1291
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0394128752421934,
      "learning_rate": 3.9238376525765915e-05,
      "loss": 0.8035,
      "step": 1292
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.524804361052651,
      "learning_rate": 3.923679255516191e-05,
      "loss": 0.8699,
      "step": 1293
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6621766218955916,
      "learning_rate": 3.9235206971191345e-05,
      "loss": 0.8007,
      "step": 1294
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.114302582202974,
      "learning_rate": 3.923361977398719e-05,
      "loss": 0.8552,
      "step": 1295
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1564639218418427,
      "learning_rate": 3.9232030963682586e-05,
      "loss": 0.7659,
      "step": 1296
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2610798433128285,
      "learning_rate": 3.923044054041077e-05,
      "loss": 0.7465,
      "step": 1297
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8086036704285449,
      "learning_rate": 3.922884850430512e-05,
      "loss": 0.8052,
      "step": 1298
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9720212457572073,
      "learning_rate": 3.9227254855499174e-05,
      "loss": 0.7867,
      "step": 1299
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0093610615209196,
      "learning_rate": 3.922565959412659e-05,
      "loss": 0.7915,
      "step": 1300
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.887939835633367,
      "learning_rate": 3.922406272032115e-05,
      "loss": 0.7765,
      "step": 1301
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7855127464014996,
      "learning_rate": 3.922246423421679e-05,
      "loss": 0.8114,
      "step": 1302
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6067704074337004,
      "learning_rate": 3.9220864135947566e-05,
      "loss": 0.7187,
      "step": 1303
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.027783624156273,
      "learning_rate": 3.921926242564768e-05,
      "loss": 0.7749,
      "step": 1304
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7681139736406308,
      "learning_rate": 3.9217659103451464e-05,
      "loss": 0.7406,
      "step": 1305
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.782513524574263,
      "learning_rate": 3.9216054169493386e-05,
      "loss": 0.7279,
      "step": 1306
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5632173130424378,
      "learning_rate": 3.921444762390806e-05,
      "loss": 0.7752,
      "step": 1307
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3956766042709348,
      "learning_rate": 3.921283946683021e-05,
      "loss": 0.7436,
      "step": 1308
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2508450785420768,
      "learning_rate": 3.921122969839472e-05,
      "loss": 0.7697,
      "step": 1309
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0353455469612305,
      "learning_rate": 3.9209618318736595e-05,
      "loss": 0.7169,
      "step": 1310
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6708148464127541,
      "learning_rate": 3.9208005327990986e-05,
      "loss": 0.7404,
      "step": 1311
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5595573608507971,
      "learning_rate": 3.920639072629316e-05,
      "loss": 0.7133,
      "step": 1312
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5377903656160318,
      "learning_rate": 3.9204774513778544e-05,
      "loss": 0.7326,
      "step": 1313
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7015751643299349,
      "learning_rate": 3.920315669058268e-05,
      "loss": 0.8096,
      "step": 1314
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4728832510774299,
      "learning_rate": 3.920153725684126e-05,
      "loss": 0.7039,
      "step": 1315
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3603759561826001,
      "learning_rate": 3.9199916212690096e-05,
      "loss": 0.7154,
      "step": 1316
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4897529089214013,
      "learning_rate": 3.919829355826515e-05,
      "loss": 0.7601,
      "step": 1317
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.78977307713275,
      "learning_rate": 3.919666929370251e-05,
      "loss": 7.7631,
      "step": 1318
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3808590377547284,
      "learning_rate": 3.9195043419138395e-05,
      "loss": 0.7238,
      "step": 1319
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5036240115030574,
      "learning_rate": 3.9193415934709176e-05,
      "loss": 0.7555,
      "step": 1320
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4607973729902788,
      "learning_rate": 3.919178684055133e-05,
      "loss": 0.7563,
      "step": 1321
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5781722869162771,
      "learning_rate": 3.9190156136801516e-05,
      "loss": 0.7654,
      "step": 1322
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4531996410407777,
      "learning_rate": 3.9188523823596476e-05,
      "loss": 0.7037,
      "step": 1323
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5179581511576965,
      "learning_rate": 3.918688990107311e-05,
      "loss": 0.7552,
      "step": 1324
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5669496395667901,
      "learning_rate": 3.918525436936847e-05,
      "loss": 0.8129,
      "step": 1325
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5680577681718927,
      "learning_rate": 3.918361722861971e-05,
      "loss": 0.729,
      "step": 1326
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4376666272008125,
      "learning_rate": 3.918197847896414e-05,
      "loss": 0.6659,
      "step": 1327
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5492243840072188,
      "learning_rate": 3.9180338120539204e-05,
      "loss": 0.7606,
      "step": 1328
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5020577769522013,
      "learning_rate": 3.917869615348248e-05,
      "loss": 0.7369,
      "step": 1329
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5630906045271907,
      "learning_rate": 3.917705257793166e-05,
      "loss": 0.7781,
      "step": 1330
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40366794723565974,
      "learning_rate": 3.9175407394024606e-05,
      "loss": 0.6936,
      "step": 1331
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45448036714628204,
      "learning_rate": 3.9173760601899286e-05,
      "loss": 0.7342,
      "step": 1332
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43557630288002613,
      "learning_rate": 3.917211220169383e-05,
      "loss": 0.7695,
      "step": 1333
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.519140720643017,
      "learning_rate": 3.9170462193546476e-05,
      "loss": 0.7452,
      "step": 1334
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4212125658812495,
      "learning_rate": 3.9168810577595606e-05,
      "loss": 0.7291,
      "step": 1335
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5284658160175685,
      "learning_rate": 3.916715735397975e-05,
      "loss": 0.8,
      "step": 1336
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42023549587524894,
      "learning_rate": 3.916550252283754e-05,
      "loss": 0.6874,
      "step": 1337
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45141286872656866,
      "learning_rate": 3.916384608430779e-05,
      "loss": 0.74,
      "step": 1338
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4329504569369694,
      "learning_rate": 3.916218803852941e-05,
      "loss": 0.7489,
      "step": 1339
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4515625224638202,
      "learning_rate": 3.9160528385641465e-05,
      "loss": 0.7635,
      "step": 1340
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.472199865020811,
      "learning_rate": 3.9158867125783144e-05,
      "loss": 0.7134,
      "step": 1341
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4364554679178414,
      "learning_rate": 3.915720425909377e-05,
      "loss": 0.7317,
      "step": 1342
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46535828917378674,
      "learning_rate": 3.915553978571281e-05,
      "loss": 0.7711,
      "step": 1343
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4647443598498267,
      "learning_rate": 3.9153873705779874e-05,
      "loss": 0.729,
      "step": 1344
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4965721445693281,
      "learning_rate": 3.9152206019434674e-05,
      "loss": 0.7203,
      "step": 1345
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4666203786076408,
      "learning_rate": 3.915053672681708e-05,
      "loss": 0.794,
      "step": 1346
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44703709352615134,
      "learning_rate": 3.914886582806711e-05,
      "loss": 0.7582,
      "step": 1347
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.48071935558418144,
      "learning_rate": 3.914719332332488e-05,
      "loss": 0.7812,
      "step": 1348
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46013427468116547,
      "learning_rate": 3.9145519212730677e-05,
      "loss": 0.7623,
      "step": 1349
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42410489660508915,
      "learning_rate": 3.9143843496424896e-05,
      "loss": 0.6878,
      "step": 1350
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4684446709888772,
      "learning_rate": 3.914216617454808e-05,
      "loss": 0.7765,
      "step": 1351
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42492147509714284,
      "learning_rate": 3.91404872472409e-05,
      "loss": 0.7459,
      "step": 1352
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45758186330928535,
      "learning_rate": 3.913880671464418e-05,
      "loss": 0.7411,
      "step": 1353
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4326247073934031,
      "learning_rate": 3.913712457689885e-05,
      "loss": 0.7557,
      "step": 1354
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39981382734189863,
      "learning_rate": 3.913544083414599e-05,
      "loss": 0.7159,
      "step": 1355
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5063730630680445,
      "learning_rate": 3.913375548652683e-05,
      "loss": 0.7463,
      "step": 1356
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4227137620543684,
      "learning_rate": 3.9132068534182694e-05,
      "loss": 0.7215,
      "step": 1357
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46339831202005455,
      "learning_rate": 3.913037997725508e-05,
      "loss": 0.7465,
      "step": 1358
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42485082463560164,
      "learning_rate": 3.9128689815885594e-05,
      "loss": 0.7386,
      "step": 1359
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43460351429159394,
      "learning_rate": 3.9126998050216e-05,
      "loss": 0.772,
      "step": 1360
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5502145565962561,
      "learning_rate": 3.912530468038818e-05,
      "loss": 0.8076,
      "step": 1361
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4152016847125865,
      "learning_rate": 3.9123609706544155e-05,
      "loss": 0.7378,
      "step": 1362
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44785430040920443,
      "learning_rate": 3.912191312882607e-05,
      "loss": 0.7267,
      "step": 1363
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4283841583967753,
      "learning_rate": 3.912021494737624e-05,
      "loss": 0.7341,
      "step": 1364
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47301843624546985,
      "learning_rate": 3.9118515162337065e-05,
      "loss": 0.774,
      "step": 1365
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38585631921025465,
      "learning_rate": 3.911681377385111e-05,
      "loss": 0.6973,
      "step": 1366
    },
    {
      "epoch": 0.12,
      "grad_norm": 132.50971379599775,
      "learning_rate": 3.911511078206108e-05,
      "loss": 8.1963,
      "step": 1367
    },
    {
      "epoch": 0.12,
      "grad_norm": 34.094892944132646,
      "learning_rate": 3.911340618710978e-05,
      "loss": 6.6165,
      "step": 1368
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2732785252148702,
      "learning_rate": 3.91116999891402e-05,
      "loss": 0.7674,
      "step": 1369
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.026153388696536,
      "learning_rate": 3.910999218829541e-05,
      "loss": 0.7464,
      "step": 1370
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.872891906811798,
      "learning_rate": 3.910828278471866e-05,
      "loss": 0.822,
      "step": 1371
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3892180941629595,
      "learning_rate": 3.910657177855331e-05,
      "loss": 0.7177,
      "step": 1372
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6656568147626662,
      "learning_rate": 3.910485916994286e-05,
      "loss": 0.7383,
      "step": 1373
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.135941102602759,
      "learning_rate": 3.910314495903095e-05,
      "loss": 7.704,
      "step": 1374
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7046315089160627,
      "learning_rate": 3.910142914596134e-05,
      "loss": 0.7553,
      "step": 1375
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3939950258172182,
      "learning_rate": 3.9099711730877935e-05,
      "loss": 0.7984,
      "step": 1376
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3113793876340154,
      "learning_rate": 3.909799271392477e-05,
      "loss": 0.7724,
      "step": 1377
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0626979909763987,
      "learning_rate": 3.9096272095246026e-05,
      "loss": 0.784,
      "step": 1378
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7974509461761123,
      "learning_rate": 3.9094549874986e-05,
      "loss": 0.7731,
      "step": 1379
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6719383884163053,
      "learning_rate": 3.909282605328914e-05,
      "loss": 0.7825,
      "step": 1380
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5273673362925191,
      "learning_rate": 3.9091100630300014e-05,
      "loss": 0.7214,
      "step": 1381
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4300291823769751,
      "learning_rate": 3.908937360616333e-05,
      "loss": 0.7092,
      "step": 1382
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42221918958150495,
      "learning_rate": 3.9087644981023946e-05,
      "loss": 0.6974,
      "step": 1383
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6596927582160934,
      "learning_rate": 3.908591475502682e-05,
      "loss": 0.7719,
      "step": 1384
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6090144580591216,
      "learning_rate": 3.908418292831709e-05,
      "loss": 0.7035,
      "step": 1385
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8655014316298274,
      "learning_rate": 3.908244950103996e-05,
      "loss": 0.7853,
      "step": 1386
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6486525932330933,
      "learning_rate": 3.9080714473340854e-05,
      "loss": 0.7614,
      "step": 1387
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6679828011456029,
      "learning_rate": 3.907897784536527e-05,
      "loss": 0.7791,
      "step": 1388
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5319308855913211,
      "learning_rate": 3.907723961725885e-05,
      "loss": 0.7337,
      "step": 1389
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6176221554336262,
      "learning_rate": 3.907549978916738e-05,
      "loss": 0.7778,
      "step": 1390
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47437574382645364,
      "learning_rate": 3.907375836123679e-05,
      "loss": 0.7546,
      "step": 1391
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44667989510128114,
      "learning_rate": 3.907201533361311e-05,
      "loss": 0.6951,
      "step": 1392
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5488338282847869,
      "learning_rate": 3.907027070644254e-05,
      "loss": 0.7543,
      "step": 1393
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6076240474498422,
      "learning_rate": 3.90685244798714e-05,
      "loss": 0.7746,
      "step": 1394
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4620401577444093,
      "learning_rate": 3.9066776654046144e-05,
      "loss": 0.7355,
      "step": 1395
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46769867990456027,
      "learning_rate": 3.9065027229113346e-05,
      "loss": 0.7368,
      "step": 1396
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49275662693219974,
      "learning_rate": 3.906327620521975e-05,
      "loss": 0.7329,
      "step": 1397
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4225702647791585,
      "learning_rate": 3.9061523582512196e-05,
      "loss": 0.7216,
      "step": 1398
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5798539786287488,
      "learning_rate": 3.905976936113768e-05,
      "loss": 0.7958,
      "step": 1399
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46524343935904183,
      "learning_rate": 3.9058013541243326e-05,
      "loss": 0.7333,
      "step": 1400
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4702492783751641,
      "learning_rate": 3.905625612297639e-05,
      "loss": 0.7404,
      "step": 1401
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45019372157260396,
      "learning_rate": 3.9054497106484275e-05,
      "loss": 0.7573,
      "step": 1402
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5633378546627253,
      "learning_rate": 3.90527364919145e-05,
      "loss": 0.7683,
      "step": 1403
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42654786335901484,
      "learning_rate": 3.9050974279414714e-05,
      "loss": 0.743,
      "step": 1404
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7683677144337824,
      "learning_rate": 3.904921046913273e-05,
      "loss": 0.7761,
      "step": 1405
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43931702393101973,
      "learning_rate": 3.9047445061216465e-05,
      "loss": 0.727,
      "step": 1406
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5262655812315867,
      "learning_rate": 3.9045678055813994e-05,
      "loss": 0.7513,
      "step": 1407
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49424587138514553,
      "learning_rate": 3.90439094530735e-05,
      "loss": 0.7659,
      "step": 1408
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4515027976882591,
      "learning_rate": 3.904213925314332e-05,
      "loss": 0.744,
      "step": 1409
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46833619826009104,
      "learning_rate": 3.904036745617192e-05,
      "loss": 0.724,
      "step": 1410
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4722727099162052,
      "learning_rate": 3.903859406230789e-05,
      "loss": 0.7496,
      "step": 1411
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41253927118429534,
      "learning_rate": 3.903681907169998e-05,
      "loss": 0.7205,
      "step": 1412
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4024309332282148,
      "learning_rate": 3.9035042484497035e-05,
      "loss": 0.7473,
      "step": 1413
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.47466394806248746,
      "learning_rate": 3.903326430084807e-05,
      "loss": 0.748,
      "step": 1414
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44441713581321063,
      "learning_rate": 3.903148452090221e-05,
      "loss": 0.7614,
      "step": 1415
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3828995401066384,
      "learning_rate": 3.902970314480874e-05,
      "loss": 0.6839,
      "step": 1416
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.09613583955194,
      "learning_rate": 3.902792017271704e-05,
      "loss": 7.8068,
      "step": 1417
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5592438637031766,
      "learning_rate": 3.902613560477665e-05,
      "loss": 0.7352,
      "step": 1418
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6870735848421596,
      "learning_rate": 3.902434944113726e-05,
      "loss": 0.7376,
      "step": 1419
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6539853185939952,
      "learning_rate": 3.902256168194864e-05,
      "loss": 0.7215,
      "step": 1420
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5244073054458516,
      "learning_rate": 3.902077232736075e-05,
      "loss": 0.6875,
      "step": 1421
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8807214801412176,
      "learning_rate": 3.9018981377523664e-05,
      "loss": 0.7501,
      "step": 1422
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8730257670537396,
      "learning_rate": 3.9017188832587576e-05,
      "loss": 0.724,
      "step": 1423
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5671517878817698,
      "learning_rate": 3.901539469270283e-05,
      "loss": 0.6866,
      "step": 1424
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6750719482492943,
      "learning_rate": 3.901359895801989e-05,
      "loss": 0.7171,
      "step": 1425
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.47284090317806576,
      "learning_rate": 3.901180162868937e-05,
      "loss": 0.7103,
      "step": 1426
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5157109717329627,
      "learning_rate": 3.9010002704862004e-05,
      "loss": 0.7479,
      "step": 1427
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5844168060583506,
      "learning_rate": 3.900820218668868e-05,
      "loss": 0.7762,
      "step": 1428
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43377441593639915,
      "learning_rate": 3.9006400074320385e-05,
      "loss": 0.7337,
      "step": 1429
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48342938635368515,
      "learning_rate": 3.900459636790827e-05,
      "loss": 0.7805,
      "step": 1430
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46667037093411046,
      "learning_rate": 3.900279106760362e-05,
      "loss": 0.7391,
      "step": 1431
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44462685276915537,
      "learning_rate": 3.900098417355782e-05,
      "loss": 0.691,
      "step": 1432
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4453655122956493,
      "learning_rate": 3.899917568592243e-05,
      "loss": 0.6998,
      "step": 1433
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.517290495471187,
      "learning_rate": 3.899736560484912e-05,
      "loss": 0.751,
      "step": 1434
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5418780375033296,
      "learning_rate": 3.89955539304897e-05,
      "loss": 0.8061,
      "step": 1435
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5370338902332776,
      "learning_rate": 3.8993740662996115e-05,
      "loss": 0.7305,
      "step": 1436
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.38971232601221856,
      "learning_rate": 3.8991925802520434e-05,
      "loss": 0.7257,
      "step": 1437
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4645359362406887,
      "learning_rate": 3.8990109349214875e-05,
      "loss": 0.7297,
      "step": 1438
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5051152985479197,
      "learning_rate": 3.8988291303231774e-05,
      "loss": 0.7244,
      "step": 1439
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4905325207783151,
      "learning_rate": 3.8986471664723624e-05,
      "loss": 0.7413,
      "step": 1440
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4895334407312211,
      "learning_rate": 3.8984650433843014e-05,
      "loss": 0.7638,
      "step": 1441
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45689246041930504,
      "learning_rate": 3.898282761074271e-05,
      "loss": 0.7031,
      "step": 1442
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4890011897193438,
      "learning_rate": 3.898100319557558e-05,
      "loss": 0.772,
      "step": 1443
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49379313084019094,
      "learning_rate": 3.8979177188494624e-05,
      "loss": 0.7806,
      "step": 1444
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4802672517552498,
      "learning_rate": 3.897734958965301e-05,
      "loss": 0.7209,
      "step": 1445
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42118817536378067,
      "learning_rate": 3.8975520399204e-05,
      "loss": 0.6758,
      "step": 1446
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.38853387273351414,
      "learning_rate": 3.8973689617301e-05,
      "loss": 0.7111,
      "step": 1447
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42663703628371924,
      "learning_rate": 3.897185724409758e-05,
      "loss": 0.7251,
      "step": 1448
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4720654054590996,
      "learning_rate": 3.897002327974741e-05,
      "loss": 0.7211,
      "step": 1449
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4157994229070869,
      "learning_rate": 3.896818772440428e-05,
      "loss": 0.7133,
      "step": 1450
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.39106263215315423,
      "learning_rate": 3.896635057822217e-05,
      "loss": 0.6901,
      "step": 1451
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4185793755069349,
      "learning_rate": 3.8964511841355136e-05,
      "loss": 0.7089,
      "step": 1452
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46494900932061395,
      "learning_rate": 3.896267151395739e-05,
      "loss": 0.7205,
      "step": 1453
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4204566792509373,
      "learning_rate": 3.89608295961833e-05,
      "loss": 0.716,
      "step": 1454
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42172040886603057,
      "learning_rate": 3.895898608818732e-05,
      "loss": 0.7494,
      "step": 1455
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44516896660937477,
      "learning_rate": 3.8957140990124076e-05,
      "loss": 0.7647,
      "step": 1456
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49405519829500105,
      "learning_rate": 3.895529430214831e-05,
      "loss": 0.7503,
      "step": 1457
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45486641484901325,
      "learning_rate": 3.89534460244149e-05,
      "loss": 0.7319,
      "step": 1458
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4463638652964958,
      "learning_rate": 3.895159615707887e-05,
      "loss": 0.761,
      "step": 1459
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4281705631178093,
      "learning_rate": 3.894974470029535e-05,
      "loss": 0.7234,
      "step": 1460
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49121216879986584,
      "learning_rate": 3.894789165421964e-05,
      "loss": 0.7819,
      "step": 1461
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.39944635149906693,
      "learning_rate": 3.894603701900713e-05,
      "loss": 0.6547,
      "step": 1462
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4231088186525573,
      "learning_rate": 3.894418079481337e-05,
      "loss": 0.716,
      "step": 1463
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40643200852681843,
      "learning_rate": 3.8942322981794055e-05,
      "loss": 0.7183,
      "step": 1464
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4541772868141762,
      "learning_rate": 3.894046358010499e-05,
      "loss": 0.7287,
      "step": 1465
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4129950895492292,
      "learning_rate": 3.893860258990212e-05,
      "loss": 0.6957,
      "step": 1466
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4521995796627358,
      "learning_rate": 3.893674001134152e-05,
      "loss": 0.755,
      "step": 1467
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44339333645395984,
      "learning_rate": 3.8934875844579395e-05,
      "loss": 0.7442,
      "step": 1468
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.480047228160539,
      "learning_rate": 3.893301008977211e-05,
      "loss": 0.7374,
      "step": 1469
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4331658396412457,
      "learning_rate": 3.893114274707613e-05,
      "loss": 0.69,
      "step": 1470
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4427975637498984,
      "learning_rate": 3.892927381664808e-05,
      "loss": 0.7247,
      "step": 1471
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40092980970671743,
      "learning_rate": 3.8927403298644685e-05,
      "loss": 0.7385,
      "step": 1472
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44995763291718477,
      "learning_rate": 3.892553119322284e-05,
      "loss": 0.7499,
      "step": 1473
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3598125072348258,
      "learning_rate": 3.892365750053955e-05,
      "loss": 0.6982,
      "step": 1474
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45163404489633835,
      "learning_rate": 3.892178222075196e-05,
      "loss": 0.7269,
      "step": 1475
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.391529704758664,
      "learning_rate": 3.8919905354017346e-05,
      "loss": 0.6983,
      "step": 1476
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4132308302255694,
      "learning_rate": 3.8918026900493113e-05,
      "loss": 0.732,
      "step": 1477
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4538081695522588,
      "learning_rate": 3.891614686033682e-05,
      "loss": 0.7075,
      "step": 1478
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.39733738053885387,
      "learning_rate": 3.891426523370613e-05,
      "loss": 0.6821,
      "step": 1479
    },
    {
      "epoch": 0.13,
      "grad_norm": 62.90657166500232,
      "learning_rate": 3.8912382020758854e-05,
      "loss": 7.0623,
      "step": 1480
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6190457702952259,
      "learning_rate": 3.891049722165295e-05,
      "loss": 0.7245,
      "step": 1481
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8443327778533808,
      "learning_rate": 3.890861083654647e-05,
      "loss": 0.7637,
      "step": 1482
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.75088874908477,
      "learning_rate": 3.8906722865597644e-05,
      "loss": 0.7685,
      "step": 1483
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9403477888117067,
      "learning_rate": 3.89048333089648e-05,
      "loss": 0.7297,
      "step": 1484
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2343896055013093,
      "learning_rate": 3.890294216680642e-05,
      "loss": 0.8008,
      "step": 1485
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0145019533235342,
      "learning_rate": 3.8901049439281105e-05,
      "loss": 0.7207,
      "step": 1486
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5900663140914839,
      "learning_rate": 3.88991551265476e-05,
      "loss": 0.7373,
      "step": 1487
    },
    {
      "epoch": 0.13,
      "grad_norm": 97.0622512922828,
      "learning_rate": 3.889725922876479e-05,
      "loss": 10.193,
      "step": 1488
    },
    {
      "epoch": 0.13,
      "grad_norm": 57.63890034058442,
      "learning_rate": 3.889536174609166e-05,
      "loss": 2.0728,
      "step": 1489
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3954959910555313,
      "learning_rate": 3.8893462678687364e-05,
      "loss": 0.8203,
      "step": 1490
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.218233164611881,
      "learning_rate": 3.8891562026711165e-05,
      "loss": 0.8059,
      "step": 1491
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.152778519423947,
      "learning_rate": 3.888965979032248e-05,
      "loss": 0.8298,
      "step": 1492
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.8771606588061407,
      "learning_rate": 3.888775596968084e-05,
      "loss": 0.8876,
      "step": 1493
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4884223270246224,
      "learning_rate": 3.8885850564945914e-05,
      "loss": 0.7777,
      "step": 1494
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4476617634825881,
      "learning_rate": 3.888394357627752e-05,
      "loss": 0.804,
      "step": 1495
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5005662099659578,
      "learning_rate": 3.888203500383557e-05,
      "loss": 0.8015,
      "step": 1496
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.305693171979275,
      "learning_rate": 3.888012484778016e-05,
      "loss": 0.833,
      "step": 1497
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.540938525561358,
      "learning_rate": 3.887821310827147e-05,
      "loss": 0.8854,
      "step": 1498
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.280419504935925,
      "learning_rate": 3.887629978546986e-05,
      "loss": 0.8596,
      "step": 1499
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8452702088624187,
      "learning_rate": 3.887438487953577e-05,
      "loss": 0.7526,
      "step": 1500
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8552530254497116,
      "learning_rate": 3.887246839062981e-05,
      "loss": 0.8258,
      "step": 1501
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7111623349122682,
      "learning_rate": 3.887055031891273e-05,
      "loss": 0.7634,
      "step": 1502
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7182214115519018,
      "learning_rate": 3.886863066454539e-05,
      "loss": 0.7933,
      "step": 1503
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6425239405552405,
      "learning_rate": 3.886670942768877e-05,
      "loss": 0.7671,
      "step": 1504
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7282280129555566,
      "learning_rate": 3.886478660850403e-05,
      "loss": 0.8365,
      "step": 1505
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8416049546078732,
      "learning_rate": 3.88628622071524e-05,
      "loss": 0.7921,
      "step": 1506
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6152792799812167,
      "learning_rate": 3.886093622379531e-05,
      "loss": 0.7603,
      "step": 1507
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5303753703794079,
      "learning_rate": 3.8859008658594283e-05,
      "loss": 0.7846,
      "step": 1508
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41365103018111843,
      "learning_rate": 3.885707951171097e-05,
      "loss": 0.6539,
      "step": 1509
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6459368795296215,
      "learning_rate": 3.885514878330717e-05,
      "loss": 0.7979,
      "step": 1510
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45991271034285497,
      "learning_rate": 3.885321647354481e-05,
      "loss": 0.7438,
      "step": 1511
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5044039211830776,
      "learning_rate": 3.885128258258596e-05,
      "loss": 0.7696,
      "step": 1512
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5538744569601126,
      "learning_rate": 3.88493471105928e-05,
      "loss": 0.7453,
      "step": 1513
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5444276866965054,
      "learning_rate": 3.8847410057727664e-05,
      "loss": 0.7606,
      "step": 1514
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5578353471961685,
      "learning_rate": 3.884547142415301e-05,
      "loss": 0.7691,
      "step": 1515
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5303128078569341,
      "learning_rate": 3.884353121003143e-05,
      "loss": 0.7605,
      "step": 1516
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5572391185685373,
      "learning_rate": 3.884158941552564e-05,
      "loss": 0.7509,
      "step": 1517
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5152345445556699,
      "learning_rate": 3.88396460407985e-05,
      "loss": 0.7438,
      "step": 1518
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.512240585112702,
      "learning_rate": 3.8837701086013e-05,
      "loss": 0.7429,
      "step": 1519
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.49568202372335146,
      "learning_rate": 3.883575455133226e-05,
      "loss": 0.7212,
      "step": 1520
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4643768112957315,
      "learning_rate": 3.883380643691954e-05,
      "loss": 0.6973,
      "step": 1521
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4894881860901471,
      "learning_rate": 3.883185674293821e-05,
      "loss": 0.7539,
      "step": 1522
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.49132562165590765,
      "learning_rate": 3.8829905469551806e-05,
      "loss": 0.7314,
      "step": 1523
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48207333226929716,
      "learning_rate": 3.882795261692397e-05,
      "loss": 0.707,
      "step": 1524
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7387356437402087,
      "learning_rate": 3.882599818521848e-05,
      "loss": 0.7527,
      "step": 1525
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5304760975778514,
      "learning_rate": 3.882404217459926e-05,
      "loss": 0.7835,
      "step": 1526
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6618829146540474,
      "learning_rate": 3.882208458523036e-05,
      "loss": 0.8224,
      "step": 1527
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40213298408368603,
      "learning_rate": 3.882012541727596e-05,
      "loss": 0.698,
      "step": 1528
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4660518937634644,
      "learning_rate": 3.881816467090036e-05,
      "loss": 0.7397,
      "step": 1529
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3606342948225546,
      "learning_rate": 3.8816202346268026e-05,
      "loss": 0.7019,
      "step": 1530
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4959817749534798,
      "learning_rate": 3.881423844354352e-05,
      "loss": 0.8027,
      "step": 1531
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43718576837545453,
      "learning_rate": 3.881227296289156e-05,
      "loss": 0.6906,
      "step": 1532
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4270162745303915,
      "learning_rate": 3.8810305904476984e-05,
      "loss": 0.7096,
      "step": 1533
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4486690990701428,
      "learning_rate": 3.880833726846477e-05,
      "loss": 0.7721,
      "step": 1534
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37702552954883,
      "learning_rate": 3.880636705502002e-05,
      "loss": 0.7269,
      "step": 1535
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46019503170384785,
      "learning_rate": 3.880439526430798e-05,
      "loss": 0.7246,
      "step": 1536
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.49267835847673797,
      "learning_rate": 3.8802421896494024e-05,
      "loss": 0.7398,
      "step": 1537
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4807480185270275,
      "learning_rate": 3.880044695174365e-05,
      "loss": 0.7658,
      "step": 1538
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38902295080746574,
      "learning_rate": 3.879847043022249e-05,
      "loss": 0.6984,
      "step": 1539
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46372929779157035,
      "learning_rate": 3.879649233209632e-05,
      "loss": 0.7451,
      "step": 1540
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4010494449778613,
      "learning_rate": 3.879451265753104e-05,
      "loss": 0.7173,
      "step": 1541
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3568967744991311,
      "learning_rate": 3.879253140669268e-05,
      "loss": 0.6788,
      "step": 1542
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.428510218099397,
      "learning_rate": 3.879054857974741e-05,
      "loss": 0.73,
      "step": 1543
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4234718694448765,
      "learning_rate": 3.8788564176861514e-05,
      "loss": 0.6932,
      "step": 1544
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4271853581592397,
      "learning_rate": 3.878657819820144e-05,
      "loss": 0.7281,
      "step": 1545
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42115567196489767,
      "learning_rate": 3.878459064393374e-05,
      "loss": 0.7077,
      "step": 1546
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45087172875281856,
      "learning_rate": 3.8782601514225116e-05,
      "loss": 0.7571,
      "step": 1547
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47196930642674534,
      "learning_rate": 3.878061080924237e-05,
      "loss": 0.758,
      "step": 1548
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.469843968757126,
      "learning_rate": 3.8778618529152484e-05,
      "loss": 0.765,
      "step": 1549
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4202413014702421,
      "learning_rate": 3.877662467412255e-05,
      "loss": 0.7045,
      "step": 1550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40132416351614325,
      "learning_rate": 3.877462924431977e-05,
      "loss": 0.7271,
      "step": 1551
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39704576977605094,
      "learning_rate": 3.8772632239911514e-05,
      "loss": 0.6604,
      "step": 1552
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44995218206935716,
      "learning_rate": 3.877063366106526e-05,
      "loss": 0.7377,
      "step": 1553
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46805524726010117,
      "learning_rate": 3.876863350794863e-05,
      "loss": 0.7561,
      "step": 1554
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4475137634976633,
      "learning_rate": 3.876663178072938e-05,
      "loss": 0.7813,
      "step": 1555
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4584312592070683,
      "learning_rate": 3.8764628479575386e-05,
      "loss": 0.6985,
      "step": 1556
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40216714725193803,
      "learning_rate": 3.8762623604654656e-05,
      "loss": 0.7027,
      "step": 1557
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45702848438574095,
      "learning_rate": 3.876061715613535e-05,
      "loss": 0.7504,
      "step": 1558
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47302606012305515,
      "learning_rate": 3.8758609134185736e-05,
      "loss": 0.7483,
      "step": 1559
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4594098639875215,
      "learning_rate": 3.875659953897423e-05,
      "loss": 0.7176,
      "step": 1560
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44306867385481313,
      "learning_rate": 3.875458837066938e-05,
      "loss": 0.7342,
      "step": 1561
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39766974468271954,
      "learning_rate": 3.875257562943984e-05,
      "loss": 0.6968,
      "step": 1562
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4089233956800588,
      "learning_rate": 3.875056131545444e-05,
      "loss": 0.7698,
      "step": 1563
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40884394252286627,
      "learning_rate": 3.87485454288821e-05,
      "loss": 0.6795,
      "step": 1564
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47791375027955774,
      "learning_rate": 3.8746527969891905e-05,
      "loss": 0.7693,
      "step": 1565
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4183519781933935,
      "learning_rate": 3.874450893865305e-05,
      "loss": 0.6968,
      "step": 1566
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46634885903127865,
      "learning_rate": 3.874248833533486e-05,
      "loss": 0.7558,
      "step": 1567
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48812296847215053,
      "learning_rate": 3.874046616010681e-05,
      "loss": 0.7435,
      "step": 1568
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4574785454337576,
      "learning_rate": 3.87384424131385e-05,
      "loss": 0.7177,
      "step": 1569
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39952187361680197,
      "learning_rate": 3.873641709459965e-05,
      "loss": 0.7163,
      "step": 1570
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43209870173889825,
      "learning_rate": 3.8734390204660134e-05,
      "loss": 0.7459,
      "step": 1571
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4708910708423934,
      "learning_rate": 3.8732361743489934e-05,
      "loss": 0.6923,
      "step": 1572
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4067539451827364,
      "learning_rate": 3.873033171125918e-05,
      "loss": 0.7066,
      "step": 1573
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44625743172902294,
      "learning_rate": 3.872830010813812e-05,
      "loss": 0.6935,
      "step": 1574
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4681496227931311,
      "learning_rate": 3.8726266934297144e-05,
      "loss": 0.715,
      "step": 1575
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3737198593158874,
      "learning_rate": 3.8724232189906784e-05,
      "loss": 0.6975,
      "step": 1576
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40464193862214515,
      "learning_rate": 3.872219587513768e-05,
      "loss": 0.7063,
      "step": 1577
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.49217618755701725,
      "learning_rate": 3.8720157990160615e-05,
      "loss": 0.7564,
      "step": 1578
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4271157900158598,
      "learning_rate": 3.871811853514652e-05,
      "loss": 0.7468,
      "step": 1579
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4038304508275699,
      "learning_rate": 3.871607751026642e-05,
      "loss": 0.6894,
      "step": 1580
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46599740932090383,
      "learning_rate": 3.87140349156915e-05,
      "loss": 0.8012,
      "step": 1581
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42434725506646687,
      "learning_rate": 3.871199075159308e-05,
      "loss": 0.7091,
      "step": 1582
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46891697768165713,
      "learning_rate": 3.870994501814259e-05,
      "loss": 0.7529,
      "step": 1583
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38453027496366593,
      "learning_rate": 3.87078977155116e-05,
      "loss": 0.7121,
      "step": 1584
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3779116315893138,
      "learning_rate": 3.8705848843871837e-05,
      "loss": 0.707,
      "step": 1585
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3773949764222041,
      "learning_rate": 3.870379840339511e-05,
      "loss": 0.7071,
      "step": 1586
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.450904250029696,
      "learning_rate": 3.870174639425341e-05,
      "loss": 0.7173,
      "step": 1587
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4221077710564387,
      "learning_rate": 3.8699692816618816e-05,
      "loss": 0.7169,
      "step": 1588
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4234705900804463,
      "learning_rate": 3.869763767066358e-05,
      "loss": 0.6977,
      "step": 1589
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39727384590140585,
      "learning_rate": 3.869558095656004e-05,
      "loss": 0.7405,
      "step": 1590
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43245823796668037,
      "learning_rate": 3.869352267448071e-05,
      "loss": 0.75,
      "step": 1591
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4128053750462542,
      "learning_rate": 3.869146282459821e-05,
      "loss": 0.7051,
      "step": 1592
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3744190600923603,
      "learning_rate": 3.86894014070853e-05,
      "loss": 0.6911,
      "step": 1593
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4696072226981361,
      "learning_rate": 3.8687338422114867e-05,
      "loss": 0.7974,
      "step": 1594
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4150620777735388,
      "learning_rate": 3.8685273869859916e-05,
      "loss": 0.6693,
      "step": 1595
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4407090169361657,
      "learning_rate": 3.868320775049362e-05,
      "loss": 0.7547,
      "step": 1596
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40589399313759056,
      "learning_rate": 3.8681140064189255e-05,
      "loss": 0.6705,
      "step": 1597
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42704161315618205,
      "learning_rate": 3.867907081112023e-05,
      "loss": 0.7439,
      "step": 1598
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4731853418914745,
      "learning_rate": 3.86769999914601e-05,
      "loss": 0.7512,
      "step": 1599
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4656534544083553,
      "learning_rate": 3.867492760538253e-05,
      "loss": 0.7581,
      "step": 1600
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4793860415987727,
      "learning_rate": 3.867285365306134e-05,
      "loss": 0.7559,
      "step": 1601
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4799434723803088,
      "learning_rate": 3.867077813467046e-05,
      "loss": 0.7276,
      "step": 1602
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4121307351795611,
      "learning_rate": 3.8668701050383973e-05,
      "loss": 0.6573,
      "step": 1603
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4226546591830006,
      "learning_rate": 3.8666622400376076e-05,
      "loss": 0.7122,
      "step": 1604
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47119276866248766,
      "learning_rate": 3.8664542184821095e-05,
      "loss": 0.7637,
      "step": 1605
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4564895497797981,
      "learning_rate": 3.86624604038935e-05,
      "loss": 0.7235,
      "step": 1606
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4497603654395264,
      "learning_rate": 3.8660377057767895e-05,
      "loss": 0.7414,
      "step": 1607
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4578259810645336,
      "learning_rate": 3.8658292146619005e-05,
      "loss": 0.7375,
      "step": 1608
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37482357930895766,
      "learning_rate": 3.8656205670621685e-05,
      "loss": 0.6829,
      "step": 1609
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3915606993897981,
      "learning_rate": 3.8654117629950925e-05,
      "loss": 0.7278,
      "step": 1610
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4105252723824667,
      "learning_rate": 3.8652028024781845e-05,
      "loss": 0.7327,
      "step": 1611
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4400394015853486,
      "learning_rate": 3.86499368552897e-05,
      "loss": 0.7298,
      "step": 1612
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42197053849454336,
      "learning_rate": 3.864784412164988e-05,
      "loss": 0.7176,
      "step": 1613
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4346201166228162,
      "learning_rate": 3.864574982403789e-05,
      "loss": 0.7187,
      "step": 1614
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42529667310234226,
      "learning_rate": 3.8643653962629386e-05,
      "loss": 0.7227,
      "step": 1615
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44197357918807434,
      "learning_rate": 3.8641556537600135e-05,
      "loss": 0.7085,
      "step": 1616
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43565709995733065,
      "learning_rate": 3.863945754912605e-05,
      "loss": 0.7281,
      "step": 1617
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3756170742588803,
      "learning_rate": 3.8637356997383176e-05,
      "loss": 0.7164,
      "step": 1618
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5033170752463338,
      "learning_rate": 3.863525488254767e-05,
      "loss": 0.7705,
      "step": 1619
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5153768791378249,
      "learning_rate": 3.863315120479585e-05,
      "loss": 0.7203,
      "step": 1620
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.445168525017866,
      "learning_rate": 3.8631045964304144e-05,
      "loss": 0.7475,
      "step": 1621
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38346450485618855,
      "learning_rate": 3.8628939161249105e-05,
      "loss": 0.6813,
      "step": 1622
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4692819678300989,
      "learning_rate": 3.862683079580744e-05,
      "loss": 0.7645,
      "step": 1623
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4356615870092533,
      "learning_rate": 3.8624720868155975e-05,
      "loss": 0.744,
      "step": 1624
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3981244267253573,
      "learning_rate": 3.862260937847166e-05,
      "loss": 0.6642,
      "step": 1625
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5085051494819554,
      "learning_rate": 3.8620496326931587e-05,
      "loss": 0.7564,
      "step": 1626
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44468618069908245,
      "learning_rate": 3.8618381713712974e-05,
      "loss": 0.7504,
      "step": 1627
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43152021659068446,
      "learning_rate": 3.861626553899318e-05,
      "loss": 0.7172,
      "step": 1628
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4455777836250921,
      "learning_rate": 3.861414780294967e-05,
      "loss": 0.757,
      "step": 1629
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.48291545508239964,
      "learning_rate": 3.861202850576007e-05,
      "loss": 0.7559,
      "step": 1630
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39780795500614907,
      "learning_rate": 3.860990764760211e-05,
      "loss": 0.7263,
      "step": 1631
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38887821458641986,
      "learning_rate": 3.860778522865367e-05,
      "loss": 0.7154,
      "step": 1632
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40732271692566097,
      "learning_rate": 3.860566124909277e-05,
      "loss": 0.7,
      "step": 1633
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3856678911220655,
      "learning_rate": 3.8603535709097514e-05,
      "loss": 0.7028,
      "step": 1634
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4090002674764174,
      "learning_rate": 3.8601408608846186e-05,
      "loss": 0.704,
      "step": 1635
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4281741025247614,
      "learning_rate": 3.8599279948517195e-05,
      "loss": 0.6983,
      "step": 1636
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4449323980686434,
      "learning_rate": 3.859714972828905e-05,
      "loss": 0.7389,
      "step": 1637
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40269474504790675,
      "learning_rate": 3.859501794834041e-05,
      "loss": 0.6853,
      "step": 1638
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3649929015886851,
      "learning_rate": 3.859288460885008e-05,
      "loss": 0.6452,
      "step": 1639
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4420124083486353,
      "learning_rate": 3.859074970999697e-05,
      "loss": 0.7464,
      "step": 1640
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47628338784378066,
      "learning_rate": 3.858861325196014e-05,
      "loss": 0.7105,
      "step": 1641
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3662725905707998,
      "learning_rate": 3.8586475234918766e-05,
      "loss": 0.6778,
      "step": 1642
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4677152406103988,
      "learning_rate": 3.858433565905215e-05,
      "loss": 0.7345,
      "step": 1643
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4543776308811037,
      "learning_rate": 3.858219452453975e-05,
      "loss": 0.7392,
      "step": 1644
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3857067291247378,
      "learning_rate": 3.858005183156114e-05,
      "loss": 0.7078,
      "step": 1645
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5013803317432391,
      "learning_rate": 3.8577907580296026e-05,
      "loss": 0.7713,
      "step": 1646
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3964470430665952,
      "learning_rate": 3.857576177092423e-05,
      "loss": 0.6929,
      "step": 1647
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4125760998015171,
      "learning_rate": 3.857361440362573e-05,
      "loss": 0.6959,
      "step": 1648
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40000136581734186,
      "learning_rate": 3.857146547858063e-05,
      "loss": 0.7089,
      "step": 1649
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39982285118552047,
      "learning_rate": 3.8569314995969144e-05,
      "loss": 0.7124,
      "step": 1650
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47555213285794556,
      "learning_rate": 3.856716295597164e-05,
      "loss": 0.7089,
      "step": 1651
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39717958949529647,
      "learning_rate": 3.856500935876858e-05,
      "loss": 0.7338,
      "step": 1652
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4639173496680093,
      "learning_rate": 3.856285420454063e-05,
      "loss": 0.7138,
      "step": 1653
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42659409119249625,
      "learning_rate": 3.856069749346851e-05,
      "loss": 0.7052,
      "step": 1654
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.452981983017023,
      "learning_rate": 3.855853922573311e-05,
      "loss": 0.7284,
      "step": 1655
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3868204017013688,
      "learning_rate": 3.8556379401515435e-05,
      "loss": 0.6786,
      "step": 1656
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4230262482669223,
      "learning_rate": 3.8554218020996626e-05,
      "loss": 0.7396,
      "step": 1657
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4466566606783735,
      "learning_rate": 3.855205508435796e-05,
      "loss": 0.7794,
      "step": 1658
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5228255967974159,
      "learning_rate": 3.854989059178084e-05,
      "loss": 0.7219,
      "step": 1659
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.36119679075585015,
      "learning_rate": 3.85477245434468e-05,
      "loss": 0.683,
      "step": 1660
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44823520832804536,
      "learning_rate": 3.85455569395375e-05,
      "loss": 0.7272,
      "step": 1661
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4294343602986109,
      "learning_rate": 3.854338778023475e-05,
      "loss": 0.7136,
      "step": 1662
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38963518182698403,
      "learning_rate": 3.854121706572045e-05,
      "loss": 0.7358,
      "step": 1663
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44052544394192833,
      "learning_rate": 3.853904479617666e-05,
      "loss": 0.7246,
      "step": 1664
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46122521989838877,
      "learning_rate": 3.8536870971785584e-05,
      "loss": 0.728,
      "step": 1665
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4367736526148371,
      "learning_rate": 3.853469559272952e-05,
      "loss": 0.716,
      "step": 1666
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.35417126608383065,
      "learning_rate": 3.853251865919092e-05,
      "loss": 0.6878,
      "step": 1667
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.37495595170560375,
      "learning_rate": 3.853034017135237e-05,
      "loss": 0.6994,
      "step": 1668
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43144251597316646,
      "learning_rate": 3.852816012939657e-05,
      "loss": 0.7108,
      "step": 1669
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43023349687617635,
      "learning_rate": 3.852597853350635e-05,
      "loss": 0.7337,
      "step": 1670
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3993373655044943,
      "learning_rate": 3.8523795383864685e-05,
      "loss": 0.6977,
      "step": 1671
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.3384868783631,
      "learning_rate": 3.852161068065467e-05,
      "loss": 6.0522,
      "step": 1672
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42217715553696716,
      "learning_rate": 3.851942442405953e-05,
      "loss": 0.7379,
      "step": 1673
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.435099163376891,
      "learning_rate": 3.851723661426264e-05,
      "loss": 0.6452,
      "step": 1674
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6444405569931587,
      "learning_rate": 3.851504725144748e-05,
      "loss": 0.7716,
      "step": 1675
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5853066741957168,
      "learning_rate": 3.851285633579766e-05,
      "loss": 0.7548,
      "step": 1676
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.964444171444499,
      "learning_rate": 3.851066386749694e-05,
      "loss": 0.725,
      "step": 1677
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6773608391300834,
      "learning_rate": 3.8508469846729195e-05,
      "loss": 0.7576,
      "step": 1678
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.550605042087254,
      "learning_rate": 3.850627427367844e-05,
      "loss": 0.735,
      "step": 1679
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46505395590130066,
      "learning_rate": 3.8504077148528806e-05,
      "loss": 0.7026,
      "step": 1680
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41020581035287923,
      "learning_rate": 3.8501878471464574e-05,
      "loss": 0.6836,
      "step": 1681
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47941569804150785,
      "learning_rate": 3.849967824267014e-05,
      "loss": 0.6886,
      "step": 1682
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40409711187585673,
      "learning_rate": 3.8497476462330034e-05,
      "loss": 0.6761,
      "step": 1683
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46824825370507267,
      "learning_rate": 3.849527313062891e-05,
      "loss": 0.7634,
      "step": 1684
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4647523888761401,
      "learning_rate": 3.8493068247751565e-05,
      "loss": 0.6957,
      "step": 1685
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4831802865119682,
      "learning_rate": 3.849086181388293e-05,
      "loss": 0.7024,
      "step": 1686
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43429159341630924,
      "learning_rate": 3.848865382920804e-05,
      "loss": 0.6865,
      "step": 1687
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5031802307735836,
      "learning_rate": 3.848644429391208e-05,
      "loss": 0.6853,
      "step": 1688
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6020861076927349,
      "learning_rate": 3.8484233208180366e-05,
      "loss": 0.7042,
      "step": 1689
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4322135353080257,
      "learning_rate": 3.848202057219833e-05,
      "loss": 0.7087,
      "step": 1690
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.420048430961562,
      "learning_rate": 3.847980638615155e-05,
      "loss": 0.69,
      "step": 1691
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4681460351532896,
      "learning_rate": 3.8477590650225735e-05,
      "loss": 0.7645,
      "step": 1692
    },
    {
      "epoch": 0.15,
      "grad_norm": 48.76327339676947,
      "learning_rate": 3.84753733646067e-05,
      "loss": 6.0098,
      "step": 1693
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4704198927919589,
      "learning_rate": 3.847315452948042e-05,
      "loss": 0.7357,
      "step": 1694
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5474839420921809,
      "learning_rate": 3.847093414503298e-05,
      "loss": 0.7627,
      "step": 1695
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4633792201856479,
      "learning_rate": 3.8468712211450594e-05,
      "loss": 0.6862,
      "step": 1696
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44373197615514154,
      "learning_rate": 3.8466488728919626e-05,
      "loss": 0.7139,
      "step": 1697
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5044948269795947,
      "learning_rate": 3.846426369762655e-05,
      "loss": 0.6993,
      "step": 1698
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.610142868490128,
      "learning_rate": 3.846203711775797e-05,
      "loss": 0.7647,
      "step": 1699
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5044641734554792,
      "learning_rate": 3.845980898950063e-05,
      "loss": 0.7391,
      "step": 1700
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5108807032035457,
      "learning_rate": 3.8457579313041414e-05,
      "loss": 0.76,
      "step": 1701
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4905595677189153,
      "learning_rate": 3.845534808856731e-05,
      "loss": 0.6901,
      "step": 1702
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47530875459752525,
      "learning_rate": 3.845311531626545e-05,
      "loss": 0.6928,
      "step": 1703
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4343928239788858,
      "learning_rate": 3.845088099632309e-05,
      "loss": 0.6951,
      "step": 1704
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5978063051424808,
      "learning_rate": 3.844864512892762e-05,
      "loss": 0.81,
      "step": 1705
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3926395663043713,
      "learning_rate": 3.8446407714266575e-05,
      "loss": 0.7,
      "step": 1706
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43304004501060706,
      "learning_rate": 3.844416875252759e-05,
      "loss": 0.6879,
      "step": 1707
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5868128544439059,
      "learning_rate": 3.844192824389844e-05,
      "loss": 0.7751,
      "step": 1708
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42050083006185374,
      "learning_rate": 3.843968618856705e-05,
      "loss": 0.6749,
      "step": 1709
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5081220708378184,
      "learning_rate": 3.843744258672144e-05,
      "loss": 0.7564,
      "step": 1710
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45904841228501536,
      "learning_rate": 3.84351974385498e-05,
      "loss": 0.7132,
      "step": 1711
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4739430841345266,
      "learning_rate": 3.8432950744240396e-05,
      "loss": 0.6868,
      "step": 1712
    },
    {
      "epoch": 0.15,
      "grad_norm": 30.932817728023007,
      "learning_rate": 3.843070250398168e-05,
      "loss": 7.0332,
      "step": 1713
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4997045839706997,
      "learning_rate": 3.842845271796221e-05,
      "loss": 0.7565,
      "step": 1714
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8684420959983745,
      "learning_rate": 3.842620138637067e-05,
      "loss": 0.7481,
      "step": 1715
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4340491446393771,
      "learning_rate": 3.8423948509395865e-05,
      "loss": 0.7001,
      "step": 1716
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5087460585313066,
      "learning_rate": 3.842169408722675e-05,
      "loss": 0.7372,
      "step": 1717
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6170123684510247,
      "learning_rate": 3.8419438120052396e-05,
      "loss": 0.7215,
      "step": 1718
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.528820601425869,
      "learning_rate": 3.841718060806202e-05,
      "loss": 0.7163,
      "step": 1719
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.526715068134163,
      "learning_rate": 3.8414921551444945e-05,
      "loss": 0.7159,
      "step": 1720
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5833836470266784,
      "learning_rate": 3.841266095039064e-05,
      "loss": 0.7131,
      "step": 1721
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.49629893300867717,
      "learning_rate": 3.84103988050887e-05,
      "loss": 0.7264,
      "step": 1722
    },
    {
      "epoch": 0.15,
      "grad_norm": 36.80795345755121,
      "learning_rate": 3.840813511572884e-05,
      "loss": 4.9857,
      "step": 1723
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4636165082994351,
      "learning_rate": 3.840586988250093e-05,
      "loss": 0.6828,
      "step": 1724
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.48885635814448486,
      "learning_rate": 3.840360310559493e-05,
      "loss": 0.7474,
      "step": 1725
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4249559151812786,
      "learning_rate": 3.8401334785200966e-05,
      "loss": 0.6798,
      "step": 1726
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46993834552075997,
      "learning_rate": 3.839906492150928e-05,
      "loss": 0.7299,
      "step": 1727
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46429348678233784,
      "learning_rate": 3.8396793514710235e-05,
      "loss": 0.728,
      "step": 1728
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4119218266288677,
      "learning_rate": 3.839452056499434e-05,
      "loss": 0.655,
      "step": 1729
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.9733286843069,
      "learning_rate": 3.839224607255222e-05,
      "loss": 3.5805,
      "step": 1730
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46277635778972903,
      "learning_rate": 3.8389970037574634e-05,
      "loss": 0.7164,
      "step": 1731
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5393008839432933,
      "learning_rate": 3.8387692460252465e-05,
      "loss": 0.7384,
      "step": 1732
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5595835121907333,
      "learning_rate": 3.8385413340776746e-05,
      "loss": 0.7733,
      "step": 1733
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4355057523164286,
      "learning_rate": 3.838313267933861e-05,
      "loss": 0.7026,
      "step": 1734
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5307478738871452,
      "learning_rate": 3.8380850476129336e-05,
      "loss": 0.7606,
      "step": 1735
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5029628830623151,
      "learning_rate": 3.837856673134033e-05,
      "loss": 0.7008,
      "step": 1736
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41931236492485,
      "learning_rate": 3.837628144516313e-05,
      "loss": 0.7306,
      "step": 1737
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5192302941812209,
      "learning_rate": 3.8373994617789395e-05,
      "loss": 0.7333,
      "step": 1738
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.495474990931602,
      "learning_rate": 3.8371706249410925e-05,
      "loss": 0.7599,
      "step": 1739
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4766916604485063,
      "learning_rate": 3.8369416340219636e-05,
      "loss": 0.7454,
      "step": 1740
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4967260588469493,
      "learning_rate": 3.8367124890407586e-05,
      "loss": 0.7455,
      "step": 1741
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43980579734181785,
      "learning_rate": 3.836483190016695e-05,
      "loss": 0.7075,
      "step": 1742
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4665153853556714,
      "learning_rate": 3.836253736969004e-05,
      "loss": 0.7231,
      "step": 1743
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48063640054198453,
      "learning_rate": 3.836024129916931e-05,
      "loss": 0.7718,
      "step": 1744
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42653803580023925,
      "learning_rate": 3.8357943688797305e-05,
      "loss": 0.71,
      "step": 1745
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4131301164061963,
      "learning_rate": 3.8355644538766734e-05,
      "loss": 0.7055,
      "step": 1746
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5179250874447261,
      "learning_rate": 3.835334384927043e-05,
      "loss": 0.7414,
      "step": 1747
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4581988886060528,
      "learning_rate": 3.8351041620501336e-05,
      "loss": 0.7177,
      "step": 1748
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4792511581489343,
      "learning_rate": 3.8348737852652546e-05,
      "loss": 0.7022,
      "step": 1749
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44509838914065564,
      "learning_rate": 3.834643254591727e-05,
      "loss": 0.6964,
      "step": 1750
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4462934817761845,
      "learning_rate": 3.8344125700488854e-05,
      "loss": 0.7485,
      "step": 1751
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42852478622851403,
      "learning_rate": 3.8341817316560776e-05,
      "loss": 0.7086,
      "step": 1752
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4897557594305605,
      "learning_rate": 3.833950739432662e-05,
      "loss": 0.7999,
      "step": 1753
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46452296172702967,
      "learning_rate": 3.833719593398013e-05,
      "loss": 0.7185,
      "step": 1754
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44037980588562536,
      "learning_rate": 3.8334882935715164e-05,
      "loss": 0.6982,
      "step": 1755
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41902770483697516,
      "learning_rate": 3.8332568399725716e-05,
      "loss": 0.6924,
      "step": 1756
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.370583762596387,
      "learning_rate": 3.833025232620589e-05,
      "loss": 0.7086,
      "step": 1757
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.37336896829211713,
      "learning_rate": 3.832793471534994e-05,
      "loss": 0.7106,
      "step": 1758
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4641964419728218,
      "learning_rate": 3.8325615567352235e-05,
      "loss": 0.7213,
      "step": 1759
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38252074524167956,
      "learning_rate": 3.8323294882407296e-05,
      "loss": 0.699,
      "step": 1760
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42057344478282743,
      "learning_rate": 3.832097266070974e-05,
      "loss": 0.7167,
      "step": 1761
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3649551415486426,
      "learning_rate": 3.831864890245433e-05,
      "loss": 0.6898,
      "step": 1762
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4226718282066028,
      "learning_rate": 3.831632360783595e-05,
      "loss": 0.7195,
      "step": 1763
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.39583721026204854,
      "learning_rate": 3.8313996777049644e-05,
      "loss": 0.7438,
      "step": 1764
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4057892751464334,
      "learning_rate": 3.831166841029054e-05,
      "loss": 0.7075,
      "step": 1765
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43365718979557555,
      "learning_rate": 3.830933850775393e-05,
      "loss": 0.6965,
      "step": 1766
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4346283544835485,
      "learning_rate": 3.83070070696352e-05,
      "loss": 0.7282,
      "step": 1767
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38338734834155547,
      "learning_rate": 3.83046740961299e-05,
      "loss": 0.701,
      "step": 1768
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.37486505707740087,
      "learning_rate": 3.830233958743369e-05,
      "loss": 0.6858,
      "step": 1769
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4685600627077527,
      "learning_rate": 3.830000354374236e-05,
      "loss": 0.745,
      "step": 1770
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3681291320315216,
      "learning_rate": 3.829766596525183e-05,
      "loss": 0.6577,
      "step": 1771
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4497839129969689,
      "learning_rate": 3.829532685215817e-05,
      "loss": 0.7285,
      "step": 1772
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40781251084333225,
      "learning_rate": 3.829298620465753e-05,
      "loss": 0.6568,
      "step": 1773
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4390546257034335,
      "learning_rate": 3.8290644022946234e-05,
      "loss": 0.7053,
      "step": 1774
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45085044724048146,
      "learning_rate": 3.828830030722071e-05,
      "loss": 0.7188,
      "step": 1775
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.408046740773219,
      "learning_rate": 3.8285955057677525e-05,
      "loss": 0.71,
      "step": 1776
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4433049619464013,
      "learning_rate": 3.828360827451337e-05,
      "loss": 0.7071,
      "step": 1777
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4330881447037364,
      "learning_rate": 3.828125995792508e-05,
      "loss": 0.6881,
      "step": 1778
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42600877684369376,
      "learning_rate": 3.827891010810959e-05,
      "loss": 0.6649,
      "step": 1779
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41601851405810686,
      "learning_rate": 3.827655872526398e-05,
      "loss": 0.7348,
      "step": 1780
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44190057762815776,
      "learning_rate": 3.827420580958547e-05,
      "loss": 0.7289,
      "step": 1781
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5216725602444825,
      "learning_rate": 3.827185136127139e-05,
      "loss": 0.7174,
      "step": 1782
    },
    {
      "epoch": 0.16,
      "grad_norm": 96.78605964250316,
      "learning_rate": 3.8269495380519194e-05,
      "loss": 3.1408,
      "step": 1783
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4886343513703032,
      "learning_rate": 3.826713786752649e-05,
      "loss": 0.7583,
      "step": 1784
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5377275999861306,
      "learning_rate": 3.8264778822490994e-05,
      "loss": 0.7868,
      "step": 1785
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5755818853350623,
      "learning_rate": 3.8262418245610556e-05,
      "loss": 0.6899,
      "step": 1786
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4674251164128945,
      "learning_rate": 3.826005613708316e-05,
      "loss": 0.6919,
      "step": 1787
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.574743641682229,
      "learning_rate": 3.82576924971069e-05,
      "loss": 0.7054,
      "step": 1788
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5407126086786602,
      "learning_rate": 3.825532732588002e-05,
      "loss": 0.7243,
      "step": 1789
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4713100627449483,
      "learning_rate": 3.825296062360089e-05,
      "loss": 0.6781,
      "step": 1790
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5976371144051301,
      "learning_rate": 3.825059239046799e-05,
      "loss": 0.7415,
      "step": 1791
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5200095346169809,
      "learning_rate": 3.824822262667995e-05,
      "loss": 0.6822,
      "step": 1792
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1360418610588228,
      "learning_rate": 3.824585133243552e-05,
      "loss": 0.7237,
      "step": 1793
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4743241952907032,
      "learning_rate": 3.824347850793357e-05,
      "loss": 0.6551,
      "step": 1794
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4851175085134896,
      "learning_rate": 3.824110415337311e-05,
      "loss": 0.6814,
      "step": 1795
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6750641809992685,
      "learning_rate": 3.823872826895328e-05,
      "loss": 0.7811,
      "step": 1796
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5139931111512707,
      "learning_rate": 3.823635085487334e-05,
      "loss": 0.7102,
      "step": 1797
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46658063489464124,
      "learning_rate": 3.823397191133266e-05,
      "loss": 0.7209,
      "step": 1798
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6789872108678876,
      "learning_rate": 3.8231591438530794e-05,
      "loss": 0.7775,
      "step": 1799
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47356462487896134,
      "learning_rate": 3.822920943666736e-05,
      "loss": 0.7005,
      "step": 1800
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6022465332317831,
      "learning_rate": 3.822682590594216e-05,
      "loss": 0.7361,
      "step": 1801
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4355724793949034,
      "learning_rate": 3.822444084655507e-05,
      "loss": 0.6771,
      "step": 1802
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6667311040964348,
      "learning_rate": 3.822205425870614e-05,
      "loss": 0.7568,
      "step": 1803
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4765022395162634,
      "learning_rate": 3.8219666142595526e-05,
      "loss": 0.7085,
      "step": 1804
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43000059064347457,
      "learning_rate": 3.821727649842352e-05,
      "loss": 0.732,
      "step": 1805
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4874096878269392,
      "learning_rate": 3.821488532639053e-05,
      "loss": 0.71,
      "step": 1806
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4688768721191298,
      "learning_rate": 3.821249262669711e-05,
      "loss": 0.7172,
      "step": 1807
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45697483914699555,
      "learning_rate": 3.8210098399543927e-05,
      "loss": 0.7383,
      "step": 1808
    },
    {
      "epoch": 0.16,
      "grad_norm": 28.662086247644577,
      "learning_rate": 3.8207702645131784e-05,
      "loss": 3.2547,
      "step": 1809
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47769396884622517,
      "learning_rate": 3.820530536366161e-05,
      "loss": 0.7374,
      "step": 1810
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5361926272802613,
      "learning_rate": 3.8202906555334454e-05,
      "loss": 0.7396,
      "step": 1811
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5805014458583715,
      "learning_rate": 3.820050622035152e-05,
      "loss": 0.7101,
      "step": 1812
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7976185704793332,
      "learning_rate": 3.8198104358914105e-05,
      "loss": 0.7548,
      "step": 1813
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5651807236125348,
      "learning_rate": 3.8195700971223655e-05,
      "loss": 0.7118,
      "step": 1814
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6167015143048108,
      "learning_rate": 3.819329605748174e-05,
      "loss": 0.7248,
      "step": 1815
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5560887043876236,
      "learning_rate": 3.8190889617890054e-05,
      "loss": 0.715,
      "step": 1816
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4249404680022538,
      "learning_rate": 3.818848165265043e-05,
      "loss": 0.6622,
      "step": 1817
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.689964033337288,
      "learning_rate": 3.818607216196481e-05,
      "loss": 0.7366,
      "step": 1818
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5548783246318911,
      "learning_rate": 3.818366114603528e-05,
      "loss": 0.7332,
      "step": 1819
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.49778481652202905,
      "learning_rate": 3.818124860506406e-05,
      "loss": 0.7274,
      "step": 1820
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.49256467437200924,
      "learning_rate": 3.8178834539253464e-05,
      "loss": 0.7396,
      "step": 1821
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5228753879908555,
      "learning_rate": 3.817641894880599e-05,
      "loss": 0.7431,
      "step": 1822
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5680845198264101,
      "learning_rate": 3.817400183392419e-05,
      "loss": 0.7324,
      "step": 1823
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47430081908602983,
      "learning_rate": 3.817158319481081e-05,
      "loss": 0.7237,
      "step": 1824
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4523802437050787,
      "learning_rate": 3.81691630316687e-05,
      "loss": 0.6862,
      "step": 1825
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4327007628069618,
      "learning_rate": 3.816674134470082e-05,
      "loss": 0.73,
      "step": 1826
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5383616735410803,
      "learning_rate": 3.8164318134110285e-05,
      "loss": 0.7078,
      "step": 1827
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3892576762139742,
      "learning_rate": 3.816189340010033e-05,
      "loss": 0.6756,
      "step": 1828
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47806227562996695,
      "learning_rate": 3.815946714287431e-05,
      "loss": 0.7192,
      "step": 1829
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.49982939033463875,
      "learning_rate": 3.8157039362635703e-05,
      "loss": 0.7719,
      "step": 1830
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45198799791547467,
      "learning_rate": 3.815461005958814e-05,
      "loss": 0.723,
      "step": 1831
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4380225770076252,
      "learning_rate": 3.8152179233935355e-05,
      "loss": 0.7041,
      "step": 1832
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5270232076171532,
      "learning_rate": 3.8149746885881216e-05,
      "loss": 0.7395,
      "step": 1833
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4658447351723755,
      "learning_rate": 3.814731301562973e-05,
      "loss": 0.7285,
      "step": 1834
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4320076379676664,
      "learning_rate": 3.8144877623385016e-05,
      "loss": 0.7165,
      "step": 1835
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42736671303756785,
      "learning_rate": 3.814244070935133e-05,
      "loss": 0.7222,
      "step": 1836
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5022103413292837,
      "learning_rate": 3.814000227373305e-05,
      "loss": 0.7164,
      "step": 1837
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42286400780883215,
      "learning_rate": 3.813756231673469e-05,
      "loss": 0.7424,
      "step": 1838
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45149155209248515,
      "learning_rate": 3.8135120838560885e-05,
      "loss": 0.7172,
      "step": 1839
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5092152482329992,
      "learning_rate": 3.81326778394164e-05,
      "loss": 0.7147,
      "step": 1840
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41472157963126927,
      "learning_rate": 3.813023331950611e-05,
      "loss": 0.7143,
      "step": 1841
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40642950668465544,
      "learning_rate": 3.812778727903506e-05,
      "loss": 0.6951,
      "step": 1842
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40449300455514775,
      "learning_rate": 3.8125339718208376e-05,
      "loss": 0.7201,
      "step": 1843
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48867428638285604,
      "learning_rate": 3.812289063723134e-05,
      "loss": 0.6955,
      "step": 1844
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4094955116417097,
      "learning_rate": 3.812044003630936e-05,
      "loss": 0.6541,
      "step": 1845
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43998134861432364,
      "learning_rate": 3.811798791564795e-05,
      "loss": 0.6891,
      "step": 1846
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4463434549083084,
      "learning_rate": 3.8115534275452774e-05,
      "loss": 0.6736,
      "step": 1847
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.438400699447903,
      "learning_rate": 3.811307911592963e-05,
      "loss": 0.696,
      "step": 1848
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45665153098747635,
      "learning_rate": 3.81106224372844e-05,
      "loss": 0.7476,
      "step": 1849
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5108148963627767,
      "learning_rate": 3.810816423972314e-05,
      "loss": 0.755,
      "step": 1850
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.470119180231187,
      "learning_rate": 3.810570452345202e-05,
      "loss": 0.7273,
      "step": 1851
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39903954128332086,
      "learning_rate": 3.810324328867733e-05,
      "loss": 0.7341,
      "step": 1852
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48007279241036405,
      "learning_rate": 3.810078053560548e-05,
      "loss": 0.7481,
      "step": 1853
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.46267620750345556,
      "learning_rate": 3.8098316264443033e-05,
      "loss": 0.7178,
      "step": 1854
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40856974068838053,
      "learning_rate": 3.809585047539666e-05,
      "loss": 0.7324,
      "step": 1855
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4199733891060807,
      "learning_rate": 3.809338316867316e-05,
      "loss": 0.6908,
      "step": 1856
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40486027795128865,
      "learning_rate": 3.809091434447946e-05,
      "loss": 0.7316,
      "step": 1857
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42222320652231415,
      "learning_rate": 3.808844400302263e-05,
      "loss": 0.7289,
      "step": 1858
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4166132078345821,
      "learning_rate": 3.8085972144509836e-05,
      "loss": 0.6976,
      "step": 1859
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4205100937307827,
      "learning_rate": 3.808349876914841e-05,
      "loss": 0.7147,
      "step": 1860
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3972133380095631,
      "learning_rate": 3.808102387714578e-05,
      "loss": 0.7374,
      "step": 1861
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42196006694984883,
      "learning_rate": 3.807854746870951e-05,
      "loss": 0.6626,
      "step": 1862
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43274126066725915,
      "learning_rate": 3.80760695440473e-05,
      "loss": 0.7616,
      "step": 1863
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41968745915310873,
      "learning_rate": 3.807359010336697e-05,
      "loss": 0.6772,
      "step": 1864
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41849864664812647,
      "learning_rate": 3.8071109146876455e-05,
      "loss": 0.674,
      "step": 1865
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4206627273915415,
      "learning_rate": 3.806862667478385e-05,
      "loss": 0.69,
      "step": 1866
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.417057601030624,
      "learning_rate": 3.8066142687297346e-05,
      "loss": 0.7077,
      "step": 1867
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4592970024812581,
      "learning_rate": 3.806365718462527e-05,
      "loss": 0.7139,
      "step": 1868
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41688445556821224,
      "learning_rate": 3.8061170166976085e-05,
      "loss": 0.7098,
      "step": 1869
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38280896739553916,
      "learning_rate": 3.805868163455837e-05,
      "loss": 0.688,
      "step": 1870
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4231703775802834,
      "learning_rate": 3.805619158758084e-05,
      "loss": 0.6968,
      "step": 1871
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4135613763801111,
      "learning_rate": 3.805370002625232e-05,
      "loss": 0.7198,
      "step": 1872
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3821237901400197,
      "learning_rate": 3.805120695078178e-05,
      "loss": 0.7162,
      "step": 1873
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4146320529132493,
      "learning_rate": 3.804871236137832e-05,
      "loss": 0.7036,
      "step": 1874
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42927965224129006,
      "learning_rate": 3.804621625825115e-05,
      "loss": 0.7457,
      "step": 1875
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4979333645573049,
      "learning_rate": 3.8043718641609615e-05,
      "loss": 0.7684,
      "step": 1876
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3836256713239746,
      "learning_rate": 3.804121951166319e-05,
      "loss": 0.7052,
      "step": 1877
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40952441899412423,
      "learning_rate": 3.8038718868621474e-05,
      "loss": 0.6967,
      "step": 1878
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3846632822327194,
      "learning_rate": 3.803621671269419e-05,
      "loss": 0.6706,
      "step": 1879
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.422222637829152,
      "learning_rate": 3.803371304409119e-05,
      "loss": 0.6751,
      "step": 1880
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38688559197110595,
      "learning_rate": 3.8031207863022465e-05,
      "loss": 0.6646,
      "step": 1881
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44032499586424245,
      "learning_rate": 3.8028701169698104e-05,
      "loss": 0.7297,
      "step": 1882
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44042618335218425,
      "learning_rate": 3.802619296432835e-05,
      "loss": 0.7272,
      "step": 1883
    },
    {
      "epoch": 0.17,
      "grad_norm": 102.93743629685854,
      "learning_rate": 3.802368324712357e-05,
      "loss": 4.5602,
      "step": 1884
    },
    {
      "epoch": 0.17,
      "grad_norm": 26.031449589687956,
      "learning_rate": 3.8021172018294236e-05,
      "loss": 2.7614,
      "step": 1885
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5533858591740344,
      "learning_rate": 3.801865927805097e-05,
      "loss": 0.7021,
      "step": 1886
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9868504371208143,
      "learning_rate": 3.801614502660451e-05,
      "loss": 0.7588,
      "step": 1887
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.022354110592178,
      "learning_rate": 3.801362926416573e-05,
      "loss": 0.7591,
      "step": 1888
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.389289790011044,
      "learning_rate": 3.801111199094561e-05,
      "loss": 0.7933,
      "step": 1889
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5717396773208747,
      "learning_rate": 3.800859320715529e-05,
      "loss": 0.6864,
      "step": 1890
    },
    {
      "epoch": 0.17,
      "grad_norm": 117.84572704261859,
      "learning_rate": 3.8006072913006e-05,
      "loss": 3.8626,
      "step": 1891
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.575050063167222,
      "learning_rate": 3.800355110870912e-05,
      "loss": 0.7145,
      "step": 1892
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4999334465146623,
      "learning_rate": 3.800102779447615e-05,
      "loss": 0.7638,
      "step": 1893
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4286631892794457,
      "learning_rate": 3.799850297051873e-05,
      "loss": 0.693,
      "step": 1894
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4471751664042283,
      "learning_rate": 3.799597663704859e-05,
      "loss": 0.7292,
      "step": 1895
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4819515694396782,
      "learning_rate": 3.799344879427763e-05,
      "loss": 0.7387,
      "step": 1896
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5150220168872438,
      "learning_rate": 3.7990919442417846e-05,
      "loss": 0.6912,
      "step": 1897
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5741967465400083,
      "learning_rate": 3.7988388581681375e-05,
      "loss": 0.751,
      "step": 1898
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4866744794684133,
      "learning_rate": 3.798585621228048e-05,
      "loss": 0.7034,
      "step": 1899
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5583011334294633,
      "learning_rate": 3.798332233442755e-05,
      "loss": 0.7768,
      "step": 1900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5261751739046098,
      "learning_rate": 3.798078694833509e-05,
      "loss": 0.7425,
      "step": 1901
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5600917621067697,
      "learning_rate": 3.7978250054215745e-05,
      "loss": 0.7439,
      "step": 1902
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4455188279648149,
      "learning_rate": 3.797571165228227e-05,
      "loss": 0.7183,
      "step": 1903
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47741847120965253,
      "learning_rate": 3.7973171742747584e-05,
      "loss": 0.6958,
      "step": 1904
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4874364319251531,
      "learning_rate": 3.797063032582468e-05,
      "loss": 0.7355,
      "step": 1905
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43248148296552025,
      "learning_rate": 3.796808740172671e-05,
      "loss": 0.6703,
      "step": 1906
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4994875970265528,
      "learning_rate": 3.796554297066696e-05,
      "loss": 0.7411,
      "step": 1907
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4126358972366985,
      "learning_rate": 3.7962997032858814e-05,
      "loss": 0.6793,
      "step": 1908
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4451839357781574,
      "learning_rate": 3.7960449588515806e-05,
      "loss": 0.7451,
      "step": 1909
    },
    {
      "epoch": 0.17,
      "grad_norm": 43.024991287759256,
      "learning_rate": 3.795790063785158e-05,
      "loss": 3.1654,
      "step": 1910
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48505716056897347,
      "learning_rate": 3.795535018107991e-05,
      "loss": 0.7495,
      "step": 1911
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4679872577706204,
      "learning_rate": 3.7952798218414704e-05,
      "loss": 0.7313,
      "step": 1912
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5663903130135868,
      "learning_rate": 3.795024475007e-05,
      "loss": 0.7534,
      "step": 1913
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4399095207556826,
      "learning_rate": 3.794768977625994e-05,
      "loss": 0.7244,
      "step": 1914
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47810210495451383,
      "learning_rate": 3.794513329719882e-05,
      "loss": 0.7074,
      "step": 1915
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3851835735943853,
      "learning_rate": 3.7942575313101036e-05,
      "loss": 0.6671,
      "step": 1916
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5449737322939109,
      "learning_rate": 3.794001582418113e-05,
      "loss": 0.7503,
      "step": 1917
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4930788575276329,
      "learning_rate": 3.793745483065377e-05,
      "loss": 0.7462,
      "step": 1918
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4796530596184397,
      "learning_rate": 3.793489233273373e-05,
      "loss": 0.7315,
      "step": 1919
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4585528237604794,
      "learning_rate": 3.7932328330635936e-05,
      "loss": 0.7425,
      "step": 1920
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42134154065751817,
      "learning_rate": 3.7929762824575414e-05,
      "loss": 0.7019,
      "step": 1921
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4754122811156046,
      "learning_rate": 3.7927195814767346e-05,
      "loss": 0.726,
      "step": 1922
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40344236175056664,
      "learning_rate": 3.792462730142701e-05,
      "loss": 0.6886,
      "step": 1923
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4869533065878173,
      "learning_rate": 3.792205728476983e-05,
      "loss": 0.7166,
      "step": 1924
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4258323391027113,
      "learning_rate": 3.7919485765011346e-05,
      "loss": 0.7217,
      "step": 1925
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.417723618168533,
      "learning_rate": 3.791691274236724e-05,
      "loss": 0.6969,
      "step": 1926
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3861511963187301,
      "learning_rate": 3.791433821705329e-05,
      "loss": 0.6871,
      "step": 1927
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45877231132035917,
      "learning_rate": 3.7911762189285444e-05,
      "loss": 0.7174,
      "step": 1928
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45195321311723297,
      "learning_rate": 3.790918465927973e-05,
      "loss": 0.7143,
      "step": 1929
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44448532156705034,
      "learning_rate": 3.790660562725232e-05,
      "loss": 0.7043,
      "step": 1930
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41579650157492626,
      "learning_rate": 3.7904025093419525e-05,
      "loss": 0.6797,
      "step": 1931
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3621483190396711,
      "learning_rate": 3.790144305799777e-05,
      "loss": 0.6545,
      "step": 1932
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5020940511763994,
      "learning_rate": 3.78988595212036e-05,
      "loss": 0.7721,
      "step": 1933
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4574125073957334,
      "learning_rate": 3.789627448325371e-05,
      "loss": 0.7082,
      "step": 1934
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4235214294516968,
      "learning_rate": 3.7893687944364884e-05,
      "loss": 0.6909,
      "step": 1935
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4213780895033943,
      "learning_rate": 3.789109990475406e-05,
      "loss": 0.6893,
      "step": 1936
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47405670972181757,
      "learning_rate": 3.78885103646383e-05,
      "loss": 0.7352,
      "step": 1937
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45108519587635304,
      "learning_rate": 3.788591932423477e-05,
      "loss": 0.7721,
      "step": 1938
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4296191441319744,
      "learning_rate": 3.7883326783760794e-05,
      "loss": 0.6725,
      "step": 1939
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4355584788835477,
      "learning_rate": 3.788073274343379e-05,
      "loss": 0.7092,
      "step": 1940
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4527353499383179,
      "learning_rate": 3.7878137203471334e-05,
      "loss": 0.7413,
      "step": 1941
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45757743725336036,
      "learning_rate": 3.78755401640911e-05,
      "loss": 0.7145,
      "step": 1942
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4621551069717173,
      "learning_rate": 3.78729416255109e-05,
      "loss": 0.7017,
      "step": 1943
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3452823255385728,
      "learning_rate": 3.787034158794867e-05,
      "loss": 0.6862,
      "step": 1944
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40550587896101636,
      "learning_rate": 3.7867740051622477e-05,
      "loss": 0.6982,
      "step": 1945
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44678596509844853,
      "learning_rate": 3.78651370167505e-05,
      "loss": 0.7208,
      "step": 1946
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38975004133672964,
      "learning_rate": 3.786253248355106e-05,
      "loss": 0.6996,
      "step": 1947
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3932866874694595,
      "learning_rate": 3.785992645224259e-05,
      "loss": 0.7245,
      "step": 1948
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45953694984856275,
      "learning_rate": 3.785731892304366e-05,
      "loss": 0.7263,
      "step": 1949
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41674629445628747,
      "learning_rate": 3.785470989617296e-05,
      "loss": 0.7338,
      "step": 1950
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.31517012162546193,
      "learning_rate": 3.78520993718493e-05,
      "loss": 0.6466,
      "step": 1951
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.49810283452819204,
      "learning_rate": 3.784948735029163e-05,
      "loss": 0.6782,
      "step": 1952
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40628841463903576,
      "learning_rate": 3.784687383171901e-05,
      "loss": 0.6586,
      "step": 1953
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.544175053146365,
      "learning_rate": 3.7844258816350634e-05,
      "loss": 2.2969,
      "step": 1954
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4106226287457788,
      "learning_rate": 3.784164230440582e-05,
      "loss": 0.7272,
      "step": 1955
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48877338921234725,
      "learning_rate": 3.783902429610401e-05,
      "loss": 0.7381,
      "step": 1956
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4272882413104178,
      "learning_rate": 3.783640479166478e-05,
      "loss": 0.7201,
      "step": 1957
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4756079629502366,
      "learning_rate": 3.783378379130783e-05,
      "loss": 0.7423,
      "step": 1958
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42772377891111246,
      "learning_rate": 3.783116129525296e-05,
      "loss": 0.7053,
      "step": 1959
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44852972386262485,
      "learning_rate": 3.7828537303720136e-05,
      "loss": 0.6916,
      "step": 1960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4847603122307005,
      "learning_rate": 3.782591181692941e-05,
      "loss": 0.728,
      "step": 1961
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5231802617892657,
      "learning_rate": 3.7823284835100994e-05,
      "loss": 0.7533,
      "step": 1962
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.325344657084871,
      "learning_rate": 3.782065635845521e-05,
      "loss": 1.7695,
      "step": 1963
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45263474020940103,
      "learning_rate": 3.7818026387212486e-05,
      "loss": 0.7369,
      "step": 1964
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5130038705671417,
      "learning_rate": 3.7815394921593416e-05,
      "loss": 0.7111,
      "step": 1965
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4258354911144242,
      "learning_rate": 3.781276196181868e-05,
      "loss": 0.7103,
      "step": 1966
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4216921218090062,
      "learning_rate": 3.7810127508109125e-05,
      "loss": 0.7021,
      "step": 1967
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4815121906847216,
      "learning_rate": 3.7807491560685676e-05,
      "loss": 0.7289,
      "step": 1968
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4828457462375156,
      "learning_rate": 3.780485411976942e-05,
      "loss": 0.7193,
      "step": 1969
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42662240497523324,
      "learning_rate": 3.780221518558155e-05,
      "loss": 0.7005,
      "step": 1970
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4347180460593013,
      "learning_rate": 3.779957475834339e-05,
      "loss": 0.6759,
      "step": 1971
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4450352048973943,
      "learning_rate": 3.779693283827639e-05,
      "loss": 0.7287,
      "step": 1972
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.46043349424655766,
      "learning_rate": 3.7794289425602135e-05,
      "loss": 0.7121,
      "step": 1973
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4144528027652025,
      "learning_rate": 3.779164452054231e-05,
      "loss": 0.6921,
      "step": 1974
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.434062357881247,
      "learning_rate": 3.778899812331874e-05,
      "loss": 0.7388,
      "step": 1975
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4045121362117341,
      "learning_rate": 3.7786350234153386e-05,
      "loss": 0.695,
      "step": 1976
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.47170352486207234,
      "learning_rate": 3.7783700853268325e-05,
      "loss": 0.7636,
      "step": 1977
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40659518689906826,
      "learning_rate": 3.7781049980885746e-05,
      "loss": 0.6763,
      "step": 1978
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4097763062344127,
      "learning_rate": 3.777839761722797e-05,
      "loss": 0.6872,
      "step": 1979
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4489041207915961,
      "learning_rate": 3.777574376251747e-05,
      "loss": 0.7666,
      "step": 1980
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4178392216070387,
      "learning_rate": 3.77730884169768e-05,
      "loss": 0.6764,
      "step": 1981
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4408473942332054,
      "learning_rate": 3.777043158082868e-05,
      "loss": 0.7086,
      "step": 1982
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4146046430198484,
      "learning_rate": 3.7767773254295914e-05,
      "loss": 0.719,
      "step": 1983
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42014395982202596,
      "learning_rate": 3.7765113437601464e-05,
      "loss": 0.6933,
      "step": 1984
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40814172298713824,
      "learning_rate": 3.776245213096841e-05,
      "loss": 0.6911,
      "step": 1985
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4373930411769075,
      "learning_rate": 3.775978933461994e-05,
      "loss": 0.7338,
      "step": 1986
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43493439556163377,
      "learning_rate": 3.7757125048779394e-05,
      "loss": 0.7417,
      "step": 1987
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4281476006898092,
      "learning_rate": 3.775445927367021e-05,
      "loss": 0.7005,
      "step": 1988
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42292633849443173,
      "learning_rate": 3.775179200951597e-05,
      "loss": 0.7535,
      "step": 1989
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3831492364422464,
      "learning_rate": 3.774912325654038e-05,
      "loss": 0.7048,
      "step": 1990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4449785181726118,
      "learning_rate": 3.774645301496725e-05,
      "loss": 0.7821,
      "step": 1991
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3708009519019661,
      "learning_rate": 3.774378128502054e-05,
      "loss": 0.6556,
      "step": 1992
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.37916810047022115,
      "learning_rate": 3.7741108066924316e-05,
      "loss": 0.7301,
      "step": 1993
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.48173201526834536,
      "learning_rate": 3.77384333609028e-05,
      "loss": 0.7386,
      "step": 1994
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4195892130155925,
      "learning_rate": 3.773575716718028e-05,
      "loss": 0.6932,
      "step": 1995
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4292457511134677,
      "learning_rate": 3.7733079485981246e-05,
      "loss": 0.7279,
      "step": 1996
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38555474637591985,
      "learning_rate": 3.773040031753024e-05,
      "loss": 0.6641,
      "step": 1997
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34820515551315495,
      "learning_rate": 3.7727719662051977e-05,
      "loss": 0.6616,
      "step": 1998
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40358897759452045,
      "learning_rate": 3.7725037519771275e-05,
      "loss": 0.6777,
      "step": 1999
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4224404567365767,
      "learning_rate": 3.7722353890913084e-05,
      "loss": 0.7262,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39112645724791717,
      "learning_rate": 3.771966877570247e-05,
      "loss": 0.6924,
      "step": 2001
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38411550833510905,
      "learning_rate": 3.7716982174364635e-05,
      "loss": 0.7007,
      "step": 2002
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41054638408235034,
      "learning_rate": 3.7714294087124904e-05,
      "loss": 0.7103,
      "step": 2003
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3681018486541451,
      "learning_rate": 3.771160451420873e-05,
      "loss": 0.6622,
      "step": 2004
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45545234540883456,
      "learning_rate": 3.770891345584166e-05,
      "loss": 0.734,
      "step": 2005
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38649595462393516,
      "learning_rate": 3.770622091224942e-05,
      "loss": 0.7061,
      "step": 2006
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3624864692313925,
      "learning_rate": 3.7703526883657805e-05,
      "loss": 0.6935,
      "step": 2007
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3944971163382798,
      "learning_rate": 3.7700831370292774e-05,
      "loss": 0.7034,
      "step": 2008
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4026444844584009,
      "learning_rate": 3.7698134372380394e-05,
      "loss": 0.6919,
      "step": 2009
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.365906790673325,
      "learning_rate": 3.769543589014686e-05,
      "loss": 0.7167,
      "step": 2010
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4042465929474885,
      "learning_rate": 3.769273592381849e-05,
      "loss": 0.6999,
      "step": 2011
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4443392233986467,
      "learning_rate": 3.7690034473621725e-05,
      "loss": 0.7199,
      "step": 2012
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.359596491941641,
      "learning_rate": 3.768733153978312e-05,
      "loss": 0.6712,
      "step": 2013
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44625217562883507,
      "learning_rate": 3.768462712252939e-05,
      "loss": 0.7192,
      "step": 2014
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3824910511087871,
      "learning_rate": 3.768192122208734e-05,
      "loss": 0.7041,
      "step": 2015
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42828879933531594,
      "learning_rate": 3.767921383868392e-05,
      "loss": 0.7666,
      "step": 2016
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3673538536667121,
      "learning_rate": 3.7676504972546176e-05,
      "loss": 0.6868,
      "step": 2017
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3557226006176314,
      "learning_rate": 3.767379462390131e-05,
      "loss": 0.6888,
      "step": 2018
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3762726151737514,
      "learning_rate": 3.767108279297663e-05,
      "loss": 0.6656,
      "step": 2019
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3392651572224488,
      "learning_rate": 3.766836947999959e-05,
      "loss": 0.6574,
      "step": 2020
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3871672424527455,
      "learning_rate": 3.766565468519772e-05,
      "loss": 0.7022,
      "step": 2021
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4262133144726188,
      "learning_rate": 3.766293840879874e-05,
      "loss": 0.7075,
      "step": 2022
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3406218705843418,
      "learning_rate": 3.766022065103045e-05,
      "loss": 0.6808,
      "step": 2023
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3937879110231063,
      "learning_rate": 3.765750141212077e-05,
      "loss": 0.7091,
      "step": 2024
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40888561144196667,
      "learning_rate": 3.765478069229778e-05,
      "loss": 0.6916,
      "step": 2025
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4272202012725263,
      "learning_rate": 3.765205849178966e-05,
      "loss": 0.7421,
      "step": 2026
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.46088818467105885,
      "learning_rate": 3.76493348108247e-05,
      "loss": 0.7333,
      "step": 2027
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4022840161276622,
      "learning_rate": 3.764660964963135e-05,
      "loss": 0.7142,
      "step": 2028
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39264244538184584,
      "learning_rate": 3.7643883008438156e-05,
      "loss": 0.6342,
      "step": 2029
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4026902503657626,
      "learning_rate": 3.764115488747381e-05,
      "loss": 0.7084,
      "step": 2030
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43484796912657026,
      "learning_rate": 3.76384252869671e-05,
      "loss": 0.7063,
      "step": 2031
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3978245916658217,
      "learning_rate": 3.763569420714697e-05,
      "loss": 0.6864,
      "step": 2032
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.37456659989713376,
      "learning_rate": 3.763296164824246e-05,
      "loss": 0.6877,
      "step": 2033
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41889726262626326,
      "learning_rate": 3.763022761048276e-05,
      "loss": 0.6875,
      "step": 2034
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.46179224587719847,
      "learning_rate": 3.762749209409716e-05,
      "loss": 0.7184,
      "step": 2035
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4029522389589981,
      "learning_rate": 3.7624755099315084e-05,
      "loss": 0.668,
      "step": 2036
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44825414859306967,
      "learning_rate": 3.762201662636608e-05,
      "loss": 0.7353,
      "step": 2037
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4051685462577793,
      "learning_rate": 3.761927667547983e-05,
      "loss": 0.7199,
      "step": 2038
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4233266699116615,
      "learning_rate": 3.761653524688612e-05,
      "loss": 0.7083,
      "step": 2039
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3939818946833482,
      "learning_rate": 3.7613792340814874e-05,
      "loss": 0.7047,
      "step": 2040
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40116030119169566,
      "learning_rate": 3.761104795749614e-05,
      "loss": 0.711,
      "step": 2041
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40355701933876487,
      "learning_rate": 3.7608302097160076e-05,
      "loss": 0.6857,
      "step": 2042
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39877913128031495,
      "learning_rate": 3.760555476003699e-05,
      "loss": 0.7129,
      "step": 2043
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3819090485271241,
      "learning_rate": 3.760280594635728e-05,
      "loss": 0.6803,
      "step": 2044
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3585101588637689,
      "learning_rate": 3.76000556563515e-05,
      "loss": 0.6795,
      "step": 2045
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.35328148328725445,
      "learning_rate": 3.7597303890250305e-05,
      "loss": 0.6563,
      "step": 2046
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4139695082703085,
      "learning_rate": 3.759455064828449e-05,
      "loss": 0.7136,
      "step": 2047
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3713877892055064,
      "learning_rate": 3.7591795930684946e-05,
      "loss": 0.6996,
      "step": 2048
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4183404052991828,
      "learning_rate": 3.758903973768274e-05,
      "loss": 0.7049,
      "step": 2049
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41595366665236716,
      "learning_rate": 3.7586282069509e-05,
      "loss": 0.6849,
      "step": 2050
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3678585139369113,
      "learning_rate": 3.7583522926395025e-05,
      "loss": 0.6957,
      "step": 2051
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4484863562950423,
      "learning_rate": 3.7580762308572226e-05,
      "loss": 0.6777,
      "step": 2052
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4323429221521488,
      "learning_rate": 3.757800021627212e-05,
      "loss": 0.716,
      "step": 2053
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3922024933581036,
      "learning_rate": 3.7575236649726364e-05,
      "loss": 0.7029,
      "step": 2054
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.48542440423873745,
      "learning_rate": 3.757247160916673e-05,
      "loss": 0.7064,
      "step": 2055
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3991188562674923,
      "learning_rate": 3.7569705094825136e-05,
      "loss": 0.7439,
      "step": 2056
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45281557871752737,
      "learning_rate": 3.756693710693358e-05,
      "loss": 0.7157,
      "step": 2057
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.378030426804588,
      "learning_rate": 3.756416764572423e-05,
      "loss": 0.6778,
      "step": 2058
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.46795706082619104,
      "learning_rate": 3.756139671142936e-05,
      "loss": 0.7047,
      "step": 2059
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39596880127901274,
      "learning_rate": 3.755862430428135e-05,
      "loss": 0.6979,
      "step": 2060
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3787185712530035,
      "learning_rate": 3.755585042451273e-05,
      "loss": 0.6931,
      "step": 2061
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42933725954653335,
      "learning_rate": 3.755307507235613e-05,
      "loss": 0.6825,
      "step": 2062
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4229065293640785,
      "learning_rate": 3.755029824804433e-05,
      "loss": 0.7,
      "step": 2063
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41889852590428317,
      "learning_rate": 3.7547519951810196e-05,
      "loss": 0.6884,
      "step": 2064
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38589847980581193,
      "learning_rate": 3.754474018388677e-05,
      "loss": 0.6913,
      "step": 2065
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43511289084716737,
      "learning_rate": 3.754195894450717e-05,
      "loss": 0.7289,
      "step": 2066
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40617772495726595,
      "learning_rate": 3.7539176233904654e-05,
      "loss": 0.6695,
      "step": 2067
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39035673554455447,
      "learning_rate": 3.753639205231261e-05,
      "loss": 0.6877,
      "step": 2068
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4083093523272005,
      "learning_rate": 3.753360639996455e-05,
      "loss": 0.6547,
      "step": 2069
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4501811263112458,
      "learning_rate": 3.75308192770941e-05,
      "loss": 0.7213,
      "step": 2070
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40429953477070657,
      "learning_rate": 3.7528030683935e-05,
      "loss": 0.6634,
      "step": 2071
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4169294406980923,
      "learning_rate": 3.752524062072113e-05,
      "loss": 0.6994,
      "step": 2072
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4332973466318051,
      "learning_rate": 3.752244908768651e-05,
      "loss": 0.7549,
      "step": 2073
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4261461276227378,
      "learning_rate": 3.7519656085065234e-05,
      "loss": 0.6509,
      "step": 2074
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38633256468918353,
      "learning_rate": 3.751686161309157e-05,
      "loss": 0.7273,
      "step": 2075
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47082317484048714,
      "learning_rate": 3.751406567199987e-05,
      "loss": 0.7281,
      "step": 2076
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40576729258760746,
      "learning_rate": 3.7511268262024635e-05,
      "loss": 0.7228,
      "step": 2077
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45827590529386375,
      "learning_rate": 3.750846938340048e-05,
      "loss": 0.7666,
      "step": 2078
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4080424942775559,
      "learning_rate": 3.750566903636214e-05,
      "loss": 0.7008,
      "step": 2079
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42960021990480624,
      "learning_rate": 3.7502867221144486e-05,
      "loss": 0.7329,
      "step": 2080
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4125028997609703,
      "learning_rate": 3.750006393798249e-05,
      "loss": 0.7143,
      "step": 2081
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.49316533635977217,
      "learning_rate": 3.749725918711127e-05,
      "loss": 0.7157,
      "step": 2082
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4390810778984298,
      "learning_rate": 3.749445296876605e-05,
      "loss": 0.6686,
      "step": 2083
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3792784604484925,
      "learning_rate": 3.749164528318219e-05,
      "loss": 0.7057,
      "step": 2084
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4204554646019221,
      "learning_rate": 3.7488836130595165e-05,
      "loss": 0.7038,
      "step": 2085
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39649882994677843,
      "learning_rate": 3.7486025511240574e-05,
      "loss": 0.6866,
      "step": 2086
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4211950606306853,
      "learning_rate": 3.7483213425354145e-05,
      "loss": 0.7187,
      "step": 2087
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34395481720509596,
      "learning_rate": 3.748039987317171e-05,
      "loss": 0.6945,
      "step": 2088
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38151917311314176,
      "learning_rate": 3.747758485492926e-05,
      "loss": 0.6855,
      "step": 2089
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42161952265044733,
      "learning_rate": 3.747476837086287e-05,
      "loss": 0.6864,
      "step": 2090
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40529270423738367,
      "learning_rate": 3.7471950421208756e-05,
      "loss": 0.7272,
      "step": 2091
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.439151619393569,
      "learning_rate": 3.746913100620327e-05,
      "loss": 0.7385,
      "step": 2092
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37000349402078037,
      "learning_rate": 3.746631012608286e-05,
      "loss": 0.649,
      "step": 2093
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34407429208900076,
      "learning_rate": 3.7463487781084116e-05,
      "loss": 0.6862,
      "step": 2094
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40990732798348034,
      "learning_rate": 3.746066397144374e-05,
      "loss": 0.6889,
      "step": 2095
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.434961748145281,
      "learning_rate": 3.745783869739857e-05,
      "loss": 0.725,
      "step": 2096
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37929775061096904,
      "learning_rate": 3.745501195918554e-05,
      "loss": 0.7211,
      "step": 2097
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41274290027717386,
      "learning_rate": 3.745218375704174e-05,
      "loss": 0.6826,
      "step": 2098
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38691762578122746,
      "learning_rate": 3.7449354091204376e-05,
      "loss": 0.6705,
      "step": 2099
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37713029930243797,
      "learning_rate": 3.744652296191075e-05,
      "loss": 0.6958,
      "step": 2100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3886744819265904,
      "learning_rate": 3.744369036939832e-05,
      "loss": 0.664,
      "step": 2101
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41058937041509164,
      "learning_rate": 3.744085631390463e-05,
      "loss": 0.6535,
      "step": 2102
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4121221774516319,
      "learning_rate": 3.7438020795667397e-05,
      "loss": 0.6912,
      "step": 2103
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4110588537119342,
      "learning_rate": 3.743518381492442e-05,
      "loss": 0.6528,
      "step": 2104
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47695937223236157,
      "learning_rate": 3.743234537191363e-05,
      "loss": 0.6844,
      "step": 2105
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3661982170707247,
      "learning_rate": 3.742950546687309e-05,
      "loss": 0.6611,
      "step": 2106
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4717309138850401,
      "learning_rate": 3.742666410004097e-05,
      "loss": 0.719,
      "step": 2107
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40744537208034265,
      "learning_rate": 3.742382127165558e-05,
      "loss": 0.6648,
      "step": 2108
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46053111619168163,
      "learning_rate": 3.742097698195534e-05,
      "loss": 0.739,
      "step": 2109
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44247011633014627,
      "learning_rate": 3.74181312311788e-05,
      "loss": 0.6951,
      "step": 2110
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4073601966178644,
      "learning_rate": 3.7415284019564635e-05,
      "loss": 0.7393,
      "step": 2111
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4855246358611801,
      "learning_rate": 3.741243534735163e-05,
      "loss": 0.7544,
      "step": 2112
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40516853987630663,
      "learning_rate": 3.740958521477869e-05,
      "loss": 0.7321,
      "step": 2113
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3856441883173399,
      "learning_rate": 3.7406733622084875e-05,
      "loss": 0.6604,
      "step": 2114
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41279029825122243,
      "learning_rate": 3.740388056950933e-05,
      "loss": 0.7023,
      "step": 2115
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.35692858139130573,
      "learning_rate": 3.7401026057291344e-05,
      "loss": 0.6895,
      "step": 2116
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.35717834711550894,
      "learning_rate": 3.739817008567031e-05,
      "loss": 0.6779,
      "step": 2117
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.61147005896562,
      "learning_rate": 3.739531265488578e-05,
      "loss": 1.6353,
      "step": 2118
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48767154628757103,
      "learning_rate": 3.739245376517737e-05,
      "loss": 0.6964,
      "step": 2119
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40676264628446157,
      "learning_rate": 3.738959341678487e-05,
      "loss": 0.6808,
      "step": 2120
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43193746549354256,
      "learning_rate": 3.738673160994817e-05,
      "loss": 0.7145,
      "step": 2121
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4433671690994566,
      "learning_rate": 3.7383868344907295e-05,
      "loss": 0.7009,
      "step": 2122
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4714029152768074,
      "learning_rate": 3.7381003621902366e-05,
      "loss": 0.6916,
      "step": 2123
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39912256999418566,
      "learning_rate": 3.737813744117366e-05,
      "loss": 0.6672,
      "step": 2124
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3659450134721109,
      "learning_rate": 3.7375269802961555e-05,
      "loss": 0.6718,
      "step": 2125
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5263471634183529,
      "learning_rate": 3.737240070750655e-05,
      "loss": 0.6996,
      "step": 2126
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.438035395427372,
      "learning_rate": 3.736953015504929e-05,
      "loss": 0.6948,
      "step": 2127
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4056160689630578,
      "learning_rate": 3.7366658145830506e-05,
      "loss": 0.6985,
      "step": 2128
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4302954966828868,
      "learning_rate": 3.7363784680091074e-05,
      "loss": 0.7421,
      "step": 2129
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48100976008688484,
      "learning_rate": 3.7360909758072e-05,
      "loss": 0.7147,
      "step": 2130
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4178165559714372,
      "learning_rate": 3.735803338001438e-05,
      "loss": 0.6653,
      "step": 2131
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41370372746084455,
      "learning_rate": 3.7355155546159466e-05,
      "loss": 0.706,
      "step": 2132
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4500199711303268,
      "learning_rate": 3.735227625674863e-05,
      "loss": 0.7082,
      "step": 2133
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48288782542700726,
      "learning_rate": 3.734939551202332e-05,
      "loss": 0.7231,
      "step": 2134
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4560365072442727,
      "learning_rate": 3.7346513312225175e-05,
      "loss": 0.7061,
      "step": 2135
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4255865798440332,
      "learning_rate": 3.73436296575959e-05,
      "loss": 0.715,
      "step": 2136
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.717182084856228,
      "learning_rate": 3.7340744548377357e-05,
      "loss": 1.4654,
      "step": 2137
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43772241526052,
      "learning_rate": 3.733785798481151e-05,
      "loss": 0.6629,
      "step": 2138
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4660593024419903,
      "learning_rate": 3.7334969967140447e-05,
      "loss": 0.7143,
      "step": 2139
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4200301386309045,
      "learning_rate": 3.733208049560639e-05,
      "loss": 0.7266,
      "step": 2140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4298900321523518,
      "learning_rate": 3.7329189570451684e-05,
      "loss": 0.6363,
      "step": 2141
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5039700079372817,
      "learning_rate": 3.732629719191876e-05,
      "loss": 0.7061,
      "step": 2142
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3883887306129039,
      "learning_rate": 3.732340336025023e-05,
      "loss": 0.6733,
      "step": 2143
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4255245458530472,
      "learning_rate": 3.732050807568878e-05,
      "loss": 0.7437,
      "step": 2144
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4574679391228096,
      "learning_rate": 3.7317611338477225e-05,
      "loss": 0.727,
      "step": 2145
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38976891235294925,
      "learning_rate": 3.731471314885853e-05,
      "loss": 0.6792,
      "step": 2146
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46227970964049064,
      "learning_rate": 3.731181350707576e-05,
      "loss": 0.6971,
      "step": 2147
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4366445835354314,
      "learning_rate": 3.7308912413372086e-05,
      "loss": 0.6885,
      "step": 2148
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44056245837894564,
      "learning_rate": 3.7306009867990836e-05,
      "loss": 0.7109,
      "step": 2149
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.452534568372585,
      "learning_rate": 3.730310587117544e-05,
      "loss": 0.7011,
      "step": 2150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45381777806945367,
      "learning_rate": 3.7300200423169453e-05,
      "loss": 0.6774,
      "step": 2151
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41247760291293706,
      "learning_rate": 3.729729352421655e-05,
      "loss": 0.6255,
      "step": 2152
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4122407496216053,
      "learning_rate": 3.729438517456052e-05,
      "loss": 0.693,
      "step": 2153
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4361014457234381,
      "learning_rate": 3.72914753744453e-05,
      "loss": 0.6939,
      "step": 2154
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4014388005019365,
      "learning_rate": 3.728856412411492e-05,
      "loss": 0.6844,
      "step": 2155
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40652720217326316,
      "learning_rate": 3.7285651423813545e-05,
      "loss": 0.6931,
      "step": 2156
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40377548521582457,
      "learning_rate": 3.7282737273785466e-05,
      "loss": 0.6715,
      "step": 2157
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4596147632919657,
      "learning_rate": 3.727982167427509e-05,
      "loss": 0.7449,
      "step": 2158
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38738343590698976,
      "learning_rate": 3.7276904625526926e-05,
      "loss": 0.6594,
      "step": 2159
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48169970756462366,
      "learning_rate": 3.7273986127785644e-05,
      "loss": 0.7364,
      "step": 2160
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3994552795336683,
      "learning_rate": 3.7271066181296e-05,
      "loss": 0.6815,
      "step": 2161
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4178210950236561,
      "learning_rate": 3.72681447863029e-05,
      "loss": 0.711,
      "step": 2162
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38806047699750273,
      "learning_rate": 3.726522194305135e-05,
      "loss": 0.6814,
      "step": 2163
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.458359146497278,
      "learning_rate": 3.726229765178648e-05,
      "loss": 0.714,
      "step": 2164
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3942095964912883,
      "learning_rate": 3.725937191275356e-05,
      "loss": 0.6871,
      "step": 2165
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4462624747463696,
      "learning_rate": 3.725644472619796e-05,
      "loss": 0.7389,
      "step": 2166
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4672351371589056,
      "learning_rate": 3.7253516092365184e-05,
      "loss": 0.7155,
      "step": 2167
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4260454098729856,
      "learning_rate": 3.725058601150085e-05,
      "loss": 0.6895,
      "step": 2168
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4679950664928605,
      "learning_rate": 3.724765448385069e-05,
      "loss": 0.7131,
      "step": 2169
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4048505144024726,
      "learning_rate": 3.72447215096606e-05,
      "loss": 0.7199,
      "step": 2170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4006091204973998,
      "learning_rate": 3.724178708917653e-05,
      "loss": 0.6903,
      "step": 2171
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3962026905275931,
      "learning_rate": 3.72388512226446e-05,
      "loss": 0.6659,
      "step": 2172
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4501267139375513,
      "learning_rate": 3.7235913910311034e-05,
      "loss": 0.7209,
      "step": 2173
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4155291158179424,
      "learning_rate": 3.723297515242219e-05,
      "loss": 0.7465,
      "step": 2174
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3968913408484223,
      "learning_rate": 3.723003494922453e-05,
      "loss": 0.7182,
      "step": 2175
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4022792505191911,
      "learning_rate": 3.7227093300964646e-05,
      "loss": 0.7196,
      "step": 2176
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45059388619513996,
      "learning_rate": 3.722415020788925e-05,
      "loss": 0.7442,
      "step": 2177
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38860076554357587,
      "learning_rate": 3.7221205670245186e-05,
      "loss": 0.6758,
      "step": 2178
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43661628779749,
      "learning_rate": 3.721825968827939e-05,
      "loss": 0.7411,
      "step": 2179
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3892303853081912,
      "learning_rate": 3.7215312262238956e-05,
      "loss": 0.6842,
      "step": 2180
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.432149214061412,
      "learning_rate": 3.7212363392371074e-05,
      "loss": 0.7058,
      "step": 2181
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42746568598783774,
      "learning_rate": 3.720941307892305e-05,
      "loss": 0.6971,
      "step": 2182
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38846109101244775,
      "learning_rate": 3.720646132214235e-05,
      "loss": 0.7115,
      "step": 2183
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37696341234803,
      "learning_rate": 3.7203508122276517e-05,
      "loss": 0.7,
      "step": 2184
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4485779310024115,
      "learning_rate": 3.7200553479573225e-05,
      "loss": 0.6986,
      "step": 2185
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3656441805258742,
      "learning_rate": 3.7197597394280294e-05,
      "loss": 0.6825,
      "step": 2186
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3828918613445849,
      "learning_rate": 3.719463986664563e-05,
      "loss": 0.6758,
      "step": 2187
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3724219504538448,
      "learning_rate": 3.7191680896917296e-05,
      "loss": 0.6462,
      "step": 2188
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.392971508753026,
      "learning_rate": 3.718872048534344e-05,
      "loss": 0.7109,
      "step": 2189
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.2129221088031845,
      "learning_rate": 3.7185758632172365e-05,
      "loss": 1.1851,
      "step": 2190
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4337229549621782,
      "learning_rate": 3.718279533765246e-05,
      "loss": 0.7325,
      "step": 2191
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3790291793589371,
      "learning_rate": 3.717983060203226e-05,
      "loss": 0.6777,
      "step": 2192
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4007423603566881,
      "learning_rate": 3.7176864425560426e-05,
      "loss": 0.6793,
      "step": 2193
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3816979263743634,
      "learning_rate": 3.717389680848571e-05,
      "loss": 0.6866,
      "step": 2194
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4242224044623732,
      "learning_rate": 3.717092775105701e-05,
      "loss": 0.6932,
      "step": 2195
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3993104806242072,
      "learning_rate": 3.7167957253523334e-05,
      "loss": 0.7117,
      "step": 2196
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46426162911370705,
      "learning_rate": 3.716498531613382e-05,
      "loss": 0.7497,
      "step": 2197
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38680468164717496,
      "learning_rate": 3.7162011939137715e-05,
      "loss": 0.6583,
      "step": 2198
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4489605575381144,
      "learning_rate": 3.71590371227844e-05,
      "loss": 0.7385,
      "step": 2199
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.47288733541980754,
      "learning_rate": 3.7156060867323356e-05,
      "loss": 0.7106,
      "step": 2200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46305398284285104,
      "learning_rate": 3.715308317300421e-05,
      "loss": 0.754,
      "step": 2201
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4159701333572372,
      "learning_rate": 3.71501040400767e-05,
      "loss": 0.6998,
      "step": 2202
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.350278918283499,
      "learning_rate": 3.7147123468790674e-05,
      "loss": 1.0491,
      "step": 2203
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4132027741969631,
      "learning_rate": 3.7144141459396106e-05,
      "loss": 0.7111,
      "step": 2204
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40675120439877266,
      "learning_rate": 3.7141158012143106e-05,
      "loss": 0.6998,
      "step": 2205
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38503201631089506,
      "learning_rate": 3.7138173127281875e-05,
      "loss": 0.674,
      "step": 2206
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4003268465923423,
      "learning_rate": 3.713518680506277e-05,
      "loss": 0.6801,
      "step": 2207
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.45351490210521533,
      "learning_rate": 3.7132199045736236e-05,
      "loss": 0.7103,
      "step": 2208
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43509408691876367,
      "learning_rate": 3.7129209849552865e-05,
      "loss": 0.7583,
      "step": 2209
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44015006631098524,
      "learning_rate": 3.712621921676335e-05,
      "loss": 0.7136,
      "step": 2210
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4300349063401402,
      "learning_rate": 3.712322714761851e-05,
      "loss": 0.6589,
      "step": 2211
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42438984207827884,
      "learning_rate": 3.712023364236928e-05,
      "loss": 0.7456,
      "step": 2212
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43292536900004736,
      "learning_rate": 3.7117238701266744e-05,
      "loss": 0.6428,
      "step": 2213
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4106594389867036,
      "learning_rate": 3.7114242324562066e-05,
      "loss": 0.6659,
      "step": 2214
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.475003268538812,
      "learning_rate": 3.711124451250655e-05,
      "loss": 0.7153,
      "step": 2215
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39255088521470166,
      "learning_rate": 3.710824526535163e-05,
      "loss": 0.6776,
      "step": 2216
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3967394183332808,
      "learning_rate": 3.710524458334883e-05,
      "loss": 0.6871,
      "step": 2217
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38140694739192293,
      "learning_rate": 3.710224246674983e-05,
      "loss": 0.7014,
      "step": 2218
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39627356664052216,
      "learning_rate": 3.7099238915806415e-05,
      "loss": 0.6681,
      "step": 2219
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40813835458897546,
      "learning_rate": 3.709623393077047e-05,
      "loss": 0.6883,
      "step": 2220
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4532117628944013,
      "learning_rate": 3.709322751189404e-05,
      "loss": 0.7209,
      "step": 2221
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5203790772564676,
      "learning_rate": 3.709021965942926e-05,
      "loss": 0.965,
      "step": 2222
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4173082728261861,
      "learning_rate": 3.70872103736284e-05,
      "loss": 0.6775,
      "step": 2223
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37150739451619885,
      "learning_rate": 3.708419965474384e-05,
      "loss": 0.6562,
      "step": 2224
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4370158806123767,
      "learning_rate": 3.7081187503028084e-05,
      "loss": 0.7152,
      "step": 2225
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4264928556693015,
      "learning_rate": 3.7078173918733764e-05,
      "loss": 0.6779,
      "step": 2226
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5105660899182518,
      "learning_rate": 3.707515890211363e-05,
      "loss": 0.6795,
      "step": 2227
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.118171860631477,
      "learning_rate": 3.707214245342053e-05,
      "loss": 0.8472,
      "step": 2228
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3935013958403679,
      "learning_rate": 3.706912457290746e-05,
      "loss": 0.6914,
      "step": 2229
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41886714327995456,
      "learning_rate": 3.7066105260827534e-05,
      "loss": 0.6809,
      "step": 2230
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4015814900953008,
      "learning_rate": 3.7063084517433966e-05,
      "loss": 0.6563,
      "step": 2231
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35916275276640186,
      "learning_rate": 3.706006234298011e-05,
      "loss": 0.7031,
      "step": 2232
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48289178766318064,
      "learning_rate": 3.705703873771942e-05,
      "loss": 0.7477,
      "step": 2233
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42316893363460206,
      "learning_rate": 3.70540137019055e-05,
      "loss": 0.6932,
      "step": 2234
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.396226151337796,
      "learning_rate": 3.7050987235792044e-05,
      "loss": 0.726,
      "step": 2235
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4721743282338376,
      "learning_rate": 3.704795933963288e-05,
      "loss": 0.703,
      "step": 2236
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5203619313698302,
      "learning_rate": 3.704493001368195e-05,
      "loss": 0.7344,
      "step": 2237
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4161591372387634,
      "learning_rate": 3.704189925819333e-05,
      "loss": 0.7309,
      "step": 2238
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39629218057695204,
      "learning_rate": 3.70388670734212e-05,
      "loss": 0.6925,
      "step": 2239
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41370212443062127,
      "learning_rate": 3.703583345961986e-05,
      "loss": 0.7209,
      "step": 2240
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4000988382699026,
      "learning_rate": 3.703279841704376e-05,
      "loss": 0.6546,
      "step": 2241
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43125655913582794,
      "learning_rate": 3.702976194594741e-05,
      "loss": 0.7141,
      "step": 2242
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3750228180337602,
      "learning_rate": 3.7026724046585506e-05,
      "loss": 0.6718,
      "step": 2243
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4308470998279503,
      "learning_rate": 3.702368471921281e-05,
      "loss": 0.7438,
      "step": 2244
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4540812809695273,
      "learning_rate": 3.7020643964084234e-05,
      "loss": 0.7181,
      "step": 2245
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.495200796965306,
      "learning_rate": 3.701760178145481e-05,
      "loss": 0.9926,
      "step": 2246
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46885141399643226,
      "learning_rate": 3.701455817157967e-05,
      "loss": 0.7459,
      "step": 2247
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.413586695503774,
      "learning_rate": 3.70115131347141e-05,
      "loss": 0.6986,
      "step": 2248
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4239948745818163,
      "learning_rate": 3.7008466671113444e-05,
      "loss": 0.7188,
      "step": 2249
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4144769908881029,
      "learning_rate": 3.700541878103325e-05,
      "loss": 0.6952,
      "step": 2250
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44278206933064346,
      "learning_rate": 3.700236946472911e-05,
      "loss": 0.6963,
      "step": 2251
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39756700242649445,
      "learning_rate": 3.699931872245678e-05,
      "loss": 0.6754,
      "step": 2252
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4278917122919504,
      "learning_rate": 3.6996266554472115e-05,
      "loss": 0.6817,
      "step": 2253
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48100624397045644,
      "learning_rate": 3.69932129610311e-05,
      "loss": 0.7171,
      "step": 2254
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4360203160855476,
      "learning_rate": 3.699015794238984e-05,
      "loss": 0.6952,
      "step": 2255
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38297410755891403,
      "learning_rate": 3.6987101498804556e-05,
      "loss": 0.6553,
      "step": 2256
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39769031612953576,
      "learning_rate": 3.698404363053158e-05,
      "loss": 0.6881,
      "step": 2257
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3677042835277016,
      "learning_rate": 3.698098433782738e-05,
      "loss": 0.6901,
      "step": 2258
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.5175899776546506,
      "learning_rate": 3.697792362094853e-05,
      "loss": 1.011,
      "step": 2259
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44647311274910556,
      "learning_rate": 3.697486148015172e-05,
      "loss": 0.7354,
      "step": 2260
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4137067975450348,
      "learning_rate": 3.697179791569379e-05,
      "loss": 0.6706,
      "step": 2261
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3951292099873708,
      "learning_rate": 3.696873292783166e-05,
      "loss": 0.6781,
      "step": 2262
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4220876548176611,
      "learning_rate": 3.6965666516822395e-05,
      "loss": 0.7325,
      "step": 2263
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3946515430568121,
      "learning_rate": 3.696259868292317e-05,
      "loss": 0.6714,
      "step": 2264
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38713954297199105,
      "learning_rate": 3.695952942639128e-05,
      "loss": 0.6967,
      "step": 2265
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4292388782724411,
      "learning_rate": 3.6956458747484136e-05,
      "loss": 0.7012,
      "step": 2266
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40328713995064475,
      "learning_rate": 3.695338664645927e-05,
      "loss": 0.6974,
      "step": 2267
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4376377863794046,
      "learning_rate": 3.695031312357435e-05,
      "loss": 0.7291,
      "step": 2268
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39366719562036884,
      "learning_rate": 3.6947238179087134e-05,
      "loss": 0.6739,
      "step": 2269
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39591606313292943,
      "learning_rate": 3.6944161813255514e-05,
      "loss": 0.7211,
      "step": 2270
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4270040284213737,
      "learning_rate": 3.6941084026337514e-05,
      "loss": 0.6914,
      "step": 2271
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.47608928119418314,
      "learning_rate": 3.693800481859126e-05,
      "loss": 0.756,
      "step": 2272
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3945671862909401,
      "learning_rate": 3.693492419027499e-05,
      "loss": 0.7076,
      "step": 2273
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40058787823179953,
      "learning_rate": 3.693184214164708e-05,
      "loss": 0.7053,
      "step": 2274
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4033347092529326,
      "learning_rate": 3.692875867296602e-05,
      "loss": 0.7193,
      "step": 2275
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4022324032498003,
      "learning_rate": 3.692567378449042e-05,
      "loss": 0.7115,
      "step": 2276
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3969071708012514,
      "learning_rate": 3.692258747647899e-05,
      "loss": 0.7193,
      "step": 2277
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9975338831140523,
      "learning_rate": 3.6919499749190594e-05,
      "loss": 0.9562,
      "step": 2278
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42856856880469857,
      "learning_rate": 3.691641060288419e-05,
      "loss": 0.7046,
      "step": 2279
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4140678862022238,
      "learning_rate": 3.691332003781885e-05,
      "loss": 0.6768,
      "step": 2280
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41532939865279667,
      "learning_rate": 3.691022805425378e-05,
      "loss": 0.7466,
      "step": 2281
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.396816132980489,
      "learning_rate": 3.690713465244832e-05,
      "loss": 0.6945,
      "step": 2282
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4162078809823484,
      "learning_rate": 3.690403983266189e-05,
      "loss": 0.6782,
      "step": 2283
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41756017092044545,
      "learning_rate": 3.6900943595154055e-05,
      "loss": 0.7237,
      "step": 2284
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39140567062247245,
      "learning_rate": 3.689784594018449e-05,
      "loss": 0.6741,
      "step": 2285
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37959550861412095,
      "learning_rate": 3.6894746868012995e-05,
      "loss": 0.7284,
      "step": 2286
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3689793850398602,
      "learning_rate": 3.6891646378899484e-05,
      "loss": 0.6738,
      "step": 2287
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4072317917957591,
      "learning_rate": 3.688854447310398e-05,
      "loss": 0.7137,
      "step": 2288
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40637849547574456,
      "learning_rate": 3.6885441150886664e-05,
      "loss": 0.7006,
      "step": 2289
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3946693567829432,
      "learning_rate": 3.6882336412507786e-05,
      "loss": 0.6318,
      "step": 2290
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.468204128531039,
      "learning_rate": 3.687923025822774e-05,
      "loss": 0.7525,
      "step": 2291
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5270074958686872,
      "learning_rate": 3.6876122688307034e-05,
      "loss": 0.7944,
      "step": 2292
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.45183053962527087,
      "learning_rate": 3.6873013703006304e-05,
      "loss": 0.7197,
      "step": 2293
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.529020293068388,
      "learning_rate": 3.6869903302586283e-05,
      "loss": 0.8783,
      "step": 2294
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44241645226418086,
      "learning_rate": 3.686679148730786e-05,
      "loss": 0.6732,
      "step": 2295
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3532282265354707,
      "learning_rate": 3.6863678257432e-05,
      "loss": 0.681,
      "step": 2296
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3762706084874086,
      "learning_rate": 3.6860563613219805e-05,
      "loss": 0.6666,
      "step": 2297
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4616009940416087,
      "learning_rate": 3.68574475549325e-05,
      "loss": 0.7226,
      "step": 2298
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46707577620665647,
      "learning_rate": 3.685433008283144e-05,
      "loss": 0.7293,
      "step": 2299
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3991712677340374,
      "learning_rate": 3.685121119717806e-05,
      "loss": 0.7083,
      "step": 2300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42376155708292457,
      "learning_rate": 3.684809089823396e-05,
      "loss": 0.7546,
      "step": 2301
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4227931152498334,
      "learning_rate": 3.684496918626081e-05,
      "loss": 0.6845,
      "step": 2302
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3982970825944915,
      "learning_rate": 3.6841846061520446e-05,
      "loss": 0.6942,
      "step": 2303
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4235872940286285,
      "learning_rate": 3.683872152427479e-05,
      "loss": 0.7325,
      "step": 2304
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.304894797331049,
      "learning_rate": 3.6835595574785895e-05,
      "loss": 0.8072,
      "step": 2305
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3632926499034497,
      "learning_rate": 3.6832468213315926e-05,
      "loss": 0.7359,
      "step": 2306
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4399647705793527,
      "learning_rate": 3.6829339440127186e-05,
      "loss": 0.6751,
      "step": 2307
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.418562423728413,
      "learning_rate": 3.682620925548207e-05,
      "loss": 0.6519,
      "step": 2308
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43879177069139064,
      "learning_rate": 3.68230776596431e-05,
      "loss": 0.6861,
      "step": 2309
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4733190179448224,
      "learning_rate": 3.681994465287293e-05,
      "loss": 0.7468,
      "step": 2310
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4444367607025674,
      "learning_rate": 3.6816810235434316e-05,
      "loss": 0.7037,
      "step": 2311
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.966707969798519,
      "learning_rate": 3.681367440759013e-05,
      "loss": 0.7438,
      "step": 2312
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43885057543531236,
      "learning_rate": 3.681053716960338e-05,
      "loss": 0.6835,
      "step": 2313
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40580034254917086,
      "learning_rate": 3.680739852173718e-05,
      "loss": 0.6794,
      "step": 2314
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3856779499582733,
      "learning_rate": 3.680425846425477e-05,
      "loss": 0.6496,
      "step": 2315
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5036466834977043,
      "learning_rate": 3.6801116997419486e-05,
      "loss": 0.7019,
      "step": 2316
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42830993493603015,
      "learning_rate": 3.679797412149482e-05,
      "loss": 0.6759,
      "step": 2317
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37376832981248986,
      "learning_rate": 3.6794829836744344e-05,
      "loss": 0.6885,
      "step": 2318
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4733031507035349,
      "learning_rate": 3.6791684143431774e-05,
      "loss": 0.7467,
      "step": 2319
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4012390849619354,
      "learning_rate": 3.6788537041820934e-05,
      "loss": 0.6372,
      "step": 2320
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.388357898985907,
      "learning_rate": 3.6785388532175773e-05,
      "loss": 0.669,
      "step": 2321
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4222577206993855,
      "learning_rate": 3.6782238614760344e-05,
      "loss": 0.7182,
      "step": 2322
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4662009615455932,
      "learning_rate": 3.677908728983883e-05,
      "loss": 0.7117,
      "step": 2323
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4473823034330259,
      "learning_rate": 3.677593455767553e-05,
      "loss": 0.7316,
      "step": 2324
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41064772693133067,
      "learning_rate": 3.6772780418534856e-05,
      "loss": 0.7122,
      "step": 2325
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.448704465538236,
      "learning_rate": 3.676962487268134e-05,
      "loss": 0.6971,
      "step": 2326
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36837230400457616,
      "learning_rate": 3.676646792037965e-05,
      "loss": 0.6711,
      "step": 2327
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3813940124240374,
      "learning_rate": 3.6763309561894544e-05,
      "loss": 0.6272,
      "step": 2328
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4128009467370319,
      "learning_rate": 3.67601497974909e-05,
      "loss": 0.6844,
      "step": 2329
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40350159569887706,
      "learning_rate": 3.675698862743374e-05,
      "loss": 0.6939,
      "step": 2330
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41133141577489085,
      "learning_rate": 3.675382605198818e-05,
      "loss": 0.7108,
      "step": 2331
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44069055991575323,
      "learning_rate": 3.675066207141946e-05,
      "loss": 0.7083,
      "step": 2332
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3841256592412094,
      "learning_rate": 3.6747496685992945e-05,
      "loss": 0.6992,
      "step": 2333
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46920408941014746,
      "learning_rate": 3.674432989597411e-05,
      "loss": 0.6808,
      "step": 2334
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3676395142306903,
      "learning_rate": 3.674116170162855e-05,
      "loss": 0.6794,
      "step": 2335
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3991310035106075,
      "learning_rate": 3.6737992103221974e-05,
      "loss": 0.665,
      "step": 2336
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4002972335878382,
      "learning_rate": 3.673482110102021e-05,
      "loss": 0.6815,
      "step": 2337
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4004877365535815,
      "learning_rate": 3.6731648695289214e-05,
      "loss": 0.7134,
      "step": 2338
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41391494530049255,
      "learning_rate": 3.6728474886295053e-05,
      "loss": 0.6754,
      "step": 2339
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44202549350521414,
      "learning_rate": 3.67252996743039e-05,
      "loss": 0.7044,
      "step": 2340
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40026626537434273,
      "learning_rate": 3.672212305958206e-05,
      "loss": 0.6814,
      "step": 2341
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42064533891419903,
      "learning_rate": 3.671894504239596e-05,
      "loss": 0.7093,
      "step": 2342
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4170617061169785,
      "learning_rate": 3.671576562301213e-05,
      "loss": 0.6774,
      "step": 2343
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43716286148069605,
      "learning_rate": 3.6712584801697224e-05,
      "loss": 0.7402,
      "step": 2344
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44004183541391056,
      "learning_rate": 3.670940257871801e-05,
      "loss": 0.7042,
      "step": 2345
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37706218217430176,
      "learning_rate": 3.670621895434139e-05,
      "loss": 0.7096,
      "step": 2346
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3688775746214641,
      "learning_rate": 3.670303392883436e-05,
      "loss": 0.6862,
      "step": 2347
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3954285303198304,
      "learning_rate": 3.669984750246404e-05,
      "loss": 0.6716,
      "step": 2348
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4196807699958916,
      "learning_rate": 3.669665967549769e-05,
      "loss": 0.7057,
      "step": 2349
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4733810899696256,
      "learning_rate": 3.6693470448202654e-05,
      "loss": 0.6811,
      "step": 2350
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41582111395820004,
      "learning_rate": 3.66902798208464e-05,
      "loss": 0.7088,
      "step": 2351
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41715961068563073,
      "learning_rate": 3.668708779369655e-05,
      "loss": 0.683,
      "step": 2352
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3765666609658116,
      "learning_rate": 3.6683894367020785e-05,
      "loss": 0.6675,
      "step": 2353
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4681196480383615,
      "learning_rate": 3.6680699541086964e-05,
      "loss": 0.7051,
      "step": 2354
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4167558270050878,
      "learning_rate": 3.667750331616301e-05,
      "loss": 0.6825,
      "step": 2355
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4293881889531413,
      "learning_rate": 3.667430569251699e-05,
      "loss": 0.7228,
      "step": 2356
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4237347210491897,
      "learning_rate": 3.6671106670417096e-05,
      "loss": 0.6714,
      "step": 2357
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4688089957019511,
      "learning_rate": 3.666790625013161e-05,
      "loss": 0.7222,
      "step": 2358
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4368351888119853,
      "learning_rate": 3.666470443192897e-05,
      "loss": 0.6588,
      "step": 2359
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.412314269243427,
      "learning_rate": 3.666150121607769e-05,
      "loss": 0.718,
      "step": 2360
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4031284022693454,
      "learning_rate": 3.665829660284642e-05,
      "loss": 0.645,
      "step": 2361
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38488513187974804,
      "learning_rate": 3.665509059250394e-05,
      "loss": 0.7193,
      "step": 2362
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37394383290143246,
      "learning_rate": 3.6651883185319114e-05,
      "loss": 0.6875,
      "step": 2363
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41768703478654035,
      "learning_rate": 3.664867438156096e-05,
      "loss": 0.7032,
      "step": 2364
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42813255048763005,
      "learning_rate": 3.6645464181498596e-05,
      "loss": 0.6555,
      "step": 2365
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46717439041744196,
      "learning_rate": 3.6642252585401257e-05,
      "loss": 0.7058,
      "step": 2366
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43483564471002417,
      "learning_rate": 3.6639039593538286e-05,
      "loss": 0.7321,
      "step": 2367
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4172066760295946,
      "learning_rate": 3.6635825206179164e-05,
      "loss": 0.7128,
      "step": 2368
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.35901360343458966,
      "learning_rate": 3.663260942359347e-05,
      "loss": 0.6837,
      "step": 2369
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3713456629709821,
      "learning_rate": 3.662939224605091e-05,
      "loss": 0.7124,
      "step": 2370
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4406386620289006,
      "learning_rate": 3.66261736738213e-05,
      "loss": 0.697,
      "step": 2371
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.379255185935842,
      "learning_rate": 3.66229537071746e-05,
      "loss": 0.7179,
      "step": 2372
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3691933512326881,
      "learning_rate": 3.6619732346380834e-05,
      "loss": 0.6423,
      "step": 2373
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4394004258466504,
      "learning_rate": 3.6616509591710195e-05,
      "loss": 0.7362,
      "step": 2374
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3491019495731975,
      "learning_rate": 3.661328544343295e-05,
      "loss": 0.676,
      "step": 2375
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37778366212882464,
      "learning_rate": 3.661005990181954e-05,
      "loss": 0.6627,
      "step": 2376
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.39855699564235786,
      "learning_rate": 3.6606832967140455e-05,
      "loss": 0.6916,
      "step": 2377
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3761501662240675,
      "learning_rate": 3.660360463966634e-05,
      "loss": 0.6938,
      "step": 2378
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4420231166751253,
      "learning_rate": 3.6600374919667966e-05,
      "loss": 0.6948,
      "step": 2379
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4243723952810464,
      "learning_rate": 3.6597143807416186e-05,
      "loss": 0.6469,
      "step": 2380
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.39302609683011375,
      "learning_rate": 3.6593911303182005e-05,
      "loss": 0.7135,
      "step": 2381
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3677902864836887,
      "learning_rate": 3.659067740723652e-05,
      "loss": 0.6684,
      "step": 2382
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3647542243582995,
      "learning_rate": 3.658744211985096e-05,
      "loss": 0.6674,
      "step": 2383
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.404933522410332,
      "learning_rate": 3.6584205441296666e-05,
      "loss": 0.6912,
      "step": 2384
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3922429588409425,
      "learning_rate": 3.658096737184508e-05,
      "loss": 0.6796,
      "step": 2385
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4538782184073001,
      "learning_rate": 3.6577727911767794e-05,
      "loss": 0.6899,
      "step": 2386
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4351346997672204,
      "learning_rate": 3.6574487061336486e-05,
      "loss": 0.7006,
      "step": 2387
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.414731982333409,
      "learning_rate": 3.657124482082297e-05,
      "loss": 0.6831,
      "step": 2388
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3878276099820932,
      "learning_rate": 3.6568001190499156e-05,
      "loss": 0.6791,
      "step": 2389
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41714680288138234,
      "learning_rate": 3.656475617063709e-05,
      "loss": 0.693,
      "step": 2390
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40348673096316545,
      "learning_rate": 3.656150976150894e-05,
      "loss": 0.6684,
      "step": 2391
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4054998477135224,
      "learning_rate": 3.655826196338696e-05,
      "loss": 0.6475,
      "step": 2392
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3858236847314057,
      "learning_rate": 3.655501277654355e-05,
      "loss": 0.7115,
      "step": 2393
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4453466267864611,
      "learning_rate": 3.65517622012512e-05,
      "loss": 0.7213,
      "step": 2394
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3971563650573086,
      "learning_rate": 3.6548510237782557e-05,
      "loss": 0.7202,
      "step": 2395
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4035052986097353,
      "learning_rate": 3.654525688641034e-05,
      "loss": 0.6797,
      "step": 2396
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3515983862353754,
      "learning_rate": 3.654200214740741e-05,
      "loss": 0.664,
      "step": 2397
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38888029702831695,
      "learning_rate": 3.653874602104674e-05,
      "loss": 0.6915,
      "step": 2398
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43528684790127087,
      "learning_rate": 3.653548850760142e-05,
      "loss": 0.6976,
      "step": 2399
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3979676957421543,
      "learning_rate": 3.653222960734464e-05,
      "loss": 0.715,
      "step": 2400
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4420344863653906,
      "learning_rate": 3.652896932054973e-05,
      "loss": 0.7136,
      "step": 2401
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4161226501888408,
      "learning_rate": 3.6525707647490126e-05,
      "loss": 0.6943,
      "step": 2402
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34522317569470906,
      "learning_rate": 3.652244458843938e-05,
      "loss": 0.6375,
      "step": 2403
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40303388623417247,
      "learning_rate": 3.651918014367116e-05,
      "loss": 0.7529,
      "step": 2404
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46264289988096,
      "learning_rate": 3.651591431345925e-05,
      "loss": 0.7201,
      "step": 2405
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9846915643851026,
      "learning_rate": 3.6512647098077554e-05,
      "loss": 0.7566,
      "step": 2406
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3834461200437024,
      "learning_rate": 3.650937849780009e-05,
      "loss": 0.6839,
      "step": 2407
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3767018297656274,
      "learning_rate": 3.650610851290099e-05,
      "loss": 0.6795,
      "step": 2408
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3749167631113613,
      "learning_rate": 3.6502837143654504e-05,
      "loss": 0.6942,
      "step": 2409
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4050730801001741,
      "learning_rate": 3.6499564390334995e-05,
      "loss": 0.6851,
      "step": 2410
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4141276439030176,
      "learning_rate": 3.6496290253216946e-05,
      "loss": 0.7098,
      "step": 2411
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4375792979488752,
      "learning_rate": 3.649301473257495e-05,
      "loss": 0.7911,
      "step": 2412
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.39727641270944714,
      "learning_rate": 3.648973782868374e-05,
      "loss": 0.6696,
      "step": 2413
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43540753395536225,
      "learning_rate": 3.648645954181812e-05,
      "loss": 0.7042,
      "step": 2414
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37479051539528924,
      "learning_rate": 3.648317987225306e-05,
      "loss": 0.6601,
      "step": 2415
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38713966898454183,
      "learning_rate": 3.647989882026361e-05,
      "loss": 0.672,
      "step": 2416
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43889940645047737,
      "learning_rate": 3.647661638612494e-05,
      "loss": 0.694,
      "step": 2417
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4599579604244912,
      "learning_rate": 3.647333257011236e-05,
      "loss": 0.7299,
      "step": 2418
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37002445865030775,
      "learning_rate": 3.647004737250127e-05,
      "loss": 0.6731,
      "step": 2419
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4489504838918971,
      "learning_rate": 3.646676079356719e-05,
      "loss": 0.6745,
      "step": 2420
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0272487587450816,
      "learning_rate": 3.646347283358578e-05,
      "loss": 0.6624,
      "step": 2421
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41734790617997136,
      "learning_rate": 3.646018349283278e-05,
      "loss": 0.7302,
      "step": 2422
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3756230877948856,
      "learning_rate": 3.6456892771584075e-05,
      "loss": 0.6982,
      "step": 2423
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43585103350648446,
      "learning_rate": 3.645360067011564e-05,
      "loss": 0.726,
      "step": 2424
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3848466235701773,
      "learning_rate": 3.645030718870359e-05,
      "loss": 0.6863,
      "step": 2425
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4312950312112905,
      "learning_rate": 3.6447012327624156e-05,
      "loss": 0.6749,
      "step": 2426
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3797124982600244,
      "learning_rate": 3.644371608715364e-05,
      "loss": 0.6745,
      "step": 2427
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4565313471201158,
      "learning_rate": 3.644041846756853e-05,
      "loss": 0.7176,
      "step": 2428
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4333057159920741,
      "learning_rate": 3.6437119469145375e-05,
      "loss": 0.7134,
      "step": 2429
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3567329538950361,
      "learning_rate": 3.643381909216085e-05,
      "loss": 0.6947,
      "step": 2430
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.39900205287170404,
      "learning_rate": 3.6430517336891786e-05,
      "loss": 0.694,
      "step": 2431
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33582969735129137,
      "learning_rate": 3.642721420361506e-05,
      "loss": 0.6762,
      "step": 2432
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44508313431907987,
      "learning_rate": 3.642390969260772e-05,
      "loss": 0.7451,
      "step": 2433
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3780050164447599,
      "learning_rate": 3.642060380414691e-05,
      "loss": 0.7061,
      "step": 2434
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4102954473096131,
      "learning_rate": 3.6417296538509894e-05,
      "loss": 0.6827,
      "step": 2435
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42688208910764264,
      "learning_rate": 3.641398789597404e-05,
      "loss": 0.6703,
      "step": 2436
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4025884242340759,
      "learning_rate": 3.641067787681685e-05,
      "loss": 0.6816,
      "step": 2437
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4230879059836441,
      "learning_rate": 3.6407366481315915e-05,
      "loss": 0.721,
      "step": 2438
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37618057996485804,
      "learning_rate": 3.6404053709748975e-05,
      "loss": 0.6863,
      "step": 2439
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38326854152688977,
      "learning_rate": 3.6400739562393867e-05,
      "loss": 0.6481,
      "step": 2440
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42211005450017347,
      "learning_rate": 3.639742403952853e-05,
      "loss": 0.7037,
      "step": 2441
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.36710070135170253,
      "learning_rate": 3.6394107141431046e-05,
      "loss": 0.6798,
      "step": 2442
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4169745590699899,
      "learning_rate": 3.639078886837959e-05,
      "loss": 0.7057,
      "step": 2443
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42579246414495436,
      "learning_rate": 3.638746922065248e-05,
      "loss": 0.7245,
      "step": 2444
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41662078689322257,
      "learning_rate": 3.6384148198528105e-05,
      "loss": 0.6845,
      "step": 2445
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38454580071097344,
      "learning_rate": 3.638082580228501e-05,
      "loss": 0.7194,
      "step": 2446
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42027295959049077,
      "learning_rate": 3.6377502032201835e-05,
      "loss": 0.6888,
      "step": 2447
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3902070120882539,
      "learning_rate": 3.637417688855735e-05,
      "loss": 0.7137,
      "step": 2448
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4053288941828581,
      "learning_rate": 3.6370850371630424e-05,
      "loss": 0.7234,
      "step": 2449
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4086742658992696,
      "learning_rate": 3.636752248170004e-05,
      "loss": 0.708,
      "step": 2450
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3462121740291441,
      "learning_rate": 3.636419321904532e-05,
      "loss": 0.6639,
      "step": 2451
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3696848833917525,
      "learning_rate": 3.636086258394548e-05,
      "loss": 0.6894,
      "step": 2452
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33425768619045676,
      "learning_rate": 3.6357530576679845e-05,
      "loss": 0.6386,
      "step": 2453
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.39705854956699993,
      "learning_rate": 3.6354197197527884e-05,
      "loss": 0.6926,
      "step": 2454
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4151457865874263,
      "learning_rate": 3.6350862446769155e-05,
      "loss": 0.6877,
      "step": 2455
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5116638006493393,
      "learning_rate": 3.6347526324683336e-05,
      "loss": 0.6734,
      "step": 2456
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40913505739014344,
      "learning_rate": 3.6344188831550226e-05,
      "loss": 0.7089,
      "step": 2457
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4085329228631364,
      "learning_rate": 3.634084996764974e-05,
      "loss": 0.7122,
      "step": 2458
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40935027234034177,
      "learning_rate": 3.6337509733261894e-05,
      "loss": 0.7072,
      "step": 2459
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4473711829492437,
      "learning_rate": 3.633416812866685e-05,
      "loss": 0.652,
      "step": 2460
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5006469442760151,
      "learning_rate": 3.633082515414484e-05,
      "loss": 0.6992,
      "step": 2461
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4259504728224732,
      "learning_rate": 3.632748080997625e-05,
      "loss": 0.7174,
      "step": 2462
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4151824189721333,
      "learning_rate": 3.632413509644157e-05,
      "loss": 0.6813,
      "step": 2463
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38653560934359044,
      "learning_rate": 3.632078801382139e-05,
      "loss": 0.6712,
      "step": 2464
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40330280206390356,
      "learning_rate": 3.631743956239643e-05,
      "loss": 0.6626,
      "step": 2465
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.45180838217204955,
      "learning_rate": 3.631408974244752e-05,
      "loss": 0.7183,
      "step": 2466
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46364517878162126,
      "learning_rate": 3.6310738554255604e-05,
      "loss": 0.7171,
      "step": 2467
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4413409136577625,
      "learning_rate": 3.6307385998101744e-05,
      "loss": 0.7023,
      "step": 2468
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4010421062196441,
      "learning_rate": 3.630403207426712e-05,
      "loss": 0.7062,
      "step": 2469
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3736024285709433,
      "learning_rate": 3.630067678303301e-05,
      "loss": 0.6884,
      "step": 2470
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3836635258186161,
      "learning_rate": 3.629732012468083e-05,
      "loss": 0.683,
      "step": 2471
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3763041515114877,
      "learning_rate": 3.629396209949208e-05,
      "loss": 0.6904,
      "step": 2472
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40395359831975497,
      "learning_rate": 3.629060270774842e-05,
      "loss": 0.6904,
      "step": 2473
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0205445354020584,
      "learning_rate": 3.6287241949731585e-05,
      "loss": 0.6761,
      "step": 2474
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4196868590545136,
      "learning_rate": 3.628387982572343e-05,
      "loss": 0.6776,
      "step": 2475
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42431043479047864,
      "learning_rate": 3.6280516336005945e-05,
      "loss": 0.657,
      "step": 2476
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4507758398699071,
      "learning_rate": 3.627715148086122e-05,
      "loss": 0.7204,
      "step": 2477
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3949217537746534,
      "learning_rate": 3.6273785260571455e-05,
      "loss": 0.6824,
      "step": 2478
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4212910626121072,
      "learning_rate": 3.6270417675418974e-05,
      "loss": 0.6937,
      "step": 2479
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.36856318766188395,
      "learning_rate": 3.626704872568621e-05,
      "loss": 0.7,
      "step": 2480
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.317375288911053,
      "learning_rate": 3.626367841165571e-05,
      "loss": 0.6124,
      "step": 2481
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40561776013351813,
      "learning_rate": 3.626030673361015e-05,
      "loss": 0.6793,
      "step": 2482
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43077664729554627,
      "learning_rate": 3.625693369183231e-05,
      "loss": 0.6886,
      "step": 2483
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3829523519384761,
      "learning_rate": 3.625355928660506e-05,
      "loss": 0.6509,
      "step": 2484
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4080859884040615,
      "learning_rate": 3.625018351821143e-05,
      "loss": 0.7006,
      "step": 2485
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3644339689676114,
      "learning_rate": 3.624680638693453e-05,
      "loss": 0.6604,
      "step": 2486
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4357805037237418,
      "learning_rate": 3.62434278930576e-05,
      "loss": 0.7265,
      "step": 2487
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.388367013528859,
      "learning_rate": 3.6240048036863995e-05,
      "loss": 0.6842,
      "step": 2488
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33973792891534416,
      "learning_rate": 3.623666681863717e-05,
      "loss": 0.638,
      "step": 2489
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.45641661542835726,
      "learning_rate": 3.623328423866071e-05,
      "loss": 0.7094,
      "step": 2490
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.45950276898313175,
      "learning_rate": 3.62299002972183e-05,
      "loss": 0.7204,
      "step": 2491
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44778893806239345,
      "learning_rate": 3.622651499459376e-05,
      "loss": 0.7145,
      "step": 2492
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3882389303978925,
      "learning_rate": 3.6223128331071e-05,
      "loss": 0.6512,
      "step": 2493
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4084314364981872,
      "learning_rate": 3.621974030693406e-05,
      "loss": 0.7231,
      "step": 2494
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42969103969079825,
      "learning_rate": 3.621635092246709e-05,
      "loss": 0.7385,
      "step": 2495
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4333397998986271,
      "learning_rate": 3.621296017795435e-05,
      "loss": 0.6974,
      "step": 2496
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43690022376126175,
      "learning_rate": 3.6209568073680224e-05,
      "loss": 0.6931,
      "step": 2497
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43446674300266236,
      "learning_rate": 3.6206174609929196e-05,
      "loss": 0.7026,
      "step": 2498
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42967481514302136,
      "learning_rate": 3.6202779786985875e-05,
      "loss": 0.7225,
      "step": 2499
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.641258014830284,
      "learning_rate": 3.6199383605134986e-05,
      "loss": 0.601,
      "step": 2500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43300400210989887,
      "learning_rate": 3.6195986064661356e-05,
      "loss": 0.7261,
      "step": 2501
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40654186277199483,
      "learning_rate": 3.6192587165849926e-05,
      "loss": 0.6877,
      "step": 2502
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.48228842457546833,
      "learning_rate": 3.6189186908985775e-05,
      "loss": 0.728,
      "step": 2503
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5189272037297791,
      "learning_rate": 3.6185785294354074e-05,
      "loss": 0.7367,
      "step": 2504
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44561158891247993,
      "learning_rate": 3.61823823222401e-05,
      "loss": 0.714,
      "step": 2505
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4707115616832597,
      "learning_rate": 3.617897799292927e-05,
      "loss": 0.7089,
      "step": 2506
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4103523169668237,
      "learning_rate": 3.617557230670709e-05,
      "loss": 0.6675,
      "step": 2507
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4729881422579595,
      "learning_rate": 3.617216526385919e-05,
      "loss": 0.7137,
      "step": 2508
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4107429099004062,
      "learning_rate": 3.6168756864671336e-05,
      "loss": 0.6738,
      "step": 2509
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44555858272957705,
      "learning_rate": 3.616534710942936e-05,
      "loss": 0.6707,
      "step": 2510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.49110395621165054,
      "learning_rate": 3.616193599841924e-05,
      "loss": 0.6872,
      "step": 2511
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.45039458662801046,
      "learning_rate": 3.6158523531927074e-05,
      "loss": 0.7444,
      "step": 2512
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40612123023675634,
      "learning_rate": 3.615510971023906e-05,
      "loss": 0.6961,
      "step": 2513
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40938010212750503,
      "learning_rate": 3.6151694533641496e-05,
      "loss": 0.6829,
      "step": 2514
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4523197496169971,
      "learning_rate": 3.614827800242082e-05,
      "loss": 0.6636,
      "step": 2515
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4032621070645777,
      "learning_rate": 3.614486011686357e-05,
      "loss": 0.7083,
      "step": 2516
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40766172439845816,
      "learning_rate": 3.61414408772564e-05,
      "loss": 0.6453,
      "step": 2517
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.42532603762297727,
      "learning_rate": 3.6138020283886074e-05,
      "loss": 0.7055,
      "step": 2518
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38997193906321365,
      "learning_rate": 3.6134598337039484e-05,
      "loss": 0.6781,
      "step": 2519
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.436016247073968,
      "learning_rate": 3.613117503700361e-05,
      "loss": 0.7195,
      "step": 2520
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40526084218660274,
      "learning_rate": 3.6127750384065575e-05,
      "loss": 0.6704,
      "step": 2521
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37951726142486863,
      "learning_rate": 3.6124324378512585e-05,
      "loss": 0.6476,
      "step": 2522
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3928049160026097,
      "learning_rate": 3.612089702063199e-05,
      "loss": 0.6572,
      "step": 2523
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39798288427283857,
      "learning_rate": 3.611746831071123e-05,
      "loss": 0.6958,
      "step": 2524
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4130796754060056,
      "learning_rate": 3.6114038249037855e-05,
      "loss": 0.7139,
      "step": 2525
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3870276943953596,
      "learning_rate": 3.611060683589957e-05,
      "loss": 0.6904,
      "step": 2526
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3943665344778248,
      "learning_rate": 3.610717407158413e-05,
      "loss": 0.6725,
      "step": 2527
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3633645344686335,
      "learning_rate": 3.6103739956379464e-05,
      "loss": 0.6958,
      "step": 2528
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4212027313001505,
      "learning_rate": 3.610030449057358e-05,
      "loss": 0.7447,
      "step": 2529
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3691201854978461,
      "learning_rate": 3.609686767445459e-05,
      "loss": 0.6304,
      "step": 2530
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3852531011878223,
      "learning_rate": 3.609342950831076e-05,
      "loss": 0.6442,
      "step": 2531
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.402616297182493,
      "learning_rate": 3.608998999243043e-05,
      "loss": 0.6982,
      "step": 2532
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41193585012079886,
      "learning_rate": 3.6086549127102066e-05,
      "loss": 0.6874,
      "step": 2533
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8283340226501805,
      "learning_rate": 3.608310691261426e-05,
      "loss": 0.6557,
      "step": 2534
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43627564120174356,
      "learning_rate": 3.60796633492557e-05,
      "loss": 0.6965,
      "step": 2535
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4385745724913996,
      "learning_rate": 3.6076218437315184e-05,
      "loss": 0.689,
      "step": 2536
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43921368265063304,
      "learning_rate": 3.607277217708165e-05,
      "loss": 0.7312,
      "step": 2537
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3986879832649528,
      "learning_rate": 3.6069324568844135e-05,
      "loss": 0.6691,
      "step": 2538
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40146370695376354,
      "learning_rate": 3.6065875612891765e-05,
      "loss": 0.7012,
      "step": 2539
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41955123662303945,
      "learning_rate": 3.606242530951381e-05,
      "loss": 0.6986,
      "step": 2540
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4156658849578884,
      "learning_rate": 3.6058973658999644e-05,
      "loss": 0.7259,
      "step": 2541
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4276507508065261,
      "learning_rate": 3.605552066163875e-05,
      "loss": 0.7106,
      "step": 2542
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.42535878347513223,
      "learning_rate": 3.605206631772073e-05,
      "loss": 0.7205,
      "step": 2543
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4257061466224844,
      "learning_rate": 3.6048610627535296e-05,
      "loss": 0.7139,
      "step": 2544
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3675289074156952,
      "learning_rate": 3.6045153591372264e-05,
      "loss": 0.6611,
      "step": 2545
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3789245901842752,
      "learning_rate": 3.6041695209521586e-05,
      "loss": 0.7088,
      "step": 2546
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.563406509535206,
      "learning_rate": 3.60382354822733e-05,
      "loss": 0.6122,
      "step": 2547
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40471220532540253,
      "learning_rate": 3.603477440991757e-05,
      "loss": 0.6798,
      "step": 2548
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4080401923817346,
      "learning_rate": 3.603131199274468e-05,
      "loss": 0.7,
      "step": 2549
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4361389967366053,
      "learning_rate": 3.602784823104501e-05,
      "loss": 0.7086,
      "step": 2550
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38306525238139305,
      "learning_rate": 3.602438312510908e-05,
      "loss": 0.6583,
      "step": 2551
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3861132800738278,
      "learning_rate": 3.6020916675227476e-05,
      "loss": 0.6914,
      "step": 2552
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45910564313487723,
      "learning_rate": 3.6017448881690936e-05,
      "loss": 0.7509,
      "step": 2553
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45240851283147054,
      "learning_rate": 3.601397974479031e-05,
      "loss": 0.7071,
      "step": 2554
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3826715278794324,
      "learning_rate": 3.6010509264816545e-05,
      "loss": 0.6654,
      "step": 2555
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4235136036551006,
      "learning_rate": 3.60070374420607e-05,
      "loss": 0.6396,
      "step": 2556
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4166492393958962,
      "learning_rate": 3.600356427681395e-05,
      "loss": 0.6941,
      "step": 2557
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38694742950383726,
      "learning_rate": 3.60000897693676e-05,
      "loss": 0.6715,
      "step": 2558
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38474353553847757,
      "learning_rate": 3.599661392001304e-05,
      "loss": 0.692,
      "step": 2559
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40903806938460546,
      "learning_rate": 3.599313672904179e-05,
      "loss": 0.7107,
      "step": 2560
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4093607114040109,
      "learning_rate": 3.5989658196745476e-05,
      "loss": 0.6961,
      "step": 2561
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37366488774788986,
      "learning_rate": 3.5986178323415845e-05,
      "loss": 0.7136,
      "step": 2562
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37986076325562146,
      "learning_rate": 3.5982697109344746e-05,
      "loss": 0.6892,
      "step": 2563
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.44207479638300573,
      "learning_rate": 3.597921455482413e-05,
      "loss": 0.7219,
      "step": 2564
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3814075598772529,
      "learning_rate": 3.5975730660146095e-05,
      "loss": 0.6935,
      "step": 2565
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.665398809434779,
      "learning_rate": 3.597224542560283e-05,
      "loss": 0.5698,
      "step": 2566
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4631512561015942,
      "learning_rate": 3.596875885148662e-05,
      "loss": 0.704,
      "step": 2567
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37394865412935485,
      "learning_rate": 3.59652709380899e-05,
      "loss": 0.6259,
      "step": 2568
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38054758382059684,
      "learning_rate": 3.596178168570518e-05,
      "loss": 0.6702,
      "step": 2569
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41806038339375473,
      "learning_rate": 3.595829109462511e-05,
      "loss": 0.7123,
      "step": 2570
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45030887009711384,
      "learning_rate": 3.595479916514244e-05,
      "loss": 0.7288,
      "step": 2571
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46161712497297824,
      "learning_rate": 3.595130589755003e-05,
      "loss": 0.7431,
      "step": 2572
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43338134592249233,
      "learning_rate": 3.594781129214086e-05,
      "loss": 0.7343,
      "step": 2573
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48717106922483794,
      "learning_rate": 3.5944315349208016e-05,
      "loss": 0.7145,
      "step": 2574
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3849660820150107,
      "learning_rate": 3.59408180690447e-05,
      "loss": 0.7038,
      "step": 2575
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45883554925896375,
      "learning_rate": 3.5937319451944225e-05,
      "loss": 0.6902,
      "step": 2576
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38873288819177176,
      "learning_rate": 3.5933819498200016e-05,
      "loss": 0.679,
      "step": 2577
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4395020739720219,
      "learning_rate": 3.593031820810561e-05,
      "loss": 0.6756,
      "step": 2578
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.42662388759328873,
      "learning_rate": 3.592681558195465e-05,
      "loss": 0.7433,
      "step": 2579
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.515381920668478,
      "learning_rate": 3.59233116200409e-05,
      "loss": 0.6163,
      "step": 2580
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49378843747649676,
      "learning_rate": 3.591980632265824e-05,
      "loss": 0.7193,
      "step": 2581
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43857363576281666,
      "learning_rate": 3.5916299690100643e-05,
      "loss": 0.6995,
      "step": 2582
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.36860940655895286,
      "learning_rate": 3.591279172266222e-05,
      "loss": 0.687,
      "step": 2583
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.44541390546797927,
      "learning_rate": 3.5909282420637174e-05,
      "loss": 0.7049,
      "step": 2584
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4340980641982526,
      "learning_rate": 3.590577178431982e-05,
      "loss": 0.6948,
      "step": 2585
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3716688619159113,
      "learning_rate": 3.59022598140046e-05,
      "loss": 0.6935,
      "step": 2586
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38573563858598486,
      "learning_rate": 3.589874650998606e-05,
      "loss": 0.6875,
      "step": 2587
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4311001229276016,
      "learning_rate": 3.589523187255884e-05,
      "loss": 0.6979,
      "step": 2588
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3875873981133232,
      "learning_rate": 3.589171590201772e-05,
      "loss": 0.6794,
      "step": 2589
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.44281405317679196,
      "learning_rate": 3.5888198598657585e-05,
      "loss": 0.7,
      "step": 2590
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4093986221910737,
      "learning_rate": 3.588467996277342e-05,
      "loss": 0.6907,
      "step": 2591
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4060721898659938,
      "learning_rate": 3.588115999466033e-05,
      "loss": 0.6987,
      "step": 2592
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3874318502316161,
      "learning_rate": 3.587763869461353e-05,
      "loss": 0.6884,
      "step": 2593
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4266595433575104,
      "learning_rate": 3.587411606292835e-05,
      "loss": 0.7238,
      "step": 2594
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3912158710257818,
      "learning_rate": 3.5870592099900224e-05,
      "loss": 0.6861,
      "step": 2595
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.44606655046558824,
      "learning_rate": 3.586706680582471e-05,
      "loss": 0.7312,
      "step": 2596
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.454752262587095,
      "learning_rate": 3.586354018099746e-05,
      "loss": 0.7281,
      "step": 2597
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4441622183973647,
      "learning_rate": 3.5860012225714255e-05,
      "loss": 0.6978,
      "step": 2598
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4868014040811365,
      "learning_rate": 3.585648294027098e-05,
      "loss": 0.6927,
      "step": 2599
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4290355045055915,
      "learning_rate": 3.5852952324963634e-05,
      "loss": 0.6687,
      "step": 2600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3718655655653913,
      "learning_rate": 3.584942038008831e-05,
      "loss": 0.7004,
      "step": 2601
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.431301789140166,
      "learning_rate": 3.584588710594125e-05,
      "loss": 0.6965,
      "step": 2602
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4529148904120257,
      "learning_rate": 3.584235250281878e-05,
      "loss": 0.6841,
      "step": 2603
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4361202153191413,
      "learning_rate": 3.5838816571017324e-05,
      "loss": 0.7063,
      "step": 2604
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3994883194166158,
      "learning_rate": 3.583527931083346e-05,
      "loss": 0.6541,
      "step": 2605
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39528952149764784,
      "learning_rate": 3.583174072256385e-05,
      "loss": 0.7014,
      "step": 2606
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4159909448267426,
      "learning_rate": 3.582820080650526e-05,
      "loss": 0.7419,
      "step": 2607
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3725867373334623,
      "learning_rate": 3.5824659562954574e-05,
      "loss": 0.6974,
      "step": 2608
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37628656742261757,
      "learning_rate": 3.582111699220881e-05,
      "loss": 0.6983,
      "step": 2609
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3819620140229539,
      "learning_rate": 3.5817573094565076e-05,
      "loss": 0.7083,
      "step": 2610
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40816013231214404,
      "learning_rate": 3.581402787032058e-05,
      "loss": 0.6901,
      "step": 2611
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3964800554913116,
      "learning_rate": 3.581048131977267e-05,
      "loss": 0.6763,
      "step": 2612
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39765049038645056,
      "learning_rate": 3.5806933443218787e-05,
      "loss": 0.7261,
      "step": 2613
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4425042959024057,
      "learning_rate": 3.580338424095648e-05,
      "loss": 0.665,
      "step": 2614
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41422986544509854,
      "learning_rate": 3.579983371328343e-05,
      "loss": 0.7012,
      "step": 2615
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4239765875513511,
      "learning_rate": 3.5796281860497406e-05,
      "loss": 0.7447,
      "step": 2616
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3880888892812283,
      "learning_rate": 3.5792728682896296e-05,
      "loss": 0.6506,
      "step": 2617
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3750405074046167,
      "learning_rate": 3.57891741807781e-05,
      "loss": 0.6781,
      "step": 2618
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40479469008278285,
      "learning_rate": 3.5785618354440945e-05,
      "loss": 0.7085,
      "step": 2619
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4631581973981995,
      "learning_rate": 3.5782061204183036e-05,
      "loss": 0.7245,
      "step": 2620
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4204858135838592,
      "learning_rate": 3.577850273030272e-05,
      "loss": 0.7296,
      "step": 2621
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45192249840328724,
      "learning_rate": 3.577494293309843e-05,
      "loss": 0.722,
      "step": 2622
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3814901708219678,
      "learning_rate": 3.577138181286873e-05,
      "loss": 0.6783,
      "step": 2623
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40564019437896903,
      "learning_rate": 3.5767819369912284e-05,
      "loss": 0.6972,
      "step": 2624
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40890463680028144,
      "learning_rate": 3.576425560452787e-05,
      "loss": 0.722,
      "step": 2625
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3593896497658619,
      "learning_rate": 3.576069051701437e-05,
      "loss": 0.6542,
      "step": 2626
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39441503768897046,
      "learning_rate": 3.5757124107670806e-05,
      "loss": 0.6952,
      "step": 2627
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3793288894189497,
      "learning_rate": 3.575355637679626e-05,
      "loss": 0.671,
      "step": 2628
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3634580775680815,
      "learning_rate": 3.574998732468997e-05,
      "loss": 0.6569,
      "step": 2629
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.35610418987064435,
      "learning_rate": 3.5746416951651265e-05,
      "loss": 0.6673,
      "step": 2630
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39962811249984925,
      "learning_rate": 3.574284525797958e-05,
      "loss": 0.6867,
      "step": 2631
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4180624326449103,
      "learning_rate": 3.5739272243974486e-05,
      "loss": 0.685,
      "step": 2632
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40184348479035376,
      "learning_rate": 3.5735697909935636e-05,
      "loss": 0.702,
      "step": 2633
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3570762886650564,
      "learning_rate": 3.57321222561628e-05,
      "loss": 0.6324,
      "step": 2634
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3944259266171717,
      "learning_rate": 3.5728545282955874e-05,
      "loss": 0.6781,
      "step": 2635
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.380122926670402,
      "learning_rate": 3.5724966990614845e-05,
      "loss": 0.6693,
      "step": 2636
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38959442460572624,
      "learning_rate": 3.572138737943983e-05,
      "loss": 0.674,
      "step": 2637
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4572418139933899,
      "learning_rate": 3.571780644973105e-05,
      "loss": 0.7315,
      "step": 2638
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42297938042827266,
      "learning_rate": 3.571422420178881e-05,
      "loss": 0.7217,
      "step": 2639
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.37598657209860936,
      "learning_rate": 3.571064063591357e-05,
      "loss": 0.6885,
      "step": 2640
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42440525049370276,
      "learning_rate": 3.570705575240588e-05,
      "loss": 0.6843,
      "step": 2641
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3815130745645982,
      "learning_rate": 3.5703469551566384e-05,
      "loss": 0.6829,
      "step": 2642
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42796174231887646,
      "learning_rate": 3.569988203369587e-05,
      "loss": 0.6743,
      "step": 2643
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39504504835957327,
      "learning_rate": 3.56962931990952e-05,
      "loss": 0.6809,
      "step": 2644
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4315503064286679,
      "learning_rate": 3.569270304806539e-05,
      "loss": 0.6734,
      "step": 2645
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4626792560500021,
      "learning_rate": 3.568911158090752e-05,
      "loss": 0.7131,
      "step": 2646
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4343425240045454,
      "learning_rate": 3.568551879792281e-05,
      "loss": 0.7197,
      "step": 2647
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.35843565938216015,
      "learning_rate": 3.5681924699412574e-05,
      "loss": 0.6765,
      "step": 2648
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42260571450400986,
      "learning_rate": 3.5678329285678266e-05,
      "loss": 0.6851,
      "step": 2649
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.37653204172078475,
      "learning_rate": 3.56747325570214e-05,
      "loss": 0.6909,
      "step": 2650
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4135401664387288,
      "learning_rate": 3.5671134513743654e-05,
      "loss": 0.682,
      "step": 2651
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4041956014400724,
      "learning_rate": 3.566753515614679e-05,
      "loss": 0.6671,
      "step": 2652
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4149878206443112,
      "learning_rate": 3.5663934484532664e-05,
      "loss": 0.6898,
      "step": 2653
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40631603460567156,
      "learning_rate": 3.5660332499203274e-05,
      "loss": 0.709,
      "step": 2654
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43940061369447947,
      "learning_rate": 3.565672920046071e-05,
      "loss": 0.687,
      "step": 2655
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4688643353995133,
      "learning_rate": 3.565312458860717e-05,
      "loss": 0.7062,
      "step": 2656
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3899174039762822,
      "learning_rate": 3.564951866394498e-05,
      "loss": 0.7181,
      "step": 2657
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4219749359905737,
      "learning_rate": 3.564591142677656e-05,
      "loss": 0.6663,
      "step": 2658
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4032600487517589,
      "learning_rate": 3.564230287740445e-05,
      "loss": 0.6767,
      "step": 2659
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42385246791891507,
      "learning_rate": 3.563869301613127e-05,
      "loss": 0.7038,
      "step": 2660
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42340629892173776,
      "learning_rate": 3.5635081843259803e-05,
      "loss": 0.7113,
      "step": 2661
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40562079697355824,
      "learning_rate": 3.563146935909291e-05,
      "loss": 0.6767,
      "step": 2662
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4024677754193632,
      "learning_rate": 3.562785556393356e-05,
      "loss": 0.645,
      "step": 2663
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42203380891972214,
      "learning_rate": 3.562424045808483e-05,
      "loss": 0.6718,
      "step": 2664
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4667148049277803,
      "learning_rate": 3.562062404184993e-05,
      "loss": 0.7088,
      "step": 2665
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4150558974400902,
      "learning_rate": 3.561700631553215e-05,
      "loss": 0.6968,
      "step": 2666
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39467671717466857,
      "learning_rate": 3.5613387279434914e-05,
      "loss": 0.7192,
      "step": 2667
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4215820001596473,
      "learning_rate": 3.5609766933861745e-05,
      "loss": 0.6874,
      "step": 2668
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.430846003753851,
      "learning_rate": 3.560614527911627e-05,
      "loss": 0.6857,
      "step": 2669
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3903849664757159,
      "learning_rate": 3.560252231550224e-05,
      "loss": 0.6684,
      "step": 2670
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36575054906972476,
      "learning_rate": 3.55988980433235e-05,
      "loss": 0.6674,
      "step": 2671
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3691418515574907,
      "learning_rate": 3.5595272462884035e-05,
      "loss": 0.6411,
      "step": 2672
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4292000388760112,
      "learning_rate": 3.559164557448789e-05,
      "loss": 0.703,
      "step": 2673
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4069520781825226,
      "learning_rate": 3.558801737843926e-05,
      "loss": 0.7183,
      "step": 2674
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44591812288152305,
      "learning_rate": 3.558438787504245e-05,
      "loss": 0.718,
      "step": 2675
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.47888507536788,
      "learning_rate": 3.5580757064601847e-05,
      "loss": 0.7317,
      "step": 2676
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42078003140227827,
      "learning_rate": 3.557712494742196e-05,
      "loss": 0.6989,
      "step": 2677
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39222919143431495,
      "learning_rate": 3.5573491523807414e-05,
      "loss": 0.6634,
      "step": 2678
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3956252896567667,
      "learning_rate": 3.556985679406295e-05,
      "loss": 0.686,
      "step": 2679
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40031355789314027,
      "learning_rate": 3.556622075849339e-05,
      "loss": 0.6368,
      "step": 2680
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4902011979441201,
      "learning_rate": 3.5562583417403705e-05,
      "loss": 0.7147,
      "step": 2681
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46363128079802635,
      "learning_rate": 3.555894477109894e-05,
      "loss": 0.7212,
      "step": 2682
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4373632549615729,
      "learning_rate": 3.5555304819884267e-05,
      "loss": 0.6828,
      "step": 2683
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40773238736422035,
      "learning_rate": 3.555166356406496e-05,
      "loss": 0.7257,
      "step": 2684
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40400967360918777,
      "learning_rate": 3.554802100394642e-05,
      "loss": 0.6962,
      "step": 2685
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.375543420430806,
      "learning_rate": 3.554437713983413e-05,
      "loss": 0.5623,
      "step": 2686
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39980759676705135,
      "learning_rate": 3.554073197203371e-05,
      "loss": 0.6773,
      "step": 2687
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4385753204727226,
      "learning_rate": 3.5537085500850854e-05,
      "loss": 0.6873,
      "step": 2688
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43226565667725453,
      "learning_rate": 3.553343772659141e-05,
      "loss": 0.6909,
      "step": 2689
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4565358456142096,
      "learning_rate": 3.5529788649561304e-05,
      "loss": 0.7406,
      "step": 2690
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4983551801577884,
      "learning_rate": 3.552613827006658e-05,
      "loss": 0.6894,
      "step": 2691
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41193898333465656,
      "learning_rate": 3.5522486588413384e-05,
      "loss": 0.6733,
      "step": 2692
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43598657722631273,
      "learning_rate": 3.5518833604907996e-05,
      "loss": 0.6752,
      "step": 2693
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4664404148457599,
      "learning_rate": 3.551517931985676e-05,
      "loss": 0.7127,
      "step": 2694
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4677350563561677,
      "learning_rate": 3.551152373356618e-05,
      "loss": 0.7004,
      "step": 2695
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.441562713056172,
      "learning_rate": 3.5507866846342845e-05,
      "loss": 0.7343,
      "step": 2696
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.47975208303917893,
      "learning_rate": 3.5504208658493435e-05,
      "loss": 0.7105,
      "step": 2697
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36949277656893675,
      "learning_rate": 3.550054917032477e-05,
      "loss": 0.6734,
      "step": 2698
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4225256435216491,
      "learning_rate": 3.549688838214377e-05,
      "loss": 0.698,
      "step": 2699
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.37852427876056377,
      "learning_rate": 3.5493226294257456e-05,
      "loss": 0.6893,
      "step": 2700
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4405473188395101,
      "learning_rate": 3.5489562906972965e-05,
      "loss": 0.5957,
      "step": 2701
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4375250234831465,
      "learning_rate": 3.5485898220597534e-05,
      "loss": 0.6556,
      "step": 2702
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39813798019457997,
      "learning_rate": 3.5482232235438525e-05,
      "loss": 0.6733,
      "step": 2703
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4484215716306611,
      "learning_rate": 3.54785649518034e-05,
      "loss": 0.7465,
      "step": 2704
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4152334462449785,
      "learning_rate": 3.5474896369999716e-05,
      "loss": 0.6473,
      "step": 2705
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5042315438604874,
      "learning_rate": 3.5471226490335174e-05,
      "loss": 0.6983,
      "step": 2706
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5339518874489194,
      "learning_rate": 3.546755531311754e-05,
      "loss": 0.7044,
      "step": 2707
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.37875534183556503,
      "learning_rate": 3.546388283865473e-05,
      "loss": 0.6822,
      "step": 2708
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3841423523469602,
      "learning_rate": 3.546020906725474e-05,
      "loss": 0.6824,
      "step": 2709
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38865269413957065,
      "learning_rate": 3.545653399922569e-05,
      "loss": 0.6762,
      "step": 2710
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39612098167597176,
      "learning_rate": 3.54528576348758e-05,
      "loss": 0.6862,
      "step": 2711
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4093754573946096,
      "learning_rate": 3.544917997451339e-05,
      "loss": 0.6825,
      "step": 2712
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3991855815447308,
      "learning_rate": 3.544550101844693e-05,
      "loss": 0.7093,
      "step": 2713
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4431382620744281,
      "learning_rate": 3.5441820766984944e-05,
      "loss": 0.7135,
      "step": 2714
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4062332754253204,
      "learning_rate": 3.54381392204361e-05,
      "loss": 0.6945,
      "step": 2715
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43299944026595927,
      "learning_rate": 3.5434456379109166e-05,
      "loss": 0.7152,
      "step": 2716
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3794909022026176,
      "learning_rate": 3.543077224331301e-05,
      "loss": 0.6836,
      "step": 2717
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3748186359620317,
      "learning_rate": 3.542708681335663e-05,
      "loss": 0.6709,
      "step": 2718
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41542327336355095,
      "learning_rate": 3.54234000895491e-05,
      "loss": 0.686,
      "step": 2719
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4018669306310842,
      "learning_rate": 3.541971207219964e-05,
      "loss": 0.6811,
      "step": 2720
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3684267831528723,
      "learning_rate": 3.541602276161755e-05,
      "loss": 0.6763,
      "step": 2721
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39207404652161776,
      "learning_rate": 3.5412332158112246e-05,
      "loss": 0.679,
      "step": 2722
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43348229339310734,
      "learning_rate": 3.540864026199326e-05,
      "loss": 0.7554,
      "step": 2723
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38792212764823825,
      "learning_rate": 3.5404947073570225e-05,
      "loss": 0.6992,
      "step": 2724
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4294569496403762,
      "learning_rate": 3.540125259315288e-05,
      "loss": 0.7149,
      "step": 2725
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39650743137962996,
      "learning_rate": 3.539755682105108e-05,
      "loss": 0.6962,
      "step": 2726
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4194851016498127,
      "learning_rate": 3.539385975757479e-05,
      "loss": 0.6973,
      "step": 2727
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38824054474335085,
      "learning_rate": 3.5390161403034064e-05,
      "loss": 0.6865,
      "step": 2728
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7171796439168223,
      "learning_rate": 3.53864617577391e-05,
      "loss": 0.6137,
      "step": 2729
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3743540975940824,
      "learning_rate": 3.5382760822000156e-05,
      "loss": 0.6852,
      "step": 2730
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41911453966834233,
      "learning_rate": 3.5379058596127646e-05,
      "loss": 0.6773,
      "step": 2731
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3966883158773197,
      "learning_rate": 3.537535508043207e-05,
      "loss": 0.6915,
      "step": 2732
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4003255664120138,
      "learning_rate": 3.5371650275224026e-05,
      "loss": 0.6897,
      "step": 2733
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41633305974684537,
      "learning_rate": 3.5367944180814244e-05,
      "loss": 0.6763,
      "step": 2734
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42673538439165154,
      "learning_rate": 3.536423679751353e-05,
      "loss": 0.6504,
      "step": 2735
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4083879589953831,
      "learning_rate": 3.536052812563284e-05,
      "loss": 0.7132,
      "step": 2736
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.45805901908278696,
      "learning_rate": 3.535681816548321e-05,
      "loss": 0.7079,
      "step": 2737
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3951936129463504,
      "learning_rate": 3.535310691737578e-05,
      "loss": 0.6549,
      "step": 2738
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39662924014502676,
      "learning_rate": 3.5349394381621825e-05,
      "loss": 0.7029,
      "step": 2739
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8102285197584718,
      "learning_rate": 3.534568055853269e-05,
      "loss": 0.5694,
      "step": 2740
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44914339860533065,
      "learning_rate": 3.5341965448419865e-05,
      "loss": 0.7042,
      "step": 2741
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4114152846888866,
      "learning_rate": 3.5338249051594926e-05,
      "loss": 0.7076,
      "step": 2742
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4129797059511372,
      "learning_rate": 3.533453136836956e-05,
      "loss": 0.6755,
      "step": 2743
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4122415473156248,
      "learning_rate": 3.5330812399055575e-05,
      "loss": 0.7113,
      "step": 2744
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40029102371012454,
      "learning_rate": 3.5327092143964865e-05,
      "loss": 0.6793,
      "step": 2745
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3865076062012881,
      "learning_rate": 3.532337060340944e-05,
      "loss": 0.6537,
      "step": 2746
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.409988238174476,
      "learning_rate": 3.531964777770143e-05,
      "loss": 0.7081,
      "step": 2747
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4595068489861511,
      "learning_rate": 3.5315923667153075e-05,
      "loss": 0.7111,
      "step": 2748
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4324789908764716,
      "learning_rate": 3.5312198272076685e-05,
      "loss": 0.725,
      "step": 2749
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42003216399749926,
      "learning_rate": 3.530847159278472e-05,
      "loss": 0.6762,
      "step": 2750
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42077046009457675,
      "learning_rate": 3.530474362958974e-05,
      "loss": 0.6792,
      "step": 2751
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3805875788781081,
      "learning_rate": 3.530101438280439e-05,
      "loss": 0.6929,
      "step": 2752
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3341438116227193,
      "learning_rate": 3.529728385274144e-05,
      "loss": 0.651,
      "step": 2753
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41782236988264365,
      "learning_rate": 3.529355203971377e-05,
      "loss": 0.6861,
      "step": 2754
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45495014959759056,
      "learning_rate": 3.528981894403436e-05,
      "loss": 0.7079,
      "step": 2755
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3940435588892468,
      "learning_rate": 3.52860845660163e-05,
      "loss": 0.7331,
      "step": 2756
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3219313456853886,
      "learning_rate": 3.528234890597279e-05,
      "loss": 0.6722,
      "step": 2757
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3822142517142836,
      "learning_rate": 3.527861196421714e-05,
      "loss": 0.6831,
      "step": 2758
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4267634540699137,
      "learning_rate": 3.527487374106274e-05,
      "loss": 0.7257,
      "step": 2759
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.36381131075417383,
      "learning_rate": 3.5271134236823136e-05,
      "loss": 0.6883,
      "step": 2760
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.36148882038239155,
      "learning_rate": 3.526739345181195e-05,
      "loss": 0.6459,
      "step": 2761
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0301328665284917,
      "learning_rate": 3.5263651386342907e-05,
      "loss": 0.609,
      "step": 2762
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44263170628493254,
      "learning_rate": 3.525990804072986e-05,
      "loss": 0.668,
      "step": 2763
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4073146886271602,
      "learning_rate": 3.525616341528675e-05,
      "loss": 0.6671,
      "step": 2764
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4276615818117853,
      "learning_rate": 3.525241751032765e-05,
      "loss": 0.7025,
      "step": 2765
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4198196437781352,
      "learning_rate": 3.5248670326166704e-05,
      "loss": 0.6652,
      "step": 2766
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4027570995798744,
      "learning_rate": 3.5244921863118194e-05,
      "loss": 0.7061,
      "step": 2767
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45185771726605906,
      "learning_rate": 3.52411721214965e-05,
      "loss": 0.7314,
      "step": 2768
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40649672884545235,
      "learning_rate": 3.523742110161611e-05,
      "loss": 0.6593,
      "step": 2769
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42176509928980427,
      "learning_rate": 3.523366880379161e-05,
      "loss": 0.7,
      "step": 2770
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.34141549315750397,
      "learning_rate": 3.522991522833771e-05,
      "loss": 0.6846,
      "step": 2771
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42744854469618526,
      "learning_rate": 3.522616037556921e-05,
      "loss": 0.7209,
      "step": 2772
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.37264314017925426,
      "learning_rate": 3.522240424580103e-05,
      "loss": 0.6733,
      "step": 2773
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.36523554856112517,
      "learning_rate": 3.5218646839348184e-05,
      "loss": 0.6666,
      "step": 2774
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4261546474563831,
      "learning_rate": 3.521488815652581e-05,
      "loss": 0.669,
      "step": 2775
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3964315155249818,
      "learning_rate": 3.5211128197649144e-05,
      "loss": 0.7142,
      "step": 2776
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38976675288598556,
      "learning_rate": 3.5207366963033525e-05,
      "loss": 0.6729,
      "step": 2777
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39212826669317064,
      "learning_rate": 3.52036044529944e-05,
      "loss": 0.6636,
      "step": 2778
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40457450429882064,
      "learning_rate": 3.519984066784733e-05,
      "loss": 0.6505,
      "step": 2779
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3950256247682361,
      "learning_rate": 3.519607560790799e-05,
      "loss": 0.6956,
      "step": 2780
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40242325392939243,
      "learning_rate": 3.5192309273492136e-05,
      "loss": 0.6712,
      "step": 2781
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45429172657521405,
      "learning_rate": 3.518854166491565e-05,
      "loss": 0.718,
      "step": 2782
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4530217468875772,
      "learning_rate": 3.518477278249452e-05,
      "loss": 0.73,
      "step": 2783
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38357131750857454,
      "learning_rate": 3.518100262654483e-05,
      "loss": 0.7017,
      "step": 2784
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38391186775818964,
      "learning_rate": 3.517723119738279e-05,
      "loss": 0.6857,
      "step": 2785
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3822038470451318,
      "learning_rate": 3.517345849532469e-05,
      "loss": 0.7088,
      "step": 2786
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4239923857775956,
      "learning_rate": 3.516968452068696e-05,
      "loss": 0.7281,
      "step": 2787
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.35819171463873223,
      "learning_rate": 3.516590927378611e-05,
      "loss": 0.6717,
      "step": 2788
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.35937887917941486,
      "learning_rate": 3.516213275493876e-05,
      "loss": 0.6465,
      "step": 2789
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.456795286297931,
      "learning_rate": 3.515835496446165e-05,
      "loss": 0.7393,
      "step": 2790
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41833187638816316,
      "learning_rate": 3.515457590267161e-05,
      "loss": 0.676,
      "step": 2791
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4189933344006255,
      "learning_rate": 3.51507955698856e-05,
      "loss": 0.7087,
      "step": 2792
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3772551729968005,
      "learning_rate": 3.514701396642066e-05,
      "loss": 0.6489,
      "step": 2793
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4397883177368971,
      "learning_rate": 3.5143231092593946e-05,
      "loss": 0.6919,
      "step": 2794
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39431748173007874,
      "learning_rate": 3.513944694872273e-05,
      "loss": 0.705,
      "step": 2795
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3571076231963024,
      "learning_rate": 3.51356615351244e-05,
      "loss": 0.6931,
      "step": 2796
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4296105534585052,
      "learning_rate": 3.5131874852116396e-05,
      "loss": 0.7236,
      "step": 2797
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39905393889787566,
      "learning_rate": 3.512808690001633e-05,
      "loss": 0.6894,
      "step": 2798
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44648143608602187,
      "learning_rate": 3.512429767914188e-05,
      "loss": 0.7017,
      "step": 2799
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9943325184513763,
      "learning_rate": 3.512050718981085e-05,
      "loss": 0.6023,
      "step": 2800
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3604282718953225,
      "learning_rate": 3.5116715432341144e-05,
      "loss": 0.6391,
      "step": 2801
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4097907546715315,
      "learning_rate": 3.511292240705078e-05,
      "loss": 0.6737,
      "step": 2802
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43224138932713674,
      "learning_rate": 3.5109128114257855e-05,
      "loss": 0.6681,
      "step": 2803
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38324076073523167,
      "learning_rate": 3.510533255428061e-05,
      "loss": 0.6828,
      "step": 2804
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.37408187174311697,
      "learning_rate": 3.510153572743736e-05,
      "loss": 0.6652,
      "step": 2805
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38704446900076245,
      "learning_rate": 3.509773763404655e-05,
      "loss": 0.6873,
      "step": 2806
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43385255823018637,
      "learning_rate": 3.509393827442672e-05,
      "loss": 0.6945,
      "step": 2807
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4008917835822329,
      "learning_rate": 3.509013764889651e-05,
      "loss": 0.6808,
      "step": 2808
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40277437575484537,
      "learning_rate": 3.508633575777468e-05,
      "loss": 0.6565,
      "step": 2809
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3643059346071237,
      "learning_rate": 3.508253260138009e-05,
      "loss": 0.6667,
      "step": 2810
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39218114143677146,
      "learning_rate": 3.507872818003171e-05,
      "loss": 0.7138,
      "step": 2811
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40548049234690225,
      "learning_rate": 3.507492249404861e-05,
      "loss": 0.644,
      "step": 2812
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3920070681152005,
      "learning_rate": 3.507111554374996e-05,
      "loss": 0.7209,
      "step": 2813
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4936776570356592,
      "learning_rate": 3.506730732945506e-05,
      "loss": 0.6847,
      "step": 2814
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40826363326701687,
      "learning_rate": 3.5063497851483285e-05,
      "loss": 0.728,
      "step": 2815
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4414980267725923,
      "learning_rate": 3.505968711015413e-05,
      "loss": 0.7303,
      "step": 2816
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38558591754932575,
      "learning_rate": 3.5055875105787215e-05,
      "loss": 0.6412,
      "step": 2817
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43875896668560516,
      "learning_rate": 3.5052061838702225e-05,
      "loss": 0.7042,
      "step": 2818
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42327625140718644,
      "learning_rate": 3.5048247309218996e-05,
      "loss": 0.7145,
      "step": 2819
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.744083665350106,
      "learning_rate": 3.504443151765745e-05,
      "loss": 0.5932,
      "step": 2820
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5365691767682726,
      "learning_rate": 3.5040614464337576e-05,
      "loss": 0.7282,
      "step": 2821
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4474726410621523,
      "learning_rate": 3.503679614957955e-05,
      "loss": 0.6907,
      "step": 2822
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4058391693921556,
      "learning_rate": 3.503297657370359e-05,
      "loss": 0.6891,
      "step": 2823
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3848939310021853,
      "learning_rate": 3.5029155737030036e-05,
      "loss": 0.6874,
      "step": 2824
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44395020518669964,
      "learning_rate": 3.502533363987934e-05,
      "loss": 0.6955,
      "step": 2825
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4156567353733531,
      "learning_rate": 3.502151028257206e-05,
      "loss": 0.6605,
      "step": 2826
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38384082567944294,
      "learning_rate": 3.5017685665428854e-05,
      "loss": 0.6717,
      "step": 2827
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40662256904403055,
      "learning_rate": 3.501385978877048e-05,
      "loss": 0.7128,
      "step": 2828
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42800459599489954,
      "learning_rate": 3.5010032652917823e-05,
      "loss": 0.7165,
      "step": 2829
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4062497714026745,
      "learning_rate": 3.500620425819186e-05,
      "loss": 0.7203,
      "step": 2830
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38992772585805124,
      "learning_rate": 3.5002374604913666e-05,
      "loss": 0.688,
      "step": 2831
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43451774072867344,
      "learning_rate": 3.499854369340443e-05,
      "loss": 0.6777,
      "step": 2832
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4015275742108636,
      "learning_rate": 3.499471152398545e-05,
      "loss": 0.6754,
      "step": 2833
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39653551560270106,
      "learning_rate": 3.499087809697813e-05,
      "loss": 0.7166,
      "step": 2834
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4565232230756723,
      "learning_rate": 3.4987043412703956e-05,
      "loss": 0.7269,
      "step": 2835
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4053487834476132,
      "learning_rate": 3.498320747148455e-05,
      "loss": 0.6589,
      "step": 2836
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42593861107428965,
      "learning_rate": 3.497937027364164e-05,
      "loss": 0.6886,
      "step": 2837
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3610599742678747,
      "learning_rate": 3.4975531819497027e-05,
      "loss": 0.6608,
      "step": 2838
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4223159086814587,
      "learning_rate": 3.4971692109372656e-05,
      "loss": 0.7175,
      "step": 2839
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39258534724343175,
      "learning_rate": 3.496785114359054e-05,
      "loss": 0.6524,
      "step": 2840
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38752113809087446,
      "learning_rate": 3.496400892247283e-05,
      "loss": 0.7164,
      "step": 2841
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.35549157783807317,
      "learning_rate": 3.496016544634176e-05,
      "loss": 0.6822,
      "step": 2842
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3750654407797229,
      "learning_rate": 3.4956320715519675e-05,
      "loss": 0.6697,
      "step": 2843
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.428172066730733,
      "learning_rate": 3.495247473032905e-05,
      "loss": 0.6983,
      "step": 2844
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4113341634127456,
      "learning_rate": 3.494862749109241e-05,
      "loss": 0.6958,
      "step": 2845
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3658841049808469,
      "learning_rate": 3.494477899813245e-05,
      "loss": 0.6884,
      "step": 2846
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4219917521671425,
      "learning_rate": 3.494092925177192e-05,
      "loss": 0.6901,
      "step": 2847
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4263203941686721,
      "learning_rate": 3.49370782523337e-05,
      "loss": 0.7187,
      "step": 2848
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4396536950312629,
      "learning_rate": 3.493322600014076e-05,
      "loss": 0.6944,
      "step": 2849
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40055867171039716,
      "learning_rate": 3.492937249551619e-05,
      "loss": 0.7114,
      "step": 2850
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45554348854434973,
      "learning_rate": 3.492551773878318e-05,
      "loss": 0.7051,
      "step": 2851
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3981804632605841,
      "learning_rate": 3.4921661730265025e-05,
      "loss": 0.7136,
      "step": 2852
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4255167838902657,
      "learning_rate": 3.4917804470285124e-05,
      "loss": 0.7052,
      "step": 2853
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4023333439078671,
      "learning_rate": 3.491394595916697e-05,
      "loss": 0.6982,
      "step": 2854
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40784347058474696,
      "learning_rate": 3.4910086197234195e-05,
      "loss": 0.7206,
      "step": 2855
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3841812557645175,
      "learning_rate": 3.490622518481049e-05,
      "loss": 0.6886,
      "step": 2856
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4051210257034244,
      "learning_rate": 3.4902362922219673e-05,
      "loss": 0.7023,
      "step": 2857
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.34214749927938526,
      "learning_rate": 3.4898499409785685e-05,
      "loss": 0.6856,
      "step": 2858
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41498053773284155,
      "learning_rate": 3.489463464783254e-05,
      "loss": 0.6985,
      "step": 2859
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3799636996547415,
      "learning_rate": 3.489076863668437e-05,
      "loss": 0.6926,
      "step": 2860
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41851075235853236,
      "learning_rate": 3.4886901376665424e-05,
      "loss": 0.66,
      "step": 2861
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3969734610986186,
      "learning_rate": 3.488303286810004e-05,
      "loss": 0.6807,
      "step": 2862
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35801663987735904,
      "learning_rate": 3.487916311131266e-05,
      "loss": 0.6728,
      "step": 2863
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3825025591403029,
      "learning_rate": 3.487529210662784e-05,
      "loss": 0.6767,
      "step": 2864
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4323915544313304,
      "learning_rate": 3.487141985437024e-05,
      "loss": 0.7592,
      "step": 2865
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42038867005325714,
      "learning_rate": 3.486754635486461e-05,
      "loss": 0.6918,
      "step": 2866
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36009253839059135,
      "learning_rate": 3.4863671608435825e-05,
      "loss": 0.6628,
      "step": 2867
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3568726709299466,
      "learning_rate": 3.4859795615408855e-05,
      "loss": 0.6337,
      "step": 2868
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39201346800595044,
      "learning_rate": 3.4855918376108773e-05,
      "loss": 0.7105,
      "step": 2869
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.374003327549852,
      "learning_rate": 3.485203989086076e-05,
      "loss": 0.6784,
      "step": 2870
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39212537001429953,
      "learning_rate": 3.484816015999009e-05,
      "loss": 0.6788,
      "step": 2871
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4090610377617328,
      "learning_rate": 3.484427918382217e-05,
      "loss": 0.7273,
      "step": 2872
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3998142990215189,
      "learning_rate": 3.484039696268248e-05,
      "loss": 0.7024,
      "step": 2873
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37012211905778,
      "learning_rate": 3.483651349689661e-05,
      "loss": 0.6523,
      "step": 2874
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4484076308089402,
      "learning_rate": 3.4832628786790276e-05,
      "loss": 0.6773,
      "step": 2875
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.414813887878861,
      "learning_rate": 3.482874283268929e-05,
      "loss": 0.7121,
      "step": 2876
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3874549350864514,
      "learning_rate": 3.4824855634919535e-05,
      "loss": 0.6751,
      "step": 2877
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4958899261354048,
      "learning_rate": 3.482096719380705e-05,
      "loss": 0.561,
      "step": 2878
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41941100741331966,
      "learning_rate": 3.481707750967794e-05,
      "loss": 0.6905,
      "step": 2879
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4486723945270597,
      "learning_rate": 3.481318658285844e-05,
      "loss": 0.6991,
      "step": 2880
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37193357349985984,
      "learning_rate": 3.480929441367487e-05,
      "loss": 0.6747,
      "step": 2881
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4016614404379527,
      "learning_rate": 3.480540100245366e-05,
      "loss": 0.7074,
      "step": 2882
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36346796171973583,
      "learning_rate": 3.480150634952135e-05,
      "loss": 0.6569,
      "step": 2883
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4216043974097413,
      "learning_rate": 3.479761045520458e-05,
      "loss": 0.6872,
      "step": 2884
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3925069800663631,
      "learning_rate": 3.479371331983009e-05,
      "loss": 0.6481,
      "step": 2885
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35752791439066534,
      "learning_rate": 3.478981494372473e-05,
      "loss": 0.6596,
      "step": 2886
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47106195109225124,
      "learning_rate": 3.4785915327215444e-05,
      "loss": 0.7073,
      "step": 2887
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4372968236579642,
      "learning_rate": 3.478201447062931e-05,
      "loss": 0.7087,
      "step": 2888
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4049264303125144,
      "learning_rate": 3.477811237429347e-05,
      "loss": 0.6908,
      "step": 2889
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4209553272451031,
      "learning_rate": 3.477420903853519e-05,
      "loss": 0.72,
      "step": 2890
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3930854016496685,
      "learning_rate": 3.477030446368185e-05,
      "loss": 0.6499,
      "step": 2891
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40133064058205825,
      "learning_rate": 3.47663986500609e-05,
      "loss": 0.708,
      "step": 2892
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36683439357396563,
      "learning_rate": 3.4762491597999935e-05,
      "loss": 0.6694,
      "step": 2893
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38493665923276843,
      "learning_rate": 3.475858330782662e-05,
      "loss": 0.6773,
      "step": 2894
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4063181734937667,
      "learning_rate": 3.475467377986876e-05,
      "loss": 0.6669,
      "step": 2895
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3825817442535429,
      "learning_rate": 3.4750763014454216e-05,
      "loss": 0.6795,
      "step": 2896
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37800479908468054,
      "learning_rate": 3.4746851011911e-05,
      "loss": 0.6538,
      "step": 2897
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42352060961021293,
      "learning_rate": 3.47429377725672e-05,
      "loss": 0.7165,
      "step": 2898
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.378851615546077,
      "learning_rate": 3.473902329675102e-05,
      "loss": 0.7064,
      "step": 2899
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4118784707219385,
      "learning_rate": 3.4735107584790745e-05,
      "loss": 0.687,
      "step": 2900
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3926901248393396,
      "learning_rate": 3.47311906370148e-05,
      "loss": 0.6804,
      "step": 2901
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3732740011032627,
      "learning_rate": 3.472727245375167e-05,
      "loss": 0.6727,
      "step": 2902
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.24842916571403,
      "learning_rate": 3.472335303533001e-05,
      "loss": 0.5477,
      "step": 2903
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4336803186513168,
      "learning_rate": 3.4719432382078496e-05,
      "loss": 0.6883,
      "step": 2904
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40962301166551507,
      "learning_rate": 3.4715510494325965e-05,
      "loss": 0.7178,
      "step": 2905
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45609904071418705,
      "learning_rate": 3.471158737240135e-05,
      "loss": 0.7429,
      "step": 2906
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4286852072619979,
      "learning_rate": 3.4707663016633666e-05,
      "loss": 0.6844,
      "step": 2907
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47207885105841946,
      "learning_rate": 3.470373742735205e-05,
      "loss": 0.6802,
      "step": 2908
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43139969411861695,
      "learning_rate": 3.4699810604885726e-05,
      "loss": 0.6961,
      "step": 2909
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43700444526259685,
      "learning_rate": 3.469588254956404e-05,
      "loss": 0.7194,
      "step": 2910
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42326170532629886,
      "learning_rate": 3.469195326171644e-05,
      "loss": 0.6919,
      "step": 2911
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41918703623773756,
      "learning_rate": 3.468802274167246e-05,
      "loss": 0.7147,
      "step": 2912
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3822222698525924,
      "learning_rate": 3.468409098976175e-05,
      "loss": 0.6666,
      "step": 2913
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4664394368549099,
      "learning_rate": 3.468015800631406e-05,
      "loss": 0.6962,
      "step": 2914
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4241276555149059,
      "learning_rate": 3.467622379165925e-05,
      "loss": 0.6633,
      "step": 2915
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40977119330272227,
      "learning_rate": 3.467228834612727e-05,
      "loss": 0.7154,
      "step": 2916
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40600814198784485,
      "learning_rate": 3.466835167004819e-05,
      "loss": 0.7103,
      "step": 2917
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4072570004395038,
      "learning_rate": 3.466441376375217e-05,
      "loss": 0.7038,
      "step": 2918
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4357149500669239,
      "learning_rate": 3.466047462756948e-05,
      "loss": 0.7487,
      "step": 2919
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38285258014842866,
      "learning_rate": 3.465653426183049e-05,
      "loss": 0.6505,
      "step": 2920
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44211546824424347,
      "learning_rate": 3.465259266686567e-05,
      "loss": 0.6822,
      "step": 2921
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41505818671013883,
      "learning_rate": 3.4648649843005597e-05,
      "loss": 0.6802,
      "step": 2922
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35444267738466045,
      "learning_rate": 3.464470579058096e-05,
      "loss": 0.6672,
      "step": 2923
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4200375895149909,
      "learning_rate": 3.464076050992252e-05,
      "loss": 0.6863,
      "step": 2924
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43082969030794654,
      "learning_rate": 3.463681400136119e-05,
      "loss": 0.7076,
      "step": 2925
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4455771200265374,
      "learning_rate": 3.463286626522794e-05,
      "loss": 0.7118,
      "step": 2926
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3885676290127596,
      "learning_rate": 3.462891730185387e-05,
      "loss": 0.6775,
      "step": 2927
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39594188023981136,
      "learning_rate": 3.4624967111570175e-05,
      "loss": 0.6876,
      "step": 2928
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3925618761698086,
      "learning_rate": 3.462101569470815e-05,
      "loss": 0.6801,
      "step": 2929
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3407968042449339,
      "learning_rate": 3.461706305159919e-05,
      "loss": 0.6542,
      "step": 2930
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4111171524088088,
      "learning_rate": 3.461310918257482e-05,
      "loss": 0.6689,
      "step": 2931
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42605108980029766,
      "learning_rate": 3.460915408796662e-05,
      "loss": 0.69,
      "step": 2932
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38561150039305053,
      "learning_rate": 3.4605197768106306e-05,
      "loss": 0.6855,
      "step": 2933
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3832374882085912,
      "learning_rate": 3.4601240223325704e-05,
      "loss": 0.6344,
      "step": 2934
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4351138884676951,
      "learning_rate": 3.459728145395671e-05,
      "loss": 0.7078,
      "step": 2935
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4289384857342199,
      "learning_rate": 3.459332146033136e-05,
      "loss": 0.6787,
      "step": 2936
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4094174190556171,
      "learning_rate": 3.4589360242781755e-05,
      "loss": 0.6705,
      "step": 2937
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3913282994621205,
      "learning_rate": 3.458539780164013e-05,
      "loss": 0.6779,
      "step": 2938
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39083688658235843,
      "learning_rate": 3.4581434137238806e-05,
      "loss": 0.6607,
      "step": 2939
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4402348502151743,
      "learning_rate": 3.457746924991021e-05,
      "loss": 0.7079,
      "step": 2940
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4583687946745662,
      "learning_rate": 3.457350313998687e-05,
      "loss": 0.7124,
      "step": 2941
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38436011479956184,
      "learning_rate": 3.456953580780142e-05,
      "loss": 0.7256,
      "step": 2942
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4088208280540416,
      "learning_rate": 3.456556725368661e-05,
      "loss": 0.6894,
      "step": 2943
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4464234448580757,
      "learning_rate": 3.4561597477975254e-05,
      "loss": 0.6813,
      "step": 2944
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3721333351387515,
      "learning_rate": 3.455762648100031e-05,
      "loss": 0.627,
      "step": 2945
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4536473984803142,
      "learning_rate": 3.4553654263094805e-05,
      "loss": 0.6951,
      "step": 2946
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39143015468535375,
      "learning_rate": 3.45496808245919e-05,
      "loss": 0.6911,
      "step": 2947
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3726742894215327,
      "learning_rate": 3.454570616582483e-05,
      "loss": 0.6558,
      "step": 2948
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4568058534497944,
      "learning_rate": 3.4541730287126954e-05,
      "loss": 0.6992,
      "step": 2949
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4316594468987722,
      "learning_rate": 3.453775318883173e-05,
      "loss": 0.6905,
      "step": 2950
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40127996831589413,
      "learning_rate": 3.453377487127269e-05,
      "loss": 0.6671,
      "step": 2951
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3870787872279221,
      "learning_rate": 3.4529795334783513e-05,
      "loss": 0.66,
      "step": 2952
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4353101943123843,
      "learning_rate": 3.4525814579697944e-05,
      "loss": 0.6901,
      "step": 2953
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4917775545286613,
      "learning_rate": 3.452183260634985e-05,
      "loss": 0.752,
      "step": 2954
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42142494078705606,
      "learning_rate": 3.451784941507319e-05,
      "loss": 0.6968,
      "step": 2955
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35143015210702694,
      "learning_rate": 3.4513865006202035e-05,
      "loss": 0.6987,
      "step": 2956
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3820679924446912,
      "learning_rate": 3.450987938007056e-05,
      "loss": 0.6902,
      "step": 2957
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4502439836553396,
      "learning_rate": 3.450589253701301e-05,
      "loss": 0.7683,
      "step": 2958
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44848825452813845,
      "learning_rate": 3.4501904477363774e-05,
      "loss": 0.7149,
      "step": 2959
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37692460883402384,
      "learning_rate": 3.449791520145733e-05,
      "loss": 0.6853,
      "step": 2960
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45356731029851377,
      "learning_rate": 3.449392470962824e-05,
      "loss": 0.696,
      "step": 2961
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37755336851383214,
      "learning_rate": 3.44899330022112e-05,
      "loss": 0.6864,
      "step": 2962
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4568653378922341,
      "learning_rate": 3.4485940079540965e-05,
      "loss": 0.7232,
      "step": 2963
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4226282581705468,
      "learning_rate": 3.448194594195244e-05,
      "loss": 0.7392,
      "step": 2964
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3686777656739771,
      "learning_rate": 3.447795058978059e-05,
      "loss": 0.6804,
      "step": 2965
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3871266741910619,
      "learning_rate": 3.4473954023360515e-05,
      "loss": 0.6572,
      "step": 2966
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4104206332079068,
      "learning_rate": 3.44699562430274e-05,
      "loss": 0.7055,
      "step": 2967
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4040689208609329,
      "learning_rate": 3.4465957249116524e-05,
      "loss": 0.7021,
      "step": 2968
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.401613675074435,
      "learning_rate": 3.4461957041963285e-05,
      "loss": 0.6785,
      "step": 2969
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4056099197439715,
      "learning_rate": 3.4457955621903175e-05,
      "loss": 0.6785,
      "step": 2970
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.416551744859555,
      "learning_rate": 3.445395298927179e-05,
      "loss": 0.6618,
      "step": 2971
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38908410491656775,
      "learning_rate": 3.444994914440482e-05,
      "loss": 0.6966,
      "step": 2972
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4306227571410921,
      "learning_rate": 3.444594408763806e-05,
      "loss": 0.6598,
      "step": 2973
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41469396524168667,
      "learning_rate": 3.444193781930743e-05,
      "loss": 0.6553,
      "step": 2974
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3812626473618644,
      "learning_rate": 3.443793033974891e-05,
      "loss": 0.7036,
      "step": 2975
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3912594938918046,
      "learning_rate": 3.443392164929861e-05,
      "loss": 0.6846,
      "step": 2976
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3959828877919994,
      "learning_rate": 3.442991174829273e-05,
      "loss": 0.7106,
      "step": 2977
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39912518852323947,
      "learning_rate": 3.442590063706758e-05,
      "loss": 0.7009,
      "step": 2978
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4458479718935068,
      "learning_rate": 3.442188831595956e-05,
      "loss": 0.7003,
      "step": 2979
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4596241881858234,
      "learning_rate": 3.4417874785305194e-05,
      "loss": 0.7403,
      "step": 2980
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4203913740788394,
      "learning_rate": 3.441386004544108e-05,
      "loss": 0.7084,
      "step": 2981
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4302064735858868,
      "learning_rate": 3.440984409670393e-05,
      "loss": 0.6775,
      "step": 2982
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38622389554142356,
      "learning_rate": 3.440582693943056e-05,
      "loss": 0.6988,
      "step": 2983
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4073380978132409,
      "learning_rate": 3.440180857395788e-05,
      "loss": 0.7136,
      "step": 2984
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4621863870718708,
      "learning_rate": 3.439778900062291e-05,
      "loss": 0.7103,
      "step": 2985
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37498001613670756,
      "learning_rate": 3.439376821976276e-05,
      "loss": 0.6818,
      "step": 2986
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36969021475421004,
      "learning_rate": 3.438974623171466e-05,
      "loss": 0.6932,
      "step": 2987
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3711448444476949,
      "learning_rate": 3.4385723036815916e-05,
      "loss": 0.6764,
      "step": 2988
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42956071767478526,
      "learning_rate": 3.438169863540395e-05,
      "loss": 0.6901,
      "step": 2989
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.375321259506802,
      "learning_rate": 3.4377673027816305e-05,
      "loss": 0.6922,
      "step": 2990
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3950317831422347,
      "learning_rate": 3.4373646214390576e-05,
      "loss": 0.7018,
      "step": 2991
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43923126901029774,
      "learning_rate": 3.4369618195464494e-05,
      "loss": 0.709,
      "step": 2992
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38573027055598574,
      "learning_rate": 3.436558897137589e-05,
      "loss": 0.6901,
      "step": 2993
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.384065902578573,
      "learning_rate": 3.436155854246269e-05,
      "loss": 0.6882,
      "step": 2994
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3990806181066842,
      "learning_rate": 3.435752690906293e-05,
      "loss": 0.6913,
      "step": 2995
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3605483823337594,
      "learning_rate": 3.435349407151471e-05,
      "loss": 0.6866,
      "step": 2996
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4447739051556688,
      "learning_rate": 3.434946003015628e-05,
      "loss": 0.7092,
      "step": 2997
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4316719480560012,
      "learning_rate": 3.434542478532598e-05,
      "loss": 0.7441,
      "step": 2998
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4124835696423802,
      "learning_rate": 3.4341388337362215e-05,
      "loss": 0.7045,
      "step": 2999
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4399141049861426,
      "learning_rate": 3.4337350686603536e-05,
      "loss": 0.7042,
      "step": 3000
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41672653215780925,
      "learning_rate": 3.433331183338857e-05,
      "loss": 0.7197,
      "step": 3001
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4371164739922415,
      "learning_rate": 3.432927177805605e-05,
      "loss": 0.6965,
      "step": 3002
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4048589530897964,
      "learning_rate": 3.432523052094481e-05,
      "loss": 0.7182,
      "step": 3003
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4096850505829695,
      "learning_rate": 3.432118806239379e-05,
      "loss": 0.6738,
      "step": 3004
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4701541318600713,
      "learning_rate": 3.431714440274202e-05,
      "loss": 0.7226,
      "step": 3005
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.34654342889599515,
      "learning_rate": 3.431309954232864e-05,
      "loss": 0.69,
      "step": 3006
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3815838130844042,
      "learning_rate": 3.430905348149288e-05,
      "loss": 0.693,
      "step": 3007
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37737004602174623,
      "learning_rate": 3.4305006220574096e-05,
      "loss": 0.6817,
      "step": 3008
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41295981944230564,
      "learning_rate": 3.430095775991172e-05,
      "loss": 0.6895,
      "step": 3009
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39889066566298365,
      "learning_rate": 3.429690809984528e-05,
      "loss": 0.6723,
      "step": 3010
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.356825022323878,
      "learning_rate": 3.4292857240714425e-05,
      "loss": 0.705,
      "step": 3011
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41125880709550605,
      "learning_rate": 3.428880518285889e-05,
      "loss": 0.6771,
      "step": 3012
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3661524842942576,
      "learning_rate": 3.428475192661853e-05,
      "loss": 0.679,
      "step": 3013
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39678547955404186,
      "learning_rate": 3.428069747233327e-05,
      "loss": 0.695,
      "step": 3014
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38253832513182223,
      "learning_rate": 3.427664182034317e-05,
      "loss": 0.7,
      "step": 3015
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4300064982885952,
      "learning_rate": 3.4272584970988356e-05,
      "loss": 0.6994,
      "step": 3016
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37079444723884897,
      "learning_rate": 3.4268526924609066e-05,
      "loss": 0.6925,
      "step": 3017
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3964735641072779,
      "learning_rate": 3.4264467681545674e-05,
      "loss": 0.6826,
      "step": 3018
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38286969693806444,
      "learning_rate": 3.426040724213859e-05,
      "loss": 0.6461,
      "step": 3019
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42623957973081433,
      "learning_rate": 3.425634560672838e-05,
      "loss": 0.7187,
      "step": 3020
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4012258095641024,
      "learning_rate": 3.4252282775655685e-05,
      "loss": 0.6818,
      "step": 3021
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4208063118450097,
      "learning_rate": 3.424821874926123e-05,
      "loss": 0.6903,
      "step": 3022
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42058559567993714,
      "learning_rate": 3.424415352788589e-05,
      "loss": 0.7059,
      "step": 3023
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3966768432959187,
      "learning_rate": 3.42400871118706e-05,
      "loss": 0.6695,
      "step": 3024
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4080544856906523,
      "learning_rate": 3.423601950155639e-05,
      "loss": 0.6737,
      "step": 3025
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38230528308087575,
      "learning_rate": 3.4231950697284424e-05,
      "loss": 0.6857,
      "step": 3026
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35623248537397995,
      "learning_rate": 3.4227880699395935e-05,
      "loss": 0.6616,
      "step": 3027
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4446881627233456,
      "learning_rate": 3.4223809508232285e-05,
      "loss": 0.7279,
      "step": 3028
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3282611264750742,
      "learning_rate": 3.42197371241349e-05,
      "loss": 0.6719,
      "step": 3029
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42180812632439935,
      "learning_rate": 3.421566354744534e-05,
      "loss": 0.6974,
      "step": 3030
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4378404935785187,
      "learning_rate": 3.421158877850525e-05,
      "loss": 0.6759,
      "step": 3031
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4441642307603048,
      "learning_rate": 3.420751281765637e-05,
      "loss": 0.7101,
      "step": 3032
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3879989735523547,
      "learning_rate": 3.420343566524054e-05,
      "loss": 0.6932,
      "step": 3033
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.40147005289323556,
      "learning_rate": 3.419935732159972e-05,
      "loss": 0.6525,
      "step": 3034
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39961511966260815,
      "learning_rate": 3.419527778707595e-05,
      "loss": 0.6845,
      "step": 3035
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36766267874267444,
      "learning_rate": 3.4191197062011374e-05,
      "loss": 0.6659,
      "step": 3036
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3872493408272944,
      "learning_rate": 3.418711514674824e-05,
      "loss": 0.6784,
      "step": 3037
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43481586658032767,
      "learning_rate": 3.418303204162889e-05,
      "loss": 0.7203,
      "step": 3038
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41079908653957997,
      "learning_rate": 3.417894774699578e-05,
      "loss": 0.7065,
      "step": 3039
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38601115200253605,
      "learning_rate": 3.4174862263191446e-05,
      "loss": 0.694,
      "step": 3040
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43183295195393445,
      "learning_rate": 3.417077559055852e-05,
      "loss": 0.6846,
      "step": 3041
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.404025861538248,
      "learning_rate": 3.4166687729439774e-05,
      "loss": 0.7489,
      "step": 3042
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42050873968641855,
      "learning_rate": 3.4162598680178026e-05,
      "loss": 0.6942,
      "step": 3043
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3611589496283211,
      "learning_rate": 3.415850844311624e-05,
      "loss": 0.6453,
      "step": 3044
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4019320844517033,
      "learning_rate": 3.415441701859744e-05,
      "loss": 0.6995,
      "step": 3045
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4566711990152414,
      "learning_rate": 3.415032440696478e-05,
      "loss": 0.715,
      "step": 3046
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4218826611249515,
      "learning_rate": 3.414623060856151e-05,
      "loss": 0.6379,
      "step": 3047
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4149158715459493,
      "learning_rate": 3.4142135623730954e-05,
      "loss": 0.7127,
      "step": 3048
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4572229774609052,
      "learning_rate": 3.413803945281657e-05,
      "loss": 0.6929,
      "step": 3049
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45512864000042025,
      "learning_rate": 3.413394209616188e-05,
      "loss": 0.7315,
      "step": 3050
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42307111162268585,
      "learning_rate": 3.412984355411055e-05,
      "loss": 0.6568,
      "step": 3051
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4472211753560678,
      "learning_rate": 3.41257438270063e-05,
      "loss": 0.6816,
      "step": 3052
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.44690627478233064,
      "learning_rate": 3.4121642915192974e-05,
      "loss": 0.6564,
      "step": 3053
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4130818813310843,
      "learning_rate": 3.411754081901451e-05,
      "loss": 0.7202,
      "step": 3054
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35165818245176095,
      "learning_rate": 3.411343753881495e-05,
      "loss": 0.6708,
      "step": 3055
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4182288346078891,
      "learning_rate": 3.410933307493843e-05,
      "loss": 0.7317,
      "step": 3056
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.40396508423918137,
      "learning_rate": 3.4105227427729176e-05,
      "loss": 0.7125,
      "step": 3057
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38419838761343417,
      "learning_rate": 3.410112059753154e-05,
      "loss": 0.6845,
      "step": 3058
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3672006801692474,
      "learning_rate": 3.409701258468995e-05,
      "loss": 0.6457,
      "step": 3059
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42569957381131035,
      "learning_rate": 3.4092903389548936e-05,
      "loss": 0.6822,
      "step": 3060
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3778226256793487,
      "learning_rate": 3.408879301245313e-05,
      "loss": 0.6875,
      "step": 3061
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3836337846845606,
      "learning_rate": 3.408468145374728e-05,
      "loss": 0.6773,
      "step": 3062
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41299727055605395,
      "learning_rate": 3.4080568713776206e-05,
      "loss": 0.6926,
      "step": 3063
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4373991972166924,
      "learning_rate": 3.407645479288483e-05,
      "loss": 0.7321,
      "step": 3064
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42224880865197273,
      "learning_rate": 3.407233969141819e-05,
      "loss": 0.6873,
      "step": 3065
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4317531716275489,
      "learning_rate": 3.406822340972142e-05,
      "loss": 0.7631,
      "step": 3066
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48973659715365137,
      "learning_rate": 3.406410594813975e-05,
      "loss": 0.6746,
      "step": 3067
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3994079596951947,
      "learning_rate": 3.405998730701849e-05,
      "loss": 0.6832,
      "step": 3068
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3659396661364973,
      "learning_rate": 3.4055867486703075e-05,
      "loss": 0.6439,
      "step": 3069
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.40178022525699747,
      "learning_rate": 3.4051746487539036e-05,
      "loss": 0.6652,
      "step": 3070
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4574687086036439,
      "learning_rate": 3.404762430987198e-05,
      "loss": 0.7487,
      "step": 3071
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4034266177934742,
      "learning_rate": 3.4043500954047646e-05,
      "loss": 0.6947,
      "step": 3072
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42900551047600827,
      "learning_rate": 3.403937642041184e-05,
      "loss": 0.69,
      "step": 3073
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.497062432072088,
      "learning_rate": 3.403525070931049e-05,
      "loss": 0.6906,
      "step": 3074
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4039779244039803,
      "learning_rate": 3.403112382108962e-05,
      "loss": 0.7059,
      "step": 3075
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.491702392175304,
      "learning_rate": 3.402699575609534e-05,
      "loss": 0.762,
      "step": 3076
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39586515261774174,
      "learning_rate": 3.4022866514673864e-05,
      "loss": 0.7137,
      "step": 3077
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4465348615544572,
      "learning_rate": 3.40187360971715e-05,
      "loss": 0.7086,
      "step": 3078
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4062155179730297,
      "learning_rate": 3.401460450393468e-05,
      "loss": 0.7198,
      "step": 3079
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42407279635616413,
      "learning_rate": 3.40104717353099e-05,
      "loss": 0.6548,
      "step": 3080
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4044608811364438,
      "learning_rate": 3.400633779164378e-05,
      "loss": 0.6471,
      "step": 3081
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3993475085596615,
      "learning_rate": 3.400220267328303e-05,
      "loss": 0.6762,
      "step": 3082
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4198007200491046,
      "learning_rate": 3.399806638057444e-05,
      "loss": 0.7415,
      "step": 3083
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.34985093440551457,
      "learning_rate": 3.3993928913864934e-05,
      "loss": 0.6899,
      "step": 3084
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.385959433215788,
      "learning_rate": 3.398979027350151e-05,
      "loss": 0.6738,
      "step": 3085
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3672873408223706,
      "learning_rate": 3.398565045983127e-05,
      "loss": 0.6616,
      "step": 3086
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3882167961922963,
      "learning_rate": 3.398150947320142e-05,
      "loss": 0.6872,
      "step": 3087
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38684126658836493,
      "learning_rate": 3.397736731395925e-05,
      "loss": 0.6978,
      "step": 3088
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43273808517898094,
      "learning_rate": 3.397322398245217e-05,
      "loss": 0.659,
      "step": 3089
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4274061973722448,
      "learning_rate": 3.396907947902767e-05,
      "loss": 0.6905,
      "step": 3090
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3781902553840013,
      "learning_rate": 3.396493380403334e-05,
      "loss": 0.6694,
      "step": 3091
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3944612141721664,
      "learning_rate": 3.396078695781688e-05,
      "loss": 0.6812,
      "step": 3092
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43524046632910673,
      "learning_rate": 3.395663894072607e-05,
      "loss": 0.7113,
      "step": 3093
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4285738170161883,
      "learning_rate": 3.395248975310882e-05,
      "loss": 0.6572,
      "step": 3094
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3753617975642525,
      "learning_rate": 3.39483393953131e-05,
      "loss": 0.7112,
      "step": 3095
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40173170830774935,
      "learning_rate": 3.3944187867687e-05,
      "loss": 0.6581,
      "step": 3096
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4288825976714959,
      "learning_rate": 3.39400351705787e-05,
      "loss": 0.6762,
      "step": 3097
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38849563439772816,
      "learning_rate": 3.3935881304336495e-05,
      "loss": 0.68,
      "step": 3098
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.34561761233645844,
      "learning_rate": 3.393172626930874e-05,
      "loss": 0.647,
      "step": 3099
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4166249109473709,
      "learning_rate": 3.3927570065843945e-05,
      "loss": 0.6893,
      "step": 3100
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3852472479329609,
      "learning_rate": 3.392341269429066e-05,
      "loss": 0.6521,
      "step": 3101
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4119476237414874,
      "learning_rate": 3.391925415499756e-05,
      "loss": 0.7139,
      "step": 3102
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43945003604062854,
      "learning_rate": 3.391509444831344e-05,
      "loss": 0.6807,
      "step": 3103
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.34882165589665387,
      "learning_rate": 3.391093357458715e-05,
      "loss": 0.6497,
      "step": 3104
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3591731213302816,
      "learning_rate": 3.390677153416765e-05,
      "loss": 0.6664,
      "step": 3105
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4593993310126632,
      "learning_rate": 3.390260832740403e-05,
      "loss": 0.6937,
      "step": 3106
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.46775389271141576,
      "learning_rate": 3.389844395464543e-05,
      "loss": 0.7459,
      "step": 3107
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45721375845342604,
      "learning_rate": 3.389427841624113e-05,
      "loss": 0.7041,
      "step": 3108
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4145167856156381,
      "learning_rate": 3.389011171254047e-05,
      "loss": 0.6988,
      "step": 3109
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3711952398196838,
      "learning_rate": 3.388594384389292e-05,
      "loss": 0.5448,
      "step": 3110
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4192915513552445,
      "learning_rate": 3.388177481064803e-05,
      "loss": 0.6995,
      "step": 3111
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4118601615153543,
      "learning_rate": 3.3877604613155455e-05,
      "loss": 0.6807,
      "step": 3112
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4507373974988738,
      "learning_rate": 3.3873433251764935e-05,
      "loss": 0.7359,
      "step": 3113
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38188462791398,
      "learning_rate": 3.386926072682632e-05,
      "loss": 0.6439,
      "step": 3114
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3781243205049435,
      "learning_rate": 3.386508703868957e-05,
      "loss": 0.6781,
      "step": 3115
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4290861937975335,
      "learning_rate": 3.38609121877047e-05,
      "loss": 0.7103,
      "step": 3116
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37377577904376685,
      "learning_rate": 3.385673617422187e-05,
      "loss": 0.6446,
      "step": 3117
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.431398477802779,
      "learning_rate": 3.385255899859132e-05,
      "loss": 0.6981,
      "step": 3118
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4122879912833896,
      "learning_rate": 3.384838066116337e-05,
      "loss": 0.7022,
      "step": 3119
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37574538011210107,
      "learning_rate": 3.3844201162288456e-05,
      "loss": 0.6243,
      "step": 3120
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3749456704154309,
      "learning_rate": 3.3840020502317124e-05,
      "loss": 0.6446,
      "step": 3121
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37766310100547046,
      "learning_rate": 3.3835838681599975e-05,
      "loss": 0.6811,
      "step": 3122
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3806010940636295,
      "learning_rate": 3.383165570048774e-05,
      "loss": 0.6483,
      "step": 3123
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.419752427337954,
      "learning_rate": 3.3827471559331256e-05,
      "loss": 0.6826,
      "step": 3124
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4146829347523694,
      "learning_rate": 3.382328625848143e-05,
      "loss": 0.7314,
      "step": 3125
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3784919894879714,
      "learning_rate": 3.381909979828927e-05,
      "loss": 0.6654,
      "step": 3126
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38773508846300897,
      "learning_rate": 3.381491217910591e-05,
      "loss": 0.6711,
      "step": 3127
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3361413517708708,
      "learning_rate": 3.3810723401282544e-05,
      "loss": 0.6629,
      "step": 3128
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39822163488016793,
      "learning_rate": 3.380653346517049e-05,
      "loss": 0.6778,
      "step": 3129
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37322375734816765,
      "learning_rate": 3.380234237112115e-05,
      "loss": 0.6733,
      "step": 3130
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42465325440329815,
      "learning_rate": 3.379815011948602e-05,
      "loss": 0.6552,
      "step": 3131
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3854851688477519,
      "learning_rate": 3.37939567106167e-05,
      "loss": 0.6638,
      "step": 3132
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3586241227078141,
      "learning_rate": 3.378976214486489e-05,
      "loss": 0.6884,
      "step": 3133
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3863764948924527,
      "learning_rate": 3.378556642258239e-05,
      "loss": 0.6535,
      "step": 3134
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3892845602820734,
      "learning_rate": 3.378136954412108e-05,
      "loss": 0.6592,
      "step": 3135
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37752685352112025,
      "learning_rate": 3.377717150983295e-05,
      "loss": 0.6856,
      "step": 3136
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3972762322858946,
      "learning_rate": 3.377297232007008e-05,
      "loss": 0.6982,
      "step": 3137
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.588639815547305,
      "learning_rate": 3.376877197518466e-05,
      "loss": 0.5063,
      "step": 3138
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.433182659036436,
      "learning_rate": 3.376457047552896e-05,
      "loss": 0.7119,
      "step": 3139
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3980303560977711,
      "learning_rate": 3.3760367821455364e-05,
      "loss": 0.6813,
      "step": 3140
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4695233915014608,
      "learning_rate": 3.375616401331633e-05,
      "loss": 0.7302,
      "step": 3141
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3994685945003732,
      "learning_rate": 3.3751959051464434e-05,
      "loss": 0.6818,
      "step": 3142
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44347224551784825,
      "learning_rate": 3.374775293625234e-05,
      "loss": 0.7339,
      "step": 3143
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45044207556742627,
      "learning_rate": 3.374354566803281e-05,
      "loss": 0.7261,
      "step": 3144
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38120456674547054,
      "learning_rate": 3.373933724715871e-05,
      "loss": 0.6542,
      "step": 3145
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42461880613635417,
      "learning_rate": 3.373512767398299e-05,
      "loss": 0.6993,
      "step": 3146
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3623160449058962,
      "learning_rate": 3.3730916948858684e-05,
      "loss": 0.7094,
      "step": 3147
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42728710865052166,
      "learning_rate": 3.372670507213896e-05,
      "loss": 0.7369,
      "step": 3148
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4147458732164098,
      "learning_rate": 3.372249204417707e-05,
      "loss": 0.6824,
      "step": 3149
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3624209560681706,
      "learning_rate": 3.371827786532634e-05,
      "loss": 0.6768,
      "step": 3150
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41343313659015135,
      "learning_rate": 3.371406253594021e-05,
      "loss": 0.6916,
      "step": 3151
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3743222103745201,
      "learning_rate": 3.3709846056372224e-05,
      "loss": 0.6847,
      "step": 3152
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43483520722653474,
      "learning_rate": 3.3705628426976e-05,
      "loss": 0.7289,
      "step": 3153
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40717312700454095,
      "learning_rate": 3.370140964810527e-05,
      "loss": 0.6702,
      "step": 3154
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4090997797488736,
      "learning_rate": 3.369718972011387e-05,
      "loss": 0.6519,
      "step": 3155
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4511092657874476,
      "learning_rate": 3.3692968643355706e-05,
      "loss": 0.7332,
      "step": 3156
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4266582508651525,
      "learning_rate": 3.3688746418184795e-05,
      "loss": 0.7223,
      "step": 3157
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4374891437477289,
      "learning_rate": 3.368452304495525e-05,
      "loss": 0.6884,
      "step": 3158
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37038496077739197,
      "learning_rate": 3.3680298524021295e-05,
      "loss": 0.647,
      "step": 3159
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3796995365232026,
      "learning_rate": 3.367607285573722e-05,
      "loss": 0.6941,
      "step": 3160
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.46338525309409156,
      "learning_rate": 3.367184604045743e-05,
      "loss": 0.7334,
      "step": 3161
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38403493764559354,
      "learning_rate": 3.3667618078536426e-05,
      "loss": 0.7215,
      "step": 3162
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44426044130515907,
      "learning_rate": 3.36633889703288e-05,
      "loss": 0.7078,
      "step": 3163
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41006321283640335,
      "learning_rate": 3.365915871618923e-05,
      "loss": 0.6973,
      "step": 3164
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39820322048170004,
      "learning_rate": 3.365492731647253e-05,
      "loss": 0.6644,
      "step": 3165
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4500302534339835,
      "learning_rate": 3.365069477153356e-05,
      "loss": 0.631,
      "step": 3166
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4111555455160287,
      "learning_rate": 3.364646108172731e-05,
      "loss": 0.6593,
      "step": 3167
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37098701634629444,
      "learning_rate": 3.364222624740885e-05,
      "loss": 0.614,
      "step": 3168
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38083596060234,
      "learning_rate": 3.3637990268933345e-05,
      "loss": 0.6579,
      "step": 3169
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41091608627362736,
      "learning_rate": 3.363375314665607e-05,
      "loss": 0.6892,
      "step": 3170
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4110187023983172,
      "learning_rate": 3.362951488093238e-05,
      "loss": 0.6862,
      "step": 3171
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3464014586109916,
      "learning_rate": 3.362527547211774e-05,
      "loss": 0.6767,
      "step": 3172
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3991651928714419,
      "learning_rate": 3.362103492056771e-05,
      "loss": 0.6822,
      "step": 3173
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5018848763389587,
      "learning_rate": 3.361679322663792e-05,
      "loss": 0.5534,
      "step": 3174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43479208923935214,
      "learning_rate": 3.361255039068413e-05,
      "loss": 0.6784,
      "step": 3175
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4317659613210629,
      "learning_rate": 3.3608306413062174e-05,
      "loss": 0.7284,
      "step": 3176
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45584690283215945,
      "learning_rate": 3.3604061294128e-05,
      "loss": 0.721,
      "step": 3177
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.49197037480453476,
      "learning_rate": 3.359981503423764e-05,
      "loss": 0.6847,
      "step": 3178
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2164772077616073,
      "learning_rate": 3.359556763374721e-05,
      "loss": 0.5096,
      "step": 3179
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44653684684942335,
      "learning_rate": 3.3591319093012935e-05,
      "loss": 0.7135,
      "step": 3180
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40850554662713184,
      "learning_rate": 3.3587069412391154e-05,
      "loss": 0.7195,
      "step": 3181
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40932043863557954,
      "learning_rate": 3.358281859223826e-05,
      "loss": 0.673,
      "step": 3182
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4644431251245942,
      "learning_rate": 3.357856663291078e-05,
      "loss": 0.7218,
      "step": 3183
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4320041435602537,
      "learning_rate": 3.357431353476532e-05,
      "loss": 0.6983,
      "step": 3184
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4510558810049729,
      "learning_rate": 3.357005929815857e-05,
      "loss": 0.7022,
      "step": 3185
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37436383786426924,
      "learning_rate": 3.356580392344734e-05,
      "loss": 0.636,
      "step": 3186
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44637741840263145,
      "learning_rate": 3.356154741098852e-05,
      "loss": 0.7346,
      "step": 3187
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4014438447022777,
      "learning_rate": 3.355728976113909e-05,
      "loss": 0.6727,
      "step": 3188
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3772850267640016,
      "learning_rate": 3.355303097425615e-05,
      "loss": 0.6263,
      "step": 3189
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3985349059077577,
      "learning_rate": 3.354877105069686e-05,
      "loss": 0.6721,
      "step": 3190
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4094038017426494,
      "learning_rate": 3.354450999081851e-05,
      "loss": 0.675,
      "step": 3191
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34242872777502037,
      "learning_rate": 3.354024779497846e-05,
      "loss": 0.6517,
      "step": 3192
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3545144783187117,
      "learning_rate": 3.353598446353418e-05,
      "loss": 0.6817,
      "step": 3193
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.40953557704656196,
      "learning_rate": 3.353171999684324e-05,
      "loss": 0.677,
      "step": 3194
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34175577101266663,
      "learning_rate": 3.3527454395263284e-05,
      "loss": 0.6853,
      "step": 3195
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4071278485596011,
      "learning_rate": 3.3523187659152057e-05,
      "loss": 0.7076,
      "step": 3196
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3585866266486323,
      "learning_rate": 3.351891978886742e-05,
      "loss": 0.6601,
      "step": 3197
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3973015914856631,
      "learning_rate": 3.351465078476731e-05,
      "loss": 0.6743,
      "step": 3198
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3446024160053511,
      "learning_rate": 3.3510380647209746e-05,
      "loss": 0.6561,
      "step": 3199
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.423042953244413,
      "learning_rate": 3.3506109376552887e-05,
      "loss": 0.7042,
      "step": 3200
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4042342777653674,
      "learning_rate": 3.350183697315494e-05,
      "loss": 0.6549,
      "step": 3201
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3965174261181527,
      "learning_rate": 3.3497563437374225e-05,
      "loss": 0.6924,
      "step": 3202
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4174386663537041,
      "learning_rate": 3.3493288769569176e-05,
      "loss": 0.6848,
      "step": 3203
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2350092167431503,
      "learning_rate": 3.348901297009829e-05,
      "loss": 0.6131,
      "step": 3204
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42726750228926474,
      "learning_rate": 3.348473603932018e-05,
      "loss": 0.671,
      "step": 3205
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4247218397253032,
      "learning_rate": 3.3480457977593537e-05,
      "loss": 0.6907,
      "step": 3206
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3835057761798008,
      "learning_rate": 3.3476178785277174e-05,
      "loss": 0.6432,
      "step": 3207
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3920401678062584,
      "learning_rate": 3.347189846272996e-05,
      "loss": 0.6638,
      "step": 3208
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4217330980859491,
      "learning_rate": 3.3467617010310896e-05,
      "loss": 0.7436,
      "step": 3209
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37842516683577737,
      "learning_rate": 3.346333442837906e-05,
      "loss": 0.6774,
      "step": 3210
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41087763245112463,
      "learning_rate": 3.3459050717293625e-05,
      "loss": 0.7104,
      "step": 3211
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.36267772673986753,
      "learning_rate": 3.345476587741386e-05,
      "loss": 0.6743,
      "step": 3212
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4128046346344657,
      "learning_rate": 3.3450479909099134e-05,
      "loss": 0.652,
      "step": 3213
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3937585504182845,
      "learning_rate": 3.34461928127089e-05,
      "loss": 0.6864,
      "step": 3214
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.35546162478607335,
      "learning_rate": 3.3441904588602714e-05,
      "loss": 0.7112,
      "step": 3215
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3607479184464036,
      "learning_rate": 3.343761523714022e-05,
      "loss": 0.6829,
      "step": 3216
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.40822770869296204,
      "learning_rate": 3.343332475868116e-05,
      "loss": 0.7062,
      "step": 3217
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34044560008875546,
      "learning_rate": 3.342903315358538e-05,
      "loss": 0.6516,
      "step": 3218
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43984407111236057,
      "learning_rate": 3.342474042221282e-05,
      "loss": 0.6856,
      "step": 3219
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4259989332861658,
      "learning_rate": 3.3420446564923476e-05,
      "loss": 0.7377,
      "step": 3220
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3874103681445269,
      "learning_rate": 3.34161515820775e-05,
      "loss": 0.6668,
      "step": 3221
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42297340320346705,
      "learning_rate": 3.341185547403509e-05,
      "loss": 0.6974,
      "step": 3222
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43137947711333424,
      "learning_rate": 3.340755824115655e-05,
      "loss": 0.69,
      "step": 3223
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38543491753648806,
      "learning_rate": 3.34032598838023e-05,
      "loss": 0.6533,
      "step": 3224
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.418512539682878,
      "learning_rate": 3.339896040233283e-05,
      "loss": 0.6862,
      "step": 3225
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3665682943610953,
      "learning_rate": 3.339465979710873e-05,
      "loss": 0.6852,
      "step": 3226
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.39994054548172603,
      "learning_rate": 3.3390358068490694e-05,
      "loss": 0.6832,
      "step": 3227
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4858651769041279,
      "learning_rate": 3.338605521683949e-05,
      "loss": 0.6881,
      "step": 3228
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4440735298784618,
      "learning_rate": 3.3381751242516014e-05,
      "loss": 0.7267,
      "step": 3229
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4940766509283425,
      "learning_rate": 3.3377446145881214e-05,
      "loss": 0.7104,
      "step": 3230
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3927471194739233,
      "learning_rate": 3.3373139927296164e-05,
      "loss": 0.6867,
      "step": 3231
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41751921649814366,
      "learning_rate": 3.336883258712202e-05,
      "loss": 0.6919,
      "step": 3232
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41773935617428154,
      "learning_rate": 3.336452412572003e-05,
      "loss": 0.6288,
      "step": 3233
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.46682334014201576,
      "learning_rate": 3.3360214543451546e-05,
      "loss": 0.7004,
      "step": 3234
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43807282051449986,
      "learning_rate": 3.3355903840677996e-05,
      "loss": 0.7327,
      "step": 3235
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3704065122479828,
      "learning_rate": 3.3351592017760925e-05,
      "loss": 0.6811,
      "step": 3236
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.458400128710563,
      "learning_rate": 3.334727907506195e-05,
      "loss": 0.6507,
      "step": 3237
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41128624196864183,
      "learning_rate": 3.3342965012942806e-05,
      "loss": 0.6934,
      "step": 3238
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3870713001989359,
      "learning_rate": 3.33386498317653e-05,
      "loss": 0.6586,
      "step": 3239
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3851068782293912,
      "learning_rate": 3.333433353189134e-05,
      "loss": 0.6357,
      "step": 3240
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3962258533274322,
      "learning_rate": 3.333001611368293e-05,
      "loss": 0.6579,
      "step": 3241
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4164310569470384,
      "learning_rate": 3.332569757750217e-05,
      "loss": 0.7044,
      "step": 3242
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37271996079657355,
      "learning_rate": 3.332137792371125e-05,
      "loss": 0.6764,
      "step": 3243
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4093830923496513,
      "learning_rate": 3.3317057152672446e-05,
      "loss": 0.6984,
      "step": 3244
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45468258229227126,
      "learning_rate": 3.3312735264748136e-05,
      "loss": 0.6957,
      "step": 3245
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4372549629311486,
      "learning_rate": 3.330841226030081e-05,
      "loss": 0.7181,
      "step": 3246
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37605010988059856,
      "learning_rate": 3.330408813969301e-05,
      "loss": 0.7068,
      "step": 3247
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.36052499975594676,
      "learning_rate": 3.329976290328741e-05,
      "loss": 0.6677,
      "step": 3248
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43839701724554325,
      "learning_rate": 3.329543655144676e-05,
      "loss": 0.7171,
      "step": 3249
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44774866565575416,
      "learning_rate": 3.32911090845339e-05,
      "loss": 0.7026,
      "step": 3250
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.35576652563735484,
      "learning_rate": 3.328678050291177e-05,
      "loss": 0.6455,
      "step": 3251
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4196196483504902,
      "learning_rate": 3.32824508069434e-05,
      "loss": 0.6681,
      "step": 3252
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38251837404796074,
      "learning_rate": 3.327811999699193e-05,
      "loss": 0.6768,
      "step": 3253
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4301687694304914,
      "learning_rate": 3.327378807342057e-05,
      "loss": 0.7005,
      "step": 3254
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3966095445820361,
      "learning_rate": 3.3269455036592644e-05,
      "loss": 0.7046,
      "step": 3255
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.39975879920592844,
      "learning_rate": 3.326512088687154e-05,
      "loss": 0.6928,
      "step": 3256
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34456930445660394,
      "learning_rate": 3.326078562462077e-05,
      "loss": 0.6401,
      "step": 3257
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34832940468127044,
      "learning_rate": 3.325644925020393e-05,
      "loss": 0.6519,
      "step": 3258
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4226825418249025,
      "learning_rate": 3.325211176398469e-05,
      "loss": 0.6734,
      "step": 3259
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4711172524893324,
      "learning_rate": 3.324777316632685e-05,
      "loss": 0.6573,
      "step": 3260
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37639682425430654,
      "learning_rate": 3.324343345759427e-05,
      "loss": 0.6698,
      "step": 3261
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3947182768957293,
      "learning_rate": 3.323909263815091e-05,
      "loss": 0.7442,
      "step": 3262
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4645711634063044,
      "learning_rate": 3.323475070836085e-05,
      "loss": 0.7131,
      "step": 3263
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4461921423741582,
      "learning_rate": 3.323040766858824e-05,
      "loss": 0.7024,
      "step": 3264
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.39716984308839276,
      "learning_rate": 3.32260635191973e-05,
      "loss": 0.6685,
      "step": 3265
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4009827473393249,
      "learning_rate": 3.322171826055239e-05,
      "loss": 0.6735,
      "step": 3266
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3818420689601539,
      "learning_rate": 3.3217371893017945e-05,
      "loss": 0.6818,
      "step": 3267
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7322433343713575,
      "learning_rate": 3.3213024416958474e-05,
      "loss": 0.5905,
      "step": 3268
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4130865176153638,
      "learning_rate": 3.320867583273861e-05,
      "loss": 0.673,
      "step": 3269
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3545920163185232,
      "learning_rate": 3.320432614072305e-05,
      "loss": 0.6471,
      "step": 3270
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4356169093224098,
      "learning_rate": 3.319997534127661e-05,
      "loss": 0.6756,
      "step": 3271
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42679062166828285,
      "learning_rate": 3.319562343476417e-05,
      "loss": 0.6862,
      "step": 3272
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3010021639132001,
      "learning_rate": 3.3191270421550734e-05,
      "loss": 0.5822,
      "step": 3273
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45842081534850393,
      "learning_rate": 3.318691630200138e-05,
      "loss": 0.6674,
      "step": 3274
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42226379814684695,
      "learning_rate": 3.318256107648128e-05,
      "loss": 0.6791,
      "step": 3275
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4326969252585841,
      "learning_rate": 3.3178204745355706e-05,
      "loss": 0.6836,
      "step": 3276
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3550737856782893,
      "learning_rate": 3.317384730899001e-05,
      "loss": 0.6716,
      "step": 3277
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4513745748075638,
      "learning_rate": 3.3169488767749656e-05,
      "loss": 0.6908,
      "step": 3278
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4160912586523597,
      "learning_rate": 3.316512912200017e-05,
      "loss": 0.6912,
      "step": 3279
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42728902743860475,
      "learning_rate": 3.316076837210722e-05,
      "loss": 0.6952,
      "step": 3280
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45001184677198575,
      "learning_rate": 3.315640651843651e-05,
      "loss": 0.6987,
      "step": 3281
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42577455338567455,
      "learning_rate": 3.315204356135389e-05,
      "loss": 0.6762,
      "step": 3282
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4360673220455207,
      "learning_rate": 3.3147679501225245e-05,
      "loss": 0.6985,
      "step": 3283
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4151012462461316,
      "learning_rate": 3.31433143384166e-05,
      "loss": 0.6667,
      "step": 3284
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3785059806541501,
      "learning_rate": 3.313894807329406e-05,
      "loss": 0.6403,
      "step": 3285
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4611610887885399,
      "learning_rate": 3.313458070622381e-05,
      "loss": 0.6714,
      "step": 3286
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1835998860163113,
      "learning_rate": 3.313021223757214e-05,
      "loss": 0.4818,
      "step": 3287
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4525775363790053,
      "learning_rate": 3.312584266770543e-05,
      "loss": 0.7263,
      "step": 3288
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.383414061530228,
      "learning_rate": 3.3121471996990134e-05,
      "loss": 0.6216,
      "step": 3289
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3932457538298182,
      "learning_rate": 3.311710022579284e-05,
      "loss": 0.6809,
      "step": 3290
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41481640287732024,
      "learning_rate": 3.311272735448019e-05,
      "loss": 0.6861,
      "step": 3291
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.40071243116205557,
      "learning_rate": 3.310835338341893e-05,
      "loss": 0.6915,
      "step": 3292
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.40687895987688466,
      "learning_rate": 3.310397831297591e-05,
      "loss": 0.7067,
      "step": 3293
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34042507609689043,
      "learning_rate": 3.3099602143518054e-05,
      "loss": 0.6891,
      "step": 3294
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3654650496020079,
      "learning_rate": 3.309522487541238e-05,
      "loss": 0.65,
      "step": 3295
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4332620470042917,
      "learning_rate": 3.3090846509026006e-05,
      "loss": 0.6703,
      "step": 3296
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3831777231941374,
      "learning_rate": 3.308646704472616e-05,
      "loss": 0.6697,
      "step": 3297
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43675666145683695,
      "learning_rate": 3.3082086482880115e-05,
      "loss": 0.7166,
      "step": 3298
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3611636562240977,
      "learning_rate": 3.307770482385528e-05,
      "loss": 0.6742,
      "step": 3299
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4129586416439639,
      "learning_rate": 3.307332206801913e-05,
      "loss": 0.6951,
      "step": 3300
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.31794966355740584,
      "learning_rate": 3.306893821573926e-05,
      "loss": 0.6614,
      "step": 3301
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4235373471189618,
      "learning_rate": 3.306455326738332e-05,
      "loss": 0.6844,
      "step": 3302
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42722859158907533,
      "learning_rate": 3.3060167223319066e-05,
      "loss": 0.745,
      "step": 3303
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44676595335136826,
      "learning_rate": 3.305578008391437e-05,
      "loss": 0.7334,
      "step": 3304
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40765399103756333,
      "learning_rate": 3.3051391849537165e-05,
      "loss": 0.7209,
      "step": 3305
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4050641604726449,
      "learning_rate": 3.304700252055548e-05,
      "loss": 0.6688,
      "step": 3306
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4823133861352211,
      "learning_rate": 3.304261209733745e-05,
      "loss": 0.5401,
      "step": 3307
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33844332023396634,
      "learning_rate": 3.303822058025131e-05,
      "loss": 0.6651,
      "step": 3308
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43572181185009995,
      "learning_rate": 3.3033827969665336e-05,
      "loss": 0.6686,
      "step": 3309
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37836197919466996,
      "learning_rate": 3.302943426594797e-05,
      "loss": 0.6546,
      "step": 3310
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4410743708234899,
      "learning_rate": 3.3025039469467675e-05,
      "loss": 0.6955,
      "step": 3311
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4581520299334162,
      "learning_rate": 3.302064358059305e-05,
      "loss": 0.6971,
      "step": 3312
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0159562144175223,
      "learning_rate": 3.301624659969278e-05,
      "loss": 0.4864,
      "step": 3313
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4271870405637643,
      "learning_rate": 3.3011848527135626e-05,
      "loss": 0.7088,
      "step": 3314
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4325846769877888,
      "learning_rate": 3.3007449363290445e-05,
      "loss": 0.6751,
      "step": 3315
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3921630592075825,
      "learning_rate": 3.30030491085262e-05,
      "loss": 0.6716,
      "step": 3316
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4529574688483302,
      "learning_rate": 3.299864776321193e-05,
      "loss": 0.7074,
      "step": 3317
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4160922470899811,
      "learning_rate": 3.299424532771677e-05,
      "loss": 0.715,
      "step": 3318
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38368373823381646,
      "learning_rate": 3.298984180240994e-05,
      "loss": 0.6869,
      "step": 3319
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4641148444556784,
      "learning_rate": 3.298543718766078e-05,
      "loss": 0.7731,
      "step": 3320
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4289670338818946,
      "learning_rate": 3.298103148383867e-05,
      "loss": 0.6857,
      "step": 3321
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4107126327235219,
      "learning_rate": 3.2976624691313134e-05,
      "loss": 0.6745,
      "step": 3322
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40641379940943156,
      "learning_rate": 3.297221681045376e-05,
      "loss": 0.7008,
      "step": 3323
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4195681982430982,
      "learning_rate": 3.2967807841630224e-05,
      "loss": 0.7018,
      "step": 3324
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4091718267603353,
      "learning_rate": 3.2963397785212304e-05,
      "loss": 0.6895,
      "step": 3325
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2848205987973562,
      "learning_rate": 3.2958986641569875e-05,
      "loss": 0.4727,
      "step": 3326
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40871917949493464,
      "learning_rate": 3.295457441107288e-05,
      "loss": 0.6525,
      "step": 3327
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45850433774986804,
      "learning_rate": 3.2950161094091376e-05,
      "loss": 0.7038,
      "step": 3328
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.424497262227926,
      "learning_rate": 3.2945746690995504e-05,
      "loss": 0.6969,
      "step": 3329
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39310630458354434,
      "learning_rate": 3.2941331202155486e-05,
      "loss": 0.6867,
      "step": 3330
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8528412768956695,
      "learning_rate": 3.293691462794165e-05,
      "loss": 0.4834,
      "step": 3331
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45548474623524987,
      "learning_rate": 3.2932496968724415e-05,
      "loss": 0.6389,
      "step": 3332
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3781405100351594,
      "learning_rate": 3.292807822487428e-05,
      "loss": 0.6676,
      "step": 3333
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3816443320499542,
      "learning_rate": 3.292365839676182e-05,
      "loss": 0.6925,
      "step": 3334
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.46832375528793035,
      "learning_rate": 3.291923748475776e-05,
      "loss": 0.7079,
      "step": 3335
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41340970525627946,
      "learning_rate": 3.291481548923283e-05,
      "loss": 0.6766,
      "step": 3336
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4454004910743777,
      "learning_rate": 3.291039241055794e-05,
      "loss": 0.7162,
      "step": 3337
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39618125906250745,
      "learning_rate": 3.290596824910403e-05,
      "loss": 0.7121,
      "step": 3338
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4024460651418764,
      "learning_rate": 3.290154300524214e-05,
      "loss": 0.7118,
      "step": 3339
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36588683620829177,
      "learning_rate": 3.289711667934343e-05,
      "loss": 0.6731,
      "step": 3340
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4364110544022878,
      "learning_rate": 3.289268927177911e-05,
      "loss": 0.7149,
      "step": 3341
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4165551596438042,
      "learning_rate": 3.2888260782920514e-05,
      "loss": 0.6836,
      "step": 3342
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4137340074281892,
      "learning_rate": 3.288383121313905e-05,
      "loss": 0.7018,
      "step": 3343
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37085174816722855,
      "learning_rate": 3.2879400562806226e-05,
      "loss": 0.6858,
      "step": 3344
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.34415830028848515,
      "learning_rate": 3.287496883229363e-05,
      "loss": 0.6341,
      "step": 3345
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3790559012843507,
      "learning_rate": 3.287053602197295e-05,
      "loss": 0.7115,
      "step": 3346
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.425921656478735,
      "learning_rate": 3.286610213221596e-05,
      "loss": 0.7402,
      "step": 3347
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38413164776362213,
      "learning_rate": 3.2861667163394506e-05,
      "loss": 0.6744,
      "step": 3348
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.397510021419291,
      "learning_rate": 3.285723111588058e-05,
      "loss": 0.7047,
      "step": 3349
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.390154424350142,
      "learning_rate": 3.28527939900462e-05,
      "loss": 0.728,
      "step": 3350
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39619055947343745,
      "learning_rate": 3.28483557862635e-05,
      "loss": 0.6799,
      "step": 3351
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3640707322018113,
      "learning_rate": 3.284391650490473e-05,
      "loss": 0.673,
      "step": 3352
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.46704937633540633,
      "learning_rate": 3.283947614634219e-05,
      "loss": 0.7001,
      "step": 3353
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4075597612303124,
      "learning_rate": 3.2835034710948295e-05,
      "loss": 0.6948,
      "step": 3354
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.387386847068558,
      "learning_rate": 3.283059219909553e-05,
      "loss": 0.6997,
      "step": 3355
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4257298183005834,
      "learning_rate": 3.28261486111565e-05,
      "loss": 0.7076,
      "step": 3356
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41762361194499753,
      "learning_rate": 3.282170394750388e-05,
      "loss": 0.6819,
      "step": 3357
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44563288687331254,
      "learning_rate": 3.281725820851043e-05,
      "loss": 0.7115,
      "step": 3358
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43978798998783775,
      "learning_rate": 3.281281139454901e-05,
      "loss": 0.6782,
      "step": 3359
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40699109200440475,
      "learning_rate": 3.2808363505992574e-05,
      "loss": 0.6731,
      "step": 3360
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3990829229502069,
      "learning_rate": 3.2803914543214167e-05,
      "loss": 0.6905,
      "step": 3361
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44378206053591734,
      "learning_rate": 3.27994645065869e-05,
      "loss": 0.7224,
      "step": 3362
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45985694907662156,
      "learning_rate": 3.279501339648401e-05,
      "loss": 0.7011,
      "step": 3363
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40249381240941995,
      "learning_rate": 3.2790561213278794e-05,
      "loss": 0.642,
      "step": 3364
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3733159243831449,
      "learning_rate": 3.278610795734466e-05,
      "loss": 0.6573,
      "step": 3365
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41158562727994435,
      "learning_rate": 3.278165362905509e-05,
      "loss": 0.7242,
      "step": 3366
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38455223268443667,
      "learning_rate": 3.277719822878367e-05,
      "loss": 0.6815,
      "step": 3367
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3810175862584087,
      "learning_rate": 3.277274175690406e-05,
      "loss": 0.6576,
      "step": 3368
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43358412775187205,
      "learning_rate": 3.276828421379003e-05,
      "loss": 0.7184,
      "step": 3369
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43221730202996417,
      "learning_rate": 3.276382559981542e-05,
      "loss": 0.689,
      "step": 3370
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4489679834280587,
      "learning_rate": 3.275936591535418e-05,
      "loss": 0.6972,
      "step": 3371
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3643516590413988,
      "learning_rate": 3.275490516078032e-05,
      "loss": 0.6757,
      "step": 3372
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3744303516956486,
      "learning_rate": 3.275044333646797e-05,
      "loss": 0.7068,
      "step": 3373
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.415275140924439,
      "learning_rate": 3.274598044279134e-05,
      "loss": 0.6939,
      "step": 3374
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42512520484613836,
      "learning_rate": 3.274151648012471e-05,
      "loss": 0.7219,
      "step": 3375
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39606298797606315,
      "learning_rate": 3.2737051448842496e-05,
      "loss": 0.6709,
      "step": 3376
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39489984904208364,
      "learning_rate": 3.273258534931916e-05,
      "loss": 0.695,
      "step": 3377
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3694271021192373,
      "learning_rate": 3.272811818192926e-05,
      "loss": 0.6766,
      "step": 3378
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4020036995694818,
      "learning_rate": 3.272364994704746e-05,
      "loss": 0.66,
      "step": 3379
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4372264889876494,
      "learning_rate": 3.271918064504851e-05,
      "loss": 0.7338,
      "step": 3380
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4374532657715952,
      "learning_rate": 3.271471027630724e-05,
      "loss": 0.6757,
      "step": 3381
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40513218851313393,
      "learning_rate": 3.2710238841198565e-05,
      "loss": 0.6983,
      "step": 3382
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48665317645799067,
      "learning_rate": 3.270576634009752e-05,
      "loss": 0.6961,
      "step": 3383
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.383264118694279,
      "learning_rate": 3.270129277337919e-05,
      "loss": 0.6391,
      "step": 3384
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4018233465347085,
      "learning_rate": 3.269681814141878e-05,
      "loss": 0.7047,
      "step": 3385
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4316591128337001,
      "learning_rate": 3.269234244459156e-05,
      "loss": 0.6857,
      "step": 3386
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42131897280732233,
      "learning_rate": 3.268786568327291e-05,
      "loss": 0.7347,
      "step": 3387
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4133257887676453,
      "learning_rate": 3.268338785783829e-05,
      "loss": 0.6828,
      "step": 3388
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4162270537998946,
      "learning_rate": 3.267890896866325e-05,
      "loss": 0.6521,
      "step": 3389
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40649917642854505,
      "learning_rate": 3.267442901612342e-05,
      "loss": 0.6744,
      "step": 3390
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37578988387547274,
      "learning_rate": 3.2669948000594544e-05,
      "loss": 0.6938,
      "step": 3391
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38315270373353816,
      "learning_rate": 3.266546592245243e-05,
      "loss": 0.6302,
      "step": 3392
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4214692069876134,
      "learning_rate": 3.2660982782072976e-05,
      "loss": 0.6923,
      "step": 3393
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42298798954420785,
      "learning_rate": 3.265649857983219e-05,
      "loss": 0.6571,
      "step": 3394
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41055044917960826,
      "learning_rate": 3.265201331610616e-05,
      "loss": 0.6828,
      "step": 3395
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40864249797363966,
      "learning_rate": 3.2647526991271044e-05,
      "loss": 0.6738,
      "step": 3396
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4258126354582261,
      "learning_rate": 3.2643039605703116e-05,
      "loss": 0.6782,
      "step": 3397
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.109244783876099,
      "learning_rate": 3.263855115977873e-05,
      "loss": 0.5044,
      "step": 3398
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4310837663606903,
      "learning_rate": 3.263406165387432e-05,
      "loss": 0.6488,
      "step": 3399
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4647497678100108,
      "learning_rate": 3.262957108836641e-05,
      "loss": 0.7224,
      "step": 3400
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3938939504217381,
      "learning_rate": 3.262507946363163e-05,
      "loss": 0.6506,
      "step": 3401
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41939953176616546,
      "learning_rate": 3.2620586780046694e-05,
      "loss": 0.6536,
      "step": 3402
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41878858767822136,
      "learning_rate": 3.261609303798837e-05,
      "loss": 0.6728,
      "step": 3403
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42831496970699406,
      "learning_rate": 3.261159823783357e-05,
      "loss": 0.721,
      "step": 3404
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43358326392811297,
      "learning_rate": 3.260710237995925e-05,
      "loss": 0.7265,
      "step": 3405
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39903828403694036,
      "learning_rate": 3.260260546474248e-05,
      "loss": 0.7095,
      "step": 3406
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0441516029187614,
      "learning_rate": 3.2598107492560414e-05,
      "loss": 0.5392,
      "step": 3407
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45026718925396486,
      "learning_rate": 3.259360846379028e-05,
      "loss": 0.7079,
      "step": 3408
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3878322863110988,
      "learning_rate": 3.258910837880942e-05,
      "loss": 0.6788,
      "step": 3409
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3998341648627362,
      "learning_rate": 3.2584607237995244e-05,
      "loss": 0.6673,
      "step": 3410
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4064317407080771,
      "learning_rate": 3.2580105041725247e-05,
      "loss": 0.6882,
      "step": 3411
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4324911607024293,
      "learning_rate": 3.2575601790377044e-05,
      "loss": 0.6841,
      "step": 3412
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4894965621831124,
      "learning_rate": 3.2571097484328305e-05,
      "loss": 0.7343,
      "step": 3413
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4038960031299061,
      "learning_rate": 3.25665921239568e-05,
      "loss": 0.7007,
      "step": 3414
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4471697325191849,
      "learning_rate": 3.256208570964039e-05,
      "loss": 0.7076,
      "step": 3415
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4470967972310392,
      "learning_rate": 3.255757824175703e-05,
      "loss": 0.725,
      "step": 3416
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42841961528874095,
      "learning_rate": 3.2553069720684736e-05,
      "loss": 0.7356,
      "step": 3417
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3977026468733147,
      "learning_rate": 3.254856014680165e-05,
      "loss": 0.6661,
      "step": 3418
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3770579486425418,
      "learning_rate": 3.2544049520485984e-05,
      "loss": 0.6637,
      "step": 3419
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4473989700893095,
      "learning_rate": 3.2539537842116026e-05,
      "loss": 0.7167,
      "step": 3420
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3759929210391074,
      "learning_rate": 3.253502511207018e-05,
      "loss": 0.6526,
      "step": 3421
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41058939059185556,
      "learning_rate": 3.253051133072691e-05,
      "loss": 0.6628,
      "step": 3422
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4054945221352761,
      "learning_rate": 3.25259964984648e-05,
      "loss": 0.724,
      "step": 3423
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3871246996973386,
      "learning_rate": 3.252148061566248e-05,
      "loss": 0.6539,
      "step": 3424
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44845457375767467,
      "learning_rate": 3.25169636826987e-05,
      "loss": 0.6916,
      "step": 3425
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44032291737129803,
      "learning_rate": 3.2512445699952304e-05,
      "loss": 0.7314,
      "step": 3426
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4276973978837126,
      "learning_rate": 3.2507926667802184e-05,
      "loss": 0.6772,
      "step": 3427
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4288937527491318,
      "learning_rate": 3.250340658662737e-05,
      "loss": 0.6959,
      "step": 3428
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1610422140323018,
      "learning_rate": 3.249888545680695e-05,
      "loss": 0.501,
      "step": 3429
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46532412228305453,
      "learning_rate": 3.249436327872009e-05,
      "loss": 0.6579,
      "step": 3430
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4615519352992388,
      "learning_rate": 3.2489840052746074e-05,
      "loss": 0.7449,
      "step": 3431
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4443086915442208,
      "learning_rate": 3.248531577926427e-05,
      "loss": 0.6756,
      "step": 3432
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.413997800532301,
      "learning_rate": 3.24807904586541e-05,
      "loss": 0.6561,
      "step": 3433
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44253386788058824,
      "learning_rate": 3.24762640912951e-05,
      "loss": 0.7481,
      "step": 3434
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3828061145485183,
      "learning_rate": 3.247173667756691e-05,
      "loss": 0.6703,
      "step": 3435
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35624145806468466,
      "learning_rate": 3.246720821784922e-05,
      "loss": 0.6751,
      "step": 3436
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4384960091660322,
      "learning_rate": 3.246267871252184e-05,
      "loss": 0.6824,
      "step": 3437
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35283723834768793,
      "learning_rate": 3.245814816196464e-05,
      "loss": 0.6541,
      "step": 3438
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3383829436597684,
      "learning_rate": 3.2453616566557605e-05,
      "loss": 0.6706,
      "step": 3439
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3927211578601109,
      "learning_rate": 3.2449083926680784e-05,
      "loss": 0.7023,
      "step": 3440
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35005411834057065,
      "learning_rate": 3.2444550242714324e-05,
      "loss": 0.6853,
      "step": 3441
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37693766331396394,
      "learning_rate": 3.2440015515038474e-05,
      "loss": 0.6741,
      "step": 3442
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4098596583135836,
      "learning_rate": 3.2435479744033545e-05,
      "loss": 0.6728,
      "step": 3443
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38065570970858953,
      "learning_rate": 3.243094293007995e-05,
      "loss": 0.6739,
      "step": 3444
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.395575723400189,
      "learning_rate": 3.2426405073558176e-05,
      "loss": 0.7361,
      "step": 3445
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39135408838207714,
      "learning_rate": 3.2421866174848826e-05,
      "loss": 0.7204,
      "step": 3446
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.45176021267039496,
      "learning_rate": 3.2417326234332556e-05,
      "loss": 0.6978,
      "step": 3447
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3621956163422502,
      "learning_rate": 3.241278525239013e-05,
      "loss": 0.7051,
      "step": 3448
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40697309527731096,
      "learning_rate": 3.24082432294024e-05,
      "loss": 0.6981,
      "step": 3449
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3591009636708367,
      "learning_rate": 3.240370016575029e-05,
      "loss": 0.6682,
      "step": 3450
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3628957336468365,
      "learning_rate": 3.239915606181483e-05,
      "loss": 0.686,
      "step": 3451
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3689345343098315,
      "learning_rate": 3.2394610917977124e-05,
      "loss": 0.7047,
      "step": 3452
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34992063093449105,
      "learning_rate": 3.239006473461837e-05,
      "loss": 0.7097,
      "step": 3453
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43268196624102034,
      "learning_rate": 3.238551751211986e-05,
      "loss": 0.7354,
      "step": 3454
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4158643864775735,
      "learning_rate": 3.238096925086294e-05,
      "loss": 0.6916,
      "step": 3455
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.45248069746775504,
      "learning_rate": 3.237641995122908e-05,
      "loss": 0.7081,
      "step": 3456
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46207236305371086,
      "learning_rate": 3.237186961359983e-05,
      "loss": 0.7313,
      "step": 3457
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4107700315295825,
      "learning_rate": 3.2367318238356827e-05,
      "loss": 0.7001,
      "step": 3458
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3874832156582775,
      "learning_rate": 3.236276582588177e-05,
      "loss": 0.6523,
      "step": 3459
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3858255139285654,
      "learning_rate": 3.235821237655647e-05,
      "loss": 0.7143,
      "step": 3460
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3994692421531858,
      "learning_rate": 3.235365789076283e-05,
      "loss": 0.6632,
      "step": 3461
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4684386308896245,
      "learning_rate": 3.234910236888282e-05,
      "loss": 0.7004,
      "step": 3462
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3799188842418822,
      "learning_rate": 3.234454581129851e-05,
      "loss": 0.6703,
      "step": 3463
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4168437450333508,
      "learning_rate": 3.233998821839205e-05,
      "loss": 0.6458,
      "step": 3464
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39969091019664743,
      "learning_rate": 3.2335429590545673e-05,
      "loss": 0.7346,
      "step": 3465
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4315339938255451,
      "learning_rate": 3.2330869928141726e-05,
      "loss": 0.7057,
      "step": 3466
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41824392866307347,
      "learning_rate": 3.2326309231562606e-05,
      "loss": 0.7083,
      "step": 3467
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42566372219387555,
      "learning_rate": 3.232174750119081e-05,
      "loss": 0.6612,
      "step": 3468
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3973343555557316,
      "learning_rate": 3.2317184737408946e-05,
      "loss": 0.6859,
      "step": 3469
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3580532944051873,
      "learning_rate": 3.2312620940599663e-05,
      "loss": 0.6636,
      "step": 3470
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41449507088277265,
      "learning_rate": 3.2308056111145735e-05,
      "loss": 0.7274,
      "step": 3471
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4474397045721777,
      "learning_rate": 3.2303490249430006e-05,
      "loss": 0.7439,
      "step": 3472
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35603224303424225,
      "learning_rate": 3.22989233558354e-05,
      "loss": 0.6456,
      "step": 3473
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3689217029993414,
      "learning_rate": 3.2294355430744955e-05,
      "loss": 0.6714,
      "step": 3474
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3793636308595273,
      "learning_rate": 3.228978647454177e-05,
      "loss": 0.6685,
      "step": 3475
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3840307047350554,
      "learning_rate": 3.228521648760903e-05,
      "loss": 0.689,
      "step": 3476
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4390437400435133,
      "learning_rate": 3.228064547033002e-05,
      "loss": 0.7201,
      "step": 3477
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38325374850658667,
      "learning_rate": 3.2276073423088106e-05,
      "loss": 0.6732,
      "step": 3478
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39762085591015983,
      "learning_rate": 3.227150034626674e-05,
      "loss": 0.6511,
      "step": 3479
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44323062885675324,
      "learning_rate": 3.226692624024946e-05,
      "loss": 0.7279,
      "step": 3480
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3856693258722826,
      "learning_rate": 3.2262351105419895e-05,
      "loss": 0.6819,
      "step": 3481
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36272808101173937,
      "learning_rate": 3.225777494216175e-05,
      "loss": 0.7158,
      "step": 3482
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3703452882842301,
      "learning_rate": 3.225319775085882e-05,
      "loss": 0.6785,
      "step": 3483
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3560521523658864,
      "learning_rate": 3.2248619531895e-05,
      "loss": 0.6546,
      "step": 3484
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37694844446151027,
      "learning_rate": 3.2244040285654246e-05,
      "loss": 0.6582,
      "step": 3485
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41810742442486926,
      "learning_rate": 3.223946001252062e-05,
      "loss": 0.6863,
      "step": 3486
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4236020414240524,
      "learning_rate": 3.223487871287827e-05,
      "loss": 0.7272,
      "step": 3487
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42125031576963445,
      "learning_rate": 3.2230296387111415e-05,
      "loss": 0.6586,
      "step": 3488
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.421467853971347,
      "learning_rate": 3.222571303560437e-05,
      "loss": 0.6935,
      "step": 3489
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3617763640856828,
      "learning_rate": 3.222112865874154e-05,
      "loss": 0.697,
      "step": 3490
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44609438571001075,
      "learning_rate": 3.221654325690741e-05,
      "loss": 0.6957,
      "step": 3491
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3914662094647451,
      "learning_rate": 3.221195683048655e-05,
      "loss": 0.6726,
      "step": 3492
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3584329750760796,
      "learning_rate": 3.2207369379863615e-05,
      "loss": 0.6543,
      "step": 3493
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1032388244819793,
      "learning_rate": 3.220278090542336e-05,
      "loss": 0.4721,
      "step": 3494
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35401159378047264,
      "learning_rate": 3.21981914075506e-05,
      "loss": 0.6746,
      "step": 3495
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.392005344909366,
      "learning_rate": 3.219360088663026e-05,
      "loss": 0.6539,
      "step": 3496
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3936868215314172,
      "learning_rate": 3.2189009343047345e-05,
      "loss": 0.6666,
      "step": 3497
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37566917288518115,
      "learning_rate": 3.2184416777186936e-05,
      "loss": 0.6909,
      "step": 3498
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4030408684814097,
      "learning_rate": 3.2179823189434205e-05,
      "loss": 0.6833,
      "step": 3499
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38238275487548795,
      "learning_rate": 3.217522858017442e-05,
      "loss": 0.706,
      "step": 3500
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3813713809012717,
      "learning_rate": 3.2170632949792905e-05,
      "loss": 0.6902,
      "step": 3501
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.381537486784383,
      "learning_rate": 3.2166036298675115e-05,
      "loss": 0.6829,
      "step": 3502
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38471938483124957,
      "learning_rate": 3.216143862720654e-05,
      "loss": 0.6692,
      "step": 3503
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4004331946087641,
      "learning_rate": 3.2156839935772805e-05,
      "loss": 0.6595,
      "step": 3504
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39602629502023,
      "learning_rate": 3.2152240224759586e-05,
      "loss": 0.6882,
      "step": 3505
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36931911323065986,
      "learning_rate": 3.214763949455265e-05,
      "loss": 0.7129,
      "step": 3506
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40680641182251825,
      "learning_rate": 3.2143037745537856e-05,
      "loss": 0.662,
      "step": 3507
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35855437013926406,
      "learning_rate": 3.213843497810117e-05,
      "loss": 0.652,
      "step": 3508
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4061527252934502,
      "learning_rate": 3.2133831192628586e-05,
      "loss": 0.6859,
      "step": 3509
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38461379471693147,
      "learning_rate": 3.212922638950624e-05,
      "loss": 0.6753,
      "step": 3510
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4115578828979505,
      "learning_rate": 3.212462056912032e-05,
      "loss": 0.7021,
      "step": 3511
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36380800601214663,
      "learning_rate": 3.212001373185712e-05,
      "loss": 0.6537,
      "step": 3512
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3992596430097847,
      "learning_rate": 3.2115405878103005e-05,
      "loss": 0.7288,
      "step": 3513
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.023066602933382,
      "learning_rate": 3.211079700824443e-05,
      "loss": 0.4624,
      "step": 3514
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4209066979434657,
      "learning_rate": 3.2106187122667934e-05,
      "loss": 0.7179,
      "step": 3515
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4246243487568303,
      "learning_rate": 3.210157622176015e-05,
      "loss": 0.6667,
      "step": 3516
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46756395461194633,
      "learning_rate": 3.2096964305907774e-05,
      "loss": 0.6971,
      "step": 3517
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42541437726250975,
      "learning_rate": 3.2092351375497624e-05,
      "loss": 0.6501,
      "step": 3518
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43362720766655194,
      "learning_rate": 3.208773743091656e-05,
      "loss": 0.6881,
      "step": 3519
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36400620793480937,
      "learning_rate": 3.2083122472551555e-05,
      "loss": 0.6777,
      "step": 3520
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39605725385009566,
      "learning_rate": 3.207850650078967e-05,
      "loss": 0.6839,
      "step": 3521
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4018774963439926,
      "learning_rate": 3.207388951601802e-05,
      "loss": 0.6622,
      "step": 3522
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39871150487366896,
      "learning_rate": 3.206927151862385e-05,
      "loss": 0.673,
      "step": 3523
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38677714010616115,
      "learning_rate": 3.206465250899444e-05,
      "loss": 0.6394,
      "step": 3524
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.47854375917124947,
      "learning_rate": 3.20600324875172e-05,
      "loss": 0.6871,
      "step": 3525
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42282270132635186,
      "learning_rate": 3.2055411454579606e-05,
      "loss": 0.7103,
      "step": 3526
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0446234302359978,
      "learning_rate": 3.205078941056921e-05,
      "loss": 0.5069,
      "step": 3527
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41337589068870245,
      "learning_rate": 3.204616635587365e-05,
      "loss": 0.6617,
      "step": 3528
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4173326376484671,
      "learning_rate": 3.204154229088067e-05,
      "loss": 0.6923,
      "step": 3529
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.34511886876287967,
      "learning_rate": 3.203691721597809e-05,
      "loss": 0.6872,
      "step": 3530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3784411332716425,
      "learning_rate": 3.203229113155379e-05,
      "loss": 0.6659,
      "step": 3531
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3962825303561463,
      "learning_rate": 3.202766403799577e-05,
      "loss": 0.6783,
      "step": 3532
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37144660437020294,
      "learning_rate": 3.2023035935692086e-05,
      "loss": 0.6811,
      "step": 3533
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41115258466088384,
      "learning_rate": 3.201840682503091e-05,
      "loss": 0.6697,
      "step": 3534
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39715247679566895,
      "learning_rate": 3.2013776706400456e-05,
      "loss": 0.698,
      "step": 3535
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42656329678961974,
      "learning_rate": 3.200914558018907e-05,
      "loss": 0.6716,
      "step": 3536
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4029461930920823,
      "learning_rate": 3.2004513446785143e-05,
      "loss": 0.6804,
      "step": 3537
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41573746319696964,
      "learning_rate": 3.199988030657716e-05,
      "loss": 0.732,
      "step": 3538
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45040777624695666,
      "learning_rate": 3.199524615995372e-05,
      "loss": 0.6429,
      "step": 3539
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41441954735077136,
      "learning_rate": 3.199061100730347e-05,
      "loss": 0.6805,
      "step": 3540
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39488618148641,
      "learning_rate": 3.198597484901516e-05,
      "loss": 0.6592,
      "step": 3541
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3579211224051283,
      "learning_rate": 3.19813376854776e-05,
      "loss": 0.6656,
      "step": 3542
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40500537424740574,
      "learning_rate": 3.197669951707973e-05,
      "loss": 0.6651,
      "step": 3543
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3683443995501038,
      "learning_rate": 3.1972060344210536e-05,
      "loss": 0.6837,
      "step": 3544
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37566895567368563,
      "learning_rate": 3.196742016725909e-05,
      "loss": 0.6795,
      "step": 3545
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4275131542144686,
      "learning_rate": 3.1962778986614576e-05,
      "loss": 0.6606,
      "step": 3546
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4148333210096453,
      "learning_rate": 3.195813680266624e-05,
      "loss": 0.719,
      "step": 3547
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43841615983502114,
      "learning_rate": 3.19534936158034e-05,
      "loss": 0.693,
      "step": 3548
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41900131058164447,
      "learning_rate": 3.19488494264155e-05,
      "loss": 0.705,
      "step": 3549
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3804696064649294,
      "learning_rate": 3.194420423489202e-05,
      "loss": 0.6657,
      "step": 3550
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40930377730489004,
      "learning_rate": 3.193955804162256e-05,
      "loss": 0.7129,
      "step": 3551
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37336196447035325,
      "learning_rate": 3.1934910846996774e-05,
      "loss": 0.6623,
      "step": 3552
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4021316365950337,
      "learning_rate": 3.1930262651404445e-05,
      "loss": 0.6685,
      "step": 3553
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4063302051888622,
      "learning_rate": 3.192561345523539e-05,
      "loss": 0.6766,
      "step": 3554
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42463537142678787,
      "learning_rate": 3.1920963258879534e-05,
      "loss": 0.7082,
      "step": 3555
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41926768150027693,
      "learning_rate": 3.1916312062726894e-05,
      "loss": 0.6937,
      "step": 3556
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4190251035376802,
      "learning_rate": 3.1911659867167546e-05,
      "loss": 0.6744,
      "step": 3557
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3379266360282943,
      "learning_rate": 3.190700667259167e-05,
      "loss": 0.6925,
      "step": 3558
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3667315381593508,
      "learning_rate": 3.190235247938953e-05,
      "loss": 0.6517,
      "step": 3559
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3753446788348624,
      "learning_rate": 3.1897697287951456e-05,
      "loss": 0.6821,
      "step": 3560
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3856238123176764,
      "learning_rate": 3.189304109866788e-05,
      "loss": 0.6135,
      "step": 3561
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4282584271077155,
      "learning_rate": 3.188838391192931e-05,
      "loss": 0.6751,
      "step": 3562
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39155928599446793,
      "learning_rate": 3.188372572812634e-05,
      "loss": 0.7034,
      "step": 3563
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3587468041988973,
      "learning_rate": 3.1879066547649645e-05,
      "loss": 0.6625,
      "step": 3564
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4375613977494907,
      "learning_rate": 3.187440637088999e-05,
      "loss": 0.7128,
      "step": 3565
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9683649281530077,
      "learning_rate": 3.18697451982382e-05,
      "loss": 0.4935,
      "step": 3566
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40148795846376534,
      "learning_rate": 3.186508303008523e-05,
      "loss": 0.6956,
      "step": 3567
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39298520840772455,
      "learning_rate": 3.186041986682207e-05,
      "loss": 0.7029,
      "step": 3568
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3987000928186506,
      "learning_rate": 3.185575570883982e-05,
      "loss": 0.6573,
      "step": 3569
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4371033424447614,
      "learning_rate": 3.185109055652966e-05,
      "loss": 0.6978,
      "step": 3570
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44272876653684157,
      "learning_rate": 3.184642441028284e-05,
      "loss": 0.6965,
      "step": 3571
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40693037104497276,
      "learning_rate": 3.1841757270490715e-05,
      "loss": 0.6914,
      "step": 3572
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38511521601011783,
      "learning_rate": 3.183708913754471e-05,
      "loss": 0.683,
      "step": 3573
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40792329196878085,
      "learning_rate": 3.183242001183633e-05,
      "loss": 0.664,
      "step": 3574
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4089897679313088,
      "learning_rate": 3.1827749893757176e-05,
      "loss": 0.6485,
      "step": 3575
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39569857935681707,
      "learning_rate": 3.182307878369892e-05,
      "loss": 0.6978,
      "step": 3576
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4189317261464692,
      "learning_rate": 3.1818406682053335e-05,
      "loss": 0.6998,
      "step": 3577
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3896459567525001,
      "learning_rate": 3.181373358921225e-05,
      "loss": 0.714,
      "step": 3578
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3770094530143188,
      "learning_rate": 3.180905950556759e-05,
      "loss": 0.6706,
      "step": 3579
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3954215186822683,
      "learning_rate": 3.180438443151138e-05,
      "loss": 0.6763,
      "step": 3580
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3996999504563216,
      "learning_rate": 3.1799708367435705e-05,
      "loss": 0.6374,
      "step": 3581
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4466147335510184,
      "learning_rate": 3.179503131373273e-05,
      "loss": 0.6602,
      "step": 3582
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4866106875212187,
      "learning_rate": 3.179035327079473e-05,
      "loss": 0.7315,
      "step": 3583
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3767576884146425,
      "learning_rate": 3.1785674239014034e-05,
      "loss": 0.6615,
      "step": 3584
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4301112905437788,
      "learning_rate": 3.1780994218783084e-05,
      "loss": 0.6925,
      "step": 3585
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40198789338378255,
      "learning_rate": 3.177631321049437e-05,
      "loss": 0.6891,
      "step": 3586
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41852142074352366,
      "learning_rate": 3.177163121454048e-05,
      "loss": 0.7313,
      "step": 3587
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0483028321877739,
      "learning_rate": 3.1766948231314114e-05,
      "loss": 0.4644,
      "step": 3588
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3980208692027161,
      "learning_rate": 3.1762264261208e-05,
      "loss": 0.6939,
      "step": 3589
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.426716111554341,
      "learning_rate": 3.1757579304614994e-05,
      "loss": 0.7056,
      "step": 3590
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3727175850410604,
      "learning_rate": 3.1752893361928e-05,
      "loss": 0.6638,
      "step": 3591
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3535539666991138,
      "learning_rate": 3.174820643354004e-05,
      "loss": 0.6714,
      "step": 3592
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.36673386929747037,
      "learning_rate": 3.17435185198442e-05,
      "loss": 0.6985,
      "step": 3593
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3733361265861804,
      "learning_rate": 3.173882962123364e-05,
      "loss": 0.6925,
      "step": 3594
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4247574828617916,
      "learning_rate": 3.1734139738101615e-05,
      "loss": 0.6728,
      "step": 3595
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37581867135419394,
      "learning_rate": 3.1729448870841464e-05,
      "loss": 0.7228,
      "step": 3596
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.438914382547976,
      "learning_rate": 3.1724757019846606e-05,
      "loss": 0.6777,
      "step": 3597
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41206323782238863,
      "learning_rate": 3.1720064185510525e-05,
      "loss": 0.6927,
      "step": 3598
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42810548657335223,
      "learning_rate": 3.171537036822682e-05,
      "loss": 0.7074,
      "step": 3599
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40876491701389966,
      "learning_rate": 3.171067556838915e-05,
      "loss": 0.7023,
      "step": 3600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4054847640804742,
      "learning_rate": 3.170597978639127e-05,
      "loss": 0.7271,
      "step": 3601
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4244274491710506,
      "learning_rate": 3.1701283022626995e-05,
      "loss": 0.705,
      "step": 3602
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4140263783284988,
      "learning_rate": 3.169658527749025e-05,
      "loss": 0.7097,
      "step": 3603
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4115723959415109,
      "learning_rate": 3.169188655137502e-05,
      "loss": 0.6732,
      "step": 3604
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.47280465977916497,
      "learning_rate": 3.168718684467539e-05,
      "loss": 0.7368,
      "step": 3605
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4011117254437206,
      "learning_rate": 3.168248615778551e-05,
      "loss": 0.6612,
      "step": 3606
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42529172071114135,
      "learning_rate": 3.167778449109962e-05,
      "loss": 0.6855,
      "step": 3607
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4184111153882102,
      "learning_rate": 3.167308184501206e-05,
      "loss": 0.6987,
      "step": 3608
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3967765288326846,
      "learning_rate": 3.1668378219917214e-05,
      "loss": 0.682,
      "step": 3609
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3907881595914601,
      "learning_rate": 3.1663673616209584e-05,
      "loss": 0.7029,
      "step": 3610
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3591369832768781,
      "learning_rate": 3.165896803428372e-05,
      "loss": 0.6903,
      "step": 3611
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38346430779759666,
      "learning_rate": 3.1654261474534304e-05,
      "loss": 0.6851,
      "step": 3612
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42166290768141157,
      "learning_rate": 3.164955393735605e-05,
      "loss": 0.663,
      "step": 3613
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4094891605597271,
      "learning_rate": 3.164484542314377e-05,
      "loss": 0.6975,
      "step": 3614
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3473061093755969,
      "learning_rate": 3.1640135932292375e-05,
      "loss": 0.6607,
      "step": 3615
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.36939549997319,
      "learning_rate": 3.1635425465196824e-05,
      "loss": 0.6567,
      "step": 3616
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3689899472451777,
      "learning_rate": 3.1630714022252206e-05,
      "loss": 0.7012,
      "step": 3617
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.401048739984706,
      "learning_rate": 3.1626001603853645e-05,
      "loss": 0.6859,
      "step": 3618
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3579728338722923,
      "learning_rate": 3.1621288210396366e-05,
      "loss": 0.6809,
      "step": 3619
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3813954356192179,
      "learning_rate": 3.161657384227568e-05,
      "loss": 0.6776,
      "step": 3620
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.439256729285623,
      "learning_rate": 3.161185849988698e-05,
      "loss": 0.6871,
      "step": 3621
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38660720249000063,
      "learning_rate": 3.160714218362573e-05,
      "loss": 0.6863,
      "step": 3622
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3949652118955774,
      "learning_rate": 3.160242489388748e-05,
      "loss": 0.7173,
      "step": 3623
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38802300925525474,
      "learning_rate": 3.1597706631067864e-05,
      "loss": 0.7013,
      "step": 3624
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37940951234892745,
      "learning_rate": 3.159298739556261e-05,
      "loss": 0.6362,
      "step": 3625
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41643806610372947,
      "learning_rate": 3.15882671877675e-05,
      "loss": 0.66,
      "step": 3626
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42642496923744927,
      "learning_rate": 3.158354600807841e-05,
      "loss": 0.6778,
      "step": 3627
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4198382289055042,
      "learning_rate": 3.1578823856891306e-05,
      "loss": 0.7064,
      "step": 3628
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4050693639912109,
      "learning_rate": 3.157410073460223e-05,
      "loss": 0.6657,
      "step": 3629
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4163259415836089,
      "learning_rate": 3.156937664160731e-05,
      "loss": 0.6923,
      "step": 3630
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38022686439759024,
      "learning_rate": 3.156465157830273e-05,
      "loss": 0.7159,
      "step": 3631
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3670943483224091,
      "learning_rate": 3.155992554508481e-05,
      "loss": 0.6725,
      "step": 3632
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4006491674165863,
      "learning_rate": 3.155519854234988e-05,
      "loss": 0.677,
      "step": 3633
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3786190343409272,
      "learning_rate": 3.155047057049441e-05,
      "loss": 0.6623,
      "step": 3634
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3886803769681028,
      "learning_rate": 3.154574162991491e-05,
      "loss": 0.6584,
      "step": 3635
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3963272394312966,
      "learning_rate": 3.154101172100801e-05,
      "loss": 0.6875,
      "step": 3636
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37830271416536204,
      "learning_rate": 3.15362808441704e-05,
      "loss": 0.6551,
      "step": 3637
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40641554107959665,
      "learning_rate": 3.153154899979884e-05,
      "loss": 0.6701,
      "step": 3638
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40055634963478154,
      "learning_rate": 3.152681618829021e-05,
      "loss": 0.7367,
      "step": 3639
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41210112712944974,
      "learning_rate": 3.152208241004141e-05,
      "loss": 0.718,
      "step": 3640
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36589083825836305,
      "learning_rate": 3.151734766544948e-05,
      "loss": 0.6589,
      "step": 3641
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3615038333906932,
      "learning_rate": 3.151261195491151e-05,
      "loss": 0.6856,
      "step": 3642
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37671018906179826,
      "learning_rate": 3.1507875278824674e-05,
      "loss": 0.6645,
      "step": 3643
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4400505940113895,
      "learning_rate": 3.150313763758624e-05,
      "loss": 0.7123,
      "step": 3644
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40357467193419483,
      "learning_rate": 3.1498399031593537e-05,
      "loss": 0.7176,
      "step": 3645
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.382298576583156,
      "learning_rate": 3.1493659461244e-05,
      "loss": 0.7093,
      "step": 3646
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.412222286704373,
      "learning_rate": 3.148891892693513e-05,
      "loss": 0.7207,
      "step": 3647
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4189675717405452,
      "learning_rate": 3.1484177429064495e-05,
      "loss": 0.7016,
      "step": 3648
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38364712913041876,
      "learning_rate": 3.1479434968029776e-05,
      "loss": 0.6955,
      "step": 3649
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4326432906380104,
      "learning_rate": 3.147469154422871e-05,
      "loss": 0.6954,
      "step": 3650
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40565076013670837,
      "learning_rate": 3.146994715805912e-05,
      "loss": 0.6816,
      "step": 3651
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43880755978612884,
      "learning_rate": 3.146520180991891e-05,
      "loss": 0.6589,
      "step": 3652
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3993659058844341,
      "learning_rate": 3.146045550020608e-05,
      "loss": 0.6836,
      "step": 3653
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45021937944636364,
      "learning_rate": 3.145570822931868e-05,
      "loss": 0.7517,
      "step": 3654
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3883514044796641,
      "learning_rate": 3.145095999765487e-05,
      "loss": 0.683,
      "step": 3655
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3697945811327461,
      "learning_rate": 3.144621080561287e-05,
      "loss": 0.7115,
      "step": 3656
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41892388961087945,
      "learning_rate": 3.1441460653590996e-05,
      "loss": 0.6746,
      "step": 3657
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40371723772638657,
      "learning_rate": 3.143670954198763e-05,
      "loss": 0.7527,
      "step": 3658
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4445935364558032,
      "learning_rate": 3.143195747120125e-05,
      "loss": 0.712,
      "step": 3659
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42205739762307226,
      "learning_rate": 3.142720444163041e-05,
      "loss": 0.6651,
      "step": 3660
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4125211261846654,
      "learning_rate": 3.142245045367373e-05,
      "loss": 0.6671,
      "step": 3661
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4183477185531842,
      "learning_rate": 3.141769550772993e-05,
      "loss": 0.6773,
      "step": 3662
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37270616319496147,
      "learning_rate": 3.141293960419778e-05,
      "loss": 0.6767,
      "step": 3663
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44263254137729113,
      "learning_rate": 3.140818274347618e-05,
      "loss": 0.7004,
      "step": 3664
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3923689958655524,
      "learning_rate": 3.1403424925964076e-05,
      "loss": 0.6516,
      "step": 3665
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41860089477742235,
      "learning_rate": 3.139866615206049e-05,
      "loss": 0.6838,
      "step": 3666
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40110033703276865,
      "learning_rate": 3.1393906422164535e-05,
      "loss": 0.6874,
      "step": 3667
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3875715738163118,
      "learning_rate": 3.1389145736675414e-05,
      "loss": 0.6712,
      "step": 3668
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44754026302702077,
      "learning_rate": 3.138438409599239e-05,
      "loss": 0.6662,
      "step": 3669
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37714897389327995,
      "learning_rate": 3.137962150051483e-05,
      "loss": 0.6802,
      "step": 3670
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4529623093511375,
      "learning_rate": 3.137485795064215e-05,
      "loss": 0.7065,
      "step": 3671
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42466802920316593,
      "learning_rate": 3.137009344677388e-05,
      "loss": 0.6954,
      "step": 3672
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.424002830559664,
      "learning_rate": 3.136532798930959e-05,
      "loss": 0.6946,
      "step": 3673
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36660163684942493,
      "learning_rate": 3.136056157864898e-05,
      "loss": 0.6328,
      "step": 3674
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.46468380633258016,
      "learning_rate": 3.135579421519178e-05,
      "loss": 0.6888,
      "step": 3675
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4274334979672675,
      "learning_rate": 3.135102589933784e-05,
      "loss": 0.6673,
      "step": 3676
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4642413563274393,
      "learning_rate": 3.1346256631487065e-05,
      "loss": 0.71,
      "step": 3677
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45363080730785627,
      "learning_rate": 3.134148641203945e-05,
      "loss": 0.6724,
      "step": 3678
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4208517804834759,
      "learning_rate": 3.133671524139507e-05,
      "loss": 0.6769,
      "step": 3679
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.411677707802405,
      "learning_rate": 3.133194311995407e-05,
      "loss": 0.6731,
      "step": 3680
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41064441154158776,
      "learning_rate": 3.132717004811669e-05,
      "loss": 0.6707,
      "step": 3681
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43379666313607823,
      "learning_rate": 3.132239602628323e-05,
      "loss": 0.7242,
      "step": 3682
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.39391318180147367,
      "learning_rate": 3.1317621054854095e-05,
      "loss": 0.7185,
      "step": 3683
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41433385535198924,
      "learning_rate": 3.1312845134229744e-05,
      "loss": 0.7074,
      "step": 3684
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3695643762476352,
      "learning_rate": 3.130806826481074e-05,
      "loss": 0.6729,
      "step": 3685
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9435302634747712,
      "learning_rate": 3.130329044699771e-05,
      "loss": 0.4898,
      "step": 3686
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42764293469002684,
      "learning_rate": 3.129851168119135e-05,
      "loss": 0.707,
      "step": 3687
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.90241397449034,
      "learning_rate": 3.1293731967792476e-05,
      "loss": 0.4678,
      "step": 3688
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4884198521406795,
      "learning_rate": 3.1288951307201934e-05,
      "loss": 0.7115,
      "step": 3689
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.452648896066073,
      "learning_rate": 3.128416969982068e-05,
      "loss": 0.6914,
      "step": 3690
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38085656685759206,
      "learning_rate": 3.127938714604973e-05,
      "loss": 0.661,
      "step": 3691
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4436064465224646,
      "learning_rate": 3.127460364629021e-05,
      "loss": 0.708,
      "step": 3692
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37893840016318897,
      "learning_rate": 3.126981920094331e-05,
      "loss": 0.6772,
      "step": 3693
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41381515555108733,
      "learning_rate": 3.126503381041027e-05,
      "loss": 0.6807,
      "step": 3694
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4032237299250737,
      "learning_rate": 3.126024747509245e-05,
      "loss": 0.6821,
      "step": 3695
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44237437319857437,
      "learning_rate": 3.125546019539128e-05,
      "loss": 0.6929,
      "step": 3696
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4083606914923999,
      "learning_rate": 3.125067197170825e-05,
      "loss": 0.6568,
      "step": 3697
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3882078498816515,
      "learning_rate": 3.124588280444495e-05,
      "loss": 0.6587,
      "step": 3698
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4232448803257345,
      "learning_rate": 3.124109269400305e-05,
      "loss": 0.6825,
      "step": 3699
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4609341987240978,
      "learning_rate": 3.123630164078427e-05,
      "loss": 0.6711,
      "step": 3700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3901986806388004,
      "learning_rate": 3.123150964519045e-05,
      "loss": 0.614,
      "step": 3701
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43180042138250446,
      "learning_rate": 3.1226716707623475e-05,
      "loss": 0.7131,
      "step": 3702
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42600243097853185,
      "learning_rate": 3.122192282848533e-05,
      "loss": 0.6894,
      "step": 3703
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38729160926748757,
      "learning_rate": 3.121712800817807e-05,
      "loss": 0.6764,
      "step": 3704
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43655608951757013,
      "learning_rate": 3.1212332247103824e-05,
      "loss": 0.7001,
      "step": 3705
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3229199022650045,
      "learning_rate": 3.120753554566482e-05,
      "loss": 0.6536,
      "step": 3706
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35178021531995934,
      "learning_rate": 3.1202737904263345e-05,
      "loss": 0.6524,
      "step": 3707
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40043584149582073,
      "learning_rate": 3.119793932330177e-05,
      "loss": 0.7273,
      "step": 3708
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3489082381367906,
      "learning_rate": 3.119313980318255e-05,
      "loss": 0.647,
      "step": 3709
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.33063694841532465,
      "learning_rate": 3.1188339344308215e-05,
      "loss": 0.6179,
      "step": 3710
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3715991913938627,
      "learning_rate": 3.118353794708137e-05,
      "loss": 0.6524,
      "step": 3711
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3783479807053817,
      "learning_rate": 3.11787356119047e-05,
      "loss": 0.644,
      "step": 3712
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3831495788391353,
      "learning_rate": 3.117393233918097e-05,
      "loss": 0.6806,
      "step": 3713
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40371992647815136,
      "learning_rate": 3.116912812931304e-05,
      "loss": 0.6834,
      "step": 3714
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.136464818059191,
      "learning_rate": 3.1164322982703816e-05,
      "loss": 0.5005,
      "step": 3715
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41435793510085683,
      "learning_rate": 3.115951689975631e-05,
      "loss": 0.6852,
      "step": 3716
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4027310255919499,
      "learning_rate": 3.11547098808736e-05,
      "loss": 0.7,
      "step": 3717
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3604207427966914,
      "learning_rate": 3.114990192645884e-05,
      "loss": 0.6745,
      "step": 3718
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45422274592843553,
      "learning_rate": 3.114509303691527e-05,
      "loss": 0.725,
      "step": 3719
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.417803194384735,
      "learning_rate": 3.114028321264621e-05,
      "loss": 0.6872,
      "step": 3720
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4220432780229201,
      "learning_rate": 3.113547245405505e-05,
      "loss": 0.7218,
      "step": 3721
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.34548608902314676,
      "learning_rate": 3.113066076154527e-05,
      "loss": 0.6447,
      "step": 3722
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4154374539018539,
      "learning_rate": 3.1125848135520404e-05,
      "loss": 0.7099,
      "step": 3723
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4235735416018374,
      "learning_rate": 3.1121034576384096e-05,
      "loss": 0.6725,
      "step": 3724
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3573357481067115,
      "learning_rate": 3.111622008454005e-05,
      "loss": 0.6392,
      "step": 3725
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36543450849391196,
      "learning_rate": 3.111140466039205e-05,
      "loss": 0.662,
      "step": 3726
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3631821042573479,
      "learning_rate": 3.110658830434396e-05,
      "loss": 0.6749,
      "step": 3727
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3766954755828958,
      "learning_rate": 3.1101771016799714e-05,
      "loss": 0.6953,
      "step": 3728
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3935225591933455,
      "learning_rate": 3.1096952798163354e-05,
      "loss": 0.6535,
      "step": 3729
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43617325075966407,
      "learning_rate": 3.109213364883896e-05,
      "loss": 0.6825,
      "step": 3730
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37663382275861984,
      "learning_rate": 3.108731356923071e-05,
      "loss": 0.6967,
      "step": 3731
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4303821277499071,
      "learning_rate": 3.108249255974286e-05,
      "loss": 0.5273,
      "step": 3732
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.46595975174671816,
      "learning_rate": 3.107767062077975e-05,
      "loss": 0.7266,
      "step": 3733
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2663339359189294,
      "learning_rate": 3.107284775274577e-05,
      "loss": 0.496,
      "step": 3734
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4481105056319032,
      "learning_rate": 3.1068023956045435e-05,
      "loss": 0.6845,
      "step": 3735
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.336038664605063,
      "learning_rate": 3.1063199231083295e-05,
      "loss": 0.6234,
      "step": 3736
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3795700015210774,
      "learning_rate": 3.105837357826399e-05,
      "loss": 0.6528,
      "step": 3737
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4646861836841299,
      "learning_rate": 3.105354699799226e-05,
      "loss": 0.7014,
      "step": 3738
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47868429302225374,
      "learning_rate": 3.104871949067288e-05,
      "loss": 0.7246,
      "step": 3739
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4096066687798516,
      "learning_rate": 3.104389105671075e-05,
      "loss": 0.6611,
      "step": 3740
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1696397648999228,
      "learning_rate": 3.1039061696510814e-05,
      "loss": 0.4989,
      "step": 3741
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4336958404980537,
      "learning_rate": 3.1034231410478095e-05,
      "loss": 0.649,
      "step": 3742
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40405714863334347,
      "learning_rate": 3.102940019901772e-05,
      "loss": 0.6361,
      "step": 3743
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4193827956245149,
      "learning_rate": 3.102456806253488e-05,
      "loss": 0.71,
      "step": 3744
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3730529167485227,
      "learning_rate": 3.101973500143482e-05,
      "loss": 0.6724,
      "step": 3745
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3741932566988868,
      "learning_rate": 3.101490101612289e-05,
      "loss": 0.6769,
      "step": 3746
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39933030728368446,
      "learning_rate": 3.1010066107004524e-05,
      "loss": 0.7276,
      "step": 3747
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41809972211337615,
      "learning_rate": 3.10052302744852e-05,
      "loss": 0.6902,
      "step": 3748
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4554381629598492,
      "learning_rate": 3.100039351897051e-05,
      "loss": 0.7038,
      "step": 3749
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40073429137344474,
      "learning_rate": 3.09955558408661e-05,
      "loss": 0.709,
      "step": 3750
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43716721248168455,
      "learning_rate": 3.09907172405777e-05,
      "loss": 0.6932,
      "step": 3751
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3973935979371117,
      "learning_rate": 3.098587771851112e-05,
      "loss": 0.7268,
      "step": 3752
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3575646377887737,
      "learning_rate": 3.0981037275072244e-05,
      "loss": 0.6745,
      "step": 3753
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3382830020447714,
      "learning_rate": 3.097619591066703e-05,
      "loss": 0.634,
      "step": 3754
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.374101105087984,
      "learning_rate": 3.097135362570153e-05,
      "loss": 0.6594,
      "step": 3755
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36247793072939005,
      "learning_rate": 3.096651042058184e-05,
      "loss": 0.6752,
      "step": 3756
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4385675955071124,
      "learning_rate": 3.096166629571417e-05,
      "loss": 0.6747,
      "step": 3757
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4406506207762739,
      "learning_rate": 3.095682125150479e-05,
      "loss": 0.6837,
      "step": 3758
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4000035578312388,
      "learning_rate": 3.095197528836004e-05,
      "loss": 0.6611,
      "step": 3759
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4166937047079357,
      "learning_rate": 3.0947128406686346e-05,
      "loss": 0.7034,
      "step": 3760
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4511029816542574,
      "learning_rate": 3.094228060689022e-05,
      "loss": 0.7121,
      "step": 3761
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46779426241851635,
      "learning_rate": 3.093743188937823e-05,
      "loss": 0.706,
      "step": 3762
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33464461302304443,
      "learning_rate": 3.0932582254557036e-05,
      "loss": 0.6377,
      "step": 3763
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36305659941832163,
      "learning_rate": 3.092773170283337e-05,
      "loss": 0.6881,
      "step": 3764
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4391821368445292,
      "learning_rate": 3.092288023461405e-05,
      "loss": 0.6914,
      "step": 3765
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43609045286322534,
      "learning_rate": 3.0918027850305955e-05,
      "loss": 0.7241,
      "step": 3766
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42366739126157016,
      "learning_rate": 3.091317455031605e-05,
      "loss": 0.6601,
      "step": 3767
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42851432608499207,
      "learning_rate": 3.0908320335051375e-05,
      "loss": 0.7045,
      "step": 3768
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4362143004689032,
      "learning_rate": 3.0903465204919046e-05,
      "loss": 0.718,
      "step": 3769
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43110232734716936,
      "learning_rate": 3.089860916032625e-05,
      "loss": 0.6887,
      "step": 3770
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4330336104151022,
      "learning_rate": 3.0893752201680274e-05,
      "loss": 0.688,
      "step": 3771
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43531645064757774,
      "learning_rate": 3.088889432938846e-05,
      "loss": 0.6678,
      "step": 3772
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41123182298920474,
      "learning_rate": 3.0884035543858226e-05,
      "loss": 0.6749,
      "step": 3773
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3709913104316663,
      "learning_rate": 3.087917584549708e-05,
      "loss": 0.6445,
      "step": 3774
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3612604437029017,
      "learning_rate": 3.0874315234712584e-05,
      "loss": 0.6794,
      "step": 3775
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3951535222961328,
      "learning_rate": 3.086945371191241e-05,
      "loss": 0.6566,
      "step": 3776
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3744453971754073,
      "learning_rate": 3.086459127750428e-05,
      "loss": 0.6715,
      "step": 3777
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36319171345657325,
      "learning_rate": 3.0859727931896e-05,
      "loss": 0.6547,
      "step": 3778
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35744355409225775,
      "learning_rate": 3.0854863675495454e-05,
      "loss": 0.6866,
      "step": 3779
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4071303236423407,
      "learning_rate": 3.084999850871059e-05,
      "loss": 0.4553,
      "step": 3780
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43350918341436057,
      "learning_rate": 3.084513243194947e-05,
      "loss": 0.6795,
      "step": 3781
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4075657006164687,
      "learning_rate": 3.0840265445620186e-05,
      "loss": 0.7192,
      "step": 3782
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42404730428669163,
      "learning_rate": 3.083539755013093e-05,
      "loss": 0.7264,
      "step": 3783
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45078639676376253,
      "learning_rate": 3.083052874588996e-05,
      "loss": 0.684,
      "step": 3784
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36176942540909407,
      "learning_rate": 3.0825659033305633e-05,
      "loss": 0.6847,
      "step": 3785
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3634266254725971,
      "learning_rate": 3.082078841278636e-05,
      "loss": 0.7058,
      "step": 3786
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39074977700616653,
      "learning_rate": 3.0815916884740626e-05,
      "loss": 0.6679,
      "step": 3787
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3908850135774886,
      "learning_rate": 3.0811044449577e-05,
      "loss": 0.6744,
      "step": 3788
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4792036744117611,
      "learning_rate": 3.080617110770414e-05,
      "loss": 0.6593,
      "step": 3789
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3729414953104255,
      "learning_rate": 3.080129685953075e-05,
      "loss": 0.6407,
      "step": 3790
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3732130727853269,
      "learning_rate": 3.0796421705465654e-05,
      "loss": 0.6785,
      "step": 3791
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40204660347275456,
      "learning_rate": 3.079154564591769e-05,
      "loss": 0.6947,
      "step": 3792
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2503512185384693,
      "learning_rate": 3.078666868129584e-05,
      "loss": 0.4625,
      "step": 3793
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4085011659593796,
      "learning_rate": 3.078179081200911e-05,
      "loss": 0.6922,
      "step": 3794
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42123459976118316,
      "learning_rate": 3.0776912038466605e-05,
      "loss": 0.6567,
      "step": 3795
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.375065726862941,
      "learning_rate": 3.07720323610775e-05,
      "loss": 0.6843,
      "step": 3796
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4365874097584167,
      "learning_rate": 3.076715178025105e-05,
      "loss": 0.6947,
      "step": 3797
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36322794175831713,
      "learning_rate": 3.076227029639659e-05,
      "loss": 0.7015,
      "step": 3798
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40154984307680874,
      "learning_rate": 3.0757387909923506e-05,
      "loss": 0.6661,
      "step": 3799
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4024652957860509,
      "learning_rate": 3.0752504621241304e-05,
      "loss": 0.6666,
      "step": 3800
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4130609899002715,
      "learning_rate": 3.074762043075952e-05,
      "loss": 0.7108,
      "step": 3801
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40015273818083186,
      "learning_rate": 3.074273533888779e-05,
      "loss": 0.6648,
      "step": 3802
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3599900946349468,
      "learning_rate": 3.073784934603581e-05,
      "loss": 0.6463,
      "step": 3803
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3696778543892026,
      "learning_rate": 3.0732962452613385e-05,
      "loss": 0.6689,
      "step": 3804
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4112524749651978,
      "learning_rate": 3.072807465903036e-05,
      "loss": 0.6991,
      "step": 3805
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45142084933471543,
      "learning_rate": 3.072318596569667e-05,
      "loss": 0.7052,
      "step": 3806
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40578729199600766,
      "learning_rate": 3.071829637302233e-05,
      "loss": 0.7121,
      "step": 3807
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45619757821661416,
      "learning_rate": 3.07134058814174e-05,
      "loss": 0.7059,
      "step": 3808
    },
    {
      "epoch": 0.34,
      "grad_norm": 22.68783253854499,
      "learning_rate": 3.070851449129207e-05,
      "loss": 0.7291,
      "step": 3809
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4277441105184531,
      "learning_rate": 3.070362220305656e-05,
      "loss": 0.6868,
      "step": 3810
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38777526548385166,
      "learning_rate": 3.0698729017121176e-05,
      "loss": 0.6906,
      "step": 3811
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42705119678791686,
      "learning_rate": 3.0693834933896326e-05,
      "loss": 0.6725,
      "step": 3812
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43932802187704223,
      "learning_rate": 3.068893995379244e-05,
      "loss": 0.6932,
      "step": 3813
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36479002619884277,
      "learning_rate": 3.068404407722008e-05,
      "loss": 0.6949,
      "step": 3814
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44674271699435797,
      "learning_rate": 3.0679147304589833e-05,
      "loss": 0.7391,
      "step": 3815
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3842556536015959,
      "learning_rate": 3.0674249636312406e-05,
      "loss": 0.6633,
      "step": 3816
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.411930837250188,
      "learning_rate": 3.066935107279855e-05,
      "loss": 0.6945,
      "step": 3817
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3908649387225812,
      "learning_rate": 3.0664451614459117e-05,
      "loss": 0.7512,
      "step": 3818
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3965658821588405,
      "learning_rate": 3.065955126170499e-05,
      "loss": 0.7077,
      "step": 3819
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3982333697221644,
      "learning_rate": 3.0654650014947183e-05,
      "loss": 0.7092,
      "step": 3820
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3663515715757617,
      "learning_rate": 3.064974787459675e-05,
      "loss": 0.6643,
      "step": 3821
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38701373005278167,
      "learning_rate": 3.0644844841064816e-05,
      "loss": 0.7059,
      "step": 3822
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40924331991445734,
      "learning_rate": 3.06399409147626e-05,
      "loss": 0.6945,
      "step": 3823
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38862108324159905,
      "learning_rate": 3.0635036096101396e-05,
      "loss": 0.6973,
      "step": 3824
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3798816409018353,
      "learning_rate": 3.0630130385492554e-05,
      "loss": 0.6746,
      "step": 3825
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3738239631654029,
      "learning_rate": 3.062522378334751e-05,
      "loss": 0.6481,
      "step": 3826
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38954451583750893,
      "learning_rate": 3.062031629007779e-05,
      "loss": 0.666,
      "step": 3827
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3652058284386238,
      "learning_rate": 3.061540790609496e-05,
      "loss": 0.6654,
      "step": 3828
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45637061160929865,
      "learning_rate": 3.0610498631810694e-05,
      "loss": 0.7039,
      "step": 3829
    },
    {
      "epoch": 0.34,
      "grad_norm": 21.508012362787696,
      "learning_rate": 3.0605588467636723e-05,
      "loss": 1.5195,
      "step": 3830
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4048612617709537,
      "learning_rate": 3.060067741398485e-05,
      "loss": 0.6991,
      "step": 3831
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4260748264855857,
      "learning_rate": 3.059576547126697e-05,
      "loss": 0.6953,
      "step": 3832
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39447377743423323,
      "learning_rate": 3.059085263989503e-05,
      "loss": 0.677,
      "step": 3833
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36203616881695794,
      "learning_rate": 3.0585938920281075e-05,
      "loss": 0.6489,
      "step": 3834
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3489202182396395,
      "learning_rate": 3.05810243128372e-05,
      "loss": 0.6548,
      "step": 3835
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41149142755311197,
      "learning_rate": 3.0576108817975606e-05,
      "loss": 0.6846,
      "step": 3836
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43902004828784513,
      "learning_rate": 3.0571192436108534e-05,
      "loss": 0.7032,
      "step": 3837
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3904841796442657,
      "learning_rate": 3.0566275167648316e-05,
      "loss": 0.6969,
      "step": 3838
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40668749285008604,
      "learning_rate": 3.056135701300736e-05,
      "loss": 0.6663,
      "step": 3839
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3922898410722824,
      "learning_rate": 3.055643797259815e-05,
      "loss": 0.6534,
      "step": 3840
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39470274529646043,
      "learning_rate": 3.0551518046833234e-05,
      "loss": 0.6989,
      "step": 3841
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40703106579991305,
      "learning_rate": 3.054659723612525e-05,
      "loss": 0.7037,
      "step": 3842
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39083329929738664,
      "learning_rate": 3.054167554088688e-05,
      "loss": 0.6736,
      "step": 3843
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39382931256595,
      "learning_rate": 3.053675296153092e-05,
      "loss": 0.7045,
      "step": 3844
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3944970096245131,
      "learning_rate": 3.053182949847022e-05,
      "loss": 0.6771,
      "step": 3845
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44496399831746575,
      "learning_rate": 3.0526905152117686e-05,
      "loss": 0.6821,
      "step": 3846
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42069055559426777,
      "learning_rate": 3.052197992288634e-05,
      "loss": 0.6399,
      "step": 3847
    },
    {
      "epoch": 0.34,
      "grad_norm": 23.0645010331617,
      "learning_rate": 3.0517053811189245e-05,
      "loss": 1.4437,
      "step": 3848
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3909035816255814,
      "learning_rate": 3.0512126817439544e-05,
      "loss": 0.6819,
      "step": 3849
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4615758923141337,
      "learning_rate": 3.0507198942050463e-05,
      "loss": 0.6829,
      "step": 3850
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41714906453215456,
      "learning_rate": 3.0502270185435298e-05,
      "loss": 0.6736,
      "step": 3851
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42949847520600704,
      "learning_rate": 3.0497340548007416e-05,
      "loss": 0.721,
      "step": 3852
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4192015793183887,
      "learning_rate": 3.0492410030180255e-05,
      "loss": 0.6841,
      "step": 3853
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42752383575477576,
      "learning_rate": 3.048747863236734e-05,
      "loss": 0.6894,
      "step": 3854
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4453842183731639,
      "learning_rate": 3.048254635498226e-05,
      "loss": 0.6962,
      "step": 3855
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40697125833186165,
      "learning_rate": 3.0477613198438674e-05,
      "loss": 0.6881,
      "step": 3856
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5077558201021659,
      "learning_rate": 3.0472679163150322e-05,
      "loss": 0.7401,
      "step": 3857
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40612850299155207,
      "learning_rate": 3.0467744249531024e-05,
      "loss": 0.6919,
      "step": 3858
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4119130180229125,
      "learning_rate": 3.046280845799465e-05,
      "loss": 0.7355,
      "step": 3859
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4345582846927341,
      "learning_rate": 3.0457871788955163e-05,
      "loss": 0.6624,
      "step": 3860
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4267154851929588,
      "learning_rate": 3.045293424282661e-05,
      "loss": 0.682,
      "step": 3861
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41255084911325735,
      "learning_rate": 3.0447995820023083e-05,
      "loss": 0.7111,
      "step": 3862
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3671674492212178,
      "learning_rate": 3.0443056520958754e-05,
      "loss": 0.6073,
      "step": 3863
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4008817441742313,
      "learning_rate": 3.0438116346047897e-05,
      "loss": 0.6936,
      "step": 3864
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42031176715616175,
      "learning_rate": 3.043317529570483e-05,
      "loss": 0.6794,
      "step": 3865
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4333129585973688,
      "learning_rate": 3.042823337034394e-05,
      "loss": 0.7151,
      "step": 3866
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.32196929376498057,
      "learning_rate": 3.0423290570379715e-05,
      "loss": 0.6979,
      "step": 3867
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46746388020543334,
      "learning_rate": 3.041834689622671e-05,
      "loss": 0.7273,
      "step": 3868
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.417912223712181,
      "learning_rate": 3.0413402348299522e-05,
      "loss": 0.6821,
      "step": 3869
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40717457861383816,
      "learning_rate": 3.040845692701285e-05,
      "loss": 0.6823,
      "step": 3870
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4078168829904331,
      "learning_rate": 3.0403510632781478e-05,
      "loss": 0.7397,
      "step": 3871
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4385947071636751,
      "learning_rate": 3.039856346602023e-05,
      "loss": 0.7169,
      "step": 3872
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.507130563361972,
      "learning_rate": 3.0393615427144022e-05,
      "loss": 1.6471,
      "step": 3873
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43136447328442556,
      "learning_rate": 3.0388666516567837e-05,
      "loss": 0.7375,
      "step": 3874
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4434259694437873,
      "learning_rate": 3.0383716734706738e-05,
      "loss": 0.6669,
      "step": 3875
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3854119873548981,
      "learning_rate": 3.037876608197586e-05,
      "loss": 0.6537,
      "step": 3876
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.47905769947003807,
      "learning_rate": 3.0373814558790406e-05,
      "loss": 0.6998,
      "step": 3877
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.452251400568244,
      "learning_rate": 3.036886216556565e-05,
      "loss": 0.7052,
      "step": 3878
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4233266831737233,
      "learning_rate": 3.036390890271695e-05,
      "loss": 0.7147,
      "step": 3879
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37313568862588614,
      "learning_rate": 3.035895477065972e-05,
      "loss": 0.6597,
      "step": 3880
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4640300352668571,
      "learning_rate": 3.035399976980947e-05,
      "loss": 0.6951,
      "step": 3881
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42556355592707923,
      "learning_rate": 3.034904390058176e-05,
      "loss": 0.6882,
      "step": 3882
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.45737945203494046,
      "learning_rate": 3.0344087163392236e-05,
      "loss": 0.6884,
      "step": 3883
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40146092228324937,
      "learning_rate": 3.0339129558656618e-05,
      "loss": 0.6606,
      "step": 3884
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4833033090921307,
      "learning_rate": 3.0334171086790695e-05,
      "loss": 0.6665,
      "step": 3885
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38362458887855283,
      "learning_rate": 3.0329211748210316e-05,
      "loss": 0.6582,
      "step": 3886
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3969469378686805,
      "learning_rate": 3.032425154333142e-05,
      "loss": 0.674,
      "step": 3887
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4019566086776192,
      "learning_rate": 3.031929047257002e-05,
      "loss": 0.6485,
      "step": 3888
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39526195742485765,
      "learning_rate": 3.0314328536342193e-05,
      "loss": 0.6784,
      "step": 3889
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.47218644699119267,
      "learning_rate": 3.0309365735064085e-05,
      "loss": 0.7561,
      "step": 3890
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38603104680275624,
      "learning_rate": 3.0304402069151925e-05,
      "loss": 0.6778,
      "step": 3891
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3633773573159901,
      "learning_rate": 3.0299437539022007e-05,
      "loss": 0.6782,
      "step": 3892
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38987181958341394,
      "learning_rate": 3.0294472145090696e-05,
      "loss": 0.6683,
      "step": 3893
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.386101670198785,
      "learning_rate": 3.0289505887774445e-05,
      "loss": 0.667,
      "step": 3894
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4123983886514345,
      "learning_rate": 3.0284538767489762e-05,
      "loss": 0.6645,
      "step": 3895
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3604945101765717,
      "learning_rate": 3.027957078465323e-05,
      "loss": 0.6936,
      "step": 3896
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4345433572188,
      "learning_rate": 3.0274601939681515e-05,
      "loss": 0.6977,
      "step": 3897
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41014147490969644,
      "learning_rate": 3.026963223299134e-05,
      "loss": 0.6846,
      "step": 3898
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4516800024620545,
      "learning_rate": 3.0264661664999514e-05,
      "loss": 0.7071,
      "step": 3899
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4051933879489684,
      "learning_rate": 3.025969023612291e-05,
      "loss": 0.7005,
      "step": 3900
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41257249042955396,
      "learning_rate": 3.0254717946778473e-05,
      "loss": 0.7029,
      "step": 3901
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4135467312261066,
      "learning_rate": 3.0249744797383233e-05,
      "loss": 0.6917,
      "step": 3902
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4554005639182643,
      "learning_rate": 3.0244770788354272e-05,
      "loss": 0.7636,
      "step": 3903
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3535891697068375,
      "learning_rate": 3.0239795920108758e-05,
      "loss": 0.6804,
      "step": 3904
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40907874094505126,
      "learning_rate": 3.0234820193063927e-05,
      "loss": 0.683,
      "step": 3905
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36263318790582777,
      "learning_rate": 3.022984360763709e-05,
      "loss": 0.6272,
      "step": 3906
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4410172133132503,
      "learning_rate": 3.0224866164245617e-05,
      "loss": 0.6851,
      "step": 3907
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36653439980050617,
      "learning_rate": 3.021988786330697e-05,
      "loss": 0.6461,
      "step": 3908
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4067989927661847,
      "learning_rate": 3.0214908705238676e-05,
      "loss": 0.704,
      "step": 3909
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4096510635653331,
      "learning_rate": 3.0209928690458318e-05,
      "loss": 0.6822,
      "step": 3910
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41508514126421847,
      "learning_rate": 3.020494781938357e-05,
      "loss": 0.668,
      "step": 3911
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38755063326879846,
      "learning_rate": 3.0199966092432178e-05,
      "loss": 0.67,
      "step": 3912
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37877978340020607,
      "learning_rate": 3.019498351002195e-05,
      "loss": 0.6619,
      "step": 3913
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46138558165104315,
      "learning_rate": 3.019000007257076e-05,
      "loss": 0.6932,
      "step": 3914
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40242326156647223,
      "learning_rate": 3.018501578049658e-05,
      "loss": 0.6565,
      "step": 3915
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4134177282715509,
      "learning_rate": 3.018003063421742e-05,
      "loss": 0.6951,
      "step": 3916
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42540995888988126,
      "learning_rate": 3.0175044634151388e-05,
      "loss": 0.6888,
      "step": 3917
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41691399575885396,
      "learning_rate": 3.0170057780716647e-05,
      "loss": 0.6588,
      "step": 3918
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3939598620419509,
      "learning_rate": 3.016507007433145e-05,
      "loss": 0.6365,
      "step": 3919
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43461704475952406,
      "learning_rate": 3.016008151541409e-05,
      "loss": 0.6934,
      "step": 3920
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3674519241620497,
      "learning_rate": 3.015509210438298e-05,
      "loss": 0.6699,
      "step": 3921
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3548585865661657,
      "learning_rate": 3.015010184165655e-05,
      "loss": 0.6639,
      "step": 3922
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4502729362302114,
      "learning_rate": 3.0145110727653334e-05,
      "loss": 0.7335,
      "step": 3923
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38719980238397445,
      "learning_rate": 3.0140118762791935e-05,
      "loss": 0.6699,
      "step": 3924
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3788542477909445,
      "learning_rate": 3.013512594749102e-05,
      "loss": 0.682,
      "step": 3925
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37884745719475366,
      "learning_rate": 3.0130132282169337e-05,
      "loss": 0.6712,
      "step": 3926
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4016417044734027,
      "learning_rate": 3.012513776724569e-05,
      "loss": 0.69,
      "step": 3927
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4491376672568249,
      "learning_rate": 3.0120142403138973e-05,
      "loss": 0.7109,
      "step": 3928
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.439354458350254,
      "learning_rate": 3.0115146190268128e-05,
      "loss": 0.7453,
      "step": 3929
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4237565694197413,
      "learning_rate": 3.011014912905219e-05,
      "loss": 0.7095,
      "step": 3930
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38559267294891103,
      "learning_rate": 3.0105151219910255e-05,
      "loss": 0.7201,
      "step": 3931
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4826020433445236,
      "learning_rate": 3.0100152463261488e-05,
      "loss": 0.7514,
      "step": 3932
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.18233395735201,
      "learning_rate": 3.0095152859525133e-05,
      "loss": 1.305,
      "step": 3933
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.376055891491513,
      "learning_rate": 3.0090152409120502e-05,
      "loss": 0.6578,
      "step": 3934
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3480189750409403,
      "learning_rate": 3.0085151112466975e-05,
      "loss": 0.6694,
      "step": 3935
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38940721061009326,
      "learning_rate": 3.0080148969984e-05,
      "loss": 0.7093,
      "step": 3936
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3908761782686453,
      "learning_rate": 3.0075145982091106e-05,
      "loss": 0.7046,
      "step": 3937
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.426872648421949,
      "learning_rate": 3.007014214920789e-05,
      "loss": 0.6839,
      "step": 3938
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46946003651627155,
      "learning_rate": 3.006513747175401e-05,
      "loss": 0.7503,
      "step": 3939
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37206347274641405,
      "learning_rate": 3.0060131950149202e-05,
      "loss": 0.6809,
      "step": 3940
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4930292380729128,
      "learning_rate": 3.0055125584813286e-05,
      "loss": 0.6549,
      "step": 3941
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.370350071582546,
      "learning_rate": 3.0050118376166124e-05,
      "loss": 0.6446,
      "step": 3942
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3900907433481344,
      "learning_rate": 3.004511032462767e-05,
      "loss": 0.6954,
      "step": 3943
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37358199612601056,
      "learning_rate": 3.0040101430617947e-05,
      "loss": 0.6663,
      "step": 3944
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39262255719707884,
      "learning_rate": 3.0035091694557047e-05,
      "loss": 0.6676,
      "step": 3945
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4274298955947771,
      "learning_rate": 3.0030081116865123e-05,
      "loss": 0.7415,
      "step": 3946
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40543255944080986,
      "learning_rate": 3.0025069697962403e-05,
      "loss": 0.7247,
      "step": 3947
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4149743222609029,
      "learning_rate": 3.00200574382692e-05,
      "loss": 0.7496,
      "step": 3948
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.938171979428885,
      "learning_rate": 3.0015044338205884e-05,
      "loss": 1.2776,
      "step": 3949
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4030385202442161,
      "learning_rate": 3.001003039819289e-05,
      "loss": 0.6926,
      "step": 3950
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4128892510985167,
      "learning_rate": 3.0005015618650733e-05,
      "loss": 0.6583,
      "step": 3951
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3773257255720113,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.6608,
      "step": 3952
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4432878084227695,
      "learning_rate": 2.9994983542661348e-05,
      "loss": 0.7191,
      "step": 3953
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42570487606510526,
      "learning_rate": 2.99899662470555e-05,
      "loss": 0.7337,
      "step": 3954
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44816379834868814,
      "learning_rate": 2.9984948113603242e-05,
      "loss": 0.7007,
      "step": 3955
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46784627918217203,
      "learning_rate": 2.9979929142725445e-05,
      "loss": 0.7422,
      "step": 3956
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3832886106791358,
      "learning_rate": 2.9974909334843042e-05,
      "loss": 0.6546,
      "step": 3957
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4037630211035019,
      "learning_rate": 2.9969888690377043e-05,
      "loss": 0.6669,
      "step": 3958
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38889928053108147,
      "learning_rate": 2.996486720974852e-05,
      "loss": 0.6903,
      "step": 3959
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39232141900003775,
      "learning_rate": 2.9959844893378617e-05,
      "loss": 0.7091,
      "step": 3960
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39572697040783,
      "learning_rate": 2.9954821741688548e-05,
      "loss": 0.6906,
      "step": 3961
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3622170563541115,
      "learning_rate": 2.994979775509961e-05,
      "loss": 0.6122,
      "step": 3962
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.851601176534883,
      "learning_rate": 2.9944772934033143e-05,
      "loss": 1.2129,
      "step": 3963
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39989027871571275,
      "learning_rate": 2.9939747278910585e-05,
      "loss": 0.697,
      "step": 3964
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46616326902483196,
      "learning_rate": 2.9934720790153426e-05,
      "loss": 0.7254,
      "step": 3965
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3959778532783042,
      "learning_rate": 2.992969346818323e-05,
      "loss": 0.6721,
      "step": 3966
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4152666601067355,
      "learning_rate": 2.9924665313421638e-05,
      "loss": 0.6596,
      "step": 3967
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3838769595872265,
      "learning_rate": 2.9919636326290348e-05,
      "loss": 0.6366,
      "step": 3968
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4849428782890099,
      "learning_rate": 2.9914606507211145e-05,
      "loss": 0.6583,
      "step": 3969
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39416556315032025,
      "learning_rate": 2.9909575856605867e-05,
      "loss": 0.687,
      "step": 3970
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.456080053765969,
      "learning_rate": 2.990454437489643e-05,
      "loss": 0.7025,
      "step": 3971
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4139459267448722,
      "learning_rate": 2.9899512062504814e-05,
      "loss": 0.7247,
      "step": 3972
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3979269031552799,
      "learning_rate": 2.9894478919853077e-05,
      "loss": 0.7215,
      "step": 3973
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43005708307105955,
      "learning_rate": 2.9889444947363345e-05,
      "loss": 0.692,
      "step": 3974
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.371688108321273,
      "learning_rate": 2.9884410145457806e-05,
      "loss": 0.7051,
      "step": 3975
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4340022835584094,
      "learning_rate": 2.987937451455873e-05,
      "loss": 0.7375,
      "step": 3976
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4246570846842158,
      "learning_rate": 2.987433805508844e-05,
      "loss": 0.6692,
      "step": 3977
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39849806465106685,
      "learning_rate": 2.9869300767469346e-05,
      "loss": 0.6474,
      "step": 3978
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3938244013436708,
      "learning_rate": 2.986426265212391e-05,
      "loss": 0.6774,
      "step": 3979
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.432800954993924,
      "learning_rate": 2.9859223709474688e-05,
      "loss": 0.6875,
      "step": 3980
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3847579765762694,
      "learning_rate": 2.985418393994427e-05,
      "loss": 0.658,
      "step": 3981
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3660236879753907,
      "learning_rate": 2.9849143343955353e-05,
      "loss": 0.6789,
      "step": 3982
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36747760329198315,
      "learning_rate": 2.984410192193068e-05,
      "loss": 0.663,
      "step": 3983
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3564111599941103,
      "learning_rate": 2.9839059674293058e-05,
      "loss": 0.6937,
      "step": 3984
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4437350082095174,
      "learning_rate": 2.9834016601465393e-05,
      "loss": 0.6965,
      "step": 3985
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.733665071371842,
      "learning_rate": 2.9828972703870625e-05,
      "loss": 1.0372,
      "step": 3986
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36251836842322227,
      "learning_rate": 2.98239279819318e-05,
      "loss": 0.644,
      "step": 3987
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43401789578120314,
      "learning_rate": 2.9818882436071983e-05,
      "loss": 0.6886,
      "step": 3988
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4635394325943281,
      "learning_rate": 2.9813836066714362e-05,
      "loss": 0.6884,
      "step": 3989
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38816539137887435,
      "learning_rate": 2.9808788874282162e-05,
      "loss": 0.6824,
      "step": 3990
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3582834698935446,
      "learning_rate": 2.980374085919868e-05,
      "loss": 0.6458,
      "step": 3991
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4853596781680777,
      "learning_rate": 2.97986920218873e-05,
      "loss": 0.7626,
      "step": 3992
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4663342644862225,
      "learning_rate": 2.9793642362771452e-05,
      "loss": 0.667,
      "step": 3993
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.48465277242060995,
      "learning_rate": 2.978859188227464e-05,
      "loss": 0.7236,
      "step": 3994
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.425531112630938,
      "learning_rate": 2.9783540580820456e-05,
      "loss": 0.6507,
      "step": 3995
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4485000282258258,
      "learning_rate": 2.9778488458832537e-05,
      "loss": 0.7157,
      "step": 3996
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44128949315272153,
      "learning_rate": 2.9773435516734602e-05,
      "loss": 0.6935,
      "step": 3997
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40764367868173296,
      "learning_rate": 2.9768381754950426e-05,
      "loss": 0.7188,
      "step": 3998
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40605802903342075,
      "learning_rate": 2.9763327173903874e-05,
      "loss": 0.7176,
      "step": 3999
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4104890431221888,
      "learning_rate": 2.975827177401887e-05,
      "loss": 0.6641,
      "step": 4000
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43188553603563984,
      "learning_rate": 2.975321555571939e-05,
      "loss": 0.686,
      "step": 4001
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4130000410939252,
      "learning_rate": 2.9748158519429497e-05,
      "loss": 0.714,
      "step": 4002
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3726892229605209,
      "learning_rate": 2.9743100665573333e-05,
      "loss": 0.6336,
      "step": 4003
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38069883153902556,
      "learning_rate": 2.9738041994575076e-05,
      "loss": 0.67,
      "step": 4004
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3668511990017474,
      "learning_rate": 2.9732982506858992e-05,
      "loss": 0.6714,
      "step": 4005
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3866605560875937,
      "learning_rate": 2.9727922202849426e-05,
      "loss": 0.6756,
      "step": 4006
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39868297412975157,
      "learning_rate": 2.9722861082970777e-05,
      "loss": 0.6932,
      "step": 4007
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38963861153437584,
      "learning_rate": 2.9717799147647505e-05,
      "loss": 0.6872,
      "step": 4008
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37792371482272535,
      "learning_rate": 2.9712736397304158e-05,
      "loss": 0.6849,
      "step": 4009
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3860798875657147,
      "learning_rate": 2.9707672832365335e-05,
      "loss": 0.6562,
      "step": 4010
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3962415174444205,
      "learning_rate": 2.970260845325572e-05,
      "loss": 0.6639,
      "step": 4011
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4004092074418327,
      "learning_rate": 2.9697543260400045e-05,
      "loss": 0.6587,
      "step": 4012
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.47938619173092606,
      "learning_rate": 2.9692477254223137e-05,
      "loss": 0.6865,
      "step": 4013
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4248588812606017,
      "learning_rate": 2.9687410435149865e-05,
      "loss": 0.7202,
      "step": 4014
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3772467302201537,
      "learning_rate": 2.9682342803605175e-05,
      "loss": 0.6592,
      "step": 4015
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3935258191110788,
      "learning_rate": 2.9677274360014095e-05,
      "loss": 0.6814,
      "step": 4016
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4523122169770517,
      "learning_rate": 2.9672205104801697e-05,
      "loss": 0.7168,
      "step": 4017
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39318721078708835,
      "learning_rate": 2.9667135038393135e-05,
      "loss": 0.701,
      "step": 4018
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4119297323604448,
      "learning_rate": 2.9662064161213636e-05,
      "loss": 0.6609,
      "step": 4019
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37651096881089313,
      "learning_rate": 2.965699247368848e-05,
      "loss": 0.6847,
      "step": 4020
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4163428491625461,
      "learning_rate": 2.9651919976243034e-05,
      "loss": 0.6996,
      "step": 4021
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42964230470831904,
      "learning_rate": 2.9646846669302712e-05,
      "loss": 0.6827,
      "step": 4022
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3840590166659001,
      "learning_rate": 2.9641772553293012e-05,
      "loss": 0.6841,
      "step": 4023
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40501494154195294,
      "learning_rate": 2.9636697628639494e-05,
      "loss": 0.6716,
      "step": 4024
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4502266101299692,
      "learning_rate": 2.963162189576778e-05,
      "loss": 0.7283,
      "step": 4025
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44300988751696113,
      "learning_rate": 2.9626545355103572e-05,
      "loss": 0.6659,
      "step": 4026
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4168124810170763,
      "learning_rate": 2.9621468007072637e-05,
      "loss": 0.7102,
      "step": 4027
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4244120771500816,
      "learning_rate": 2.961638985210079e-05,
      "loss": 0.6803,
      "step": 4028
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33739875869233815,
      "learning_rate": 2.961131089061394e-05,
      "loss": 0.6321,
      "step": 4029
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.5452904986456955,
      "learning_rate": 2.960623112303806e-05,
      "loss": 0.8524,
      "step": 4030
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4071217862620117,
      "learning_rate": 2.960115054979918e-05,
      "loss": 0.6773,
      "step": 4031
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.406993402091068,
      "learning_rate": 2.9596069171323392e-05,
      "loss": 0.6581,
      "step": 4032
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4192776903946344,
      "learning_rate": 2.959098698803687e-05,
      "loss": 0.7045,
      "step": 4033
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4353311486841643,
      "learning_rate": 2.9585904000365856e-05,
      "loss": 0.705,
      "step": 4034
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3785110254109617,
      "learning_rate": 2.9580820208736648e-05,
      "loss": 0.6961,
      "step": 4035
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4465294078144305,
      "learning_rate": 2.9575735613575615e-05,
      "loss": 0.7304,
      "step": 4036
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3672207831518609,
      "learning_rate": 2.9570650215309214e-05,
      "loss": 0.6889,
      "step": 4037
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4070334167507681,
      "learning_rate": 2.956556401436393e-05,
      "loss": 0.6753,
      "step": 4038
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4276245756733852,
      "learning_rate": 2.9560477011166344e-05,
      "loss": 0.7013,
      "step": 4039
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39725383406622805,
      "learning_rate": 2.9555389206143098e-05,
      "loss": 0.723,
      "step": 4040
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41104998892256384,
      "learning_rate": 2.95503005997209e-05,
      "loss": 0.6994,
      "step": 4041
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38416881994071156,
      "learning_rate": 2.954521119232652e-05,
      "loss": 0.6632,
      "step": 4042
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4068270100104131,
      "learning_rate": 2.954012098438681e-05,
      "loss": 0.657,
      "step": 4043
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4311147781534054,
      "learning_rate": 2.9535029976328675e-05,
      "loss": 0.6831,
      "step": 4044
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36402881717237945,
      "learning_rate": 2.952993816857909e-05,
      "loss": 0.6763,
      "step": 4045
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3960413759897233,
      "learning_rate": 2.9524845561565096e-05,
      "loss": 0.6727,
      "step": 4046
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40316026204579297,
      "learning_rate": 2.9519752155713814e-05,
      "loss": 0.6581,
      "step": 4047
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3953208399319326,
      "learning_rate": 2.9514657951452414e-05,
      "loss": 0.6943,
      "step": 4048
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4299303129649733,
      "learning_rate": 2.950956294920814e-05,
      "loss": 0.6975,
      "step": 4049
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43174317311479476,
      "learning_rate": 2.9504467149408306e-05,
      "loss": 0.6975,
      "step": 4050
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4318634498090579,
      "learning_rate": 2.9499370552480298e-05,
      "loss": 0.6936,
      "step": 4051
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4308373481350626,
      "learning_rate": 2.9494273158851542e-05,
      "loss": 0.7252,
      "step": 4052
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35598235212509816,
      "learning_rate": 2.948917496894957e-05,
      "loss": 0.6722,
      "step": 4053
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4168729259934421,
      "learning_rate": 2.9484075983201953e-05,
      "loss": 0.677,
      "step": 4054
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3816326783581216,
      "learning_rate": 2.947897620203634e-05,
      "loss": 0.6569,
      "step": 4055
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4373451029063189,
      "learning_rate": 2.9473875625880435e-05,
      "loss": 0.7223,
      "step": 4056
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.45507269977434167,
      "learning_rate": 2.946877425516202e-05,
      "loss": 0.6996,
      "step": 4057
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3963374813903327,
      "learning_rate": 2.9463672090308956e-05,
      "loss": 0.7012,
      "step": 4058
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40713491919147937,
      "learning_rate": 2.945856913174913e-05,
      "loss": 0.6826,
      "step": 4059
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41227255513565314,
      "learning_rate": 2.945346537991054e-05,
      "loss": 0.7141,
      "step": 4060
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3712540988342609,
      "learning_rate": 2.9448360835221223e-05,
      "loss": 0.6399,
      "step": 4061
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4166470153410077,
      "learning_rate": 2.9443255498109295e-05,
      "loss": 0.6722,
      "step": 4062
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3627302276282655,
      "learning_rate": 2.9438149369002937e-05,
      "loss": 0.727,
      "step": 4063
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38522823557693003,
      "learning_rate": 2.9433042448330384e-05,
      "loss": 0.7114,
      "step": 4064
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38902325023218526,
      "learning_rate": 2.9427934736519962e-05,
      "loss": 0.7066,
      "step": 4065
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4046619802827245,
      "learning_rate": 2.9422826234000024e-05,
      "loss": 0.6823,
      "step": 4066
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3951344254028464,
      "learning_rate": 2.941771694119904e-05,
      "loss": 0.6858,
      "step": 4067
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40265970377878424,
      "learning_rate": 2.941260685854551e-05,
      "loss": 0.6863,
      "step": 4068
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3543714325336068,
      "learning_rate": 2.9407495986468007e-05,
      "loss": 0.6764,
      "step": 4069
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3334804505786733,
      "learning_rate": 2.9402384325395182e-05,
      "loss": 0.6341,
      "step": 4070
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40756182422749593,
      "learning_rate": 2.9397271875755734e-05,
      "loss": 0.7,
      "step": 4071
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39115676887460354,
      "learning_rate": 2.939215863797844e-05,
      "loss": 0.6675,
      "step": 4072
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38719355907460984,
      "learning_rate": 2.9387044612492155e-05,
      "loss": 0.697,
      "step": 4073
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4176823770410575,
      "learning_rate": 2.9381929799725764e-05,
      "loss": 0.7098,
      "step": 4074
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40297587557262854,
      "learning_rate": 2.937681420010826e-05,
      "loss": 0.6831,
      "step": 4075
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3918112947822158,
      "learning_rate": 2.9371697814068665e-05,
      "loss": 0.6761,
      "step": 4076
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36129662076320057,
      "learning_rate": 2.9366580642036096e-05,
      "loss": 0.6871,
      "step": 4077
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36275349494987735,
      "learning_rate": 2.936146268443972e-05,
      "loss": 0.6742,
      "step": 4078
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41992769821127196,
      "learning_rate": 2.9356343941708777e-05,
      "loss": 0.7061,
      "step": 4079
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3956443525322908,
      "learning_rate": 2.9351224414272566e-05,
      "loss": 0.6952,
      "step": 4080
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32473994272525214,
      "learning_rate": 2.9346104102560455e-05,
      "loss": 0.6609,
      "step": 4081
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34842170517446275,
      "learning_rate": 2.9340983007001882e-05,
      "loss": 0.6841,
      "step": 4082
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4097443611194442,
      "learning_rate": 2.9335861128026345e-05,
      "loss": 0.6557,
      "step": 4083
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3618672582813145,
      "learning_rate": 2.933073846606341e-05,
      "loss": 0.6575,
      "step": 4084
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4178188685800746,
      "learning_rate": 2.932561502154271e-05,
      "loss": 0.6797,
      "step": 4085
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43687334133892697,
      "learning_rate": 2.9320490794893937e-05,
      "loss": 0.7112,
      "step": 4086
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37968774923653925,
      "learning_rate": 2.931536578654686e-05,
      "loss": 0.6808,
      "step": 4087
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40070143656809903,
      "learning_rate": 2.9310239996931303e-05,
      "loss": 0.6614,
      "step": 4088
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39388615331795784,
      "learning_rate": 2.930511342647716e-05,
      "loss": 0.6999,
      "step": 4089
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4223858868725718,
      "learning_rate": 2.92999860756144e-05,
      "loss": 0.7053,
      "step": 4090
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39911185599065696,
      "learning_rate": 2.929485794477303e-05,
      "loss": 0.6503,
      "step": 4091
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40200890193696937,
      "learning_rate": 2.9289729034383156e-05,
      "loss": 0.7203,
      "step": 4092
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43326465984689105,
      "learning_rate": 2.928459934487492e-05,
      "loss": 0.7092,
      "step": 4093
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9574135007401414,
      "learning_rate": 2.9279468876678554e-05,
      "loss": 0.6875,
      "step": 4094
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43584607277675175,
      "learning_rate": 2.927433763022434e-05,
      "loss": 0.7255,
      "step": 4095
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.45552844828989975,
      "learning_rate": 2.9269205605942628e-05,
      "loss": 0.6983,
      "step": 4096
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.35496092212022423,
      "learning_rate": 2.9264072804263838e-05,
      "loss": 0.6553,
      "step": 4097
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38274543154511664,
      "learning_rate": 2.925893922561845e-05,
      "loss": 0.6903,
      "step": 4098
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3704159890995331,
      "learning_rate": 2.9253804870437008e-05,
      "loss": 0.6854,
      "step": 4099
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4187344215506257,
      "learning_rate": 2.9248669739150134e-05,
      "loss": 0.7255,
      "step": 4100
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42479970000000067,
      "learning_rate": 2.9243533832188497e-05,
      "loss": 0.6755,
      "step": 4101
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40432587761095773,
      "learning_rate": 2.9238397149982836e-05,
      "loss": 0.6682,
      "step": 4102
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36712539919126685,
      "learning_rate": 2.9233259692963964e-05,
      "loss": 0.6323,
      "step": 4103
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4088776489377427,
      "learning_rate": 2.9228121461562754e-05,
      "loss": 0.6696,
      "step": 4104
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.370565482561817,
      "learning_rate": 2.922298245621014e-05,
      "loss": 0.6252,
      "step": 4105
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39345637982317283,
      "learning_rate": 2.9217842677337125e-05,
      "loss": 0.6978,
      "step": 4106
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41104339635472403,
      "learning_rate": 2.9212702125374782e-05,
      "loss": 0.6859,
      "step": 4107
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3724327849721282,
      "learning_rate": 2.9207560800754237e-05,
      "loss": 0.6795,
      "step": 4108
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4134570578947115,
      "learning_rate": 2.920241870390669e-05,
      "loss": 0.7048,
      "step": 4109
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3279155922124161,
      "learning_rate": 2.9197275835263395e-05,
      "loss": 0.6435,
      "step": 4110
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3864654334918345,
      "learning_rate": 2.9192132195255686e-05,
      "loss": 0.6889,
      "step": 4111
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3904525927504222,
      "learning_rate": 2.9186987784314946e-05,
      "loss": 0.6738,
      "step": 4112
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.416133087058357,
      "learning_rate": 2.9181842602872638e-05,
      "loss": 0.7405,
      "step": 4113
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3667547270844445,
      "learning_rate": 2.9176696651360283e-05,
      "loss": 0.6858,
      "step": 4114
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3586099175960988,
      "learning_rate": 2.9171549930209463e-05,
      "loss": 0.6554,
      "step": 4115
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38581744144528246,
      "learning_rate": 2.9166402439851817e-05,
      "loss": 0.6963,
      "step": 4116
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39947606765933324,
      "learning_rate": 2.916125418071908e-05,
      "loss": 0.7133,
      "step": 4117
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40359715467062773,
      "learning_rate": 2.9156105153243016e-05,
      "loss": 0.7261,
      "step": 4118
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3553650030704542,
      "learning_rate": 2.915095535785547e-05,
      "loss": 0.6983,
      "step": 4119
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4038175113495886,
      "learning_rate": 2.9145804794988345e-05,
      "loss": 0.6872,
      "step": 4120
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39125263619336625,
      "learning_rate": 2.914065346507362e-05,
      "loss": 0.7081,
      "step": 4121
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4044731897113818,
      "learning_rate": 2.913550136854333e-05,
      "loss": 0.6643,
      "step": 4122
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.33886806659910956,
      "learning_rate": 2.913034850582957e-05,
      "loss": 0.6598,
      "step": 4123
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4000170233561872,
      "learning_rate": 2.912519487736451e-05,
      "loss": 0.6537,
      "step": 4124
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38614932046296463,
      "learning_rate": 2.9120040483580378e-05,
      "loss": 0.6966,
      "step": 4125
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41376225268020966,
      "learning_rate": 2.9114885324909462e-05,
      "loss": 0.6649,
      "step": 4126
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39438269712701746,
      "learning_rate": 2.9109729401784118e-05,
      "loss": 0.6871,
      "step": 4127
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3861520753587865,
      "learning_rate": 2.910457271463678e-05,
      "loss": 0.651,
      "step": 4128
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39968295152430644,
      "learning_rate": 2.9099415263899917e-05,
      "loss": 0.6668,
      "step": 4129
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3718960207068686,
      "learning_rate": 2.909425705000609e-05,
      "loss": 0.668,
      "step": 4130
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4519941249751675,
      "learning_rate": 2.9089098073387904e-05,
      "loss": 0.7234,
      "step": 4131
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4063496186387659,
      "learning_rate": 2.9083938334478042e-05,
      "loss": 0.6746,
      "step": 4132
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3496480100329123,
      "learning_rate": 2.9078777833709235e-05,
      "loss": 0.6318,
      "step": 4133
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40735283082878504,
      "learning_rate": 2.90736165715143e-05,
      "loss": 0.6723,
      "step": 4134
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39302627478439106,
      "learning_rate": 2.9068454548326106e-05,
      "loss": 0.6557,
      "step": 4135
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.35400740540000003,
      "learning_rate": 2.9063291764577575e-05,
      "loss": 0.7099,
      "step": 4136
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3801690757233998,
      "learning_rate": 2.9058128220701714e-05,
      "loss": 0.6718,
      "step": 4137
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3851879953234649,
      "learning_rate": 2.9052963917131574e-05,
      "loss": 0.7117,
      "step": 4138
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.366108070055261,
      "learning_rate": 2.9047798854300288e-05,
      "loss": 0.6715,
      "step": 4139
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3803208474409943,
      "learning_rate": 2.9042633032641034e-05,
      "loss": 0.6931,
      "step": 4140
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43335513089555056,
      "learning_rate": 2.903746645258707e-05,
      "loss": 0.6917,
      "step": 4141
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38101077301802866,
      "learning_rate": 2.9032299114571713e-05,
      "loss": 0.6723,
      "step": 4142
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38303297133489694,
      "learning_rate": 2.9027131019028334e-05,
      "loss": 0.6622,
      "step": 4143
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37774379109677414,
      "learning_rate": 2.9021962166390377e-05,
      "loss": 0.665,
      "step": 4144
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5393901776262415,
      "learning_rate": 2.9016792557091355e-05,
      "loss": 0.6471,
      "step": 4145
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4373056298328023,
      "learning_rate": 2.9011622191564824e-05,
      "loss": 0.7213,
      "step": 4146
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4254517392632128,
      "learning_rate": 2.9006451070244422e-05,
      "loss": 0.7025,
      "step": 4147
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.35746678340409477,
      "learning_rate": 2.9001279193563853e-05,
      "loss": 0.6462,
      "step": 4148
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4303058144122979,
      "learning_rate": 2.8996106561956868e-05,
      "loss": 0.6885,
      "step": 4149
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39545659802877275,
      "learning_rate": 2.899093317585728e-05,
      "loss": 0.6978,
      "step": 4150
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38146552700336384,
      "learning_rate": 2.8985759035698997e-05,
      "loss": 0.6711,
      "step": 4151
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4304105016944384,
      "learning_rate": 2.8980584141915957e-05,
      "loss": 0.6963,
      "step": 4152
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37974489783984056,
      "learning_rate": 2.897540849494216e-05,
      "loss": 0.6972,
      "step": 4153
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42261365121425626,
      "learning_rate": 2.8970232095211694e-05,
      "loss": 0.6878,
      "step": 4154
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4139675955511528,
      "learning_rate": 2.8965054943158707e-05,
      "loss": 0.6192,
      "step": 4155
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4178644522609294,
      "learning_rate": 2.8959877039217375e-05,
      "loss": 0.6846,
      "step": 4156
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3776300996084323,
      "learning_rate": 2.895469838382198e-05,
      "loss": 0.6767,
      "step": 4157
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38024371023227893,
      "learning_rate": 2.894951897740685e-05,
      "loss": 0.6857,
      "step": 4158
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3707105415707956,
      "learning_rate": 2.8944338820406376e-05,
      "loss": 0.7003,
      "step": 4159
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3612806032245561,
      "learning_rate": 2.8939157913254994e-05,
      "loss": 0.656,
      "step": 4160
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3544462937692785,
      "learning_rate": 2.893397625638724e-05,
      "loss": 0.6445,
      "step": 4161
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3634154145259612,
      "learning_rate": 2.8928793850237694e-05,
      "loss": 0.6719,
      "step": 4162
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3310966098400475,
      "learning_rate": 2.892361069524098e-05,
      "loss": 0.6398,
      "step": 4163
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3892036027444643,
      "learning_rate": 2.8918426791831815e-05,
      "loss": 0.6935,
      "step": 4164
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4074991292100098,
      "learning_rate": 2.891324214044497e-05,
      "loss": 0.6924,
      "step": 4165
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39406010560332394,
      "learning_rate": 2.8908056741515276e-05,
      "loss": 0.68,
      "step": 4166
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36586026753864115,
      "learning_rate": 2.8902870595477615e-05,
      "loss": 0.6789,
      "step": 4167
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3804263510982238,
      "learning_rate": 2.8897683702766946e-05,
      "loss": 0.6959,
      "step": 4168
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3490188139955435,
      "learning_rate": 2.88924960638183e-05,
      "loss": 0.6698,
      "step": 4169
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40028814223224746,
      "learning_rate": 2.8887307679066737e-05,
      "loss": 0.6769,
      "step": 4170
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4480841600747902,
      "learning_rate": 2.8882118548947415e-05,
      "loss": 0.6997,
      "step": 4171
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4526685217578336,
      "learning_rate": 2.8876928673895545e-05,
      "loss": 0.7467,
      "step": 4172
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46896536699141755,
      "learning_rate": 2.887173805434638e-05,
      "loss": 0.7006,
      "step": 4173
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41797805778024116,
      "learning_rate": 2.8866546690735258e-05,
      "loss": 0.6958,
      "step": 4174
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3772839761633901,
      "learning_rate": 2.8861354583497577e-05,
      "loss": 0.6811,
      "step": 4175
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.33176780706475295,
      "learning_rate": 2.8856161733068794e-05,
      "loss": 0.6486,
      "step": 4176
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3788564913749019,
      "learning_rate": 2.8850968139884416e-05,
      "loss": 0.619,
      "step": 4177
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3549457016691405,
      "learning_rate": 2.8845773804380028e-05,
      "loss": 0.6632,
      "step": 4178
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38739402956299257,
      "learning_rate": 2.8840578726991278e-05,
      "loss": 0.6812,
      "step": 4179
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4314370464598644,
      "learning_rate": 2.8835382908153865e-05,
      "loss": 0.6724,
      "step": 4180
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3471185578778505,
      "learning_rate": 2.8830186348303555e-05,
      "loss": 0.6497,
      "step": 4181
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4162009165607772,
      "learning_rate": 2.8824989047876185e-05,
      "loss": 0.6821,
      "step": 4182
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3838560617087708,
      "learning_rate": 2.8819791007307638e-05,
      "loss": 0.5985,
      "step": 4183
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.35152631562820075,
      "learning_rate": 2.881459222703387e-05,
      "loss": 0.6569,
      "step": 4184
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.484677573406484,
      "learning_rate": 2.88093927074909e-05,
      "loss": 0.5941,
      "step": 4185
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4213125043997126,
      "learning_rate": 2.8804192449114805e-05,
      "loss": 0.6701,
      "step": 4186
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3783803940907819,
      "learning_rate": 2.8798991452341715e-05,
      "loss": 0.6628,
      "step": 4187
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3608325625888926,
      "learning_rate": 2.879378971760784e-05,
      "loss": 0.6367,
      "step": 4188
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37687180559209543,
      "learning_rate": 2.8788587245349448e-05,
      "loss": 0.6674,
      "step": 4189
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38094721224571704,
      "learning_rate": 2.8783384036002853e-05,
      "loss": 0.6854,
      "step": 4190
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4426191109829665,
      "learning_rate": 2.877818009000445e-05,
      "loss": 0.6598,
      "step": 4191
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3921287965405898,
      "learning_rate": 2.877297540779068e-05,
      "loss": 0.653,
      "step": 4192
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3972190548618616,
      "learning_rate": 2.876776998979806e-05,
      "loss": 0.6968,
      "step": 4193
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3714728017467796,
      "learning_rate": 2.8762563836463155e-05,
      "loss": 0.6857,
      "step": 4194
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3864782035540937,
      "learning_rate": 2.875735694822261e-05,
      "loss": 0.6588,
      "step": 4195
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3903041515318376,
      "learning_rate": 2.875214932551311e-05,
      "loss": 0.6517,
      "step": 4196
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3635479958437173,
      "learning_rate": 2.874694096877142e-05,
      "loss": 0.694,
      "step": 4197
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3679633636175745,
      "learning_rate": 2.874173187843435e-05,
      "loss": 0.7176,
      "step": 4198
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3934087278345082,
      "learning_rate": 2.873652205493879e-05,
      "loss": 0.6916,
      "step": 4199
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3899954674845463,
      "learning_rate": 2.8731311498721667e-05,
      "loss": 0.7022,
      "step": 4200
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35494201919970597,
      "learning_rate": 2.8726100210220004e-05,
      "loss": 0.666,
      "step": 4201
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37354955397682155,
      "learning_rate": 2.8720888189870846e-05,
      "loss": 0.6934,
      "step": 4202
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3607504662477146,
      "learning_rate": 2.8715675438111334e-05,
      "loss": 0.6861,
      "step": 4203
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3655883370313586,
      "learning_rate": 2.8710461955378645e-05,
      "loss": 0.6771,
      "step": 4204
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4159264709123502,
      "learning_rate": 2.8705247742110037e-05,
      "loss": 0.6742,
      "step": 4205
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43614959577455736,
      "learning_rate": 2.870003279874281e-05,
      "loss": 0.7088,
      "step": 4206
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3695818051197759,
      "learning_rate": 2.8694817125714335e-05,
      "loss": 0.6545,
      "step": 4207
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4207565104541963,
      "learning_rate": 2.8689600723462056e-05,
      "loss": 0.7245,
      "step": 4208
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4411602381194853,
      "learning_rate": 2.8684383592423452e-05,
      "loss": 0.697,
      "step": 4209
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42953673567520273,
      "learning_rate": 2.867916573303609e-05,
      "loss": 0.7196,
      "step": 4210
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4554453771658321,
      "learning_rate": 2.8673947145737577e-05,
      "loss": 0.705,
      "step": 4211
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4521808027199821,
      "learning_rate": 2.8668727830965592e-05,
      "loss": 0.7397,
      "step": 4212
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45252896835477535,
      "learning_rate": 2.8663507789157873e-05,
      "loss": 0.6757,
      "step": 4213
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37231749251065943,
      "learning_rate": 2.8658287020752217e-05,
      "loss": 0.6813,
      "step": 4214
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3781198606604896,
      "learning_rate": 2.8653065526186492e-05,
      "loss": 0.6745,
      "step": 4215
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41557055340246685,
      "learning_rate": 2.8647843305898605e-05,
      "loss": 0.6748,
      "step": 4216
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.34168682607990336,
      "learning_rate": 2.8642620360326546e-05,
      "loss": 0.6598,
      "step": 4217
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39971583907530756,
      "learning_rate": 2.8637396689908353e-05,
      "loss": 0.6657,
      "step": 4218
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.390989558254911,
      "learning_rate": 2.8632172295082132e-05,
      "loss": 0.685,
      "step": 4219
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40416802258883894,
      "learning_rate": 2.862694717628605e-05,
      "loss": 0.698,
      "step": 4220
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3944603267675047,
      "learning_rate": 2.862172133395832e-05,
      "loss": 0.6592,
      "step": 4221
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41945638706517996,
      "learning_rate": 2.861649476853724e-05,
      "loss": 0.6475,
      "step": 4222
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40252744393126316,
      "learning_rate": 2.8611267480461148e-05,
      "loss": 0.6415,
      "step": 4223
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43701774074148136,
      "learning_rate": 2.860603947016845e-05,
      "loss": 0.667,
      "step": 4224
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.375702993983104,
      "learning_rate": 2.860081073809762e-05,
      "loss": 0.6861,
      "step": 4225
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3697606050306782,
      "learning_rate": 2.8595581284687176e-05,
      "loss": 0.6556,
      "step": 4226
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4275725240696932,
      "learning_rate": 2.859035111037571e-05,
      "loss": 0.6806,
      "step": 4227
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4037282855075002,
      "learning_rate": 2.8585120215601872e-05,
      "loss": 0.6911,
      "step": 4228
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3654441658254939,
      "learning_rate": 2.857988860080437e-05,
      "loss": 0.6607,
      "step": 4229
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40424623463508247,
      "learning_rate": 2.8574656266421975e-05,
      "loss": 0.6827,
      "step": 4230
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3750232461964951,
      "learning_rate": 2.856942321289351e-05,
      "loss": 0.6933,
      "step": 4231
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3896847166237503,
      "learning_rate": 2.8564189440657876e-05,
      "loss": 0.675,
      "step": 4232
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40683454935356833,
      "learning_rate": 2.855895495015401e-05,
      "loss": 0.7033,
      "step": 4233
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3384366742003786,
      "learning_rate": 2.8553719741820928e-05,
      "loss": 0.6559,
      "step": 4234
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32338120846364576,
      "learning_rate": 2.854848381609771e-05,
      "loss": 0.6639,
      "step": 4235
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4126536065201781,
      "learning_rate": 2.8543247173423477e-05,
      "loss": 0.6502,
      "step": 4236
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4088987133018401,
      "learning_rate": 2.8538009814237412e-05,
      "loss": 0.679,
      "step": 4237
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.334469247561339,
      "learning_rate": 2.853277173897878e-05,
      "loss": 0.6843,
      "step": 4238
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4040372478836638,
      "learning_rate": 2.852753294808689e-05,
      "loss": 0.6356,
      "step": 4239
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4717954194179103,
      "learning_rate": 2.8522293442001108e-05,
      "loss": 0.7133,
      "step": 4240
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36367982101847346,
      "learning_rate": 2.8517053221160868e-05,
      "loss": 0.6441,
      "step": 4241
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4323397453148139,
      "learning_rate": 2.8511812286005662e-05,
      "loss": 0.7235,
      "step": 4242
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4069058385239383,
      "learning_rate": 2.8506570636975038e-05,
      "loss": 0.6612,
      "step": 4243
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37259403448752304,
      "learning_rate": 2.8501328274508604e-05,
      "loss": 0.6649,
      "step": 4244
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41640227894560977,
      "learning_rate": 2.849608519904604e-05,
      "loss": 0.6899,
      "step": 4245
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4274386518693982,
      "learning_rate": 2.849084141102707e-05,
      "loss": 0.7311,
      "step": 4246
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46396615386958856,
      "learning_rate": 2.8485596910891482e-05,
      "loss": 0.7392,
      "step": 4247
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4152259830099945,
      "learning_rate": 2.8480351699079133e-05,
      "loss": 0.582,
      "step": 4248
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4152573005335094,
      "learning_rate": 2.8475105776029924e-05,
      "loss": 0.6798,
      "step": 4249
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3997394403711439,
      "learning_rate": 2.8469859142183836e-05,
      "loss": 0.6711,
      "step": 4250
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3794129326738284,
      "learning_rate": 2.8464611797980885e-05,
      "loss": 0.7088,
      "step": 4251
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4028778239799101,
      "learning_rate": 2.8459363743861166e-05,
      "loss": 0.6821,
      "step": 4252
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36701280428180777,
      "learning_rate": 2.8454114980264827e-05,
      "loss": 0.629,
      "step": 4253
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37271305462833626,
      "learning_rate": 2.8448865507632075e-05,
      "loss": 0.6787,
      "step": 4254
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3939287226417475,
      "learning_rate": 2.844361532640317e-05,
      "loss": 0.6782,
      "step": 4255
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4364053396473,
      "learning_rate": 2.8438364437018457e-05,
      "loss": 0.7372,
      "step": 4256
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4153706334112258,
      "learning_rate": 2.84331128399183e-05,
      "loss": 0.6774,
      "step": 4257
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3872868787552275,
      "learning_rate": 2.8427860535543157e-05,
      "loss": 0.7006,
      "step": 4258
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4193851469116088,
      "learning_rate": 2.842260752433353e-05,
      "loss": 0.6581,
      "step": 4259
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0249448648507493,
      "learning_rate": 2.8417353806729986e-05,
      "loss": 0.5288,
      "step": 4260
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38260917777307274,
      "learning_rate": 2.841209938317313e-05,
      "loss": 0.6593,
      "step": 4261
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4328384031563208,
      "learning_rate": 2.840684425410367e-05,
      "loss": 0.7012,
      "step": 4262
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40409401593677324,
      "learning_rate": 2.8401588419962337e-05,
      "loss": 0.6728,
      "step": 4263
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3925737769267193,
      "learning_rate": 2.8396331881189923e-05,
      "loss": 0.665,
      "step": 4264
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38528965364987167,
      "learning_rate": 2.8391074638227297e-05,
      "loss": 0.6545,
      "step": 4265
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4068058065846135,
      "learning_rate": 2.8385816691515376e-05,
      "loss": 0.6646,
      "step": 4266
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3970550287790882,
      "learning_rate": 2.8380558041495137e-05,
      "loss": 0.6845,
      "step": 4267
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3500940288944195,
      "learning_rate": 2.8375298688607605e-05,
      "loss": 0.6616,
      "step": 4268
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37990385558649536,
      "learning_rate": 2.8370038633293902e-05,
      "loss": 0.6758,
      "step": 4269
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43581302025519203,
      "learning_rate": 2.836477787599516e-05,
      "loss": 0.6625,
      "step": 4270
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4013621072376423,
      "learning_rate": 2.83595164171526e-05,
      "loss": 0.7038,
      "step": 4271
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4100421877949024,
      "learning_rate": 2.8354254257207495e-05,
      "loss": 0.717,
      "step": 4272
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3689330340177353,
      "learning_rate": 2.8348991396601176e-05,
      "loss": 0.6556,
      "step": 4273
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4056053336939598,
      "learning_rate": 2.8343727835775027e-05,
      "loss": 0.688,
      "step": 4274
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1338478322400867,
      "learning_rate": 2.83384635751705e-05,
      "loss": 0.5207,
      "step": 4275
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4116578659472129,
      "learning_rate": 2.8333198615229113e-05,
      "loss": 0.7356,
      "step": 4276
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42596846173799074,
      "learning_rate": 2.832793295639242e-05,
      "loss": 0.7035,
      "step": 4277
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35828698505711015,
      "learning_rate": 2.832266659910204e-05,
      "loss": 0.6516,
      "step": 4278
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3878777698092221,
      "learning_rate": 2.8317399543799667e-05,
      "loss": 0.6607,
      "step": 4279
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8852575469065223,
      "learning_rate": 2.8312131790927044e-05,
      "loss": 0.5329,
      "step": 4280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4410263356271098,
      "learning_rate": 2.830686334092596e-05,
      "loss": 0.6924,
      "step": 4281
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39775869491662225,
      "learning_rate": 2.830159419423828e-05,
      "loss": 0.6777,
      "step": 4282
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3728885975777016,
      "learning_rate": 2.8296324351305923e-05,
      "loss": 0.6594,
      "step": 4283
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.444854956429002,
      "learning_rate": 2.8291053812570862e-05,
      "loss": 0.6743,
      "step": 4284
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.47232047843853797,
      "learning_rate": 2.8285782578475127e-05,
      "loss": 0.6775,
      "step": 4285
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41876996719071563,
      "learning_rate": 2.828051064946082e-05,
      "loss": 0.6748,
      "step": 4286
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36441210795524526,
      "learning_rate": 2.8275238025970085e-05,
      "loss": 0.6485,
      "step": 4287
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3718492888395036,
      "learning_rate": 2.8269964708445127e-05,
      "loss": 0.6228,
      "step": 4288
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3765166662284889,
      "learning_rate": 2.8264690697328215e-05,
      "loss": 0.6701,
      "step": 4289
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4331004014888043,
      "learning_rate": 2.8259415993061678e-05,
      "loss": 0.6953,
      "step": 4290
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42487334089412415,
      "learning_rate": 2.8254140596087897e-05,
      "loss": 0.6938,
      "step": 4291
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40820942714428277,
      "learning_rate": 2.824886450684931e-05,
      "loss": 0.698,
      "step": 4292
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3579513720301129,
      "learning_rate": 2.824358772578842e-05,
      "loss": 0.6535,
      "step": 4293
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3854939044546329,
      "learning_rate": 2.8238310253347793e-05,
      "loss": 0.6526,
      "step": 4294
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40430238365550125,
      "learning_rate": 2.8233032089970023e-05,
      "loss": 0.6667,
      "step": 4295
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4521772890199007,
      "learning_rate": 2.8227753236097792e-05,
      "loss": 0.6636,
      "step": 4296
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3842264697353554,
      "learning_rate": 2.8222473692173843e-05,
      "loss": 0.6808,
      "step": 4297
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4018118375777675,
      "learning_rate": 2.8217193458640947e-05,
      "loss": 0.7117,
      "step": 4298
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36978424694747414,
      "learning_rate": 2.8211912535941964e-05,
      "loss": 0.666,
      "step": 4299
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43175188195103,
      "learning_rate": 2.8206630924519796e-05,
      "loss": 0.7094,
      "step": 4300
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39517751052745526,
      "learning_rate": 2.8201348624817403e-05,
      "loss": 0.6977,
      "step": 4301
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39548348871751365,
      "learning_rate": 2.81960656372778e-05,
      "loss": 0.6451,
      "step": 4302
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3793638914861289,
      "learning_rate": 2.8190781962344078e-05,
      "loss": 0.7031,
      "step": 4303
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42316825754323667,
      "learning_rate": 2.8185497600459368e-05,
      "loss": 0.6768,
      "step": 4304
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46000722162229174,
      "learning_rate": 2.8180212552066847e-05,
      "loss": 0.6846,
      "step": 4305
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.41110121636847724,
      "learning_rate": 2.8174926817609782e-05,
      "loss": 0.7006,
      "step": 4306
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3924168311230486,
      "learning_rate": 2.8169640397531486e-05,
      "loss": 0.6662,
      "step": 4307
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0502006043214707,
      "learning_rate": 2.8164353292275307e-05,
      "loss": 0.4945,
      "step": 4308
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9498188441098206,
      "learning_rate": 2.815906550228468e-05,
      "loss": 0.517,
      "step": 4309
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36148008394391967,
      "learning_rate": 2.8153777028003087e-05,
      "loss": 0.653,
      "step": 4310
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38286324939366095,
      "learning_rate": 2.8148487869874062e-05,
      "loss": 0.6801,
      "step": 4311
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4026558808152384,
      "learning_rate": 2.8143198028341197e-05,
      "loss": 0.6471,
      "step": 4312
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3840168550693437,
      "learning_rate": 2.813790750384814e-05,
      "loss": 0.6902,
      "step": 4313
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33111234094986786,
      "learning_rate": 2.8132616296838623e-05,
      "loss": 0.6275,
      "step": 4314
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3900373455209294,
      "learning_rate": 2.812732440775639e-05,
      "loss": 0.6758,
      "step": 4315
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38956635421708175,
      "learning_rate": 2.8122031837045274e-05,
      "loss": 0.6606,
      "step": 4316
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4040908187117285,
      "learning_rate": 2.8116738585149163e-05,
      "loss": 0.6577,
      "step": 4317
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36629663117537586,
      "learning_rate": 2.8111444652511984e-05,
      "loss": 0.6947,
      "step": 4318
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38912110317618953,
      "learning_rate": 2.8106150039577742e-05,
      "loss": 0.6676,
      "step": 4319
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4179407506282223,
      "learning_rate": 2.8100854746790482e-05,
      "loss": 0.7125,
      "step": 4320
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4192203262153997,
      "learning_rate": 2.8095558774594325e-05,
      "loss": 0.6507,
      "step": 4321
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37437895323575765,
      "learning_rate": 2.8090262123433426e-05,
      "loss": 0.6374,
      "step": 4322
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46053281951474734,
      "learning_rate": 2.808496479375201e-05,
      "loss": 0.6651,
      "step": 4323
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4363745789952116,
      "learning_rate": 2.8079666785994362e-05,
      "loss": 0.694,
      "step": 4324
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4264553415726227,
      "learning_rate": 2.807436810060482e-05,
      "loss": 0.6577,
      "step": 4325
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38254491401494944,
      "learning_rate": 2.8069068738027777e-05,
      "loss": 0.6758,
      "step": 4326
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39565713902231625,
      "learning_rate": 2.8063768698707683e-05,
      "loss": 0.6598,
      "step": 4327
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3475104940533308,
      "learning_rate": 2.8058467983089053e-05,
      "loss": 0.6653,
      "step": 4328
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4319731412077778,
      "learning_rate": 2.8053166591616435e-05,
      "loss": 0.7031,
      "step": 4329
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39584529606113505,
      "learning_rate": 2.8047864524734466e-05,
      "loss": 0.6909,
      "step": 4330
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4404380896767156,
      "learning_rate": 2.8042561782887818e-05,
      "loss": 0.7217,
      "step": 4331
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37205027402364577,
      "learning_rate": 2.8037258366521224e-05,
      "loss": 0.6321,
      "step": 4332
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3658587209293849,
      "learning_rate": 2.8031954276079486e-05,
      "loss": 0.6396,
      "step": 4333
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3938264965468915,
      "learning_rate": 2.8026649512007433e-05,
      "loss": 0.6808,
      "step": 4334
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43412442559405157,
      "learning_rate": 2.8021344074749983e-05,
      "loss": 0.7436,
      "step": 4335
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39752275629596806,
      "learning_rate": 2.801603796475209e-05,
      "loss": 0.7026,
      "step": 4336
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3713011326192339,
      "learning_rate": 2.801073118245878e-05,
      "loss": 0.6362,
      "step": 4337
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3748660183152932,
      "learning_rate": 2.8005423728315118e-05,
      "loss": 0.659,
      "step": 4338
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36640471644036127,
      "learning_rate": 2.800011560276623e-05,
      "loss": 0.6968,
      "step": 4339
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40737516205240837,
      "learning_rate": 2.799480680625732e-05,
      "loss": 0.73,
      "step": 4340
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42278098088268223,
      "learning_rate": 2.7989497339233607e-05,
      "loss": 0.6782,
      "step": 4341
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4108376962078854,
      "learning_rate": 2.7984187202140405e-05,
      "loss": 0.6864,
      "step": 4342
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4173232401691788,
      "learning_rate": 2.7978876395423066e-05,
      "loss": 0.6231,
      "step": 4343
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34695244269286324,
      "learning_rate": 2.7973564919526998e-05,
      "loss": 0.6866,
      "step": 4344
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36190192132417837,
      "learning_rate": 2.7968252774897677e-05,
      "loss": 0.643,
      "step": 4345
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42061824067481524,
      "learning_rate": 2.7962939961980607e-05,
      "loss": 0.6785,
      "step": 4346
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3744260529141902,
      "learning_rate": 2.7957626481221382e-05,
      "loss": 0.6732,
      "step": 4347
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4020426161410259,
      "learning_rate": 2.7952312333065643e-05,
      "loss": 0.6496,
      "step": 4348
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.41576086118739297,
      "learning_rate": 2.7946997517959062e-05,
      "loss": 0.7344,
      "step": 4349
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4091392388330321,
      "learning_rate": 2.7941682036347405e-05,
      "loss": 0.6837,
      "step": 4350
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42872268952996034,
      "learning_rate": 2.793636588867646e-05,
      "loss": 0.7053,
      "step": 4351
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38428145650968687,
      "learning_rate": 2.7931049075392097e-05,
      "loss": 0.6873,
      "step": 4352
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.413943171205333,
      "learning_rate": 2.7925731596940226e-05,
      "loss": 0.6961,
      "step": 4353
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4011416852572842,
      "learning_rate": 2.7920413453766815e-05,
      "loss": 0.6583,
      "step": 4354
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4236317044908771,
      "learning_rate": 2.7915094646317896e-05,
      "loss": 0.6945,
      "step": 4355
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4022665862047627,
      "learning_rate": 2.7909775175039548e-05,
      "loss": 0.6522,
      "step": 4356
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39123742875923934,
      "learning_rate": 2.7904455040377912e-05,
      "loss": 0.6742,
      "step": 4357
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39757131050612204,
      "learning_rate": 2.7899134242779173e-05,
      "loss": 0.6874,
      "step": 4358
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4342948750137312,
      "learning_rate": 2.7893812782689584e-05,
      "loss": 0.6714,
      "step": 4359
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36940735346112336,
      "learning_rate": 2.788849066055545e-05,
      "loss": 0.6367,
      "step": 4360
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.443382057419612,
      "learning_rate": 2.7883167876823134e-05,
      "loss": 0.7031,
      "step": 4361
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4255234037979406,
      "learning_rate": 2.7877844431939047e-05,
      "loss": 0.691,
      "step": 4362
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3978152341806406,
      "learning_rate": 2.7872520326349664e-05,
      "loss": 0.6582,
      "step": 4363
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3754555063252066,
      "learning_rate": 2.7867195560501508e-05,
      "loss": 0.6646,
      "step": 4364
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3578290331349639,
      "learning_rate": 2.7861870134841163e-05,
      "loss": 0.6661,
      "step": 4365
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38040163255835646,
      "learning_rate": 2.7856544049815258e-05,
      "loss": 0.694,
      "step": 4366
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42216765184533067,
      "learning_rate": 2.78512173058705e-05,
      "loss": 0.6627,
      "step": 4367
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43548193106270733,
      "learning_rate": 2.7845889903453623e-05,
      "loss": 0.7233,
      "step": 4368
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39318968481532723,
      "learning_rate": 2.784056184301143e-05,
      "loss": 0.7059,
      "step": 4369
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43273228645034006,
      "learning_rate": 2.7835233124990795e-05,
      "loss": 0.6771,
      "step": 4370
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3811756677839093,
      "learning_rate": 2.7829903749838618e-05,
      "loss": 0.676,
      "step": 4371
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3437156518009252,
      "learning_rate": 2.7824573718001868e-05,
      "loss": 0.6389,
      "step": 4372
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4142998488653346,
      "learning_rate": 2.781924302992757e-05,
      "loss": 0.6859,
      "step": 4373
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5786360602515965,
      "learning_rate": 2.7813911686062804e-05,
      "loss": 0.5574,
      "step": 4374
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37762255665592154,
      "learning_rate": 2.7808579686854696e-05,
      "loss": 0.7064,
      "step": 4375
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3553224334542055,
      "learning_rate": 2.780324703275044e-05,
      "loss": 0.6774,
      "step": 4376
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46241493692458585,
      "learning_rate": 2.779791372419729e-05,
      "loss": 0.6589,
      "step": 4377
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3765786726787564,
      "learning_rate": 2.7792579761642527e-05,
      "loss": 0.6575,
      "step": 4378
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3903878702025883,
      "learning_rate": 2.7787245145533505e-05,
      "loss": 0.6699,
      "step": 4379
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8439627839290211,
      "learning_rate": 2.7781909876317645e-05,
      "loss": 0.4836,
      "step": 4380
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4003193571678501,
      "learning_rate": 2.7776573954442397e-05,
      "loss": 0.6849,
      "step": 4381
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0004951121677936,
      "learning_rate": 2.7771237380355284e-05,
      "loss": 0.5295,
      "step": 4382
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3797460959058538,
      "learning_rate": 2.776590015450388e-05,
      "loss": 0.6537,
      "step": 4383
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40695588380608955,
      "learning_rate": 2.776056227733581e-05,
      "loss": 0.7039,
      "step": 4384
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4413141819468952,
      "learning_rate": 2.775522374929875e-05,
      "loss": 0.6907,
      "step": 4385
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38282748579485076,
      "learning_rate": 2.7749884570840443e-05,
      "loss": 0.6854,
      "step": 4386
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39172701505398944,
      "learning_rate": 2.7744544742408676e-05,
      "loss": 0.6682,
      "step": 4387
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4007011618692008,
      "learning_rate": 2.7739204264451297e-05,
      "loss": 0.6862,
      "step": 4388
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4230178533776199,
      "learning_rate": 2.7733863137416196e-05,
      "loss": 0.6621,
      "step": 4389
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40054870792440966,
      "learning_rate": 2.7728521361751334e-05,
      "loss": 0.6624,
      "step": 4390
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4222842003059501,
      "learning_rate": 2.7723178937904728e-05,
      "loss": 0.6673,
      "step": 4391
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3987709408629274,
      "learning_rate": 2.7717835866324424e-05,
      "loss": 0.6847,
      "step": 4392
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4238060113157178,
      "learning_rate": 2.7712492147458544e-05,
      "loss": 0.6981,
      "step": 4393
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4363121256387012,
      "learning_rate": 2.770714778175527e-05,
      "loss": 0.6525,
      "step": 4394
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38812220754704596,
      "learning_rate": 2.770180276966282e-05,
      "loss": 0.708,
      "step": 4395
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42575610398603475,
      "learning_rate": 2.7696457111629463e-05,
      "loss": 0.6732,
      "step": 4396
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4119249935754443,
      "learning_rate": 2.769111080810355e-05,
      "loss": 0.6431,
      "step": 4397
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.41068829234875376,
      "learning_rate": 2.7685763859533463e-05,
      "loss": 0.6732,
      "step": 4398
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37554741704990796,
      "learning_rate": 2.768041626636764e-05,
      "loss": 0.653,
      "step": 4399
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3584830944565309,
      "learning_rate": 2.7675068029054583e-05,
      "loss": 0.6614,
      "step": 4400
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4028306263703564,
      "learning_rate": 2.766971914804284e-05,
      "loss": 0.6916,
      "step": 4401
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43872272950370217,
      "learning_rate": 2.7664369623781016e-05,
      "loss": 0.7103,
      "step": 4402
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4552941771286571,
      "learning_rate": 2.7659019456717762e-05,
      "loss": 0.6964,
      "step": 4403
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4025214880486038,
      "learning_rate": 2.7653668647301797e-05,
      "loss": 0.6488,
      "step": 4404
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4580540996115395,
      "learning_rate": 2.7648317195981893e-05,
      "loss": 0.688,
      "step": 4405
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3409097636213614,
      "learning_rate": 2.764296510320685e-05,
      "loss": 0.5995,
      "step": 4406
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.41927810682392624,
      "learning_rate": 2.763761236942556e-05,
      "loss": 0.7424,
      "step": 4407
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36651066653273245,
      "learning_rate": 2.7632258995086952e-05,
      "loss": 0.6543,
      "step": 4408
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3970653223620085,
      "learning_rate": 2.7626904980639993e-05,
      "loss": 0.7192,
      "step": 4409
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4202199723750192,
      "learning_rate": 2.7621550326533722e-05,
      "loss": 0.6675,
      "step": 4410
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39170918125959414,
      "learning_rate": 2.7616195033217235e-05,
      "loss": 0.6425,
      "step": 4411
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39866566014966737,
      "learning_rate": 2.7610839101139668e-05,
      "loss": 0.7043,
      "step": 4412
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3634409627199495,
      "learning_rate": 2.760548253075021e-05,
      "loss": 0.6628,
      "step": 4413
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44124058990909854,
      "learning_rate": 2.7600125322498127e-05,
      "loss": 0.7472,
      "step": 4414
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4302383865224162,
      "learning_rate": 2.759476747683271e-05,
      "loss": 0.6927,
      "step": 4415
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33983599366961154,
      "learning_rate": 2.758940899420331e-05,
      "loss": 0.6695,
      "step": 4416
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40382695823419307,
      "learning_rate": 2.7584049875059346e-05,
      "loss": 0.6224,
      "step": 4417
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4402553949880545,
      "learning_rate": 2.7578690119850283e-05,
      "loss": 0.715,
      "step": 4418
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39666136165463867,
      "learning_rate": 2.7573329729025627e-05,
      "loss": 0.709,
      "step": 4419
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41438522896811086,
      "learning_rate": 2.7567968703034954e-05,
      "loss": 0.7364,
      "step": 4420
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42891591615014524,
      "learning_rate": 2.7562607042327888e-05,
      "loss": 0.7005,
      "step": 4421
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36921153809590646,
      "learning_rate": 2.75572447473541e-05,
      "loss": 0.6938,
      "step": 4422
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38116485381686244,
      "learning_rate": 2.7551881818563324e-05,
      "loss": 0.6993,
      "step": 4423
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3817507528801642,
      "learning_rate": 2.754651825640534e-05,
      "loss": 0.6958,
      "step": 4424
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3897587898356622,
      "learning_rate": 2.7541154061329984e-05,
      "loss": 0.6801,
      "step": 4425
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3726371702704352,
      "learning_rate": 2.7535789233787143e-05,
      "loss": 0.6745,
      "step": 4426
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42998486164899496,
      "learning_rate": 2.7530423774226758e-05,
      "loss": 0.699,
      "step": 4427
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.374209463253634,
      "learning_rate": 2.752505768309883e-05,
      "loss": 0.6676,
      "step": 4428
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3244374157029211,
      "learning_rate": 2.7519690960853402e-05,
      "loss": 0.603,
      "step": 4429
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41595686942472626,
      "learning_rate": 2.7514323607940566e-05,
      "loss": 0.6771,
      "step": 4430
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38554132066339475,
      "learning_rate": 2.7508955624810493e-05,
      "loss": 0.6667,
      "step": 4431
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43179601266871154,
      "learning_rate": 2.750358701191338e-05,
      "loss": 0.6728,
      "step": 4432
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42597560755833586,
      "learning_rate": 2.749821776969948e-05,
      "loss": 0.6852,
      "step": 4433
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3251054855456278,
      "learning_rate": 2.7492847898619115e-05,
      "loss": 0.6653,
      "step": 4434
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4395380723442368,
      "learning_rate": 2.7487477399122648e-05,
      "loss": 0.7309,
      "step": 4435
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42425841545638493,
      "learning_rate": 2.7482106271660494e-05,
      "loss": 0.6997,
      "step": 4436
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38618116503077327,
      "learning_rate": 2.7476734516683115e-05,
      "loss": 0.6704,
      "step": 4437
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4107438909372245,
      "learning_rate": 2.7471362134641056e-05,
      "loss": 0.6512,
      "step": 4438
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42415413315275613,
      "learning_rate": 2.746598912598487e-05,
      "loss": 0.6716,
      "step": 4439
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3290330861602572,
      "learning_rate": 2.7460615491165187e-05,
      "loss": 0.6201,
      "step": 4440
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3769895148662234,
      "learning_rate": 2.7455241230632702e-05,
      "loss": 0.6711,
      "step": 4441
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4294905207696554,
      "learning_rate": 2.7449866344838138e-05,
      "loss": 0.7057,
      "step": 4442
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4303611470141265,
      "learning_rate": 2.744449083423227e-05,
      "loss": 0.6747,
      "step": 4443
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3987820969601576,
      "learning_rate": 2.7439114699265954e-05,
      "loss": 0.7001,
      "step": 4444
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38881694596594235,
      "learning_rate": 2.743373794039008e-05,
      "loss": 0.6841,
      "step": 4445
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3970720740151322,
      "learning_rate": 2.7428360558055572e-05,
      "loss": 0.6659,
      "step": 4446
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4523030857686626,
      "learning_rate": 2.742298255271344e-05,
      "loss": 0.7413,
      "step": 4447
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35733572335177816,
      "learning_rate": 2.741760392481472e-05,
      "loss": 0.6553,
      "step": 4448
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3659232587801494,
      "learning_rate": 2.741222467481053e-05,
      "loss": 0.6679,
      "step": 4449
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42973209513297445,
      "learning_rate": 2.7406844803151992e-05,
      "loss": 0.6777,
      "step": 4450
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4042537753416722,
      "learning_rate": 2.7401464310290334e-05,
      "loss": 0.6948,
      "step": 4451
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.386845819766133,
      "learning_rate": 2.7396083196676805e-05,
      "loss": 0.7197,
      "step": 4452
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33418412650361096,
      "learning_rate": 2.7390701462762703e-05,
      "loss": 0.6702,
      "step": 4453
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41375424484160517,
      "learning_rate": 2.7385319108999407e-05,
      "loss": 0.6826,
      "step": 4454
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37375725658495046,
      "learning_rate": 2.7379936135838305e-05,
      "loss": 0.6649,
      "step": 4455
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9046025245214413,
      "learning_rate": 2.737455254373088e-05,
      "loss": 0.518,
      "step": 4456
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38441150433686955,
      "learning_rate": 2.7369168333128634e-05,
      "loss": 0.682,
      "step": 4457
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3802775512668922,
      "learning_rate": 2.7363783504483137e-05,
      "loss": 0.6913,
      "step": 4458
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3901305900927186,
      "learning_rate": 2.7358398058246017e-05,
      "loss": 0.7072,
      "step": 4459
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38911070843429624,
      "learning_rate": 2.7353011994868935e-05,
      "loss": 0.7038,
      "step": 4460
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37858201050279855,
      "learning_rate": 2.734762531480362e-05,
      "loss": 0.6705,
      "step": 4461
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3693520457204147,
      "learning_rate": 2.7342238018501843e-05,
      "loss": 0.6576,
      "step": 4462
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36029258918108825,
      "learning_rate": 2.7336850106415428e-05,
      "loss": 0.6826,
      "step": 4463
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4099074625382922,
      "learning_rate": 2.733146157899626e-05,
      "loss": 0.7018,
      "step": 4464
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39715097675223077,
      "learning_rate": 2.7326072436696258e-05,
      "loss": 0.6572,
      "step": 4465
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3508144081324714,
      "learning_rate": 2.7320682679967413e-05,
      "loss": 0.6669,
      "step": 4466
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3484922094183626,
      "learning_rate": 2.731529230926175e-05,
      "loss": 0.6609,
      "step": 4467
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3807166093375152,
      "learning_rate": 2.7309901325031357e-05,
      "loss": 0.6394,
      "step": 4468
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38675490741602525,
      "learning_rate": 2.730450972772837e-05,
      "loss": 0.6594,
      "step": 4469
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4034193408111514,
      "learning_rate": 2.7299117517804967e-05,
      "loss": 0.6814,
      "step": 4470
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3819884989181357,
      "learning_rate": 2.72937246957134e-05,
      "loss": 0.6939,
      "step": 4471
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3611902148293737,
      "learning_rate": 2.7288331261905944e-05,
      "loss": 0.6415,
      "step": 4472
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42083629209472423,
      "learning_rate": 2.7282937216834956e-05,
      "loss": 0.6937,
      "step": 4473
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4752381727621318,
      "learning_rate": 2.7277542560952813e-05,
      "loss": 0.5376,
      "step": 4474
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3634324172436202,
      "learning_rate": 2.7272147294711966e-05,
      "loss": 0.6365,
      "step": 4475
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4017639076291214,
      "learning_rate": 2.726675141856491e-05,
      "loss": 0.6862,
      "step": 4476
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36118685503881226,
      "learning_rate": 2.7261354932964183e-05,
      "loss": 0.6937,
      "step": 4477
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48099804246401145,
      "learning_rate": 2.7255957838362393e-05,
      "loss": 0.7086,
      "step": 4478
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35706723202732166,
      "learning_rate": 2.725056013521218e-05,
      "loss": 0.6709,
      "step": 4479
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42346506901334247,
      "learning_rate": 2.7245161823966242e-05,
      "loss": 0.7004,
      "step": 4480
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38697680525569755,
      "learning_rate": 2.723976290507734e-05,
      "loss": 0.7031,
      "step": 4481
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3885931304674952,
      "learning_rate": 2.7234363378998255e-05,
      "loss": 0.6586,
      "step": 4482
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4385070598886451,
      "learning_rate": 2.7228963246181864e-05,
      "loss": 0.7079,
      "step": 4483
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.331023467606753,
      "learning_rate": 2.7223562507081042e-05,
      "loss": 0.6216,
      "step": 4484
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4012048004429676,
      "learning_rate": 2.7218161162148767e-05,
      "loss": 0.6818,
      "step": 4485
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36807392947266093,
      "learning_rate": 2.7212759211838027e-05,
      "loss": 0.6553,
      "step": 4486
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3807535607218142,
      "learning_rate": 2.720735665660188e-05,
      "loss": 0.6725,
      "step": 4487
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3902409947166052,
      "learning_rate": 2.7201953496893443e-05,
      "loss": 0.6504,
      "step": 4488
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39987456975562097,
      "learning_rate": 2.719654973316586e-05,
      "loss": 0.6708,
      "step": 4489
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.40222547484748317,
      "learning_rate": 2.7191145365872342e-05,
      "loss": 0.7064,
      "step": 4490
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3917663534481793,
      "learning_rate": 2.7185740395466143e-05,
      "loss": 0.6812,
      "step": 4491
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1338303215751449,
      "learning_rate": 2.7180334822400586e-05,
      "loss": 0.4679,
      "step": 4492
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3952851770888506,
      "learning_rate": 2.7174928647129007e-05,
      "loss": 0.6581,
      "step": 4493
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4108401505691503,
      "learning_rate": 2.7169521870104833e-05,
      "loss": 0.6762,
      "step": 4494
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35544904421695495,
      "learning_rate": 2.7164114491781516e-05,
      "loss": 0.6694,
      "step": 4495
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4057765840391173,
      "learning_rate": 2.7158706512612574e-05,
      "loss": 0.6821,
      "step": 4496
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3581356164336298,
      "learning_rate": 2.7153297933051557e-05,
      "loss": 0.672,
      "step": 4497
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37949641924904437,
      "learning_rate": 2.7147888753552086e-05,
      "loss": 0.6566,
      "step": 4498
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3886674751930682,
      "learning_rate": 2.7142478974567814e-05,
      "loss": 0.6701,
      "step": 4499
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3953530403829404,
      "learning_rate": 2.7137068596552458e-05,
      "loss": 0.6602,
      "step": 4500
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4099181032402369,
      "learning_rate": 2.7131657619959774e-05,
      "loss": 0.6584,
      "step": 4501
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39661279408702027,
      "learning_rate": 2.7126246045243584e-05,
      "loss": 0.6871,
      "step": 4502
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37964217411465523,
      "learning_rate": 2.712083387285774e-05,
      "loss": 0.6879,
      "step": 4503
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3714595871155693,
      "learning_rate": 2.711542110325616e-05,
      "loss": 0.6901,
      "step": 4504
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39626910701392387,
      "learning_rate": 2.7110007736892808e-05,
      "loss": 0.6435,
      "step": 4505
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36544285768779083,
      "learning_rate": 2.7104593774221696e-05,
      "loss": 0.6427,
      "step": 4506
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41126146046864603,
      "learning_rate": 2.7099179215696878e-05,
      "loss": 0.6991,
      "step": 4507
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39163789740137744,
      "learning_rate": 2.7093764061772473e-05,
      "loss": 0.656,
      "step": 4508
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4569690276530737,
      "learning_rate": 2.708834831290265e-05,
      "loss": 0.7274,
      "step": 4509
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4035841861982175,
      "learning_rate": 2.7082931969541605e-05,
      "loss": 0.7297,
      "step": 4510
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36617939165080515,
      "learning_rate": 2.7077515032143617e-05,
      "loss": 0.6729,
      "step": 4511
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4011583060316892,
      "learning_rate": 2.7072097501162987e-05,
      "loss": 0.6937,
      "step": 4512
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43224896454475586,
      "learning_rate": 2.7066679377054083e-05,
      "loss": 0.6903,
      "step": 4513
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4004371119339902,
      "learning_rate": 2.706126066027131e-05,
      "loss": 0.6404,
      "step": 4514
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37460924043817895,
      "learning_rate": 2.705584135126914e-05,
      "loss": 0.6644,
      "step": 4515
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38810602708087616,
      "learning_rate": 2.7050421450502074e-05,
      "loss": 0.7036,
      "step": 4516
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3876663541571785,
      "learning_rate": 2.7045000958424674e-05,
      "loss": 0.6613,
      "step": 4517
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36072903170543946,
      "learning_rate": 2.703957987549155e-05,
      "loss": 0.6192,
      "step": 4518
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4217460434278351,
      "learning_rate": 2.703415820215737e-05,
      "loss": 0.6988,
      "step": 4519
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3821995268248028,
      "learning_rate": 2.702873593887683e-05,
      "loss": 0.664,
      "step": 4520
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4083599132504833,
      "learning_rate": 2.70233130861047e-05,
      "loss": 0.7109,
      "step": 4521
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4056493509244389,
      "learning_rate": 2.701788964429578e-05,
      "loss": 0.6917,
      "step": 4522
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3726536740352233,
      "learning_rate": 2.7012465613904937e-05,
      "loss": 0.6697,
      "step": 4523
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3816968679232872,
      "learning_rate": 2.7007040995387065e-05,
      "loss": 0.7056,
      "step": 4524
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4093777740737351,
      "learning_rate": 2.7001615789197128e-05,
      "loss": 0.6956,
      "step": 4525
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35088545119387504,
      "learning_rate": 2.6996189995790138e-05,
      "loss": 0.65,
      "step": 4526
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35913463596577194,
      "learning_rate": 2.6990763615621135e-05,
      "loss": 0.7083,
      "step": 4527
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44138571592830184,
      "learning_rate": 2.698533664914523e-05,
      "loss": 0.7103,
      "step": 4528
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3487417447693691,
      "learning_rate": 2.697990909681758e-05,
      "loss": 0.6567,
      "step": 4529
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44941022118660356,
      "learning_rate": 2.6974480959093386e-05,
      "loss": 0.7181,
      "step": 4530
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38258185217480073,
      "learning_rate": 2.6969052236427886e-05,
      "loss": 0.6632,
      "step": 4531
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42402897034592957,
      "learning_rate": 2.69636229292764e-05,
      "loss": 0.6853,
      "step": 4532
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38124689708550963,
      "learning_rate": 2.6958193038094273e-05,
      "loss": 0.6801,
      "step": 4533
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4074060131453398,
      "learning_rate": 2.6952762563336893e-05,
      "loss": 0.7024,
      "step": 4534
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40467715440204105,
      "learning_rate": 2.694733150545971e-05,
      "loss": 0.675,
      "step": 4535
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.46710038388424774,
      "learning_rate": 2.6941899864918232e-05,
      "loss": 0.6878,
      "step": 4536
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38815584449650575,
      "learning_rate": 2.6936467642167992e-05,
      "loss": 0.7097,
      "step": 4537
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3945965475725689,
      "learning_rate": 2.6931034837664583e-05,
      "loss": 0.6993,
      "step": 4538
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.380426918013406,
      "learning_rate": 2.6925601451863662e-05,
      "loss": 0.709,
      "step": 4539
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3678954666121966,
      "learning_rate": 2.692016748522091e-05,
      "loss": 0.6435,
      "step": 4540
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44370453480845756,
      "learning_rate": 2.691473293819206e-05,
      "loss": 0.7113,
      "step": 4541
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4535295808367283,
      "learning_rate": 2.690929781123292e-05,
      "loss": 0.725,
      "step": 4542
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37641306952006404,
      "learning_rate": 2.6903862104799312e-05,
      "loss": 0.6799,
      "step": 4543
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3372965592682321,
      "learning_rate": 2.6898425819347128e-05,
      "loss": 0.6437,
      "step": 4544
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.35897906767964455,
      "learning_rate": 2.6892988955332295e-05,
      "loss": 0.6634,
      "step": 4545
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36868797400612663,
      "learning_rate": 2.688755151321081e-05,
      "loss": 0.6466,
      "step": 4546
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36803769180126455,
      "learning_rate": 2.6882113493438698e-05,
      "loss": 0.6404,
      "step": 4547
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3677705352325167,
      "learning_rate": 2.6876674896472033e-05,
      "loss": 0.6923,
      "step": 4548
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3769915291599035,
      "learning_rate": 2.6871235722766956e-05,
      "loss": 0.6495,
      "step": 4549
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38935392971575106,
      "learning_rate": 2.686579597277964e-05,
      "loss": 0.6845,
      "step": 4550
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40166488315548515,
      "learning_rate": 2.6860355646966297e-05,
      "loss": 0.69,
      "step": 4551
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4078588449912638,
      "learning_rate": 2.6854914745783217e-05,
      "loss": 0.6632,
      "step": 4552
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.387074790207559,
      "learning_rate": 2.6849473269686716e-05,
      "loss": 0.6885,
      "step": 4553
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3487488199036185,
      "learning_rate": 2.6844031219133164e-05,
      "loss": 0.6771,
      "step": 4554
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3828245249251135,
      "learning_rate": 2.6838588594578973e-05,
      "loss": 0.6858,
      "step": 4555
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3968510505369535,
      "learning_rate": 2.6833145396480627e-05,
      "loss": 0.6977,
      "step": 4556
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4310988267833439,
      "learning_rate": 2.6827701625294623e-05,
      "loss": 0.6504,
      "step": 4557
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3964079713735949,
      "learning_rate": 2.6822257281477523e-05,
      "loss": 0.6789,
      "step": 4558
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41500024328762336,
      "learning_rate": 2.6816812365485955e-05,
      "loss": 0.694,
      "step": 4559
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3702297755007951,
      "learning_rate": 2.6811366877776563e-05,
      "loss": 0.6962,
      "step": 4560
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.370160039436743,
      "learning_rate": 2.6805920818806054e-05,
      "loss": 0.6761,
      "step": 4561
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34059136698124587,
      "learning_rate": 2.680047418903119e-05,
      "loss": 0.6317,
      "step": 4562
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4142008525316254,
      "learning_rate": 2.6795026988908765e-05,
      "loss": 0.6906,
      "step": 4563
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4432721168220936,
      "learning_rate": 2.6789579218895636e-05,
      "loss": 0.7075,
      "step": 4564
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41464912477957805,
      "learning_rate": 2.678413087944869e-05,
      "loss": 0.6923,
      "step": 4565
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3828595000456924,
      "learning_rate": 2.6778681971024886e-05,
      "loss": 0.729,
      "step": 4566
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3641054718142656,
      "learning_rate": 2.6773232494081214e-05,
      "loss": 0.6649,
      "step": 4567
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39197646543232734,
      "learning_rate": 2.6767782449074706e-05,
      "loss": 0.6501,
      "step": 4568
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43272489807492137,
      "learning_rate": 2.6762331836462458e-05,
      "loss": 0.7165,
      "step": 4569
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41866807875565204,
      "learning_rate": 2.6756880656701615e-05,
      "loss": 0.6712,
      "step": 4570
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3970854737106133,
      "learning_rate": 2.6751428910249343e-05,
      "loss": 0.6407,
      "step": 4571
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38017050394194746,
      "learning_rate": 2.6745976597562885e-05,
      "loss": 0.6618,
      "step": 4572
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41025047491195116,
      "learning_rate": 2.6740523719099516e-05,
      "loss": 0.7181,
      "step": 4573
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.463527132352966,
      "learning_rate": 2.6735070275316566e-05,
      "loss": 0.6687,
      "step": 4574
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40763806179650186,
      "learning_rate": 2.6729616266671406e-05,
      "loss": 0.6914,
      "step": 4575
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4324728945312577,
      "learning_rate": 2.672416169362145e-05,
      "loss": 0.6723,
      "step": 4576
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3992893779281653,
      "learning_rate": 2.671870655662418e-05,
      "loss": 0.6988,
      "step": 4577
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40488564945064154,
      "learning_rate": 2.6713250856137102e-05,
      "loss": 0.6981,
      "step": 4578
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3493860037151451,
      "learning_rate": 2.670779459261778e-05,
      "loss": 0.6904,
      "step": 4579
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41256189365929025,
      "learning_rate": 2.6702337766523832e-05,
      "loss": 0.6923,
      "step": 4580
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3738134076480701,
      "learning_rate": 2.669688037831291e-05,
      "loss": 0.6351,
      "step": 4581
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3732663571441062,
      "learning_rate": 2.669142242844272e-05,
      "loss": 0.7,
      "step": 4582
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41306356715758386,
      "learning_rate": 2.6685963917371004e-05,
      "loss": 0.6381,
      "step": 4583
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3934299247011484,
      "learning_rate": 2.6680504845555577e-05,
      "loss": 0.7052,
      "step": 4584
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.411848589574823,
      "learning_rate": 2.6675045213454265e-05,
      "loss": 0.7144,
      "step": 4585
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41055735665353016,
      "learning_rate": 2.6669585021524983e-05,
      "loss": 0.6947,
      "step": 4586
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37132889280670894,
      "learning_rate": 2.6664124270225662e-05,
      "loss": 0.6756,
      "step": 4587
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39251579292794225,
      "learning_rate": 2.665866296001428e-05,
      "loss": 0.6442,
      "step": 4588
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3750858428795868,
      "learning_rate": 2.665320109134888e-05,
      "loss": 0.6864,
      "step": 4589
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33441362261035834,
      "learning_rate": 2.6647738664687536e-05,
      "loss": 0.6523,
      "step": 4590
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38146642031451367,
      "learning_rate": 2.6642275680488383e-05,
      "loss": 0.6528,
      "step": 4591
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3068881179737858,
      "learning_rate": 2.663681213920959e-05,
      "loss": 0.6396,
      "step": 4592
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3411635295698183,
      "learning_rate": 2.6631348041309375e-05,
      "loss": 0.6604,
      "step": 4593
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36515625870361174,
      "learning_rate": 2.662588338724601e-05,
      "loss": 0.7018,
      "step": 4594
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43786654945627734,
      "learning_rate": 2.6620418177477807e-05,
      "loss": 0.7032,
      "step": 4595
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4162073919870651,
      "learning_rate": 2.6614952412463124e-05,
      "loss": 0.6919,
      "step": 4596
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4783525975006097,
      "learning_rate": 2.660948609266038e-05,
      "loss": 0.6659,
      "step": 4597
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40961704771764196,
      "learning_rate": 2.6604019218528012e-05,
      "loss": 0.707,
      "step": 4598
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.387867732034645,
      "learning_rate": 2.6598551790524532e-05,
      "loss": 0.7173,
      "step": 4599
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40341671068872303,
      "learning_rate": 2.6593083809108478e-05,
      "loss": 0.6951,
      "step": 4600
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3923874569351586,
      "learning_rate": 2.6587615274738454e-05,
      "loss": 0.701,
      "step": 4601
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43221693064908284,
      "learning_rate": 2.6582146187873087e-05,
      "loss": 0.7318,
      "step": 4602
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4035271843611054,
      "learning_rate": 2.6576676548971066e-05,
      "loss": 0.7176,
      "step": 4603
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4135731263235936,
      "learning_rate": 2.657120635849113e-05,
      "loss": 0.6574,
      "step": 4604
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40702347399500466,
      "learning_rate": 2.6565735616892054e-05,
      "loss": 0.6871,
      "step": 4605
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4253727741044398,
      "learning_rate": 2.656026432463266e-05,
      "loss": 0.6914,
      "step": 4606
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.387887878657694,
      "learning_rate": 2.6554792482171818e-05,
      "loss": 0.6547,
      "step": 4607
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39229198631744655,
      "learning_rate": 2.654932008996845e-05,
      "loss": 0.6567,
      "step": 4608
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3743569118935621,
      "learning_rate": 2.654384714848151e-05,
      "loss": 0.6423,
      "step": 4609
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3870264378924741,
      "learning_rate": 2.6538373658170017e-05,
      "loss": 0.679,
      "step": 4610
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3978492835186051,
      "learning_rate": 2.653289961949302e-05,
      "loss": 0.6953,
      "step": 4611
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4578415659651516,
      "learning_rate": 2.6527425032909623e-05,
      "loss": 0.687,
      "step": 4612
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37998414221443355,
      "learning_rate": 2.6521949898878975e-05,
      "loss": 0.6712,
      "step": 4613
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38452866979324263,
      "learning_rate": 2.651647421786026e-05,
      "loss": 0.6789,
      "step": 4614
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4214824472609249,
      "learning_rate": 2.651099799031272e-05,
      "loss": 0.6818,
      "step": 4615
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3536410310064562,
      "learning_rate": 2.6505521216695653e-05,
      "loss": 0.66,
      "step": 4616
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37226735483910955,
      "learning_rate": 2.6500043897468367e-05,
      "loss": 0.687,
      "step": 4617
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41077227169309577,
      "learning_rate": 2.6494566033090256e-05,
      "loss": 0.6914,
      "step": 4618
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4156491922494579,
      "learning_rate": 2.6489087624020727e-05,
      "loss": 0.7019,
      "step": 4619
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4064132852410933,
      "learning_rate": 2.648360867071927e-05,
      "loss": 0.6625,
      "step": 4620
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0676414576199986,
      "learning_rate": 2.6478129173645372e-05,
      "loss": 0.5029,
      "step": 4621
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31535114220010374,
      "learning_rate": 2.6472649133258603e-05,
      "loss": 0.6085,
      "step": 4622
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.371961354420971,
      "learning_rate": 2.6467168550018576e-05,
      "loss": 0.6879,
      "step": 4623
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38383939193780947,
      "learning_rate": 2.6461687424384925e-05,
      "loss": 0.6524,
      "step": 4624
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.45066577477037895,
      "learning_rate": 2.6456205756817365e-05,
      "loss": 0.7402,
      "step": 4625
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.35192601061361856,
      "learning_rate": 2.645072354777561e-05,
      "loss": 0.6436,
      "step": 4626
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3726194064980603,
      "learning_rate": 2.6445240797719477e-05,
      "loss": 0.6711,
      "step": 4627
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3672918181075235,
      "learning_rate": 2.643975750710877e-05,
      "loss": 0.652,
      "step": 4628
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3767590977511688,
      "learning_rate": 2.6434273676403383e-05,
      "loss": 0.6741,
      "step": 4629
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3956042605590708,
      "learning_rate": 2.6428789306063233e-05,
      "loss": 0.6549,
      "step": 4630
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39796367701512214,
      "learning_rate": 2.642330439654829e-05,
      "loss": 0.6757,
      "step": 4631
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4291062271303043,
      "learning_rate": 2.6417818948318557e-05,
      "loss": 0.6844,
      "step": 4632
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3913183961117002,
      "learning_rate": 2.641233296183411e-05,
      "loss": 0.6883,
      "step": 4633
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4139596603859716,
      "learning_rate": 2.640684643755504e-05,
      "loss": 0.6694,
      "step": 4634
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3699407221369228,
      "learning_rate": 2.6401359375941495e-05,
      "loss": 0.685,
      "step": 4635
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4371961386373627,
      "learning_rate": 2.6395871777453666e-05,
      "loss": 0.7018,
      "step": 4636
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.36841477271878825,
      "learning_rate": 2.63903836425518e-05,
      "loss": 0.6547,
      "step": 4637
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38952391358416627,
      "learning_rate": 2.6384894971696176e-05,
      "loss": 0.6677,
      "step": 4638
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4167991180360476,
      "learning_rate": 2.6379405765347113e-05,
      "loss": 0.6756,
      "step": 4639
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4223572499965685,
      "learning_rate": 2.6373916023965004e-05,
      "loss": 0.6935,
      "step": 4640
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1137245387470978,
      "learning_rate": 2.636842574801025e-05,
      "loss": 0.4763,
      "step": 4641
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39168414104981936,
      "learning_rate": 2.636293493794332e-05,
      "loss": 0.7134,
      "step": 4642
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3677521871971816,
      "learning_rate": 2.6357443594224723e-05,
      "loss": 0.6923,
      "step": 4643
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3139699852347278,
      "learning_rate": 2.6351951717315012e-05,
      "loss": 0.6129,
      "step": 4644
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39970011849249154,
      "learning_rate": 2.6346459307674774e-05,
      "loss": 0.6717,
      "step": 4645
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38485909199894003,
      "learning_rate": 2.6340966365764664e-05,
      "loss": 0.6785,
      "step": 4646
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3782108722967048,
      "learning_rate": 2.633547289204536e-05,
      "loss": 0.6935,
      "step": 4647
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3437771752710381,
      "learning_rate": 2.6329978886977595e-05,
      "loss": 0.6352,
      "step": 4648
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3353641243937342,
      "learning_rate": 2.632448435102215e-05,
      "loss": 0.6469,
      "step": 4649
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.349023664833726,
      "learning_rate": 2.631898928463984e-05,
      "loss": 0.7124,
      "step": 4650
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41519115499564646,
      "learning_rate": 2.6313493688291535e-05,
      "loss": 0.7176,
      "step": 4651
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3714550086099095,
      "learning_rate": 2.630799756243813e-05,
      "loss": 0.6815,
      "step": 4652
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3829221817884486,
      "learning_rate": 2.6302500907540584e-05,
      "loss": 0.6547,
      "step": 4653
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39177058923725705,
      "learning_rate": 2.629700372405991e-05,
      "loss": 0.6596,
      "step": 4654
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42001391822496914,
      "learning_rate": 2.629150601245713e-05,
      "loss": 0.7215,
      "step": 4655
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4463120168859357,
      "learning_rate": 2.628600777319334e-05,
      "loss": 0.7234,
      "step": 4656
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35116204974945364,
      "learning_rate": 2.6280509006729675e-05,
      "loss": 0.6899,
      "step": 4657
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36119732170555396,
      "learning_rate": 2.6275009713527303e-05,
      "loss": 0.6872,
      "step": 4658
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36917146100841086,
      "learning_rate": 2.626950989404744e-05,
      "loss": 0.6595,
      "step": 4659
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3902706778733762,
      "learning_rate": 2.6264009548751358e-05,
      "loss": 0.6855,
      "step": 4660
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4440254438437403,
      "learning_rate": 2.6258508678100362e-05,
      "loss": 0.6688,
      "step": 4661
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40694534160212004,
      "learning_rate": 2.6253007282555795e-05,
      "loss": 0.659,
      "step": 4662
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37141988378547536,
      "learning_rate": 2.624750536257906e-05,
      "loss": 0.6886,
      "step": 4663
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.374574833392626,
      "learning_rate": 2.62420029186316e-05,
      "loss": 0.6459,
      "step": 4664
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36668016593750163,
      "learning_rate": 2.6236499951174895e-05,
      "loss": 0.6544,
      "step": 4665
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39793573049762776,
      "learning_rate": 2.623099646067046e-05,
      "loss": 0.6916,
      "step": 4666
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40017287751799163,
      "learning_rate": 2.6225492447579886e-05,
      "loss": 0.6778,
      "step": 4667
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39432295313332333,
      "learning_rate": 2.621998791236478e-05,
      "loss": 0.6839,
      "step": 4668
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4068484281013689,
      "learning_rate": 2.6214482855486795e-05,
      "loss": 0.6931,
      "step": 4669
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.43905151067700743,
      "learning_rate": 2.620897727740764e-05,
      "loss": 0.7502,
      "step": 4670
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35387002005598306,
      "learning_rate": 2.620347117858906e-05,
      "loss": 0.6592,
      "step": 4671
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41706075537046033,
      "learning_rate": 2.6197964559492845e-05,
      "loss": 0.6845,
      "step": 4672
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37255108377945984,
      "learning_rate": 2.6192457420580823e-05,
      "loss": 0.6646,
      "step": 4673
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38334109330477417,
      "learning_rate": 2.618694976231488e-05,
      "loss": 0.666,
      "step": 4674
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.387639274842198,
      "learning_rate": 2.618144158515693e-05,
      "loss": 0.6847,
      "step": 4675
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3429083805665648,
      "learning_rate": 2.6175932889568935e-05,
      "loss": 0.6328,
      "step": 4676
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39813296527717756,
      "learning_rate": 2.6170423676012915e-05,
      "loss": 0.6421,
      "step": 4677
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3878393134961979,
      "learning_rate": 2.6164913944950918e-05,
      "loss": 0.684,
      "step": 4678
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3580077071345333,
      "learning_rate": 2.615940369684502e-05,
      "loss": 0.6747,
      "step": 4679
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37756090822040733,
      "learning_rate": 2.6153892932157382e-05,
      "loss": 0.6939,
      "step": 4680
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39181120757974597,
      "learning_rate": 2.6148381651350176e-05,
      "loss": 0.6435,
      "step": 4681
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4159604362903046,
      "learning_rate": 2.6142869854885626e-05,
      "loss": 0.7036,
      "step": 4682
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3713334622278006,
      "learning_rate": 2.613735754322599e-05,
      "loss": 0.674,
      "step": 4683
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39326597625808357,
      "learning_rate": 2.6131844716833595e-05,
      "loss": 0.7409,
      "step": 4684
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3588149164439973,
      "learning_rate": 2.6126331376170794e-05,
      "loss": 0.6812,
      "step": 4685
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38391945038485914,
      "learning_rate": 2.6120817521699975e-05,
      "loss": 0.6909,
      "step": 4686
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34784884717452086,
      "learning_rate": 2.6115303153883585e-05,
      "loss": 0.6726,
      "step": 4687
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40748066846576875,
      "learning_rate": 2.6109788273184103e-05,
      "loss": 0.69,
      "step": 4688
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3837994014528341,
      "learning_rate": 2.6104272880064058e-05,
      "loss": 0.6362,
      "step": 4689
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39850371279843605,
      "learning_rate": 2.6098756974986013e-05,
      "loss": 0.7295,
      "step": 4690
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4109819722429313,
      "learning_rate": 2.6093240558412595e-05,
      "loss": 0.678,
      "step": 4691
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3791044796912315,
      "learning_rate": 2.608772363080645e-05,
      "loss": 0.6627,
      "step": 4692
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4075400001927001,
      "learning_rate": 2.6082206192630263e-05,
      "loss": 0.695,
      "step": 4693
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39652254059271574,
      "learning_rate": 2.60766882443468e-05,
      "loss": 0.7068,
      "step": 4694
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9916780031932659,
      "learning_rate": 2.6071169786418833e-05,
      "loss": 0.457,
      "step": 4695
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.386382151174307,
      "learning_rate": 2.6065650819309186e-05,
      "loss": 0.662,
      "step": 4696
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3909947172480672,
      "learning_rate": 2.6060131343480723e-05,
      "loss": 0.6833,
      "step": 4697
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3561392382013433,
      "learning_rate": 2.6054611359396377e-05,
      "loss": 0.6767,
      "step": 4698
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4534398081333141,
      "learning_rate": 2.6049090867519073e-05,
      "loss": 0.7358,
      "step": 4699
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4321936922418268,
      "learning_rate": 2.6043569868311832e-05,
      "loss": 0.7175,
      "step": 4700
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3870371386531058,
      "learning_rate": 2.603804836223769e-05,
      "loss": 0.6876,
      "step": 4701
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39185555268398614,
      "learning_rate": 2.603252634975972e-05,
      "loss": 0.6986,
      "step": 4702
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4249222072536243,
      "learning_rate": 2.6027003831341047e-05,
      "loss": 0.6873,
      "step": 4703
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34621781725115397,
      "learning_rate": 2.602148080744484e-05,
      "loss": 0.6563,
      "step": 4704
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40525324940579943,
      "learning_rate": 2.6015957278534315e-05,
      "loss": 0.6767,
      "step": 4705
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3336395813594516,
      "learning_rate": 2.601043324507271e-05,
      "loss": 0.6232,
      "step": 4706
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3828260749058742,
      "learning_rate": 2.6004908707523332e-05,
      "loss": 0.7224,
      "step": 4707
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4012631954513498,
      "learning_rate": 2.5999383666349517e-05,
      "loss": 0.675,
      "step": 4708
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35995403720279884,
      "learning_rate": 2.5993858122014642e-05,
      "loss": 0.6711,
      "step": 4709
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4340996361289164,
      "learning_rate": 2.5988332074982115e-05,
      "loss": 0.6949,
      "step": 4710
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8411049012097257,
      "learning_rate": 2.5982805525715415e-05,
      "loss": 0.4756,
      "step": 4711
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4105402192266654,
      "learning_rate": 2.5977278474678046e-05,
      "loss": 0.6845,
      "step": 4712
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3784378851885366,
      "learning_rate": 2.597175092233354e-05,
      "loss": 0.6676,
      "step": 4713
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3970550042944936,
      "learning_rate": 2.5966222869145497e-05,
      "loss": 0.6253,
      "step": 4714
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7180801885746981,
      "learning_rate": 2.5960694315577555e-05,
      "loss": 0.4384,
      "step": 4715
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4511028398268944,
      "learning_rate": 2.5955165262093372e-05,
      "loss": 0.678,
      "step": 4716
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38372503218025994,
      "learning_rate": 2.5949635709156677e-05,
      "loss": 0.6754,
      "step": 4717
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3710807781728605,
      "learning_rate": 2.5944105657231212e-05,
      "loss": 0.6262,
      "step": 4718
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3965838761238218,
      "learning_rate": 2.5938575106780796e-05,
      "loss": 0.7024,
      "step": 4719
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38789476667134,
      "learning_rate": 2.5933044058269244e-05,
      "loss": 0.6743,
      "step": 4720
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3908520984772533,
      "learning_rate": 2.5927512512160462e-05,
      "loss": 0.7179,
      "step": 4721
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3406353745144113,
      "learning_rate": 2.5921980468918363e-05,
      "loss": 0.6641,
      "step": 4722
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41748351028174063,
      "learning_rate": 2.591644792900691e-05,
      "loss": 0.71,
      "step": 4723
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3921758176326344,
      "learning_rate": 2.5910914892890113e-05,
      "loss": 0.6396,
      "step": 4724
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3485612734737525,
      "learning_rate": 2.590538136103203e-05,
      "loss": 0.6638,
      "step": 4725
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4038560023244687,
      "learning_rate": 2.589984733389674e-05,
      "loss": 0.6976,
      "step": 4726
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40080198776934317,
      "learning_rate": 2.5894312811948378e-05,
      "loss": 0.6623,
      "step": 4727
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4629903444437987,
      "learning_rate": 2.5888777795651118e-05,
      "loss": 0.6811,
      "step": 4728
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3514905632843584,
      "learning_rate": 2.5883242285469187e-05,
      "loss": 0.6456,
      "step": 4729
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3415101684831419,
      "learning_rate": 2.5877706281866816e-05,
      "loss": 0.6601,
      "step": 4730
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.361201228916149,
      "learning_rate": 2.5872169785308323e-05,
      "loss": 0.6733,
      "step": 4731
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3447934105709553,
      "learning_rate": 2.5866632796258047e-05,
      "loss": 0.6289,
      "step": 4732
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42507459191760977,
      "learning_rate": 2.5861095315180355e-05,
      "loss": 0.7005,
      "step": 4733
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3732179253837137,
      "learning_rate": 2.5855557342539683e-05,
      "loss": 0.6591,
      "step": 4734
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8591137808716899,
      "learning_rate": 2.585001887880049e-05,
      "loss": 0.4813,
      "step": 4735
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4224549499323877,
      "learning_rate": 2.5844479924427278e-05,
      "loss": 0.7063,
      "step": 4736
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37781607886390856,
      "learning_rate": 2.5838940479884585e-05,
      "loss": 0.6521,
      "step": 4737
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37364429290713685,
      "learning_rate": 2.583340054563701e-05,
      "loss": 0.6554,
      "step": 4738
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3476453750550675,
      "learning_rate": 2.5827860122149182e-05,
      "loss": 0.6418,
      "step": 4739
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36018013229714585,
      "learning_rate": 2.5822319209885756e-05,
      "loss": 0.661,
      "step": 4740
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42801852273866914,
      "learning_rate": 2.5816777809311457e-05,
      "loss": 0.7422,
      "step": 4741
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.45283110562640344,
      "learning_rate": 2.5811235920891022e-05,
      "loss": 0.7048,
      "step": 4742
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4091947787833002,
      "learning_rate": 2.580569354508925e-05,
      "loss": 0.6857,
      "step": 4743
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37006113200540375,
      "learning_rate": 2.5800150682370976e-05,
      "loss": 0.6671,
      "step": 4744
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4253616789022199,
      "learning_rate": 2.579460733320106e-05,
      "loss": 0.6807,
      "step": 4745
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3674009101260741,
      "learning_rate": 2.578906349804443e-05,
      "loss": 0.6575,
      "step": 4746
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3483443767516101,
      "learning_rate": 2.578351917736604e-05,
      "loss": 0.6298,
      "step": 4747
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40456685866486014,
      "learning_rate": 2.5777974371630876e-05,
      "loss": 0.6424,
      "step": 4748
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4115016772181624,
      "learning_rate": 2.5772429081303982e-05,
      "loss": 0.6564,
      "step": 4749
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4141607435410833,
      "learning_rate": 2.576688330685043e-05,
      "loss": 0.6831,
      "step": 4750
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3790899804744153,
      "learning_rate": 2.5761337048735343e-05,
      "loss": 0.6594,
      "step": 4751
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.46383318802950174,
      "learning_rate": 2.575579030742387e-05,
      "loss": 0.724,
      "step": 4752
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3881700928051834,
      "learning_rate": 2.5750243083381222e-05,
      "loss": 0.6863,
      "step": 4753
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4424157871982877,
      "learning_rate": 2.5744695377072624e-05,
      "loss": 0.7053,
      "step": 4754
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9075457750396694,
      "learning_rate": 2.573914718896337e-05,
      "loss": 0.4641,
      "step": 4755
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3666939600323717,
      "learning_rate": 2.5733598519518773e-05,
      "loss": 0.6179,
      "step": 4756
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3624863675519094,
      "learning_rate": 2.5728049369204185e-05,
      "loss": 0.6537,
      "step": 4757
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4094634254374075,
      "learning_rate": 2.5722499738485018e-05,
      "loss": 0.663,
      "step": 4758
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38396220025002814,
      "learning_rate": 2.5716949627826707e-05,
      "loss": 0.7039,
      "step": 4759
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4034245347804289,
      "learning_rate": 2.5711399037694735e-05,
      "loss": 0.7082,
      "step": 4760
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38103138397731895,
      "learning_rate": 2.570584796855463e-05,
      "loss": 0.6483,
      "step": 4761
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4068135202342256,
      "learning_rate": 2.5700296420871943e-05,
      "loss": 0.7098,
      "step": 4762
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3962627940345992,
      "learning_rate": 2.5694744395112278e-05,
      "loss": 0.6768,
      "step": 4763
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37680350950873504,
      "learning_rate": 2.5689191891741274e-05,
      "loss": 0.629,
      "step": 4764
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41779868193124753,
      "learning_rate": 2.5683638911224624e-05,
      "loss": 0.6826,
      "step": 4765
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40374322459595513,
      "learning_rate": 2.567808545402804e-05,
      "loss": 0.6785,
      "step": 4766
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3903930829214635,
      "learning_rate": 2.5672531520617285e-05,
      "loss": 0.6637,
      "step": 4767
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38929228092776114,
      "learning_rate": 2.5666977111458162e-05,
      "loss": 0.6744,
      "step": 4768
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33989607625756496,
      "learning_rate": 2.5661422227016514e-05,
      "loss": 0.6517,
      "step": 4769
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39906774684600116,
      "learning_rate": 2.5655866867758224e-05,
      "loss": 0.6415,
      "step": 4770
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8296933202252194,
      "learning_rate": 2.56503110341492e-05,
      "loss": 0.4451,
      "step": 4771
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3942374844112088,
      "learning_rate": 2.5644754726655426e-05,
      "loss": 0.6999,
      "step": 4772
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3971121623229072,
      "learning_rate": 2.5639197945742886e-05,
      "loss": 0.6527,
      "step": 4773
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3983379028002534,
      "learning_rate": 2.563364069187762e-05,
      "loss": 0.6912,
      "step": 4774
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37808913764429125,
      "learning_rate": 2.562808296552572e-05,
      "loss": 0.6622,
      "step": 4775
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3901009386321713,
      "learning_rate": 2.562252476715329e-05,
      "loss": 0.6617,
      "step": 4776
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4442206787573799,
      "learning_rate": 2.5616966097226503e-05,
      "loss": 0.6523,
      "step": 4777
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4461409043565025,
      "learning_rate": 2.5611406956211554e-05,
      "loss": 0.7255,
      "step": 4778
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39215061763885567,
      "learning_rate": 2.5605847344574683e-05,
      "loss": 0.6746,
      "step": 4779
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3846154051171106,
      "learning_rate": 2.5600287262782164e-05,
      "loss": 0.6975,
      "step": 4780
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40871511495245944,
      "learning_rate": 2.5594726711300307e-05,
      "loss": 0.6345,
      "step": 4781
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38841778615167727,
      "learning_rate": 2.558916569059549e-05,
      "loss": 0.6871,
      "step": 4782
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3982985296446032,
      "learning_rate": 2.5583604201134087e-05,
      "loss": 0.6778,
      "step": 4783
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3959911578712814,
      "learning_rate": 2.5578042243382547e-05,
      "loss": 0.6817,
      "step": 4784
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39521374154493616,
      "learning_rate": 2.557247981780734e-05,
      "loss": 0.6956,
      "step": 4785
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3396124698921208,
      "learning_rate": 2.5566916924874985e-05,
      "loss": 0.6761,
      "step": 4786
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3658996540728377,
      "learning_rate": 2.5561353565052027e-05,
      "loss": 0.6559,
      "step": 4787
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3758222864208914,
      "learning_rate": 2.555578973880506e-05,
      "loss": 0.6506,
      "step": 4788
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33276858324018743,
      "learning_rate": 2.555022544660072e-05,
      "loss": 0.6721,
      "step": 4789
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.819836305109614,
      "learning_rate": 2.554466068890567e-05,
      "loss": 0.4591,
      "step": 4790
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.395352466016917,
      "learning_rate": 2.5539095466186626e-05,
      "loss": 0.6823,
      "step": 4791
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37576950571270606,
      "learning_rate": 2.553352977891034e-05,
      "loss": 0.6408,
      "step": 4792
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36810580015860717,
      "learning_rate": 2.552796362754359e-05,
      "loss": 0.6527,
      "step": 4793
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36119400606043645,
      "learning_rate": 2.5522397012553204e-05,
      "loss": 0.6233,
      "step": 4794
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4325261950653504,
      "learning_rate": 2.5516829934406055e-05,
      "loss": 0.7182,
      "step": 4795
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3842629215184278,
      "learning_rate": 2.551126239356904e-05,
      "loss": 0.6322,
      "step": 4796
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39127699734912186,
      "learning_rate": 2.550569439050911e-05,
      "loss": 0.6602,
      "step": 4797
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3521968233111817,
      "learning_rate": 2.550012592569323e-05,
      "loss": 0.6465,
      "step": 4798
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41194527162814726,
      "learning_rate": 2.549455699958844e-05,
      "loss": 0.69,
      "step": 4799
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3661471697077331,
      "learning_rate": 2.548898761266178e-05,
      "loss": 0.6481,
      "step": 4800
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4024175437372794,
      "learning_rate": 2.5483417765380364e-05,
      "loss": 0.687,
      "step": 4801
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3963396963330264,
      "learning_rate": 2.5477847458211326e-05,
      "loss": 0.6866,
      "step": 4802
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37726346903390995,
      "learning_rate": 2.5472276691621837e-05,
      "loss": 0.6978,
      "step": 4803
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3435746052043029,
      "learning_rate": 2.54667054660791e-05,
      "loss": 0.6579,
      "step": 4804
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3461422140305528,
      "learning_rate": 2.546113378205039e-05,
      "loss": 0.664,
      "step": 4805
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3973246740528095,
      "learning_rate": 2.5455561640002985e-05,
      "loss": 0.6873,
      "step": 4806
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33488740084463553,
      "learning_rate": 2.5449989040404202e-05,
      "loss": 0.6488,
      "step": 4807
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4337596591699532,
      "learning_rate": 2.5444415983721426e-05,
      "loss": 0.6926,
      "step": 4808
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3690317599979,
      "learning_rate": 2.5438842470422064e-05,
      "loss": 0.6817,
      "step": 4809
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37925130224875414,
      "learning_rate": 2.5433268500973546e-05,
      "loss": 0.6771,
      "step": 4810
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8284586485698432,
      "learning_rate": 2.5427694075843358e-05,
      "loss": 0.431,
      "step": 4811
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44871866458520254,
      "learning_rate": 2.5422119195499026e-05,
      "loss": 0.7288,
      "step": 4812
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4274244053124781,
      "learning_rate": 2.5416543860408106e-05,
      "loss": 0.6832,
      "step": 4813
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39516960812738844,
      "learning_rate": 2.541096807103819e-05,
      "loss": 0.6988,
      "step": 4814
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36554222770807165,
      "learning_rate": 2.540539182785692e-05,
      "loss": 0.6825,
      "step": 4815
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.378658653399493,
      "learning_rate": 2.5399815131331965e-05,
      "loss": 0.665,
      "step": 4816
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2990331202735497,
      "learning_rate": 2.539423798193104e-05,
      "loss": 0.6175,
      "step": 4817
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35920473758567667,
      "learning_rate": 2.5388660380121885e-05,
      "loss": 0.6842,
      "step": 4818
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3486221635744284,
      "learning_rate": 2.5383082326372296e-05,
      "loss": 0.6366,
      "step": 4819
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35045724606173406,
      "learning_rate": 2.537750382115009e-05,
      "loss": 0.6574,
      "step": 4820
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.366782866764665,
      "learning_rate": 2.5371924864923126e-05,
      "loss": 0.6807,
      "step": 4821
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4080167900486257,
      "learning_rate": 2.5366345458159326e-05,
      "loss": 0.6608,
      "step": 4822
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4098711684835645,
      "learning_rate": 2.5360765601326606e-05,
      "loss": 0.7075,
      "step": 4823
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4024140168113827,
      "learning_rate": 2.5355185294892945e-05,
      "loss": 0.7023,
      "step": 4824
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3929129295307102,
      "learning_rate": 2.5349604539326365e-05,
      "loss": 0.6395,
      "step": 4825
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4406184571680365,
      "learning_rate": 2.5344023335094914e-05,
      "loss": 0.6794,
      "step": 4826
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.42078768877743367,
      "learning_rate": 2.5338441682666685e-05,
      "loss": 0.6931,
      "step": 4827
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.45775353248676104,
      "learning_rate": 2.5332859582509786e-05,
      "loss": 0.6955,
      "step": 4828
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40917205839978854,
      "learning_rate": 2.5327277035092403e-05,
      "loss": 0.6484,
      "step": 4829
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.42582176985518405,
      "learning_rate": 2.532169404088273e-05,
      "loss": 0.6523,
      "step": 4830
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4393208674401822,
      "learning_rate": 2.5316110600348998e-05,
      "loss": 0.6707,
      "step": 4831
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4169592994784922,
      "learning_rate": 2.5310526713959496e-05,
      "loss": 0.7067,
      "step": 4832
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40228877214118963,
      "learning_rate": 2.5304942382182533e-05,
      "loss": 0.7053,
      "step": 4833
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37154024471331787,
      "learning_rate": 2.529935760548645e-05,
      "loss": 0.6628,
      "step": 4834
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3730806910760491,
      "learning_rate": 2.5293772384339653e-05,
      "loss": 0.6802,
      "step": 4835
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4086773793451287,
      "learning_rate": 2.5288186719210557e-05,
      "loss": 0.69,
      "step": 4836
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35479346209120316,
      "learning_rate": 2.5282600610567628e-05,
      "loss": 0.6517,
      "step": 4837
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4010805808171723,
      "learning_rate": 2.5277014058879368e-05,
      "loss": 0.6805,
      "step": 4838
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.398624230651156,
      "learning_rate": 2.5271427064614303e-05,
      "loss": 0.6205,
      "step": 4839
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40989890768656523,
      "learning_rate": 2.5265839628241028e-05,
      "loss": 0.6927,
      "step": 4840
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37752404679508667,
      "learning_rate": 2.5260251750228136e-05,
      "loss": 0.6818,
      "step": 4841
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43562911084056005,
      "learning_rate": 2.5254663431044285e-05,
      "loss": 0.6831,
      "step": 4842
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3613289817041357,
      "learning_rate": 2.5249074671158157e-05,
      "loss": 0.6604,
      "step": 4843
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41365105268817437,
      "learning_rate": 2.5243485471038477e-05,
      "loss": 0.6807,
      "step": 4844
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4024642285610027,
      "learning_rate": 2.5237895831154e-05,
      "loss": 0.6863,
      "step": 4845
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3722781060748921,
      "learning_rate": 2.5232305751973525e-05,
      "loss": 0.6814,
      "step": 4846
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4997930914018764,
      "learning_rate": 2.5226715233965894e-05,
      "loss": 0.6922,
      "step": 4847
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.46130419172805226,
      "learning_rate": 2.5221124277599955e-05,
      "loss": 0.7111,
      "step": 4848
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35571167757926725,
      "learning_rate": 2.5215532883344637e-05,
      "loss": 0.6612,
      "step": 4849
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.386513616477889,
      "learning_rate": 2.5209941051668874e-05,
      "loss": 0.694,
      "step": 4850
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3745011465956594,
      "learning_rate": 2.5204348783041644e-05,
      "loss": 0.7021,
      "step": 4851
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3735457539151593,
      "learning_rate": 2.519875607793197e-05,
      "loss": 0.6698,
      "step": 4852
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3933112619750676,
      "learning_rate": 2.5193162936808903e-05,
      "loss": 0.7062,
      "step": 4853
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9341697609176604,
      "learning_rate": 2.5187569360141534e-05,
      "loss": 0.4439,
      "step": 4854
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4087680686125592,
      "learning_rate": 2.5181975348398982e-05,
      "loss": 0.699,
      "step": 4855
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3664734464527367,
      "learning_rate": 2.5176380902050418e-05,
      "loss": 0.6901,
      "step": 4856
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.396496562272822,
      "learning_rate": 2.517078602156504e-05,
      "loss": 0.6598,
      "step": 4857
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38492786540693025,
      "learning_rate": 2.5165190707412085e-05,
      "loss": 0.6938,
      "step": 4858
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3743300073469873,
      "learning_rate": 2.5159594960060813e-05,
      "loss": 0.6392,
      "step": 4859
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.45049670069110975,
      "learning_rate": 2.5153998779980557e-05,
      "loss": 0.7031,
      "step": 4860
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37904705558147705,
      "learning_rate": 2.514840216764064e-05,
      "loss": 0.699,
      "step": 4861
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36852250417778987,
      "learning_rate": 2.5142805123510458e-05,
      "loss": 0.6377,
      "step": 4862
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40354978327702895,
      "learning_rate": 2.5137207648059412e-05,
      "loss": 0.6712,
      "step": 4863
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32410570735154687,
      "learning_rate": 2.5131609741756967e-05,
      "loss": 0.6849,
      "step": 4864
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.42243755660362275,
      "learning_rate": 2.5126011405072612e-05,
      "loss": 0.7078,
      "step": 4865
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3787715278170815,
      "learning_rate": 2.5120412638475872e-05,
      "loss": 0.6894,
      "step": 4866
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4075757206964787,
      "learning_rate": 2.511481344243631e-05,
      "loss": 0.7182,
      "step": 4867
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39327652055673373,
      "learning_rate": 2.5109213817423514e-05,
      "loss": 0.735,
      "step": 4868
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38615036121864516,
      "learning_rate": 2.510361376390713e-05,
      "loss": 0.672,
      "step": 4869
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3774789727597371,
      "learning_rate": 2.5098013282356825e-05,
      "loss": 0.6741,
      "step": 4870
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4269349309958372,
      "learning_rate": 2.50924123732423e-05,
      "loss": 0.7286,
      "step": 4871
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37276266775320915,
      "learning_rate": 2.5086811037033302e-05,
      "loss": 0.6813,
      "step": 4872
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.412717589136981,
      "learning_rate": 2.50812092741996e-05,
      "loss": 0.6771,
      "step": 4873
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42498140137493395,
      "learning_rate": 2.5075607085211023e-05,
      "loss": 0.6914,
      "step": 4874
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4093228914583818,
      "learning_rate": 2.50700044705374e-05,
      "loss": 0.7135,
      "step": 4875
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40627458710491143,
      "learning_rate": 2.5064401430648633e-05,
      "loss": 0.6843,
      "step": 4876
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39498538893278307,
      "learning_rate": 2.5058797966014625e-05,
      "loss": 0.6283,
      "step": 4877
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38983815800017907,
      "learning_rate": 2.5053194077105344e-05,
      "loss": 0.6779,
      "step": 4878
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4131197575764322,
      "learning_rate": 2.5047589764390786e-05,
      "loss": 0.7045,
      "step": 4879
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4140207950228728,
      "learning_rate": 2.504198502834096e-05,
      "loss": 0.6737,
      "step": 4880
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3453063909992483,
      "learning_rate": 2.503637986942595e-05,
      "loss": 0.649,
      "step": 4881
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3547593531508622,
      "learning_rate": 2.503077428811583e-05,
      "loss": 0.6685,
      "step": 4882
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3870702806446266,
      "learning_rate": 2.5025168284880756e-05,
      "loss": 0.6589,
      "step": 4883
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38088608019558773,
      "learning_rate": 2.5019561860190883e-05,
      "loss": 0.6688,
      "step": 4884
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3592841316597773,
      "learning_rate": 2.501395501451642e-05,
      "loss": 0.6584,
      "step": 4885
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3537555151774834,
      "learning_rate": 2.5008347748327603e-05,
      "loss": 0.6731,
      "step": 4886
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3414107897862758,
      "learning_rate": 2.500274006209471e-05,
      "loss": 0.7056,
      "step": 4887
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4032758242969692,
      "learning_rate": 2.499713195628805e-05,
      "loss": 0.6612,
      "step": 4888
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3118068208334738,
      "learning_rate": 2.4991523431377965e-05,
      "loss": 0.6349,
      "step": 4889
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38987488493212585,
      "learning_rate": 2.4985914487834845e-05,
      "loss": 0.6623,
      "step": 4890
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.387215946944403,
      "learning_rate": 2.4980305126129093e-05,
      "loss": 0.6872,
      "step": 4891
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43836386815470535,
      "learning_rate": 2.4974695346731165e-05,
      "loss": 0.676,
      "step": 4892
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4141354517666029,
      "learning_rate": 2.4969085150111552e-05,
      "loss": 0.6786,
      "step": 4893
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35457518795179854,
      "learning_rate": 2.4963474536740758e-05,
      "loss": 0.6596,
      "step": 4894
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38264557598990867,
      "learning_rate": 2.495786350708935e-05,
      "loss": 0.6727,
      "step": 4895
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.416621528243509,
      "learning_rate": 2.495225206162793e-05,
      "loss": 0.6707,
      "step": 4896
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38337086112898205,
      "learning_rate": 2.4946640200827103e-05,
      "loss": 0.6798,
      "step": 4897
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42154255199664414,
      "learning_rate": 2.4941027925157535e-05,
      "loss": 0.6993,
      "step": 4898
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4009591642566,
      "learning_rate": 2.4935415235089918e-05,
      "loss": 0.6526,
      "step": 4899
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37372464026256486,
      "learning_rate": 2.4929802131094996e-05,
      "loss": 0.6789,
      "step": 4900
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38140685608487596,
      "learning_rate": 2.4924188613643513e-05,
      "loss": 0.6509,
      "step": 4901
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4059702605101381,
      "learning_rate": 2.4918574683206283e-05,
      "loss": 0.7342,
      "step": 4902
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37005130882781606,
      "learning_rate": 2.4912960340254135e-05,
      "loss": 0.6859,
      "step": 4903
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4049917343562889,
      "learning_rate": 2.4907345585257938e-05,
      "loss": 0.6785,
      "step": 4904
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4029521341277068,
      "learning_rate": 2.4901730418688594e-05,
      "loss": 0.6591,
      "step": 4905
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41630286790507026,
      "learning_rate": 2.4896114841017033e-05,
      "loss": 0.7004,
      "step": 4906
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3586824875840026,
      "learning_rate": 2.489049885271424e-05,
      "loss": 0.6703,
      "step": 4907
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39443316499950754,
      "learning_rate": 2.4884882454251207e-05,
      "loss": 0.6572,
      "step": 4908
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42863091660365193,
      "learning_rate": 2.4879265646098983e-05,
      "loss": 0.712,
      "step": 4909
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34510556829389794,
      "learning_rate": 2.487364842872865e-05,
      "loss": 0.6558,
      "step": 4910
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4344588043431015,
      "learning_rate": 2.4868030802611306e-05,
      "loss": 0.7118,
      "step": 4911
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40931852730231477,
      "learning_rate": 2.486241276821809e-05,
      "loss": 0.6747,
      "step": 4912
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.46947920390722636,
      "learning_rate": 2.4856794326020196e-05,
      "loss": 0.7228,
      "step": 4913
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41209372820524537,
      "learning_rate": 2.485117547648883e-05,
      "loss": 0.679,
      "step": 4914
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38298939939182103,
      "learning_rate": 2.4845556220095227e-05,
      "loss": 0.6826,
      "step": 4915
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4086248592452329,
      "learning_rate": 2.483993655731068e-05,
      "loss": 0.6815,
      "step": 4916
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39064386055164396,
      "learning_rate": 2.4834316488606498e-05,
      "loss": 0.6722,
      "step": 4917
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39513029381286396,
      "learning_rate": 2.482869601445403e-05,
      "loss": 0.6623,
      "step": 4918
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4020124218220785,
      "learning_rate": 2.4823075135324658e-05,
      "loss": 0.6823,
      "step": 4919
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.36631088657032157,
      "learning_rate": 2.48174538516898e-05,
      "loss": 0.6675,
      "step": 4920
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3435653387428798,
      "learning_rate": 2.4811832164020913e-05,
      "loss": 0.6449,
      "step": 4921
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3830188671569398,
      "learning_rate": 2.4806210072789458e-05,
      "loss": 0.7002,
      "step": 4922
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7186369485595839,
      "learning_rate": 2.480058757846698e-05,
      "loss": 0.4327,
      "step": 4923
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4248218215266174,
      "learning_rate": 2.479496468152502e-05,
      "loss": 0.7063,
      "step": 4924
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3963750645067026,
      "learning_rate": 2.4789341382435158e-05,
      "loss": 0.6748,
      "step": 4925
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39460511043622165,
      "learning_rate": 2.478371768166902e-05,
      "loss": 0.6698,
      "step": 4926
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4066992443057022,
      "learning_rate": 2.4778093579698256e-05,
      "loss": 0.694,
      "step": 4927
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3963652283306998,
      "learning_rate": 2.4772469076994562e-05,
      "loss": 0.6442,
      "step": 4928
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3767184947010954,
      "learning_rate": 2.4766844174029636e-05,
      "loss": 0.6609,
      "step": 4929
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39206594685896784,
      "learning_rate": 2.4761218871275256e-05,
      "loss": 0.6331,
      "step": 4930
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4048144121320214,
      "learning_rate": 2.47555931692032e-05,
      "loss": 0.689,
      "step": 4931
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39273150197135853,
      "learning_rate": 2.4749967068285284e-05,
      "loss": 0.6995,
      "step": 4932
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43583586144104697,
      "learning_rate": 2.4744340568993365e-05,
      "loss": 0.7321,
      "step": 4933
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3930357189622204,
      "learning_rate": 2.4738713671799337e-05,
      "loss": 0.6794,
      "step": 4934
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4012942901048952,
      "learning_rate": 2.4733086377175112e-05,
      "loss": 0.6811,
      "step": 4935
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4219104633898046,
      "learning_rate": 2.472745868559265e-05,
      "loss": 0.718,
      "step": 4936
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3692408816096105,
      "learning_rate": 2.4721830597523943e-05,
      "loss": 0.6606,
      "step": 4937
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4196319584498854,
      "learning_rate": 2.4716202113441003e-05,
      "loss": 0.7016,
      "step": 4938
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34078680384344145,
      "learning_rate": 2.471057323381588e-05,
      "loss": 0.6545,
      "step": 4939
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35666946616633427,
      "learning_rate": 2.4704943959120677e-05,
      "loss": 0.6921,
      "step": 4940
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4022346711817869,
      "learning_rate": 2.4699314289827506e-05,
      "loss": 0.6814,
      "step": 4941
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3825870232634355,
      "learning_rate": 2.4693684226408515e-05,
      "loss": 0.7129,
      "step": 4942
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3802972905978408,
      "learning_rate": 2.46880537693359e-05,
      "loss": 0.6822,
      "step": 4943
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4115944435897479,
      "learning_rate": 2.468242291908188e-05,
      "loss": 0.6842,
      "step": 4944
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3719618362821306,
      "learning_rate": 2.4676791676118706e-05,
      "loss": 0.6588,
      "step": 4945
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38442634526578257,
      "learning_rate": 2.467116004091865e-05,
      "loss": 0.6596,
      "step": 4946
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3766614441673416,
      "learning_rate": 2.4665528013954054e-05,
      "loss": 0.6844,
      "step": 4947
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40571946303261436,
      "learning_rate": 2.465989559569726e-05,
      "loss": 0.6806,
      "step": 4948
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39809630750713526,
      "learning_rate": 2.465426278662064e-05,
      "loss": 0.6825,
      "step": 4949
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4031156415229315,
      "learning_rate": 2.464862958719663e-05,
      "loss": 0.6852,
      "step": 4950
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3630947522899515,
      "learning_rate": 2.464299599789767e-05,
      "loss": 0.6843,
      "step": 4951
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37618090852749486,
      "learning_rate": 2.4637362019196235e-05,
      "loss": 0.6627,
      "step": 4952
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35021441108494306,
      "learning_rate": 2.4631727651564852e-05,
      "loss": 0.6546,
      "step": 4953
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4009951019219294,
      "learning_rate": 2.462609289547607e-05,
      "loss": 0.703,
      "step": 4954
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3946941393348354,
      "learning_rate": 2.462045775140246e-05,
      "loss": 0.6807,
      "step": 4955
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3689406140074689,
      "learning_rate": 2.461482221981663e-05,
      "loss": 0.6844,
      "step": 4956
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35099671720553416,
      "learning_rate": 2.4609186301191245e-05,
      "loss": 0.6795,
      "step": 4957
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3532338479321422,
      "learning_rate": 2.460354999599897e-05,
      "loss": 0.667,
      "step": 4958
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.391181032534281,
      "learning_rate": 2.4597913304712516e-05,
      "loss": 0.6867,
      "step": 4959
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41268357803010663,
      "learning_rate": 2.4592276227804618e-05,
      "loss": 0.6943,
      "step": 4960
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40385958573007563,
      "learning_rate": 2.458663876574807e-05,
      "loss": 0.6777,
      "step": 4961
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4146126834057797,
      "learning_rate": 2.4581000919015664e-05,
      "loss": 0.6733,
      "step": 4962
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40593198904556393,
      "learning_rate": 2.4575362688080236e-05,
      "loss": 0.6975,
      "step": 4963
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3761600194769962,
      "learning_rate": 2.456972407341468e-05,
      "loss": 0.6717,
      "step": 4964
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3643151193724866,
      "learning_rate": 2.4564085075491876e-05,
      "loss": 0.6486,
      "step": 4965
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39030977934349065,
      "learning_rate": 2.455844569478477e-05,
      "loss": 0.6664,
      "step": 4966
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41462449652045386,
      "learning_rate": 2.4552805931766322e-05,
      "loss": 0.6658,
      "step": 4967
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40629705841316927,
      "learning_rate": 2.4547165786909548e-05,
      "loss": 0.689,
      "step": 4968
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.456105453329369,
      "learning_rate": 2.4541525260687468e-05,
      "loss": 0.7104,
      "step": 4969
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4551460333280139,
      "learning_rate": 2.4535884353573148e-05,
      "loss": 0.7168,
      "step": 4970
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3379218798955708,
      "learning_rate": 2.4530243066039688e-05,
      "loss": 0.6475,
      "step": 4971
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3583114060664663,
      "learning_rate": 2.452460139856021e-05,
      "loss": 0.6443,
      "step": 4972
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.45682516545324253,
      "learning_rate": 2.451895935160788e-05,
      "loss": 0.7013,
      "step": 4973
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3799815870496727,
      "learning_rate": 2.4513316925655882e-05,
      "loss": 0.6711,
      "step": 4974
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40256462440566715,
      "learning_rate": 2.450767412117745e-05,
      "loss": 0.7036,
      "step": 4975
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.382072273317626,
      "learning_rate": 2.450203093864583e-05,
      "loss": 0.6893,
      "step": 4976
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39038960251915067,
      "learning_rate": 2.449638737853431e-05,
      "loss": 0.6923,
      "step": 4977
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3917983069376394,
      "learning_rate": 2.4490743441316217e-05,
      "loss": 0.6773,
      "step": 4978
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33017173664360044,
      "learning_rate": 2.448509912746489e-05,
      "loss": 0.652,
      "step": 4979
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7833328745768454,
      "learning_rate": 2.4479454437453716e-05,
      "loss": 0.4349,
      "step": 4980
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4267646289062674,
      "learning_rate": 2.447380937175611e-05,
      "loss": 0.6805,
      "step": 4981
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40147931980165535,
      "learning_rate": 2.446816393084552e-05,
      "loss": 0.6439,
      "step": 4982
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3987112834848449,
      "learning_rate": 2.4462518115195408e-05,
      "loss": 0.6938,
      "step": 4983
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40403693556952935,
      "learning_rate": 2.4456871925279293e-05,
      "loss": 0.7147,
      "step": 4984
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.41193045562875097,
      "learning_rate": 2.4451225361570717e-05,
      "loss": 0.6846,
      "step": 4985
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37595157391438205,
      "learning_rate": 2.4445578424543245e-05,
      "loss": 0.6832,
      "step": 4986
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38784811233111516,
      "learning_rate": 2.443993111467047e-05,
      "loss": 0.6785,
      "step": 4987
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38209534645568666,
      "learning_rate": 2.4434283432426048e-05,
      "loss": 0.6914,
      "step": 4988
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7800510927366165,
      "learning_rate": 2.4428635378283624e-05,
      "loss": 0.4753,
      "step": 4989
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4385299221680319,
      "learning_rate": 2.4422986952716897e-05,
      "loss": 0.7086,
      "step": 4990
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39940084274965393,
      "learning_rate": 2.4417338156199596e-05,
      "loss": 0.7082,
      "step": 4991
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3909951703512148,
      "learning_rate": 2.4411688989205482e-05,
      "loss": 0.6711,
      "step": 4992
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4151341935472241,
      "learning_rate": 2.4406039452208336e-05,
      "loss": 0.6828,
      "step": 4993
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3373657802993326,
      "learning_rate": 2.4400389545681983e-05,
      "loss": 0.6482,
      "step": 4994
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42369378872356805,
      "learning_rate": 2.439473927010028e-05,
      "loss": 0.6981,
      "step": 4995
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3978447187387019,
      "learning_rate": 2.4389088625937098e-05,
      "loss": 0.6699,
      "step": 4996
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3968820793938454,
      "learning_rate": 2.4383437613666353e-05,
      "loss": 0.6618,
      "step": 4997
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3952316122506642,
      "learning_rate": 2.4377786233761985e-05,
      "loss": 0.6357,
      "step": 4998
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4174211291640352,
      "learning_rate": 2.4372134486697982e-05,
      "loss": 0.7256,
      "step": 4999
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38473480296131785,
      "learning_rate": 2.436648237294833e-05,
      "loss": 0.6815,
      "step": 5000
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46971707815514135,
      "learning_rate": 2.4360829892987078e-05,
      "loss": 0.7015,
      "step": 5001
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3586008636485142,
      "learning_rate": 2.435517704728829e-05,
      "loss": 0.6697,
      "step": 5002
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32977751717237597,
      "learning_rate": 2.434952383632607e-05,
      "loss": 0.6489,
      "step": 5003
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.398008520652578,
      "learning_rate": 2.4343870260574528e-05,
      "loss": 0.6606,
      "step": 5004
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3959905663817182,
      "learning_rate": 2.433821632050784e-05,
      "loss": 0.6579,
      "step": 5005
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40598424174579595,
      "learning_rate": 2.4332562016600186e-05,
      "loss": 0.7045,
      "step": 5006
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3548744836038275,
      "learning_rate": 2.4326907349325784e-05,
      "loss": 0.6743,
      "step": 5007
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3818703087517876,
      "learning_rate": 2.4321252319158893e-05,
      "loss": 0.6521,
      "step": 5008
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42510407178404064,
      "learning_rate": 2.4315596926573787e-05,
      "loss": 0.6723,
      "step": 5009
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3725832810000726,
      "learning_rate": 2.430994117204478e-05,
      "loss": 0.6828,
      "step": 5010
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3732947953423606,
      "learning_rate": 2.4304285056046215e-05,
      "loss": 0.657,
      "step": 5011
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3812059677842551,
      "learning_rate": 2.4298628579052456e-05,
      "loss": 0.6678,
      "step": 5012
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38602240562397494,
      "learning_rate": 2.4292971741537907e-05,
      "loss": 0.6407,
      "step": 5013
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.41073978992133714,
      "learning_rate": 2.428731454397701e-05,
      "loss": 0.701,
      "step": 5014
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3890799161342682,
      "learning_rate": 2.4281656986844217e-05,
      "loss": 0.6587,
      "step": 5015
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3726024799945501,
      "learning_rate": 2.427599907061402e-05,
      "loss": 0.6562,
      "step": 5016
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4030406126593672,
      "learning_rate": 2.427034079576095e-05,
      "loss": 0.693,
      "step": 5017
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7824549913467863,
      "learning_rate": 2.4264682162759555e-05,
      "loss": 0.4434,
      "step": 5018
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40803181857350657,
      "learning_rate": 2.4259023172084412e-05,
      "loss": 0.7025,
      "step": 5019
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6716911446217015,
      "learning_rate": 2.425336382421014e-05,
      "loss": 0.4455,
      "step": 5020
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40158252319247745,
      "learning_rate": 2.4247704119611385e-05,
      "loss": 0.6734,
      "step": 5021
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39180892806677053,
      "learning_rate": 2.4242044058762812e-05,
      "loss": 0.6532,
      "step": 5022
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37163100710140207,
      "learning_rate": 2.423638364213913e-05,
      "loss": 0.6495,
      "step": 5023
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3469679148378537,
      "learning_rate": 2.4230722870215062e-05,
      "loss": 0.6582,
      "step": 5024
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3842420979370329,
      "learning_rate": 2.422506174346538e-05,
      "loss": 0.6719,
      "step": 5025
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.395201435637471,
      "learning_rate": 2.4219400262364875e-05,
      "loss": 0.6949,
      "step": 5026
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36255809719450666,
      "learning_rate": 2.4213738427388362e-05,
      "loss": 0.6793,
      "step": 5027
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38122571952335155,
      "learning_rate": 2.42080762390107e-05,
      "loss": 0.6498,
      "step": 5028
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.398695952576594,
      "learning_rate": 2.4202413697706764e-05,
      "loss": 0.6724,
      "step": 5029
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37154045807614505,
      "learning_rate": 2.419675080395146e-05,
      "loss": 0.675,
      "step": 5030
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3958922087609287,
      "learning_rate": 2.4191087558219744e-05,
      "loss": 0.6671,
      "step": 5031
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3797850315168316,
      "learning_rate": 2.418542396098657e-05,
      "loss": 0.6507,
      "step": 5032
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37817687646861503,
      "learning_rate": 2.417976001272695e-05,
      "loss": 0.6411,
      "step": 5033
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42059357586453183,
      "learning_rate": 2.4174095713915898e-05,
      "loss": 0.7115,
      "step": 5034
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.48281930537038253,
      "learning_rate": 2.4168431065028487e-05,
      "loss": 0.7236,
      "step": 5035
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43384641451185,
      "learning_rate": 2.416276606653979e-05,
      "loss": 0.6844,
      "step": 5036
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.34177645329146167,
      "learning_rate": 2.415710071892493e-05,
      "loss": 0.6406,
      "step": 5037
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4011258654214577,
      "learning_rate": 2.415143502265906e-05,
      "loss": 0.6967,
      "step": 5038
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.396809469752543,
      "learning_rate": 2.414576897821734e-05,
      "loss": 0.6546,
      "step": 5039
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3711828566751928,
      "learning_rate": 2.4140102586074986e-05,
      "loss": 0.6514,
      "step": 5040
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38931060311038473,
      "learning_rate": 2.4134435846707232e-05,
      "loss": 0.6951,
      "step": 5041
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3562548972457339,
      "learning_rate": 2.4128768760589333e-05,
      "loss": 0.6892,
      "step": 5042
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4302452259514144,
      "learning_rate": 2.412310132819659e-05,
      "loss": 0.6568,
      "step": 5043
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36220908227640475,
      "learning_rate": 2.4117433550004304e-05,
      "loss": 0.6563,
      "step": 5044
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35669299055047854,
      "learning_rate": 2.4111765426487854e-05,
      "loss": 0.6509,
      "step": 5045
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3473105042018202,
      "learning_rate": 2.4106096958122594e-05,
      "loss": 0.6598,
      "step": 5046
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3812179410485492,
      "learning_rate": 2.410042814538394e-05,
      "loss": 0.6732,
      "step": 5047
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36612125115891553,
      "learning_rate": 2.409475898874734e-05,
      "loss": 0.7256,
      "step": 5048
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37442654396518354,
      "learning_rate": 2.4089089488688245e-05,
      "loss": 0.6292,
      "step": 5049
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3480572347265827,
      "learning_rate": 2.408341964568215e-05,
      "loss": 0.6501,
      "step": 5050
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35569612880209306,
      "learning_rate": 2.4077749460204576e-05,
      "loss": 0.648,
      "step": 5051
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42067915785969645,
      "learning_rate": 2.4072078932731088e-05,
      "loss": 0.7205,
      "step": 5052
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3589500141575633,
      "learning_rate": 2.4066408063737253e-05,
      "loss": 0.6896,
      "step": 5053
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37458940111572936,
      "learning_rate": 2.4060736853698683e-05,
      "loss": 0.6838,
      "step": 5054
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3458126980119592,
      "learning_rate": 2.4055065303091025e-05,
      "loss": 0.6393,
      "step": 5055
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39804078198856113,
      "learning_rate": 2.4049393412389935e-05,
      "loss": 0.6637,
      "step": 5056
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3702978690885662,
      "learning_rate": 2.40437211820711e-05,
      "loss": 0.6614,
      "step": 5057
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4104400100569165,
      "learning_rate": 2.4038048612610264e-05,
      "loss": 0.696,
      "step": 5058
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3680848447571055,
      "learning_rate": 2.4032375704483166e-05,
      "loss": 0.6261,
      "step": 5059
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38366222071316697,
      "learning_rate": 2.4026702458165587e-05,
      "loss": 0.7218,
      "step": 5060
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4169421317489382,
      "learning_rate": 2.402102887413333e-05,
      "loss": 0.7272,
      "step": 5061
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3860314129035029,
      "learning_rate": 2.4015354952862248e-05,
      "loss": 0.6746,
      "step": 5062
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35564378029259947,
      "learning_rate": 2.4009680694828184e-05,
      "loss": 0.6854,
      "step": 5063
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37292256946076113,
      "learning_rate": 2.4004006100507048e-05,
      "loss": 0.6594,
      "step": 5064
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.559281376290567,
      "learning_rate": 2.399833117037476e-05,
      "loss": 0.6849,
      "step": 5065
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3641776561879856,
      "learning_rate": 2.399265590490726e-05,
      "loss": 0.6658,
      "step": 5066
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3457818010871866,
      "learning_rate": 2.3986980304580533e-05,
      "loss": 0.6651,
      "step": 5067
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4018885471949905,
      "learning_rate": 2.3981304369870578e-05,
      "loss": 0.6628,
      "step": 5068
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39637436728567504,
      "learning_rate": 2.3975628101253446e-05,
      "loss": 0.7137,
      "step": 5069
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40397037706756667,
      "learning_rate": 2.3969951499205175e-05,
      "loss": 0.6562,
      "step": 5070
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3951675185126869,
      "learning_rate": 2.396427456420187e-05,
      "loss": 0.6671,
      "step": 5071
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38421456232859885,
      "learning_rate": 2.3958597296719646e-05,
      "loss": 0.6757,
      "step": 5072
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4373677274474804,
      "learning_rate": 2.3952919697234646e-05,
      "loss": 0.7298,
      "step": 5073
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4224953044039383,
      "learning_rate": 2.394724176622304e-05,
      "loss": 0.6665,
      "step": 5074
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39462530500986925,
      "learning_rate": 2.3941563504161044e-05,
      "loss": 0.6501,
      "step": 5075
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35203911521360076,
      "learning_rate": 2.3935884911524876e-05,
      "loss": 0.6597,
      "step": 5076
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4418289556128719,
      "learning_rate": 2.393020598879079e-05,
      "loss": 0.7058,
      "step": 5077
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4094807401074866,
      "learning_rate": 2.3924526736435068e-05,
      "loss": 0.6509,
      "step": 5078
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4082377564824117,
      "learning_rate": 2.391884715493404e-05,
      "loss": 0.666,
      "step": 5079
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4230261281537159,
      "learning_rate": 2.3913167244764034e-05,
      "loss": 0.7122,
      "step": 5080
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3864258059100064,
      "learning_rate": 2.3907487006401403e-05,
      "loss": 0.6492,
      "step": 5081
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4019526945469802,
      "learning_rate": 2.390180644032257e-05,
      "loss": 0.6804,
      "step": 5082
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.372374894265984,
      "learning_rate": 2.3896125547003938e-05,
      "loss": 0.6421,
      "step": 5083
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3927345566278705,
      "learning_rate": 2.389044432692196e-05,
      "loss": 0.6817,
      "step": 5084
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4296166718488976,
      "learning_rate": 2.3884762780553114e-05,
      "loss": 0.6673,
      "step": 5085
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39726753808087606,
      "learning_rate": 2.387908090837391e-05,
      "loss": 0.6689,
      "step": 5086
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37863253973045435,
      "learning_rate": 2.3873398710860873e-05,
      "loss": 0.645,
      "step": 5087
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.390058378832689,
      "learning_rate": 2.3867716188490563e-05,
      "loss": 0.6967,
      "step": 5088
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4370832643588679,
      "learning_rate": 2.386203334173957e-05,
      "loss": 0.706,
      "step": 5089
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.48145700850102935,
      "learning_rate": 2.38563501710845e-05,
      "loss": 0.6738,
      "step": 5090
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39206448413582556,
      "learning_rate": 2.3850666677002e-05,
      "loss": 0.6524,
      "step": 5091
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36966612049348246,
      "learning_rate": 2.3844982859968736e-05,
      "loss": 0.643,
      "step": 5092
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36045168518424,
      "learning_rate": 2.3839298720461405e-05,
      "loss": 0.6283,
      "step": 5093
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3997571267009239,
      "learning_rate": 2.3833614258956725e-05,
      "loss": 0.671,
      "step": 5094
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.348840097554397,
      "learning_rate": 2.3827929475931447e-05,
      "loss": 0.652,
      "step": 5095
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41814455898362995,
      "learning_rate": 2.3822244371862352e-05,
      "loss": 0.6829,
      "step": 5096
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3767228823516036,
      "learning_rate": 2.3816558947226232e-05,
      "loss": 0.6878,
      "step": 5097
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3513598648649391,
      "learning_rate": 2.381087320249992e-05,
      "loss": 0.6588,
      "step": 5098
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38979194913842735,
      "learning_rate": 2.380518713816028e-05,
      "loss": 0.6704,
      "step": 5099
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38237289273421654,
      "learning_rate": 2.3799500754684196e-05,
      "loss": 0.6536,
      "step": 5100
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4127575457339897,
      "learning_rate": 2.379381405254856e-05,
      "loss": 0.7111,
      "step": 5101
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3734930945610388,
      "learning_rate": 2.378812703223033e-05,
      "loss": 0.7075,
      "step": 5102
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34040322099323345,
      "learning_rate": 2.3782439694206456e-05,
      "loss": 0.6459,
      "step": 5103
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3690896368658065,
      "learning_rate": 2.3776752038953932e-05,
      "loss": 0.6426,
      "step": 5104
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39568644754453886,
      "learning_rate": 2.3771064066949777e-05,
      "loss": 0.7159,
      "step": 5105
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38037969780597414,
      "learning_rate": 2.3765375778671035e-05,
      "loss": 0.6932,
      "step": 5106
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37558064759080506,
      "learning_rate": 2.3759687174594778e-05,
      "loss": 0.6652,
      "step": 5107
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3429306345223944,
      "learning_rate": 2.3753998255198085e-05,
      "loss": 0.6454,
      "step": 5108
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.35361033169401973,
      "learning_rate": 2.3748309020958105e-05,
      "loss": 0.6129,
      "step": 5109
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32959669634504674,
      "learning_rate": 2.374261947235197e-05,
      "loss": 0.6014,
      "step": 5110
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40871022056545103,
      "learning_rate": 2.373692960985686e-05,
      "loss": 0.6602,
      "step": 5111
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33162939818411225,
      "learning_rate": 2.3731239433949972e-05,
      "loss": 0.6891,
      "step": 5112
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3659359633026822,
      "learning_rate": 2.3725548945108544e-05,
      "loss": 0.6853,
      "step": 5113
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3626321969623017,
      "learning_rate": 2.3719858143809824e-05,
      "loss": 0.6808,
      "step": 5114
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3739365073005409,
      "learning_rate": 2.3714167030531093e-05,
      "loss": 0.6904,
      "step": 5115
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3878956631222952,
      "learning_rate": 2.3708475605749663e-05,
      "loss": 0.6899,
      "step": 5116
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38955923870734355,
      "learning_rate": 2.370278386994286e-05,
      "loss": 0.7233,
      "step": 5117
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36617788956925446,
      "learning_rate": 2.369709182358804e-05,
      "loss": 0.6623,
      "step": 5118
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33690006380365817,
      "learning_rate": 2.36913994671626e-05,
      "loss": 0.6541,
      "step": 5119
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3802321469204746,
      "learning_rate": 2.3685706801143944e-05,
      "loss": 0.6876,
      "step": 5120
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3650471731462781,
      "learning_rate": 2.3680013826009507e-05,
      "loss": 0.6648,
      "step": 5121
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.361014023167749,
      "learning_rate": 2.3674320542236757e-05,
      "loss": 0.6686,
      "step": 5122
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40599673157656185,
      "learning_rate": 2.366862695030318e-05,
      "loss": 0.6505,
      "step": 5123
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39219993926355873,
      "learning_rate": 2.3662933050686293e-05,
      "loss": 0.6653,
      "step": 5124
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33462417199421884,
      "learning_rate": 2.3657238843863636e-05,
      "loss": 0.6823,
      "step": 5125
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3910403000998927,
      "learning_rate": 2.365154433031277e-05,
      "loss": 0.6534,
      "step": 5126
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3724017475243748,
      "learning_rate": 2.3645849510511295e-05,
      "loss": 0.6485,
      "step": 5127
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3823652805471142,
      "learning_rate": 2.364015438493682e-05,
      "loss": 0.6798,
      "step": 5128
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3623300006540185,
      "learning_rate": 2.3634458954066992e-05,
      "loss": 0.6383,
      "step": 5129
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3747402126684342,
      "learning_rate": 2.3628763218379485e-05,
      "loss": 0.6426,
      "step": 5130
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.979224659058831,
      "learning_rate": 2.362306717835199e-05,
      "loss": 0.4924,
      "step": 5131
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3980516588564587,
      "learning_rate": 2.3617370834462223e-05,
      "loss": 0.6868,
      "step": 5132
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38993830849567923,
      "learning_rate": 2.361167418718793e-05,
      "loss": 0.7023,
      "step": 5133
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.436042095218702,
      "learning_rate": 2.3605977237006898e-05,
      "loss": 0.67,
      "step": 5134
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3879322622828566,
      "learning_rate": 2.3600279984396894e-05,
      "loss": 0.6515,
      "step": 5135
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3760837120759185,
      "learning_rate": 2.3594582429835763e-05,
      "loss": 0.6483,
      "step": 5136
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9687989098561466,
      "learning_rate": 2.358888457380135e-05,
      "loss": 0.4418,
      "step": 5137
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.45973749485896526,
      "learning_rate": 2.3583186416771513e-05,
      "loss": 0.6544,
      "step": 5138
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41438111843437225,
      "learning_rate": 2.3577487959224162e-05,
      "loss": 0.6889,
      "step": 5139
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36919717627727006,
      "learning_rate": 2.3571789201637216e-05,
      "loss": 0.6634,
      "step": 5140
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3888244748098219,
      "learning_rate": 2.356609014448862e-05,
      "loss": 0.6224,
      "step": 5141
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3722353219857113,
      "learning_rate": 2.356039078825636e-05,
      "loss": 0.6399,
      "step": 5142
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37655123881793684,
      "learning_rate": 2.355469113341841e-05,
      "loss": 0.6469,
      "step": 5143
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3418641345774744,
      "learning_rate": 2.3548991180452824e-05,
      "loss": 0.6677,
      "step": 5144
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4008333347214531,
      "learning_rate": 2.3543290929837626e-05,
      "loss": 0.6673,
      "step": 5145
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37962216487192574,
      "learning_rate": 2.3537590382050893e-05,
      "loss": 0.6836,
      "step": 5146
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3739759592958746,
      "learning_rate": 2.3531889537570735e-05,
      "loss": 0.6723,
      "step": 5147
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4013824681112453,
      "learning_rate": 2.352618839687526e-05,
      "loss": 0.6538,
      "step": 5148
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.45201375444630065,
      "learning_rate": 2.3520486960442627e-05,
      "loss": 0.6976,
      "step": 5149
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39269407235590365,
      "learning_rate": 2.3514785228751e-05,
      "loss": 0.6846,
      "step": 5150
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37054942820258735,
      "learning_rate": 2.3509083202278588e-05,
      "loss": 0.6523,
      "step": 5151
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4026880638073772,
      "learning_rate": 2.3503380881503598e-05,
      "loss": 0.6673,
      "step": 5152
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36517423398727517,
      "learning_rate": 2.3497678266904287e-05,
      "loss": 0.6743,
      "step": 5153
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36028866760454814,
      "learning_rate": 2.3491975358958925e-05,
      "loss": 0.6509,
      "step": 5154
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3556464600478887,
      "learning_rate": 2.3486272158145807e-05,
      "loss": 0.659,
      "step": 5155
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3956098114866703,
      "learning_rate": 2.3480568664943255e-05,
      "loss": 0.6973,
      "step": 5156
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38806235252353133,
      "learning_rate": 2.3474864879829608e-05,
      "loss": 0.6994,
      "step": 5157
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39300255698397946,
      "learning_rate": 2.346916080328324e-05,
      "loss": 0.6945,
      "step": 5158
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42995961833353574,
      "learning_rate": 2.3463456435782554e-05,
      "loss": 0.6525,
      "step": 5159
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4782013611579781,
      "learning_rate": 2.345775177780595e-05,
      "loss": 0.71,
      "step": 5160
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4211475188247324,
      "learning_rate": 2.3452046829831884e-05,
      "loss": 0.6929,
      "step": 5161
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43026707659258323,
      "learning_rate": 2.344634159233882e-05,
      "loss": 0.6335,
      "step": 5162
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37193397309729764,
      "learning_rate": 2.344063606580525e-05,
      "loss": 0.6766,
      "step": 5163
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.498135907875003,
      "learning_rate": 2.343493025070968e-05,
      "loss": 0.4823,
      "step": 5164
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3926732487229983,
      "learning_rate": 2.3429224147530664e-05,
      "loss": 0.6374,
      "step": 5165
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37732413080668054,
      "learning_rate": 2.3423517756746764e-05,
      "loss": 0.6866,
      "step": 5166
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37241673826902183,
      "learning_rate": 2.341781107883656e-05,
      "loss": 0.677,
      "step": 5167
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4405215336468044,
      "learning_rate": 2.341210411427867e-05,
      "loss": 0.6894,
      "step": 5168
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3949804312808213,
      "learning_rate": 2.340639686355173e-05,
      "loss": 0.6706,
      "step": 5169
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3808402263661684,
      "learning_rate": 2.3400689327134395e-05,
      "loss": 0.6984,
      "step": 5170
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40696802224262624,
      "learning_rate": 2.3394981505505357e-05,
      "loss": 0.7215,
      "step": 5171
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38059510651337,
      "learning_rate": 2.3389273399143317e-05,
      "loss": 0.6889,
      "step": 5172
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38685877416291947,
      "learning_rate": 2.3383565008527008e-05,
      "loss": 0.6983,
      "step": 5173
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.35442047733936716,
      "learning_rate": 2.3377856334135194e-05,
      "loss": 0.6379,
      "step": 5174
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.35234078604321684,
      "learning_rate": 2.337214737644664e-05,
      "loss": 0.6923,
      "step": 5175
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3546666923659889,
      "learning_rate": 2.336643813594017e-05,
      "loss": 0.665,
      "step": 5176
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40795366328011684,
      "learning_rate": 2.3360728613094595e-05,
      "loss": 0.6571,
      "step": 5177
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38460719880104904,
      "learning_rate": 2.3355018808388765e-05,
      "loss": 0.6613,
      "step": 5178
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4642225398099244,
      "learning_rate": 2.334930872230156e-05,
      "loss": 0.7235,
      "step": 5179
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41154610522196156,
      "learning_rate": 2.3343598355311886e-05,
      "loss": 0.6872,
      "step": 5180
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37475629666836163,
      "learning_rate": 2.3337887707898648e-05,
      "loss": 0.6711,
      "step": 5181
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3810317467014674,
      "learning_rate": 2.33321767805408e-05,
      "loss": 0.6862,
      "step": 5182
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.35683701831858633,
      "learning_rate": 2.332646557371731e-05,
      "loss": 0.6339,
      "step": 5183
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41524946446223193,
      "learning_rate": 2.3320754087907176e-05,
      "loss": 0.6773,
      "step": 5184
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4226748580018073,
      "learning_rate": 2.33150423235894e-05,
      "loss": 0.7225,
      "step": 5185
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40277655921513544,
      "learning_rate": 2.330933028124303e-05,
      "loss": 0.6646,
      "step": 5186
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3782382946820522,
      "learning_rate": 2.3303617961347126e-05,
      "loss": 0.6827,
      "step": 5187
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4566156323171507,
      "learning_rate": 2.3297905364380773e-05,
      "loss": 0.7062,
      "step": 5188
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4175461985624497,
      "learning_rate": 2.3292192490823075e-05,
      "loss": 0.6701,
      "step": 5189
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38926024760566436,
      "learning_rate": 2.328647934115318e-05,
      "loss": 0.6547,
      "step": 5190
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.364769205425524,
      "learning_rate": 2.3280765915850224e-05,
      "loss": 0.683,
      "step": 5191
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42336068771643975,
      "learning_rate": 2.327505221539339e-05,
      "loss": 0.657,
      "step": 5192
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4231439396315833,
      "learning_rate": 2.3269338240261886e-05,
      "loss": 0.6803,
      "step": 5193
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.40353020731535616,
      "learning_rate": 2.3263623990934936e-05,
      "loss": 0.6764,
      "step": 5194
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4279208997740201,
      "learning_rate": 2.325790946789178e-05,
      "loss": 0.6959,
      "step": 5195
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36004346367516243,
      "learning_rate": 2.3252194671611686e-05,
      "loss": 0.6213,
      "step": 5196
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36831716709374307,
      "learning_rate": 2.324647960257396e-05,
      "loss": 0.626,
      "step": 5197
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3870306258222731,
      "learning_rate": 2.3240764261257908e-05,
      "loss": 0.6864,
      "step": 5198
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3585700670123206,
      "learning_rate": 2.3235048648142867e-05,
      "loss": 0.6654,
      "step": 5199
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.35164062660828366,
      "learning_rate": 2.3229332763708214e-05,
      "loss": 0.6475,
      "step": 5200
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3121819425669012,
      "learning_rate": 2.3223616608433317e-05,
      "loss": 0.651,
      "step": 5201
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3105913187640046,
      "learning_rate": 2.321790018279758e-05,
      "loss": 0.6373,
      "step": 5202
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33117403793146183,
      "learning_rate": 2.3212183487280448e-05,
      "loss": 0.6398,
      "step": 5203
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4075047449167006,
      "learning_rate": 2.320646652236137e-05,
      "loss": 0.7102,
      "step": 5204
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3911478843986713,
      "learning_rate": 2.3200749288519812e-05,
      "loss": 0.6518,
      "step": 5205
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4166011997649554,
      "learning_rate": 2.3195031786235272e-05,
      "loss": 0.6624,
      "step": 5206
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39917498291706743,
      "learning_rate": 2.3189314015987286e-05,
      "loss": 0.6491,
      "step": 5207
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42442074747829395,
      "learning_rate": 2.3183595978255382e-05,
      "loss": 0.6718,
      "step": 5208
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44305237697551014,
      "learning_rate": 2.317787767351912e-05,
      "loss": 0.7092,
      "step": 5209
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44737574084403375,
      "learning_rate": 2.3172159102258106e-05,
      "loss": 0.6825,
      "step": 5210
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40708974077011756,
      "learning_rate": 2.3166440264951935e-05,
      "loss": 0.6658,
      "step": 5211
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39332027783541856,
      "learning_rate": 2.316072116208024e-05,
      "loss": 0.6527,
      "step": 5212
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.32213123056316073,
      "learning_rate": 2.3155001794122676e-05,
      "loss": 0.644,
      "step": 5213
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3902299907959512,
      "learning_rate": 2.314928216155893e-05,
      "loss": 0.6371,
      "step": 5214
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42087697819617537,
      "learning_rate": 2.3143562264868686e-05,
      "loss": 0.6599,
      "step": 5215
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42200340789364255,
      "learning_rate": 2.313784210453167e-05,
      "loss": 0.6779,
      "step": 5216
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39938620714467415,
      "learning_rate": 2.3132121681027635e-05,
      "loss": 0.6661,
      "step": 5217
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36248608817158834,
      "learning_rate": 2.3126400994836337e-05,
      "loss": 0.6549,
      "step": 5218
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4227424102141565,
      "learning_rate": 2.3120680046437554e-05,
      "loss": 0.6887,
      "step": 5219
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3824964615831857,
      "learning_rate": 2.311495883631111e-05,
      "loss": 0.6885,
      "step": 5220
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.397055922437459,
      "learning_rate": 2.3109237364936834e-05,
      "loss": 0.6465,
      "step": 5221
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39868027108644344,
      "learning_rate": 2.310351563279457e-05,
      "loss": 0.6447,
      "step": 5222
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34882320369423186,
      "learning_rate": 2.309779364036419e-05,
      "loss": 0.632,
      "step": 5223
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4225202539447535,
      "learning_rate": 2.3092071388125614e-05,
      "loss": 0.6691,
      "step": 5224
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37437034361570265,
      "learning_rate": 2.308634887655874e-05,
      "loss": 0.6738,
      "step": 5225
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4609412277633373,
      "learning_rate": 2.3080626106143503e-05,
      "loss": 0.712,
      "step": 5226
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38116535782036104,
      "learning_rate": 2.3074903077359888e-05,
      "loss": 0.7041,
      "step": 5227
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3615566608634733,
      "learning_rate": 2.3069179790687867e-05,
      "loss": 0.6675,
      "step": 5228
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3967377423443064,
      "learning_rate": 2.3063456246607432e-05,
      "loss": 0.6506,
      "step": 5229
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37196360361614694,
      "learning_rate": 2.3057732445598625e-05,
      "loss": 0.6776,
      "step": 5230
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4206762220306008,
      "learning_rate": 2.3052008388141495e-05,
      "loss": 0.7195,
      "step": 5231
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37753470721152466,
      "learning_rate": 2.30462840747161e-05,
      "loss": 0.6989,
      "step": 5232
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43550133361229176,
      "learning_rate": 2.3040559505802545e-05,
      "loss": 0.7017,
      "step": 5233
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4187368878387388,
      "learning_rate": 2.3034834681880938e-05,
      "loss": 0.6662,
      "step": 5234
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33099856306595726,
      "learning_rate": 2.302910960343141e-05,
      "loss": 0.6501,
      "step": 5235
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41449077934268597,
      "learning_rate": 2.3023384270934107e-05,
      "loss": 0.6488,
      "step": 5236
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39889276141744145,
      "learning_rate": 2.301765868486923e-05,
      "loss": 0.6951,
      "step": 5237
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3527365116589222,
      "learning_rate": 2.3011932845716963e-05,
      "loss": 0.661,
      "step": 5238
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.372790146870431,
      "learning_rate": 2.3006206753957518e-05,
      "loss": 0.6593,
      "step": 5239
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40685920701892586,
      "learning_rate": 2.3000480410071147e-05,
      "loss": 0.6859,
      "step": 5240
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4243413305353788,
      "learning_rate": 2.2994753814538107e-05,
      "loss": 0.6905,
      "step": 5241
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3997173906597574,
      "learning_rate": 2.2989026967838684e-05,
      "loss": 0.6722,
      "step": 5242
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35393342780798376,
      "learning_rate": 2.2983299870453172e-05,
      "loss": 0.6392,
      "step": 5243
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3475942697281778,
      "learning_rate": 2.297757252286191e-05,
      "loss": 0.6682,
      "step": 5244
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39672892774795177,
      "learning_rate": 2.2971844925545234e-05,
      "loss": 0.7062,
      "step": 5245
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38973323522676767,
      "learning_rate": 2.296611707898351e-05,
      "loss": 0.7027,
      "step": 5246
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.465155233221736,
      "learning_rate": 2.2960388983657134e-05,
      "loss": 0.7006,
      "step": 5247
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3824667960941462,
      "learning_rate": 2.2954660640046507e-05,
      "loss": 0.666,
      "step": 5248
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43132814718046764,
      "learning_rate": 2.294893204863206e-05,
      "loss": 0.7077,
      "step": 5249
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3969493631890397,
      "learning_rate": 2.294320320989424e-05,
      "loss": 0.7023,
      "step": 5250
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4135251734497789,
      "learning_rate": 2.293747412431353e-05,
      "loss": 0.6329,
      "step": 5251
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4530399907120933,
      "learning_rate": 2.293174479237041e-05,
      "loss": 0.7038,
      "step": 5252
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41923083724824683,
      "learning_rate": 2.2926015214545393e-05,
      "loss": 0.6872,
      "step": 5253
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.380608226492085,
      "learning_rate": 2.2920285391319007e-05,
      "loss": 0.6869,
      "step": 5254
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4420111732023356,
      "learning_rate": 2.2914555323171824e-05,
      "loss": 0.7482,
      "step": 5255
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3544042419636452,
      "learning_rate": 2.29088250105844e-05,
      "loss": 0.6494,
      "step": 5256
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3386292148905571,
      "learning_rate": 2.2903094454037332e-05,
      "loss": 0.6528,
      "step": 5257
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42164467464158706,
      "learning_rate": 2.2897363654011246e-05,
      "loss": 0.6968,
      "step": 5258
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3627241304804903,
      "learning_rate": 2.2891632610986766e-05,
      "loss": 0.6542,
      "step": 5259
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4268490564858417,
      "learning_rate": 2.288590132544455e-05,
      "loss": 0.6982,
      "step": 5260
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8403223963791803,
      "learning_rate": 2.2880169797865273e-05,
      "loss": 0.4357,
      "step": 5261
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4308854319174922,
      "learning_rate": 2.287443802872964e-05,
      "loss": 0.6882,
      "step": 5262
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42993174398982414,
      "learning_rate": 2.2868706018518356e-05,
      "loss": 0.6509,
      "step": 5263
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.431208604204508,
      "learning_rate": 2.2862973767712165e-05,
      "loss": 0.6914,
      "step": 5264
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4041639164945816,
      "learning_rate": 2.285724127679182e-05,
      "loss": 0.6624,
      "step": 5265
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39625270792861966,
      "learning_rate": 2.28515085462381e-05,
      "loss": 0.7055,
      "step": 5266
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3704733703928652,
      "learning_rate": 2.2845775576531802e-05,
      "loss": 0.645,
      "step": 5267
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3671460929909443,
      "learning_rate": 2.2840042368153746e-05,
      "loss": 0.6448,
      "step": 5268
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3761860427212895,
      "learning_rate": 2.2834308921584763e-05,
      "loss": 0.6692,
      "step": 5269
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3972203298030315,
      "learning_rate": 2.2828575237305715e-05,
      "loss": 0.6526,
      "step": 5270
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39270469635779387,
      "learning_rate": 2.2822841315797474e-05,
      "loss": 0.6682,
      "step": 5271
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4103342198592922,
      "learning_rate": 2.2817107157540947e-05,
      "loss": 0.7101,
      "step": 5272
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38798527336313177,
      "learning_rate": 2.281137276301704e-05,
      "loss": 0.6891,
      "step": 5273
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4019590592534748,
      "learning_rate": 2.2805638132706696e-05,
      "loss": 0.6898,
      "step": 5274
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37559472451929626,
      "learning_rate": 2.2799903267090877e-05,
      "loss": 0.6484,
      "step": 5275
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38393726964480657,
      "learning_rate": 2.2794168166650543e-05,
      "loss": 0.6816,
      "step": 5276
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3351633892653914,
      "learning_rate": 2.2788432831866708e-05,
      "loss": 0.6237,
      "step": 5277
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4083990901226303,
      "learning_rate": 2.2782697263220374e-05,
      "loss": 0.6517,
      "step": 5278
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.396960240941882,
      "learning_rate": 2.277696146119258e-05,
      "loss": 0.68,
      "step": 5279
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.382320305073185,
      "learning_rate": 2.2771225426264382e-05,
      "loss": 0.6792,
      "step": 5280
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38776471094289305,
      "learning_rate": 2.2765489158916853e-05,
      "loss": 0.657,
      "step": 5281
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36799575355089875,
      "learning_rate": 2.2759752659631093e-05,
      "loss": 0.6545,
      "step": 5282
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35822260717741666,
      "learning_rate": 2.275401592888821e-05,
      "loss": 0.683,
      "step": 5283
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4408315300374255,
      "learning_rate": 2.2748278967169335e-05,
      "loss": 0.7213,
      "step": 5284
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36750109001062786,
      "learning_rate": 2.274254177495562e-05,
      "loss": 0.6634,
      "step": 5285
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34403864264317735,
      "learning_rate": 2.2736804352728243e-05,
      "loss": 0.6474,
      "step": 5286
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41108196988531037,
      "learning_rate": 2.273106670096839e-05,
      "loss": 0.6711,
      "step": 5287
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3872837812356933,
      "learning_rate": 2.2725328820157263e-05,
      "loss": 0.665,
      "step": 5288
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3762962636711575,
      "learning_rate": 2.2719590710776108e-05,
      "loss": 0.6969,
      "step": 5289
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.370938113533518,
      "learning_rate": 2.271385237330616e-05,
      "loss": 0.6557,
      "step": 5290
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3697295315387932,
      "learning_rate": 2.2708113808228697e-05,
      "loss": 0.6468,
      "step": 5291
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3719635555178075,
      "learning_rate": 2.2702375016024992e-05,
      "loss": 0.668,
      "step": 5292
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3267161789845773,
      "learning_rate": 2.2696635997176362e-05,
      "loss": 0.6618,
      "step": 5293
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38665509534199805,
      "learning_rate": 2.2690896752164128e-05,
      "loss": 0.6926,
      "step": 5294
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8132153776846114,
      "learning_rate": 2.268515728146963e-05,
      "loss": 0.4297,
      "step": 5295
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40499954711154645,
      "learning_rate": 2.2679417585574237e-05,
      "loss": 0.6939,
      "step": 5296
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3855297918358069,
      "learning_rate": 2.2673677664959322e-05,
      "loss": 0.6999,
      "step": 5297
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4094093751478122,
      "learning_rate": 2.26679375201063e-05,
      "loss": 0.6785,
      "step": 5298
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3858362935462762,
      "learning_rate": 2.266219715149657e-05,
      "loss": 0.6453,
      "step": 5299
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40007546404928424,
      "learning_rate": 2.2656456559611587e-05,
      "loss": 0.6868,
      "step": 5300
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.378616366526358,
      "learning_rate": 2.2650715744932796e-05,
      "loss": 0.6838,
      "step": 5301
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3997730537028824,
      "learning_rate": 2.2644974707941677e-05,
      "loss": 0.668,
      "step": 5302
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4347489324153409,
      "learning_rate": 2.263923344911973e-05,
      "loss": 0.6856,
      "step": 5303
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39792999219954106,
      "learning_rate": 2.2633491968948454e-05,
      "loss": 0.6691,
      "step": 5304
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40302443828640305,
      "learning_rate": 2.262775026790939e-05,
      "loss": 0.63,
      "step": 5305
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42681361499464493,
      "learning_rate": 2.2622008346484082e-05,
      "loss": 0.6899,
      "step": 5306
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3199899121380982,
      "learning_rate": 2.26162662051541e-05,
      "loss": 0.6292,
      "step": 5307
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4009434888215739,
      "learning_rate": 2.261052384440104e-05,
      "loss": 0.6831,
      "step": 5308
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8250175589673622,
      "learning_rate": 2.2604781264706482e-05,
      "loss": 0.4387,
      "step": 5309
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.410819419792439,
      "learning_rate": 2.2599038466552075e-05,
      "loss": 0.7141,
      "step": 5310
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43845663686479946,
      "learning_rate": 2.2593295450419452e-05,
      "loss": 0.6773,
      "step": 5311
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39046795339889795,
      "learning_rate": 2.2587552216790272e-05,
      "loss": 0.6999,
      "step": 5312
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4175476074076428,
      "learning_rate": 2.258180876614621e-05,
      "loss": 0.6583,
      "step": 5313
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4024980473308189,
      "learning_rate": 2.257606509896896e-05,
      "loss": 0.6374,
      "step": 5314
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3592081978370529,
      "learning_rate": 2.257032121574025e-05,
      "loss": 0.6666,
      "step": 5315
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3806111065391814,
      "learning_rate": 2.25645771169418e-05,
      "loss": 0.6773,
      "step": 5316
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39080026950518887,
      "learning_rate": 2.255883280305536e-05,
      "loss": 0.7189,
      "step": 5317
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3969089563346201,
      "learning_rate": 2.255308827456271e-05,
      "loss": 0.6847,
      "step": 5318
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42799232939417464,
      "learning_rate": 2.2547343531945626e-05,
      "loss": 0.7039,
      "step": 5319
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.389960041432127,
      "learning_rate": 2.2541598575685914e-05,
      "loss": 0.7004,
      "step": 5320
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3820378232047342,
      "learning_rate": 2.2535853406265408e-05,
      "loss": 0.6723,
      "step": 5321
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34718589226156477,
      "learning_rate": 2.2530108024165936e-05,
      "loss": 0.6447,
      "step": 5322
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3937460634333132,
      "learning_rate": 2.2524362429869356e-05,
      "loss": 0.6876,
      "step": 5323
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33863435789603186,
      "learning_rate": 2.2518616623857542e-05,
      "loss": 0.6447,
      "step": 5324
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38874133838373315,
      "learning_rate": 2.2512870606612407e-05,
      "loss": 0.652,
      "step": 5325
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4272324097935429,
      "learning_rate": 2.2507124378615842e-05,
      "loss": 0.6994,
      "step": 5326
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37457328640836857,
      "learning_rate": 2.250137794034978e-05,
      "loss": 0.7045,
      "step": 5327
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41957836192273,
      "learning_rate": 2.2495631292296176e-05,
      "loss": 0.7035,
      "step": 5328
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41405859916955784,
      "learning_rate": 2.248988443493699e-05,
      "loss": 0.719,
      "step": 5329
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34584043331190906,
      "learning_rate": 2.2484137368754203e-05,
      "loss": 0.6675,
      "step": 5330
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4121663750080992,
      "learning_rate": 2.2478390094229807e-05,
      "loss": 0.6757,
      "step": 5331
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36381265404977464,
      "learning_rate": 2.2472642611845837e-05,
      "loss": 0.6514,
      "step": 5332
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.381682893992973,
      "learning_rate": 2.2466894922084315e-05,
      "loss": 0.7138,
      "step": 5333
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3532418168851749,
      "learning_rate": 2.246114702542729e-05,
      "loss": 0.6655,
      "step": 5334
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38683041502363186,
      "learning_rate": 2.2455398922356846e-05,
      "loss": 0.7047,
      "step": 5335
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34405854394401514,
      "learning_rate": 2.2449650613355052e-05,
      "loss": 0.6691,
      "step": 5336
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3960182857953198,
      "learning_rate": 2.2443902098904018e-05,
      "loss": 0.665,
      "step": 5337
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36685004244110203,
      "learning_rate": 2.243815337948587e-05,
      "loss": 0.6706,
      "step": 5338
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4085028814492969,
      "learning_rate": 2.2432404455582745e-05,
      "loss": 0.677,
      "step": 5339
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3833613656676951,
      "learning_rate": 2.2426655327676793e-05,
      "loss": 0.6934,
      "step": 5340
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40485440033117726,
      "learning_rate": 2.242090599625019e-05,
      "loss": 0.7166,
      "step": 5341
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31841556943839494,
      "learning_rate": 2.2415156461785122e-05,
      "loss": 0.6507,
      "step": 5342
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4869571491662087,
      "learning_rate": 2.2409406724763808e-05,
      "loss": 0.7097,
      "step": 5343
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.365145284801284,
      "learning_rate": 2.240365678566845e-05,
      "loss": 0.6815,
      "step": 5344
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4078731067180505,
      "learning_rate": 2.2397906644981306e-05,
      "loss": 0.6902,
      "step": 5345
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3447702450049244,
      "learning_rate": 2.2392156303184627e-05,
      "loss": 0.6646,
      "step": 5346
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3382191384136667,
      "learning_rate": 2.2386405760760687e-05,
      "loss": 0.6714,
      "step": 5347
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4419746536233205,
      "learning_rate": 2.238065501819178e-05,
      "loss": 0.6663,
      "step": 5348
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38480724203294747,
      "learning_rate": 2.237490407596021e-05,
      "loss": 0.6651,
      "step": 5349
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39249482996219903,
      "learning_rate": 2.23691529345483e-05,
      "loss": 0.6807,
      "step": 5350
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3883973664041097,
      "learning_rate": 2.2363401594438398e-05,
      "loss": 0.6863,
      "step": 5351
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3402943841123282,
      "learning_rate": 2.235765005611286e-05,
      "loss": 0.6236,
      "step": 5352
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4012124899855139,
      "learning_rate": 2.2351898320054062e-05,
      "loss": 0.703,
      "step": 5353
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32594980001828977,
      "learning_rate": 2.234614638674438e-05,
      "loss": 0.6575,
      "step": 5354
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38101509131200245,
      "learning_rate": 2.2340394256666244e-05,
      "loss": 0.7,
      "step": 5355
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3623301128977106,
      "learning_rate": 2.2334641930302066e-05,
      "loss": 0.5848,
      "step": 5356
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4080558269707447,
      "learning_rate": 2.232888940813429e-05,
      "loss": 0.6802,
      "step": 5357
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39205270568697975,
      "learning_rate": 2.2323136690645368e-05,
      "loss": 0.6373,
      "step": 5358
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35046811573018477,
      "learning_rate": 2.2317383778317783e-05,
      "loss": 0.6533,
      "step": 5359
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4041749990973493,
      "learning_rate": 2.231163067163402e-05,
      "loss": 0.6628,
      "step": 5360
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3454239033559395,
      "learning_rate": 2.2305877371076576e-05,
      "loss": 0.6658,
      "step": 5361
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37356168539057893,
      "learning_rate": 2.2300123877127987e-05,
      "loss": 0.6351,
      "step": 5362
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41063369049068105,
      "learning_rate": 2.2294370190270792e-05,
      "loss": 0.6877,
      "step": 5363
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4147672073929694,
      "learning_rate": 2.228861631098753e-05,
      "loss": 0.6579,
      "step": 5364
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3715826058307642,
      "learning_rate": 2.2282862239760786e-05,
      "loss": 0.7225,
      "step": 5365
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38557387888047434,
      "learning_rate": 2.227710797707314e-05,
      "loss": 0.651,
      "step": 5366
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3786500480262368,
      "learning_rate": 2.22713535234072e-05,
      "loss": 0.6797,
      "step": 5367
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37198468677479696,
      "learning_rate": 2.2265598879245583e-05,
      "loss": 0.6658,
      "step": 5368
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7372192881948613,
      "learning_rate": 2.225984404507093e-05,
      "loss": 0.4346,
      "step": 5369
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3799903751625899,
      "learning_rate": 2.2254089021365882e-05,
      "loss": 0.6695,
      "step": 5370
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42359557059098335,
      "learning_rate": 2.22483338086131e-05,
      "loss": 0.6834,
      "step": 5371
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36861109175923645,
      "learning_rate": 2.224257840729529e-05,
      "loss": 0.6515,
      "step": 5372
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37110263872140314,
      "learning_rate": 2.2236822817895133e-05,
      "loss": 0.6857,
      "step": 5373
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37045502608239017,
      "learning_rate": 2.2231067040895348e-05,
      "loss": 0.6965,
      "step": 5374
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40757000958503,
      "learning_rate": 2.2225311076778663e-05,
      "loss": 0.6624,
      "step": 5375
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3848376626196642,
      "learning_rate": 2.221955492602783e-05,
      "loss": 0.612,
      "step": 5376
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4181135720801906,
      "learning_rate": 2.2213798589125603e-05,
      "loss": 0.6733,
      "step": 5377
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4410614542208181,
      "learning_rate": 2.220804206655476e-05,
      "loss": 0.722,
      "step": 5378
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.386154984179949,
      "learning_rate": 2.2202285358798096e-05,
      "loss": 0.6529,
      "step": 5379
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3863410487630295,
      "learning_rate": 2.2196528466338424e-05,
      "loss": 0.6682,
      "step": 5380
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44193167231029573,
      "learning_rate": 2.219077138965856e-05,
      "loss": 0.7342,
      "step": 5381
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37132118139368553,
      "learning_rate": 2.218501412924134e-05,
      "loss": 0.6683,
      "step": 5382
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39571955163684075,
      "learning_rate": 2.217925668556963e-05,
      "loss": 0.6735,
      "step": 5383
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7653347956921834,
      "learning_rate": 2.2173499059126285e-05,
      "loss": 0.4144,
      "step": 5384
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4160935712901345,
      "learning_rate": 2.2167741250394208e-05,
      "loss": 0.6333,
      "step": 5385
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37496994100901965,
      "learning_rate": 2.2161983259856287e-05,
      "loss": 0.6533,
      "step": 5386
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40637745779805634,
      "learning_rate": 2.2156225087995446e-05,
      "loss": 0.6614,
      "step": 5387
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4546753478544705,
      "learning_rate": 2.2150466735294605e-05,
      "loss": 0.6992,
      "step": 5388
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41050252011021393,
      "learning_rate": 2.2144708202236712e-05,
      "loss": 0.6699,
      "step": 5389
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3964838856936454,
      "learning_rate": 2.2138949489304743e-05,
      "loss": 0.6182,
      "step": 5390
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3848330318072804,
      "learning_rate": 2.2133190596981655e-05,
      "loss": 0.6543,
      "step": 5391
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39365893706724464,
      "learning_rate": 2.212743152575045e-05,
      "loss": 0.6323,
      "step": 5392
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4358984100733771,
      "learning_rate": 2.2121672276094132e-05,
      "loss": 0.7216,
      "step": 5393
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38551657889880475,
      "learning_rate": 2.2115912848495725e-05,
      "loss": 0.6295,
      "step": 5394
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.449928528064071,
      "learning_rate": 2.2110153243438264e-05,
      "loss": 0.7054,
      "step": 5395
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4042161701296582,
      "learning_rate": 2.210439346140479e-05,
      "loss": 0.7059,
      "step": 5396
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43650056301195395,
      "learning_rate": 2.2098633502878388e-05,
      "loss": 0.6333,
      "step": 5397
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4075182216236582,
      "learning_rate": 2.2092873368342126e-05,
      "loss": 0.7015,
      "step": 5398
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4183303307245924,
      "learning_rate": 2.20871130582791e-05,
      "loss": 0.6487,
      "step": 5399
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42994005167749627,
      "learning_rate": 2.2081352573172424e-05,
      "loss": 0.6628,
      "step": 5400
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36056687052771946,
      "learning_rate": 2.2075591913505218e-05,
      "loss": 0.6774,
      "step": 5401
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4618240188311809,
      "learning_rate": 2.2069831079760625e-05,
      "loss": 0.7366,
      "step": 5402
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33921758851816997,
      "learning_rate": 2.2064070072421804e-05,
      "loss": 0.6604,
      "step": 5403
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3614367171047892,
      "learning_rate": 2.205830889197192e-05,
      "loss": 0.6766,
      "step": 5404
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41355189671916776,
      "learning_rate": 2.205254753889415e-05,
      "loss": 0.6806,
      "step": 5405
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3813927651904085,
      "learning_rate": 2.2046786013671693e-05,
      "loss": 0.6752,
      "step": 5406
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3875778992370406,
      "learning_rate": 2.2041024316787775e-05,
      "loss": 0.6325,
      "step": 5407
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40297941247904256,
      "learning_rate": 2.2035262448725606e-05,
      "loss": 0.6708,
      "step": 5408
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38810461815434527,
      "learning_rate": 2.2029500409968435e-05,
      "loss": 0.6923,
      "step": 5409
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36176971712458167,
      "learning_rate": 2.2023738200999517e-05,
      "loss": 0.6828,
      "step": 5410
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3454971030660419,
      "learning_rate": 2.201797582230212e-05,
      "loss": 0.6407,
      "step": 5411
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41882634949030145,
      "learning_rate": 2.2012213274359526e-05,
      "loss": 0.6631,
      "step": 5412
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3883681075061157,
      "learning_rate": 2.2006450557655034e-05,
      "loss": 0.6552,
      "step": 5413
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3854847665719136,
      "learning_rate": 2.2000687672671963e-05,
      "loss": 0.681,
      "step": 5414
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3592266141450822,
      "learning_rate": 2.1994924619893623e-05,
      "loss": 0.6426,
      "step": 5415
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37393678189565277,
      "learning_rate": 2.198916139980337e-05,
      "loss": 0.6469,
      "step": 5416
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3984738500975824,
      "learning_rate": 2.1983398012884552e-05,
      "loss": 0.6802,
      "step": 5417
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3712505885601697,
      "learning_rate": 2.197763445962054e-05,
      "loss": 0.6577,
      "step": 5418
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39407107310935885,
      "learning_rate": 2.1971870740494712e-05,
      "loss": 0.6638,
      "step": 5419
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43062903422795623,
      "learning_rate": 2.196610685599047e-05,
      "loss": 0.6535,
      "step": 5420
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3318621759784212,
      "learning_rate": 2.196034280659122e-05,
      "loss": 0.6326,
      "step": 5421
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3911716381860799,
      "learning_rate": 2.195457859278038e-05,
      "loss": 0.686,
      "step": 5422
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34855339186271717,
      "learning_rate": 2.1948814215041395e-05,
      "loss": 0.6572,
      "step": 5423
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41202197457836426,
      "learning_rate": 2.194304967385772e-05,
      "loss": 0.7223,
      "step": 5424
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39337312574149763,
      "learning_rate": 2.1937284969712812e-05,
      "loss": 0.6978,
      "step": 5425
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42105733204552387,
      "learning_rate": 2.193152010309016e-05,
      "loss": 0.6402,
      "step": 5426
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7122846301873679,
      "learning_rate": 2.192575507447324e-05,
      "loss": 0.4274,
      "step": 5427
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.369587720144153,
      "learning_rate": 2.1919989884345575e-05,
      "loss": 0.6667,
      "step": 5428
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38195770400260615,
      "learning_rate": 2.1914224533190676e-05,
      "loss": 0.6599,
      "step": 5429
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3831888269625281,
      "learning_rate": 2.190845902149207e-05,
      "loss": 0.6595,
      "step": 5430
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40947301671316066,
      "learning_rate": 2.190269334973332e-05,
      "loss": 0.6781,
      "step": 5431
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3540466602581815,
      "learning_rate": 2.189692751839797e-05,
      "loss": 0.6614,
      "step": 5432
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3868968202417676,
      "learning_rate": 2.1891161527969603e-05,
      "loss": 0.672,
      "step": 5433
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4393698365448143,
      "learning_rate": 2.1885395378931805e-05,
      "loss": 0.7137,
      "step": 5434
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34028932813815005,
      "learning_rate": 2.1879629071768166e-05,
      "loss": 0.6658,
      "step": 5435
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38183891248541924,
      "learning_rate": 2.1873862606962316e-05,
      "loss": 0.6859,
      "step": 5436
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4249840573461583,
      "learning_rate": 2.1868095984997865e-05,
      "loss": 0.6861,
      "step": 5437
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43664365825178963,
      "learning_rate": 2.1862329206358465e-05,
      "loss": 0.7205,
      "step": 5438
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38728648781857533,
      "learning_rate": 2.185656227152776e-05,
      "loss": 0.6776,
      "step": 5439
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3807292999890118,
      "learning_rate": 2.1850795180989427e-05,
      "loss": 0.6397,
      "step": 5440
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40054976054226216,
      "learning_rate": 2.1845027935227128e-05,
      "loss": 0.6914,
      "step": 5441
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32896506753824073,
      "learning_rate": 2.1839260534724564e-05,
      "loss": 0.6328,
      "step": 5442
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39312265676822333,
      "learning_rate": 2.1833492979965446e-05,
      "loss": 0.6473,
      "step": 5443
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3730406216048185,
      "learning_rate": 2.1827725271433483e-05,
      "loss": 0.6725,
      "step": 5444
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39206388208051207,
      "learning_rate": 2.1821957409612405e-05,
      "loss": 0.664,
      "step": 5445
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37909202957527427,
      "learning_rate": 2.1816189394985967e-05,
      "loss": 0.6752,
      "step": 5446
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40950017230219266,
      "learning_rate": 2.1810421228037913e-05,
      "loss": 0.6759,
      "step": 5447
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37079323272882103,
      "learning_rate": 2.1804652909252024e-05,
      "loss": 0.6978,
      "step": 5448
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4290418291127397,
      "learning_rate": 2.1798884439112066e-05,
      "loss": 0.607,
      "step": 5449
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3789178329341047,
      "learning_rate": 2.1793115818101853e-05,
      "loss": 0.6584,
      "step": 5450
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36578703511138977,
      "learning_rate": 2.1787347046705172e-05,
      "loss": 0.6605,
      "step": 5451
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42117500311831074,
      "learning_rate": 2.1781578125405852e-05,
      "loss": 0.7226,
      "step": 5452
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4240059914551387,
      "learning_rate": 2.1775809054687734e-05,
      "loss": 0.6681,
      "step": 5453
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3979992844534693,
      "learning_rate": 2.177003983503465e-05,
      "loss": 0.6802,
      "step": 5454
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3601996393585782,
      "learning_rate": 2.1764270466930465e-05,
      "loss": 0.654,
      "step": 5455
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3844617138649847,
      "learning_rate": 2.1758500950859046e-05,
      "loss": 0.6341,
      "step": 5456
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40149924655152197,
      "learning_rate": 2.175273128730428e-05,
      "loss": 0.6965,
      "step": 5457
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4074249974328877,
      "learning_rate": 2.174696147675005e-05,
      "loss": 0.7207,
      "step": 5458
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4236152713079556,
      "learning_rate": 2.1741191519680273e-05,
      "loss": 0.6666,
      "step": 5459
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8240014230067298,
      "learning_rate": 2.1735421416578867e-05,
      "loss": 0.4259,
      "step": 5460
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3792485640713312,
      "learning_rate": 2.1729651167929763e-05,
      "loss": 0.6639,
      "step": 5461
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3455529078514839,
      "learning_rate": 2.1723880774216902e-05,
      "loss": 0.6205,
      "step": 5462
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.354800809888456,
      "learning_rate": 2.171811023592425e-05,
      "loss": 0.6593,
      "step": 5463
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3580141873836087,
      "learning_rate": 2.1712339553535766e-05,
      "loss": 0.628,
      "step": 5464
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35561197526654775,
      "learning_rate": 2.1706568727535422e-05,
      "loss": 0.6251,
      "step": 5465
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4220457465021433,
      "learning_rate": 2.1700797758407225e-05,
      "loss": 0.6549,
      "step": 5466
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40496768025017416,
      "learning_rate": 2.1695026646635177e-05,
      "loss": 0.6697,
      "step": 5467
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3959748985266854,
      "learning_rate": 2.1689255392703288e-05,
      "loss": 0.6597,
      "step": 5468
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3524230281766133,
      "learning_rate": 2.1683483997095588e-05,
      "loss": 0.6453,
      "step": 5469
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38917850816358157,
      "learning_rate": 2.1677712460296125e-05,
      "loss": 0.7049,
      "step": 5470
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.343065874594027,
      "learning_rate": 2.167194078278894e-05,
      "loss": 0.683,
      "step": 5471
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3935875568764615,
      "learning_rate": 2.1666168965058096e-05,
      "loss": 0.6847,
      "step": 5472
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.384171130126905,
      "learning_rate": 2.1660397007587682e-05,
      "loss": 0.7023,
      "step": 5473
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38161957496629206,
      "learning_rate": 2.165462491086177e-05,
      "loss": 0.6952,
      "step": 5474
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37053104425475064,
      "learning_rate": 2.164885267536447e-05,
      "loss": 0.6673,
      "step": 5475
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3180983417675049,
      "learning_rate": 2.164308030157988e-05,
      "loss": 0.6549,
      "step": 5476
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38101585587057235,
      "learning_rate": 2.1637307789992135e-05,
      "loss": 0.6781,
      "step": 5477
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3450557323036437,
      "learning_rate": 2.1631535141085362e-05,
      "loss": 0.6677,
      "step": 5478
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3410479799513869,
      "learning_rate": 2.1625762355343706e-05,
      "loss": 0.6552,
      "step": 5479
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4041841947231689,
      "learning_rate": 2.1619989433251325e-05,
      "loss": 0.6646,
      "step": 5480
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4217588354195789,
      "learning_rate": 2.1614216375292392e-05,
      "loss": 0.7072,
      "step": 5481
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3540110468479293,
      "learning_rate": 2.160844318195107e-05,
      "loss": 0.6594,
      "step": 5482
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3671395548447069,
      "learning_rate": 2.160266985371157e-05,
      "loss": 0.6905,
      "step": 5483
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37698943707419474,
      "learning_rate": 2.1596896391058082e-05,
      "loss": 0.6784,
      "step": 5484
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35998117616828484,
      "learning_rate": 2.159112279447482e-05,
      "loss": 0.635,
      "step": 5485
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35036735262827323,
      "learning_rate": 2.1585349064446013e-05,
      "loss": 0.6387,
      "step": 5486
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4039029796191325,
      "learning_rate": 2.1579575201455894e-05,
      "loss": 0.6958,
      "step": 5487
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3811579179581898,
      "learning_rate": 2.157380120598872e-05,
      "loss": 0.6657,
      "step": 5488
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39635045531643365,
      "learning_rate": 2.156802707852872e-05,
      "loss": 0.7056,
      "step": 5489
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9466071397107856,
      "learning_rate": 2.1562252819560197e-05,
      "loss": 0.4713,
      "step": 5490
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4543592673791951,
      "learning_rate": 2.1556478429567415e-05,
      "loss": 0.6833,
      "step": 5491
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4486378001987836,
      "learning_rate": 2.1550703909034666e-05,
      "loss": 0.6491,
      "step": 5492
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.389646396280321,
      "learning_rate": 2.154492925844625e-05,
      "loss": 0.6715,
      "step": 5493
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3812204483209344,
      "learning_rate": 2.1539154478286488e-05,
      "loss": 0.6688,
      "step": 5494
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40773202039565354,
      "learning_rate": 2.153337956903969e-05,
      "loss": 0.6642,
      "step": 5495
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3612295629991052,
      "learning_rate": 2.1527604531190205e-05,
      "loss": 0.6574,
      "step": 5496
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35600634774450024,
      "learning_rate": 2.1521829365222376e-05,
      "loss": 0.6601,
      "step": 5497
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3758221678961567,
      "learning_rate": 2.1516054071620552e-05,
      "loss": 0.6633,
      "step": 5498
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36580421445261435,
      "learning_rate": 2.1510278650869098e-05,
      "loss": 0.6721,
      "step": 5499
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3832503647572133,
      "learning_rate": 2.150450310345241e-05,
      "loss": 0.6895,
      "step": 5500
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40052891448167377,
      "learning_rate": 2.149872742985486e-05,
      "loss": 0.6462,
      "step": 5501
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3827948310485423,
      "learning_rate": 2.149295163056085e-05,
      "loss": 0.6983,
      "step": 5502
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4027147292937553,
      "learning_rate": 2.148717570605478e-05,
      "loss": 0.694,
      "step": 5503
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4035909611224563,
      "learning_rate": 2.1481399656821092e-05,
      "loss": 0.6436,
      "step": 5504
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4335954491613138,
      "learning_rate": 2.1475623483344204e-05,
      "loss": 0.6977,
      "step": 5505
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4091690684944432,
      "learning_rate": 2.1469847186108542e-05,
      "loss": 0.6495,
      "step": 5506
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36443041363872974,
      "learning_rate": 2.1464070765598584e-05,
      "loss": 0.6228,
      "step": 5507
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37224365727157477,
      "learning_rate": 2.1458294222298784e-05,
      "loss": 0.6483,
      "step": 5508
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3969596947348901,
      "learning_rate": 2.14525175566936e-05,
      "loss": 0.6618,
      "step": 5509
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3985114582950438,
      "learning_rate": 2.144674076926752e-05,
      "loss": 0.6687,
      "step": 5510
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35736436788167875,
      "learning_rate": 2.144096386050505e-05,
      "loss": 0.6512,
      "step": 5511
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3702181077678079,
      "learning_rate": 2.143518683089067e-05,
      "loss": 0.6633,
      "step": 5512
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3787713573373262,
      "learning_rate": 2.1429409680908905e-05,
      "loss": 0.655,
      "step": 5513
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0177786471250359,
      "learning_rate": 2.1423632411044283e-05,
      "loss": 0.4824,
      "step": 5514
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37112463343955293,
      "learning_rate": 2.141785502178133e-05,
      "loss": 0.6289,
      "step": 5515
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6542766386457542,
      "learning_rate": 2.1412077513604582e-05,
      "loss": 0.4243,
      "step": 5516
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38177161724834524,
      "learning_rate": 2.14062998869986e-05,
      "loss": 0.6687,
      "step": 5517
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3585428193963101,
      "learning_rate": 2.140052214244795e-05,
      "loss": 0.6689,
      "step": 5518
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36180692037685247,
      "learning_rate": 2.1394744280437194e-05,
      "loss": 0.6651,
      "step": 5519
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4162632725846266,
      "learning_rate": 2.1388966301450916e-05,
      "loss": 0.6417,
      "step": 5520
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35462414629123185,
      "learning_rate": 2.1383188205973723e-05,
      "loss": 0.6905,
      "step": 5521
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3669157534478619,
      "learning_rate": 2.13774099944902e-05,
      "loss": 0.6874,
      "step": 5522
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40072100640948927,
      "learning_rate": 2.1371631667484963e-05,
      "loss": 0.6995,
      "step": 5523
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4233631830884608,
      "learning_rate": 2.1365853225442633e-05,
      "loss": 0.7085,
      "step": 5524
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3964777169608953,
      "learning_rate": 2.136007466884785e-05,
      "loss": 0.6666,
      "step": 5525
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3949579874221187,
      "learning_rate": 2.1354295998185238e-05,
      "loss": 0.6839,
      "step": 5526
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36748807375793346,
      "learning_rate": 2.134851721393946e-05,
      "loss": 0.6671,
      "step": 5527
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34970780169097754,
      "learning_rate": 2.134273831659517e-05,
      "loss": 0.6599,
      "step": 5528
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3804612161183997,
      "learning_rate": 2.1336959306637044e-05,
      "loss": 0.645,
      "step": 5529
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35987441617277827,
      "learning_rate": 2.133118018454975e-05,
      "loss": 0.6534,
      "step": 5530
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3953479382116069,
      "learning_rate": 2.1325400950817988e-05,
      "loss": 0.681,
      "step": 5531
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38015360204216103,
      "learning_rate": 2.1319621605926448e-05,
      "loss": 0.6536,
      "step": 5532
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3430072829987967,
      "learning_rate": 2.131384215035983e-05,
      "loss": 0.6866,
      "step": 5533
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4049787222281488,
      "learning_rate": 2.1308062584602865e-05,
      "loss": 0.6841,
      "step": 5534
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4116565430634706,
      "learning_rate": 2.130228290914027e-05,
      "loss": 0.691,
      "step": 5535
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.339533866625961,
      "learning_rate": 2.1296503124456774e-05,
      "loss": 0.6215,
      "step": 5536
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3932632252534514,
      "learning_rate": 2.1290723231037132e-05,
      "loss": 0.7157,
      "step": 5537
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3492120479327898,
      "learning_rate": 2.128494322936609e-05,
      "loss": 0.675,
      "step": 5538
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3747488231143356,
      "learning_rate": 2.127916311992841e-05,
      "loss": 0.6767,
      "step": 5539
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.348989620412908,
      "learning_rate": 2.127338290320887e-05,
      "loss": 0.6064,
      "step": 5540
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41414128976860015,
      "learning_rate": 2.126760257969224e-05,
      "loss": 0.6724,
      "step": 5541
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3599650127949674,
      "learning_rate": 2.126182214986331e-05,
      "loss": 0.6434,
      "step": 5542
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9597270679311308,
      "learning_rate": 2.125604161420688e-05,
      "loss": 0.4592,
      "step": 5543
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45973124711741836,
      "learning_rate": 2.1250260973207757e-05,
      "loss": 0.7086,
      "step": 5544
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.384638379816656,
      "learning_rate": 2.1244480227350763e-05,
      "loss": 0.6772,
      "step": 5545
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42426749301071365,
      "learning_rate": 2.123869937712071e-05,
      "loss": 0.6988,
      "step": 5546
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4073330027402479,
      "learning_rate": 2.1232918423002437e-05,
      "loss": 0.6802,
      "step": 5547
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37437404458731016,
      "learning_rate": 2.1227137365480785e-05,
      "loss": 0.6832,
      "step": 5548
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3813765471328941,
      "learning_rate": 2.1221356205040604e-05,
      "loss": 0.6527,
      "step": 5549
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.389802528659434,
      "learning_rate": 2.121557494216675e-05,
      "loss": 0.6573,
      "step": 5550
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4392965243869933,
      "learning_rate": 2.1209793577344097e-05,
      "loss": 0.7618,
      "step": 5551
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4356168656033722,
      "learning_rate": 2.120401211105752e-05,
      "loss": 0.7076,
      "step": 5552
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3748341237106048,
      "learning_rate": 2.1198230543791895e-05,
      "loss": 0.6577,
      "step": 5553
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40925181700561336,
      "learning_rate": 2.1192448876032125e-05,
      "loss": 0.7061,
      "step": 5554
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3772136925347158,
      "learning_rate": 2.1186667108263108e-05,
      "loss": 0.6699,
      "step": 5555
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3491275716031097,
      "learning_rate": 2.118088524096975e-05,
      "loss": 0.6698,
      "step": 5556
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4027165128598146,
      "learning_rate": 2.117510327463698e-05,
      "loss": 0.7023,
      "step": 5557
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35740964813117565,
      "learning_rate": 2.1169321209749713e-05,
      "loss": 0.6562,
      "step": 5558
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4087724692536064,
      "learning_rate": 2.116353904679289e-05,
      "loss": 0.6936,
      "step": 5559
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4526232980038337,
      "learning_rate": 2.1157756786251447e-05,
      "loss": 0.6861,
      "step": 5560
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44913422854210716,
      "learning_rate": 2.115197442861035e-05,
      "loss": 0.6861,
      "step": 5561
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41062438143167523,
      "learning_rate": 2.1146191974354542e-05,
      "loss": 0.6393,
      "step": 5562
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36537236265323114,
      "learning_rate": 2.1140409423969e-05,
      "loss": 0.7072,
      "step": 5563
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3804108581502047,
      "learning_rate": 2.1134626777938698e-05,
      "loss": 0.6736,
      "step": 5564
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.39291849245836946,
      "learning_rate": 2.1128844036748616e-05,
      "loss": 0.672,
      "step": 5565
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37991322270709504,
      "learning_rate": 2.1123061200883748e-05,
      "loss": 0.7042,
      "step": 5566
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3936847274938941,
      "learning_rate": 2.111727827082909e-05,
      "loss": 0.6454,
      "step": 5567
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3538616682517649,
      "learning_rate": 2.111149524706966e-05,
      "loss": 0.649,
      "step": 5568
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7986110704872563,
      "learning_rate": 2.1105712130090464e-05,
      "loss": 0.4344,
      "step": 5569
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3301123506528188,
      "learning_rate": 2.1099928920376525e-05,
      "loss": 0.6614,
      "step": 5570
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3775922849733634,
      "learning_rate": 2.1094145618412875e-05,
      "loss": 0.6501,
      "step": 5571
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3918846222527773,
      "learning_rate": 2.1088362224684555e-05,
      "loss": 0.6762,
      "step": 5572
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37142853323037445,
      "learning_rate": 2.1082578739676607e-05,
      "loss": 0.6559,
      "step": 5573
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37853234371664857,
      "learning_rate": 2.1076795163874096e-05,
      "loss": 0.6796,
      "step": 5574
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.39009643865551547,
      "learning_rate": 2.1071011497762065e-05,
      "loss": 0.6862,
      "step": 5575
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4045108966872814,
      "learning_rate": 2.10652277418256e-05,
      "loss": 0.7105,
      "step": 5576
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36419717018148645,
      "learning_rate": 2.1059443896549766e-05,
      "loss": 0.6641,
      "step": 5577
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.38119042268348696,
      "learning_rate": 2.1053659962419654e-05,
      "loss": 0.6593,
      "step": 5578
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35556399992942617,
      "learning_rate": 2.1047875939920353e-05,
      "loss": 0.6418,
      "step": 5579
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3801545320037341,
      "learning_rate": 2.1042091829536965e-05,
      "loss": 0.6579,
      "step": 5580
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7351501027563591,
      "learning_rate": 2.103630763175459e-05,
      "loss": 0.4253,
      "step": 5581
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.38867698540199136,
      "learning_rate": 2.1030523347058348e-05,
      "loss": 0.6896,
      "step": 5582
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41181811218036973,
      "learning_rate": 2.1024738975933357e-05,
      "loss": 0.6622,
      "step": 5583
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33883021113220596,
      "learning_rate": 2.1018954518864744e-05,
      "loss": 0.6301,
      "step": 5584
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3622272391209403,
      "learning_rate": 2.101316997633765e-05,
      "loss": 0.6802,
      "step": 5585
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3667876079196622,
      "learning_rate": 2.1007385348837208e-05,
      "loss": 0.6476,
      "step": 5586
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3609365319586827,
      "learning_rate": 2.1001600636848575e-05,
      "loss": 0.6817,
      "step": 5587
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3499795281915648,
      "learning_rate": 2.099581584085691e-05,
      "loss": 0.6345,
      "step": 5588
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41032134019200256,
      "learning_rate": 2.0990030961347368e-05,
      "loss": 0.6951,
      "step": 5589
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3617350892607084,
      "learning_rate": 2.0984245998805125e-05,
      "loss": 0.6234,
      "step": 5590
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34224620635993214,
      "learning_rate": 2.0978460953715364e-05,
      "loss": 0.6238,
      "step": 5591
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3987435206254086,
      "learning_rate": 2.0972675826563266e-05,
      "loss": 0.6826,
      "step": 5592
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40454213161787084,
      "learning_rate": 2.0966890617834014e-05,
      "loss": 0.7037,
      "step": 5593
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3851119142978833,
      "learning_rate": 2.0961105328012813e-05,
      "loss": 0.6557,
      "step": 5594
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3212170753876904,
      "learning_rate": 2.0955319957584876e-05,
      "loss": 0.6427,
      "step": 5595
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35075702618928545,
      "learning_rate": 2.0949534507035402e-05,
      "loss": 0.6802,
      "step": 5596
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3611512488256497,
      "learning_rate": 2.094374897684962e-05,
      "loss": 0.6495,
      "step": 5597
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43606493188355083,
      "learning_rate": 2.093796336751275e-05,
      "loss": 0.6876,
      "step": 5598
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.32471762172862695,
      "learning_rate": 2.0932177679510027e-05,
      "loss": 0.6503,
      "step": 5599
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40358676677000815,
      "learning_rate": 2.0926391913326678e-05,
      "loss": 0.657,
      "step": 5600
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42232452443901064,
      "learning_rate": 2.0920606069447968e-05,
      "loss": 0.6888,
      "step": 5601
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34994590998500547,
      "learning_rate": 2.0914820148359135e-05,
      "loss": 0.6403,
      "step": 5602
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7675740692946439,
      "learning_rate": 2.090903415054544e-05,
      "loss": 0.4223,
      "step": 5603
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4295179650316953,
      "learning_rate": 2.0903248076492152e-05,
      "loss": 0.6628,
      "step": 5604
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42370942229748243,
      "learning_rate": 2.089746192668454e-05,
      "loss": 0.7153,
      "step": 5605
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36431688796929734,
      "learning_rate": 2.089167570160788e-05,
      "loss": 0.7016,
      "step": 5606
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4401696433892622,
      "learning_rate": 2.0885889401747445e-05,
      "loss": 0.7099,
      "step": 5607
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4250982899389747,
      "learning_rate": 2.0880103027588542e-05,
      "loss": 0.6743,
      "step": 5608
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4217127825245457,
      "learning_rate": 2.087431657961647e-05,
      "loss": 0.7078,
      "step": 5609
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3952648383392062,
      "learning_rate": 2.0868530058316508e-05,
      "loss": 0.6767,
      "step": 5610
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3740130233684372,
      "learning_rate": 2.0862743464173988e-05,
      "loss": 0.6655,
      "step": 5611
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36414998336733095,
      "learning_rate": 2.0856956797674216e-05,
      "loss": 0.6978,
      "step": 5612
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3871016217812392,
      "learning_rate": 2.0851170059302507e-05,
      "loss": 0.688,
      "step": 5613
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3879023106516302,
      "learning_rate": 2.0845383249544192e-05,
      "loss": 0.7151,
      "step": 5614
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34216821773638095,
      "learning_rate": 2.0839596368884612e-05,
      "loss": 0.6269,
      "step": 5615
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4000967230550532,
      "learning_rate": 2.08338094178091e-05,
      "loss": 0.6272,
      "step": 5616
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3987358893799404,
      "learning_rate": 2.0828022396802984e-05,
      "loss": 0.6872,
      "step": 5617
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3444718889330739,
      "learning_rate": 2.0822235306351645e-05,
      "loss": 0.6571,
      "step": 5618
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4121959147613517,
      "learning_rate": 2.081644814694042e-05,
      "loss": 0.7222,
      "step": 5619
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3990097745239136,
      "learning_rate": 2.0810660919054672e-05,
      "loss": 0.6613,
      "step": 5620
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4304648594481455,
      "learning_rate": 2.0804873623179772e-05,
      "loss": 0.7224,
      "step": 5621
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4220336354337147,
      "learning_rate": 2.07990862598011e-05,
      "loss": 0.7141,
      "step": 5622
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4224202074284843,
      "learning_rate": 2.0793298829404024e-05,
      "loss": 0.6707,
      "step": 5623
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3956174386728679,
      "learning_rate": 2.0787511332473924e-05,
      "loss": 0.6352,
      "step": 5624
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45170302106127835,
      "learning_rate": 2.0781723769496213e-05,
      "loss": 0.7184,
      "step": 5625
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3647232051911045,
      "learning_rate": 2.077593614095627e-05,
      "loss": 0.6966,
      "step": 5626
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3804995287177033,
      "learning_rate": 2.0770148447339496e-05,
      "loss": 0.6244,
      "step": 5627
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4196909619568521,
      "learning_rate": 2.0764360689131303e-05,
      "loss": 0.7109,
      "step": 5628
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40079962415264786,
      "learning_rate": 2.0758572866817107e-05,
      "loss": 0.6926,
      "step": 5629
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45764469252311524,
      "learning_rate": 2.0752784980882315e-05,
      "loss": 0.6857,
      "step": 5630
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40619033590352754,
      "learning_rate": 2.0746997031812353e-05,
      "loss": 0.7041,
      "step": 5631
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4166272694293997,
      "learning_rate": 2.0741209020092664e-05,
      "loss": 0.6626,
      "step": 5632
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3812932925270587,
      "learning_rate": 2.0735420946208663e-05,
      "loss": 0.6953,
      "step": 5633
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3627454848844982,
      "learning_rate": 2.072963281064579e-05,
      "loss": 0.6711,
      "step": 5634
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3782800537308002,
      "learning_rate": 2.0723844613889504e-05,
      "loss": 0.6417,
      "step": 5635
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36689539299570695,
      "learning_rate": 2.071805635642524e-05,
      "loss": 0.6621,
      "step": 5636
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3729335679267587,
      "learning_rate": 2.0712268038738456e-05,
      "loss": 0.6552,
      "step": 5637
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.347182147546811,
      "learning_rate": 2.070647966131461e-05,
      "loss": 0.6167,
      "step": 5638
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4267174963268392,
      "learning_rate": 2.070069122463917e-05,
      "loss": 0.7077,
      "step": 5639
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4141727832906247,
      "learning_rate": 2.069490272919761e-05,
      "loss": 0.7035,
      "step": 5640
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3234294652211836,
      "learning_rate": 2.0689114175475384e-05,
      "loss": 0.619,
      "step": 5641
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43428631714905686,
      "learning_rate": 2.0683325563957993e-05,
      "loss": 0.725,
      "step": 5642
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3795832875193927,
      "learning_rate": 2.0677536895130915e-05,
      "loss": 0.6555,
      "step": 5643
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37534794203678185,
      "learning_rate": 2.067174816947963e-05,
      "loss": 0.6322,
      "step": 5644
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7555934207988387,
      "learning_rate": 2.066595938748963e-05,
      "loss": 0.4544,
      "step": 5645
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3788328438505036,
      "learning_rate": 2.0660170549646436e-05,
      "loss": 0.6936,
      "step": 5646
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3964183892772736,
      "learning_rate": 2.0654381656435526e-05,
      "loss": 0.6849,
      "step": 5647
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.429533301118632,
      "learning_rate": 2.0648592708342418e-05,
      "loss": 0.6578,
      "step": 5648
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40639330289849146,
      "learning_rate": 2.0642803705852623e-05,
      "loss": 0.6945,
      "step": 5649
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3910006048419106,
      "learning_rate": 2.063701464945166e-05,
      "loss": 0.706,
      "step": 5650
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37519598312226116,
      "learning_rate": 2.0631225539625048e-05,
      "loss": 0.656,
      "step": 5651
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39606821772463263,
      "learning_rate": 2.0625436376858307e-05,
      "loss": 0.6921,
      "step": 5652
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4314663232177305,
      "learning_rate": 2.061964716163698e-05,
      "loss": 0.7136,
      "step": 5653
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33925235676853976,
      "learning_rate": 2.0613857894446587e-05,
      "loss": 0.6555,
      "step": 5654
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42841651509702544,
      "learning_rate": 2.0608068575772675e-05,
      "loss": 0.7009,
      "step": 5655
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4161165077081141,
      "learning_rate": 2.060227920610079e-05,
      "loss": 0.6549,
      "step": 5656
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37769336709145485,
      "learning_rate": 2.0596489785916472e-05,
      "loss": 0.6622,
      "step": 5657
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3879873688562255,
      "learning_rate": 2.0590700315705276e-05,
      "loss": 0.6702,
      "step": 5658
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3799524183369781,
      "learning_rate": 2.0584910795952758e-05,
      "loss": 0.6423,
      "step": 5659
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37762805362653734,
      "learning_rate": 2.057912122714448e-05,
      "loss": 0.6808,
      "step": 5660
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4072069397941952,
      "learning_rate": 2.0573331609766e-05,
      "loss": 0.6844,
      "step": 5661
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4107042842063119,
      "learning_rate": 2.0567541944302892e-05,
      "loss": 0.6805,
      "step": 5662
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39472111268056786,
      "learning_rate": 2.0561752231240732e-05,
      "loss": 0.6881,
      "step": 5663
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40993039720946695,
      "learning_rate": 2.0555962471065082e-05,
      "loss": 0.682,
      "step": 5664
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3780013724635888,
      "learning_rate": 2.0550172664261535e-05,
      "loss": 0.6366,
      "step": 5665
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36309044892414355,
      "learning_rate": 2.0544382811315672e-05,
      "loss": 0.6773,
      "step": 5666
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6905025679962613,
      "learning_rate": 2.0538592912713083e-05,
      "loss": 0.4246,
      "step": 5667
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5976328907776118,
      "learning_rate": 2.053280296893935e-05,
      "loss": 0.4258,
      "step": 5668
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4276748355774714,
      "learning_rate": 2.0527012980480076e-05,
      "loss": 0.7088,
      "step": 5669
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.741226930963186,
      "learning_rate": 2.052122294782087e-05,
      "loss": 0.4316,
      "step": 5670
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40407221201588045,
      "learning_rate": 2.0515432871447314e-05,
      "loss": 0.646,
      "step": 5671
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3699841265249311,
      "learning_rate": 2.0509642751845024e-05,
      "loss": 0.6307,
      "step": 5672
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37835645608345064,
      "learning_rate": 2.050385258949962e-05,
      "loss": 0.6607,
      "step": 5673
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4433022136373591,
      "learning_rate": 2.0498062384896703e-05,
      "loss": 0.691,
      "step": 5674
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.439648258091224,
      "learning_rate": 2.0492272138521902e-05,
      "loss": 0.6666,
      "step": 5675
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40499870982916025,
      "learning_rate": 2.048648185086082e-05,
      "loss": 0.6959,
      "step": 5676
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38414745068179856,
      "learning_rate": 2.0480691522399104e-05,
      "loss": 0.6468,
      "step": 5677
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3757315286879752,
      "learning_rate": 2.0474901153622363e-05,
      "loss": 0.6483,
      "step": 5678
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34691239834750637,
      "learning_rate": 2.046911074501624e-05,
      "loss": 0.6613,
      "step": 5679
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4014663819979436,
      "learning_rate": 2.0463320297066366e-05,
      "loss": 0.6953,
      "step": 5680
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9094184968383324,
      "learning_rate": 2.045752981025837e-05,
      "loss": 0.4041,
      "step": 5681
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42804860145448476,
      "learning_rate": 2.045173928507791e-05,
      "loss": 0.6812,
      "step": 5682
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39686500829862004,
      "learning_rate": 2.0445948722010618e-05,
      "loss": 0.6479,
      "step": 5683
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4146588684622822,
      "learning_rate": 2.044015812154215e-05,
      "loss": 0.6828,
      "step": 5684
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39683681496941076,
      "learning_rate": 2.0434367484158146e-05,
      "loss": 0.6797,
      "step": 5685
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4191453463906737,
      "learning_rate": 2.0428576810344265e-05,
      "loss": 0.6761,
      "step": 5686
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37089317185899706,
      "learning_rate": 2.0422786100586167e-05,
      "loss": 0.6668,
      "step": 5687
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38698281601729884,
      "learning_rate": 2.041699535536951e-05,
      "loss": 0.6261,
      "step": 5688
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3576842021921099,
      "learning_rate": 2.041120457517996e-05,
      "loss": 0.6685,
      "step": 5689
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36644735668234335,
      "learning_rate": 2.040541376050317e-05,
      "loss": 0.6601,
      "step": 5690
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.377828160587143,
      "learning_rate": 2.0399622911824822e-05,
      "loss": 0.7146,
      "step": 5691
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39444764594259835,
      "learning_rate": 2.0393832029630583e-05,
      "loss": 0.5997,
      "step": 5692
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3577883217369645,
      "learning_rate": 2.038804111440613e-05,
      "loss": 0.652,
      "step": 5693
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4160743462502789,
      "learning_rate": 2.0382250166637133e-05,
      "loss": 0.6408,
      "step": 5694
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3310617228885811,
      "learning_rate": 2.037645918680928e-05,
      "loss": 0.6684,
      "step": 5695
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35196787288559894,
      "learning_rate": 2.037066817540825e-05,
      "loss": 0.6556,
      "step": 5696
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.361852901239032,
      "learning_rate": 2.036487713291972e-05,
      "loss": 0.6448,
      "step": 5697
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45232196638803646,
      "learning_rate": 2.035908605982939e-05,
      "loss": 0.7278,
      "step": 5698
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3999704810164332,
      "learning_rate": 2.0353294956622948e-05,
      "loss": 0.6688,
      "step": 5699
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35756906852261083,
      "learning_rate": 2.0347503823786083e-05,
      "loss": 0.6432,
      "step": 5700
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3731657711212691,
      "learning_rate": 2.0341712661804495e-05,
      "loss": 0.6852,
      "step": 5701
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.446114173812539,
      "learning_rate": 2.0335921471163873e-05,
      "loss": 0.699,
      "step": 5702
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4267330614625005,
      "learning_rate": 2.0330130252349925e-05,
      "loss": 0.6397,
      "step": 5703
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3896327059704361,
      "learning_rate": 2.0324339005848354e-05,
      "loss": 0.6676,
      "step": 5704
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39520603849086827,
      "learning_rate": 2.031854773214486e-05,
      "loss": 0.6786,
      "step": 5705
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37081466712385464,
      "learning_rate": 2.0312756431725157e-05,
      "loss": 0.676,
      "step": 5706
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40133104343622283,
      "learning_rate": 2.0306965105074943e-05,
      "loss": 0.7194,
      "step": 5707
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38513498486056585,
      "learning_rate": 2.030117375267994e-05,
      "loss": 0.6654,
      "step": 5708
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4003723312138028,
      "learning_rate": 2.029538237502586e-05,
      "loss": 0.6613,
      "step": 5709
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37909192090592125,
      "learning_rate": 2.028959097259841e-05,
      "loss": 0.6399,
      "step": 5710
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4146494209983824,
      "learning_rate": 2.0283799545883323e-05,
      "loss": 0.6533,
      "step": 5711
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36738663036593644,
      "learning_rate": 2.0278008095366304e-05,
      "loss": 0.6704,
      "step": 5712
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32825820301659236,
      "learning_rate": 2.0272216621533087e-05,
      "loss": 0.6374,
      "step": 5713
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38938452961250897,
      "learning_rate": 2.0266425124869385e-05,
      "loss": 0.6697,
      "step": 5714
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4524258809911877,
      "learning_rate": 2.0260633605860933e-05,
      "loss": 0.6167,
      "step": 5715
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4358508693737424,
      "learning_rate": 2.0254842064993458e-05,
      "loss": 0.6664,
      "step": 5716
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4228945823656917,
      "learning_rate": 2.0249050502752685e-05,
      "loss": 0.6869,
      "step": 5717
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38117514766946764,
      "learning_rate": 2.0243258919624343e-05,
      "loss": 0.6777,
      "step": 5718
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42745771307662034,
      "learning_rate": 2.0237467316094177e-05,
      "loss": 0.6538,
      "step": 5719
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3976101218106431,
      "learning_rate": 2.023167569264791e-05,
      "loss": 0.6909,
      "step": 5720
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42613599444107597,
      "learning_rate": 2.022588404977128e-05,
      "loss": 0.7022,
      "step": 5721
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3932825232847145,
      "learning_rate": 2.022009238795003e-05,
      "loss": 0.675,
      "step": 5722
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3384748048232668,
      "learning_rate": 2.02143007076699e-05,
      "loss": 0.6523,
      "step": 5723
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35900924139645984,
      "learning_rate": 2.020850900941662e-05,
      "loss": 0.651,
      "step": 5724
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.41168723639165844,
      "learning_rate": 2.020271729367594e-05,
      "loss": 0.6712,
      "step": 5725
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39198036258887636,
      "learning_rate": 2.019692556093362e-05,
      "loss": 0.6268,
      "step": 5726
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.840830235174418,
      "learning_rate": 2.0191133811675384e-05,
      "loss": 0.4058,
      "step": 5727
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3771439119028797,
      "learning_rate": 2.0185342046386983e-05,
      "loss": 0.6468,
      "step": 5728
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4463245953055463,
      "learning_rate": 2.017955026555417e-05,
      "loss": 0.6976,
      "step": 5729
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3731748401551108,
      "learning_rate": 2.017375846966269e-05,
      "loss": 0.6555,
      "step": 5730
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3825697168255574,
      "learning_rate": 2.0167966659198303e-05,
      "loss": 0.6311,
      "step": 5731
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3518424068024983,
      "learning_rate": 2.0162174834646748e-05,
      "loss": 0.6385,
      "step": 5732
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39887312881567055,
      "learning_rate": 2.0156382996493794e-05,
      "loss": 0.655,
      "step": 5733
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3746869076602953,
      "learning_rate": 2.0150591145225186e-05,
      "loss": 0.6604,
      "step": 5734
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.41688333271274514,
      "learning_rate": 2.014479928132667e-05,
      "loss": 0.6604,
      "step": 5735
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38115689972610695,
      "learning_rate": 2.0139007405284022e-05,
      "loss": 0.6286,
      "step": 5736
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38285260938143295,
      "learning_rate": 2.0133215517582993e-05,
      "loss": 0.652,
      "step": 5737
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4336571417739419,
      "learning_rate": 2.012742361870933e-05,
      "loss": 0.6853,
      "step": 5738
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37362664549985075,
      "learning_rate": 2.0121631709148806e-05,
      "loss": 0.6886,
      "step": 5739
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37691434134027996,
      "learning_rate": 2.0115839789387186e-05,
      "loss": 0.6551,
      "step": 5740
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4127443164128726,
      "learning_rate": 2.011004785991021e-05,
      "loss": 0.7066,
      "step": 5741
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37856171286022894,
      "learning_rate": 2.010425592120366e-05,
      "loss": 0.6397,
      "step": 5742
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9087161628910331,
      "learning_rate": 2.0098463973753294e-05,
      "loss": 0.4119,
      "step": 5743
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40438117024336634,
      "learning_rate": 2.009267201804487e-05,
      "loss": 0.6624,
      "step": 5744
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36490938669786366,
      "learning_rate": 2.0086880054564156e-05,
      "loss": 0.6766,
      "step": 5745
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3802378848579009,
      "learning_rate": 2.0081088083796912e-05,
      "loss": 0.6544,
      "step": 5746
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3815199071363615,
      "learning_rate": 2.007529610622892e-05,
      "loss": 0.6905,
      "step": 5747
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.41007785440864664,
      "learning_rate": 2.0069504122345923e-05,
      "loss": 0.6949,
      "step": 5748
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3849918365132269,
      "learning_rate": 2.0063712132633703e-05,
      "loss": 0.6556,
      "step": 5749
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35444397731448746,
      "learning_rate": 2.0057920137578027e-05,
      "loss": 0.6204,
      "step": 5750
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4168604958169781,
      "learning_rate": 2.0052128137664657e-05,
      "loss": 0.7241,
      "step": 5751
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.41881987069422993,
      "learning_rate": 2.0046336133379354e-05,
      "loss": 0.6832,
      "step": 5752
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3872294073089985,
      "learning_rate": 2.0040544125207905e-05,
      "loss": 0.6553,
      "step": 5753
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3888211531643624,
      "learning_rate": 2.0034752113636072e-05,
      "loss": 0.6629,
      "step": 5754
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40056952027831505,
      "learning_rate": 2.0028960099149616e-05,
      "loss": 0.6497,
      "step": 5755
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.394120841463669,
      "learning_rate": 2.002316808223431e-05,
      "loss": 0.6669,
      "step": 5756
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37473222563999287,
      "learning_rate": 2.001737606337593e-05,
      "loss": 0.6918,
      "step": 5757
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.449526267890784,
      "learning_rate": 2.001158404306024e-05,
      "loss": 0.6559,
      "step": 5758
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3746954375206394,
      "learning_rate": 2.0005792021773e-05,
      "loss": 0.6615,
      "step": 5759
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.355178928448285,
      "learning_rate": 2e-05,
      "loss": 0.619,
      "step": 5760
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37054013744976205,
      "learning_rate": 1.9994207978227002e-05,
      "loss": 0.6631,
      "step": 5761
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37829288598188854,
      "learning_rate": 1.998841595693977e-05,
      "loss": 0.6828,
      "step": 5762
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36661971095893364,
      "learning_rate": 1.998262393662408e-05,
      "loss": 0.6598,
      "step": 5763
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3621929248543238,
      "learning_rate": 1.997683191776569e-05,
      "loss": 0.6329,
      "step": 5764
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3643533698020623,
      "learning_rate": 1.997103990085039e-05,
      "loss": 0.6497,
      "step": 5765
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38088537832390174,
      "learning_rate": 1.996524788636393e-05,
      "loss": 0.6772,
      "step": 5766
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3761124636180668,
      "learning_rate": 1.99594558747921e-05,
      "loss": 0.6701,
      "step": 5767
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7694771430910134,
      "learning_rate": 1.995366386662065e-05,
      "loss": 0.4149,
      "step": 5768
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4206632384025455,
      "learning_rate": 1.9947871862335353e-05,
      "loss": 0.6807,
      "step": 5769
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4011560611636546,
      "learning_rate": 1.9942079862421987e-05,
      "loss": 0.6448,
      "step": 5770
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4280213430578307,
      "learning_rate": 1.99362878673663e-05,
      "loss": 0.6988,
      "step": 5771
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37050547288502056,
      "learning_rate": 1.9930495877654084e-05,
      "loss": 0.6207,
      "step": 5772
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3891298625111287,
      "learning_rate": 1.9924703893771088e-05,
      "loss": 0.6514,
      "step": 5773
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39763636059305024,
      "learning_rate": 1.991891191620309e-05,
      "loss": 0.7163,
      "step": 5774
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4367160736114915,
      "learning_rate": 1.991311994543585e-05,
      "loss": 0.7064,
      "step": 5775
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39889041487583404,
      "learning_rate": 1.9907327981955137e-05,
      "loss": 0.6765,
      "step": 5776
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4100249605434862,
      "learning_rate": 1.9901536026246716e-05,
      "loss": 0.6874,
      "step": 5777
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.459638975479496,
      "learning_rate": 1.989574407879634e-05,
      "loss": 0.6799,
      "step": 5778
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34467874234779644,
      "learning_rate": 1.9889952140089786e-05,
      "loss": 0.6555,
      "step": 5779
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3671284868819852,
      "learning_rate": 1.988416021061282e-05,
      "loss": 0.6742,
      "step": 5780
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.380973211549813,
      "learning_rate": 1.9878368290851197e-05,
      "loss": 0.7056,
      "step": 5781
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3632156153308434,
      "learning_rate": 1.9872576381290673e-05,
      "loss": 0.6625,
      "step": 5782
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3585306061163792,
      "learning_rate": 1.9866784482417018e-05,
      "loss": 0.6536,
      "step": 5783
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4179121803869877,
      "learning_rate": 1.9860992594715988e-05,
      "loss": 0.6753,
      "step": 5784
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39490663293449874,
      "learning_rate": 1.985520071867333e-05,
      "loss": 0.6936,
      "step": 5785
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4444523250620175,
      "learning_rate": 1.984940885477482e-05,
      "loss": 0.6493,
      "step": 5786
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.405562570402159,
      "learning_rate": 1.9843617003506212e-05,
      "loss": 0.6343,
      "step": 5787
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4338678652293522,
      "learning_rate": 1.9837825165353256e-05,
      "loss": 0.6724,
      "step": 5788
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3603767162907171,
      "learning_rate": 1.9832033340801703e-05,
      "loss": 0.6625,
      "step": 5789
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3822931799658384,
      "learning_rate": 1.9826241530337315e-05,
      "loss": 0.6406,
      "step": 5790
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3735244784844885,
      "learning_rate": 1.9820449734445844e-05,
      "loss": 0.6787,
      "step": 5791
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39444450491650623,
      "learning_rate": 1.9814657953613024e-05,
      "loss": 0.6666,
      "step": 5792
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3777291579352823,
      "learning_rate": 1.9808866188324623e-05,
      "loss": 0.6697,
      "step": 5793
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39578052680138925,
      "learning_rate": 1.9803074439066388e-05,
      "loss": 0.6408,
      "step": 5794
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3636911216587506,
      "learning_rate": 1.9797282706324063e-05,
      "loss": 0.6708,
      "step": 5795
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3952072078468976,
      "learning_rate": 1.9791490990583386e-05,
      "loss": 0.6941,
      "step": 5796
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40146191539721904,
      "learning_rate": 1.9785699292330115e-05,
      "loss": 0.6421,
      "step": 5797
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3799236763614889,
      "learning_rate": 1.977990761204998e-05,
      "loss": 0.6906,
      "step": 5798
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4000235849842237,
      "learning_rate": 1.9774115950228725e-05,
      "loss": 0.6375,
      "step": 5799
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8662384741135629,
      "learning_rate": 1.9768324307352096e-05,
      "loss": 0.4241,
      "step": 5800
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38352784116243444,
      "learning_rate": 1.976253268390583e-05,
      "loss": 0.6736,
      "step": 5801
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40639253836973777,
      "learning_rate": 1.9756741080375663e-05,
      "loss": 0.6466,
      "step": 5802
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3894807785715256,
      "learning_rate": 1.9750949497247322e-05,
      "loss": 0.6988,
      "step": 5803
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3910050923961294,
      "learning_rate": 1.9745157935006552e-05,
      "loss": 0.7181,
      "step": 5804
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37859306462119674,
      "learning_rate": 1.9739366394139067e-05,
      "loss": 0.6587,
      "step": 5805
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4166244320621278,
      "learning_rate": 1.9733574875130618e-05,
      "loss": 0.6894,
      "step": 5806
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3697806783790277,
      "learning_rate": 1.972778337846692e-05,
      "loss": 0.6415,
      "step": 5807
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4038211766438618,
      "learning_rate": 1.9721991904633702e-05,
      "loss": 0.6383,
      "step": 5808
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4236569154864236,
      "learning_rate": 1.9716200454116683e-05,
      "loss": 0.6978,
      "step": 5809
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36707145721122814,
      "learning_rate": 1.9710409027401592e-05,
      "loss": 0.6847,
      "step": 5810
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40751623266499687,
      "learning_rate": 1.970461762497415e-05,
      "loss": 0.6844,
      "step": 5811
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3490093008282769,
      "learning_rate": 1.969882624732006e-05,
      "loss": 0.6608,
      "step": 5812
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4148739187217766,
      "learning_rate": 1.9693034894925056e-05,
      "loss": 0.6888,
      "step": 5813
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37267221609924367,
      "learning_rate": 1.968724356827485e-05,
      "loss": 0.6458,
      "step": 5814
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4087916031209384,
      "learning_rate": 1.9681452267855142e-05,
      "loss": 0.6789,
      "step": 5815
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40519198050279853,
      "learning_rate": 1.967566099415165e-05,
      "loss": 0.6622,
      "step": 5816
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37846105952842707,
      "learning_rate": 1.9669869747650078e-05,
      "loss": 0.668,
      "step": 5817
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32954069260702257,
      "learning_rate": 1.9664078528836134e-05,
      "loss": 0.6663,
      "step": 5818
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3927573545610986,
      "learning_rate": 1.965828733819551e-05,
      "loss": 0.6875,
      "step": 5819
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3903028872066974,
      "learning_rate": 1.9652496176213916e-05,
      "loss": 0.6771,
      "step": 5820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44061699483230143,
      "learning_rate": 1.9646705043377055e-05,
      "loss": 0.6666,
      "step": 5821
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34256134012736955,
      "learning_rate": 1.9640913940170612e-05,
      "loss": 0.6727,
      "step": 5822
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42623476624702655,
      "learning_rate": 1.9635122867080284e-05,
      "loss": 0.6868,
      "step": 5823
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39652130006501446,
      "learning_rate": 1.962933182459176e-05,
      "loss": 0.6826,
      "step": 5824
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39372842179629336,
      "learning_rate": 1.962354081319073e-05,
      "loss": 0.7039,
      "step": 5825
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3363385836547554,
      "learning_rate": 1.961774983336287e-05,
      "loss": 0.6581,
      "step": 5826
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4393054856908983,
      "learning_rate": 1.9611958885593875e-05,
      "loss": 0.7051,
      "step": 5827
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40482623524987016,
      "learning_rate": 1.960616797036942e-05,
      "loss": 0.7074,
      "step": 5828
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3916759168740821,
      "learning_rate": 1.9600377088175185e-05,
      "loss": 0.686,
      "step": 5829
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39141737696081796,
      "learning_rate": 1.9594586239496837e-05,
      "loss": 0.655,
      "step": 5830
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35272582523557977,
      "learning_rate": 1.958879542482005e-05,
      "loss": 0.6405,
      "step": 5831
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3624267860350741,
      "learning_rate": 1.9583004644630497e-05,
      "loss": 0.6653,
      "step": 5832
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3644162341348085,
      "learning_rate": 1.9577213899413836e-05,
      "loss": 0.6962,
      "step": 5833
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37943893921381155,
      "learning_rate": 1.9571423189655735e-05,
      "loss": 0.6857,
      "step": 5834
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3365920973891702,
      "learning_rate": 1.956563251584186e-05,
      "loss": 0.6644,
      "step": 5835
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37042848081442525,
      "learning_rate": 1.9559841878457857e-05,
      "loss": 0.6446,
      "step": 5836
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35535425150295874,
      "learning_rate": 1.955405127798939e-05,
      "loss": 0.642,
      "step": 5837
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37823736130524394,
      "learning_rate": 1.9548260714922097e-05,
      "loss": 0.6495,
      "step": 5838
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4331000719405423,
      "learning_rate": 1.9542470189741634e-05,
      "loss": 0.7366,
      "step": 5839
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4346843628799342,
      "learning_rate": 1.9536679702933637e-05,
      "loss": 0.701,
      "step": 5840
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3564432058538821,
      "learning_rate": 1.9530889254983764e-05,
      "loss": 0.6561,
      "step": 5841
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3881521726604771,
      "learning_rate": 1.952509884637764e-05,
      "loss": 0.6817,
      "step": 5842
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3862906637684625,
      "learning_rate": 1.9519308477600903e-05,
      "loss": 0.6986,
      "step": 5843
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3808654807912006,
      "learning_rate": 1.9513518149139183e-05,
      "loss": 0.6446,
      "step": 5844
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39698788717564887,
      "learning_rate": 1.950772786147811e-05,
      "loss": 0.7047,
      "step": 5845
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3409211867963827,
      "learning_rate": 1.95019376151033e-05,
      "loss": 0.6562,
      "step": 5846
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4041983985301644,
      "learning_rate": 1.949614741050038e-05,
      "loss": 0.6542,
      "step": 5847
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37673006263613595,
      "learning_rate": 1.9490357248154976e-05,
      "loss": 0.6677,
      "step": 5848
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3535369202873018,
      "learning_rate": 1.9484567128552692e-05,
      "loss": 0.6397,
      "step": 5849
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3549676942258591,
      "learning_rate": 1.9478777052179138e-05,
      "loss": 0.6357,
      "step": 5850
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35217037715755434,
      "learning_rate": 1.9472987019519928e-05,
      "loss": 0.663,
      "step": 5851
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3778707656384127,
      "learning_rate": 1.9467197031060658e-05,
      "loss": 0.6738,
      "step": 5852
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3763944352721374,
      "learning_rate": 1.9461407087286927e-05,
      "loss": 0.6817,
      "step": 5853
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3881651687657838,
      "learning_rate": 1.9455617188684328e-05,
      "loss": 0.6842,
      "step": 5854
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3888173941775806,
      "learning_rate": 1.9449827335738468e-05,
      "loss": 0.6768,
      "step": 5855
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3739105797272609,
      "learning_rate": 1.944403752893492e-05,
      "loss": 0.6723,
      "step": 5856
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.41323881933315365,
      "learning_rate": 1.9438247768759275e-05,
      "loss": 0.6829,
      "step": 5857
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40531513711936523,
      "learning_rate": 1.9432458055697115e-05,
      "loss": 0.6623,
      "step": 5858
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3900102197872326,
      "learning_rate": 1.942666839023401e-05,
      "loss": 0.7128,
      "step": 5859
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3629058278528399,
      "learning_rate": 1.942087877285553e-05,
      "loss": 0.6827,
      "step": 5860
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37163735017810345,
      "learning_rate": 1.9415089204047245e-05,
      "loss": 0.6735,
      "step": 5861
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4031907864621503,
      "learning_rate": 1.9409299684294728e-05,
      "loss": 0.6971,
      "step": 5862
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.41607319771608625,
      "learning_rate": 1.9403510214083538e-05,
      "loss": 0.6434,
      "step": 5863
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4107450430759047,
      "learning_rate": 1.9397720793899218e-05,
      "loss": 0.7091,
      "step": 5864
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3974256292615606,
      "learning_rate": 1.9391931424227332e-05,
      "loss": 0.6563,
      "step": 5865
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42958946861150005,
      "learning_rate": 1.938614210555342e-05,
      "loss": 0.7051,
      "step": 5866
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3681409206666517,
      "learning_rate": 1.9380352838363033e-05,
      "loss": 0.6981,
      "step": 5867
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4027831854675857,
      "learning_rate": 1.9374563623141693e-05,
      "loss": 0.7033,
      "step": 5868
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3656712503123091,
      "learning_rate": 1.936877446037496e-05,
      "loss": 0.6668,
      "step": 5869
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3618451205794163,
      "learning_rate": 1.9362985350548342e-05,
      "loss": 0.6564,
      "step": 5870
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3897310726818506,
      "learning_rate": 1.935719629414738e-05,
      "loss": 0.6972,
      "step": 5871
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37419388167815826,
      "learning_rate": 1.935140729165759e-05,
      "loss": 0.6474,
      "step": 5872
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3960286885243881,
      "learning_rate": 1.934561834356448e-05,
      "loss": 0.6194,
      "step": 5873
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36930355210273913,
      "learning_rate": 1.9339829450353574e-05,
      "loss": 0.6401,
      "step": 5874
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38676019277924734,
      "learning_rate": 1.933404061251037e-05,
      "loss": 0.6515,
      "step": 5875
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.40231885488526486,
      "learning_rate": 1.932825183052038e-05,
      "loss": 0.6694,
      "step": 5876
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35902277769351326,
      "learning_rate": 1.9322463104869095e-05,
      "loss": 0.6179,
      "step": 5877
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36063865830693226,
      "learning_rate": 1.9316674436042014e-05,
      "loss": 0.6886,
      "step": 5878
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3902603413541188,
      "learning_rate": 1.9310885824524623e-05,
      "loss": 0.7003,
      "step": 5879
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39646254508246637,
      "learning_rate": 1.93050972708024e-05,
      "loss": 0.661,
      "step": 5880
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4144365130473838,
      "learning_rate": 1.929930877536084e-05,
      "loss": 0.711,
      "step": 5881
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37921250198521333,
      "learning_rate": 1.929352033868539e-05,
      "loss": 0.6459,
      "step": 5882
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35664534830214256,
      "learning_rate": 1.928773196126155e-05,
      "loss": 0.6955,
      "step": 5883
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3616378351052787,
      "learning_rate": 1.9281943643574766e-05,
      "loss": 0.6242,
      "step": 5884
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38480469764736625,
      "learning_rate": 1.9276155386110503e-05,
      "loss": 0.6691,
      "step": 5885
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39244717829286196,
      "learning_rate": 1.9270367189354216e-05,
      "loss": 0.6532,
      "step": 5886
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3956346491745052,
      "learning_rate": 1.9264579053791347e-05,
      "loss": 0.6718,
      "step": 5887
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4002971959439724,
      "learning_rate": 1.925879097990734e-05,
      "loss": 0.6734,
      "step": 5888
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39034671415634703,
      "learning_rate": 1.9253002968187647e-05,
      "loss": 0.6588,
      "step": 5889
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38558266537443786,
      "learning_rate": 1.924721501911769e-05,
      "loss": 0.6772,
      "step": 5890
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38918867820825187,
      "learning_rate": 1.92414271331829e-05,
      "loss": 0.6898,
      "step": 5891
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.454803834608682,
      "learning_rate": 1.92356393108687e-05,
      "loss": 0.7334,
      "step": 5892
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3567200968066474,
      "learning_rate": 1.9229851552660507e-05,
      "loss": 0.6267,
      "step": 5893
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3584365995211179,
      "learning_rate": 1.922406385904374e-05,
      "loss": 0.6715,
      "step": 5894
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.371602253369784,
      "learning_rate": 1.921827623050379e-05,
      "loss": 0.6302,
      "step": 5895
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35886149460503186,
      "learning_rate": 1.9212488667526076e-05,
      "loss": 0.6669,
      "step": 5896
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4108068564448041,
      "learning_rate": 1.9206701170595983e-05,
      "loss": 0.7093,
      "step": 5897
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3507496281586584,
      "learning_rate": 1.9200913740198908e-05,
      "loss": 0.6539,
      "step": 5898
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3634113310173221,
      "learning_rate": 1.919512637682023e-05,
      "loss": 0.6732,
      "step": 5899
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3805097234789944,
      "learning_rate": 1.9189339080945334e-05,
      "loss": 0.6404,
      "step": 5900
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3731587325056007,
      "learning_rate": 1.9183551853059586e-05,
      "loss": 0.6645,
      "step": 5901
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39054210367394604,
      "learning_rate": 1.9177764693648355e-05,
      "loss": 0.6186,
      "step": 5902
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3858267974865906,
      "learning_rate": 1.9171977603197013e-05,
      "loss": 0.6295,
      "step": 5903
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3930754246291058,
      "learning_rate": 1.9166190582190907e-05,
      "loss": 0.6647,
      "step": 5904
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39916895857234275,
      "learning_rate": 1.916040363111539e-05,
      "loss": 0.689,
      "step": 5905
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3732182251426923,
      "learning_rate": 1.915461675045581e-05,
      "loss": 0.6919,
      "step": 5906
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3810914869798143,
      "learning_rate": 1.91488299406975e-05,
      "loss": 0.706,
      "step": 5907
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.423882683060152,
      "learning_rate": 1.9143043202325794e-05,
      "loss": 0.7058,
      "step": 5908
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3492374544202581,
      "learning_rate": 1.9137256535826016e-05,
      "loss": 0.6159,
      "step": 5909
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34722618008419603,
      "learning_rate": 1.9131469941683495e-05,
      "loss": 0.6443,
      "step": 5910
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3247040375083922,
      "learning_rate": 1.912568342038354e-05,
      "loss": 0.6319,
      "step": 5911
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3719951395423438,
      "learning_rate": 1.911989697241146e-05,
      "loss": 0.6781,
      "step": 5912
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4185622450706399,
      "learning_rate": 1.9114110598252562e-05,
      "loss": 0.6887,
      "step": 5913
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.373283319963109,
      "learning_rate": 1.9108324298392133e-05,
      "loss": 0.6694,
      "step": 5914
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3696086876690316,
      "learning_rate": 1.910253807331547e-05,
      "loss": 0.6378,
      "step": 5915
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8346452312255307,
      "learning_rate": 1.909675192350785e-05,
      "loss": 0.4078,
      "step": 5916
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38396303913639745,
      "learning_rate": 1.9090965849454562e-05,
      "loss": 0.7134,
      "step": 5917
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42887451276216576,
      "learning_rate": 1.9085179851640868e-05,
      "loss": 0.6984,
      "step": 5918
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39636502163863024,
      "learning_rate": 1.907939393055204e-05,
      "loss": 0.6934,
      "step": 5919
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38637226603031066,
      "learning_rate": 1.907360808667333e-05,
      "loss": 0.6585,
      "step": 5920
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39442718627968915,
      "learning_rate": 1.9067822320489987e-05,
      "loss": 0.6825,
      "step": 5921
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41670694406611475,
      "learning_rate": 1.906203663248726e-05,
      "loss": 0.7119,
      "step": 5922
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3896029314710803,
      "learning_rate": 1.9056251023150385e-05,
      "loss": 0.6467,
      "step": 5923
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3570935946530052,
      "learning_rate": 1.9050465492964604e-05,
      "loss": 0.6509,
      "step": 5924
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3782401078511043,
      "learning_rate": 1.904468004241513e-05,
      "loss": 0.6526,
      "step": 5925
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4203367819795735,
      "learning_rate": 1.903889467198719e-05,
      "loss": 0.6616,
      "step": 5926
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3389813494598931,
      "learning_rate": 1.903310938216599e-05,
      "loss": 0.6166,
      "step": 5927
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3601817333485714,
      "learning_rate": 1.9027324173436744e-05,
      "loss": 0.6522,
      "step": 5928
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39396442597403686,
      "learning_rate": 1.9021539046284646e-05,
      "loss": 0.6517,
      "step": 5929
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4245370706143543,
      "learning_rate": 1.9015754001194875e-05,
      "loss": 0.6731,
      "step": 5930
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4404930549064852,
      "learning_rate": 1.9009969038652635e-05,
      "loss": 0.6913,
      "step": 5931
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4119688169868639,
      "learning_rate": 1.9004184159143094e-05,
      "loss": 0.655,
      "step": 5932
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4076264909697443,
      "learning_rate": 1.8998399363151432e-05,
      "loss": 0.6764,
      "step": 5933
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37665650871787204,
      "learning_rate": 1.89926146511628e-05,
      "loss": 0.6691,
      "step": 5934
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4096994124074957,
      "learning_rate": 1.898683002366236e-05,
      "loss": 0.6471,
      "step": 5935
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3551437012143748,
      "learning_rate": 1.8981045481135266e-05,
      "loss": 0.6582,
      "step": 5936
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.40130551421924404,
      "learning_rate": 1.897526102406665e-05,
      "loss": 0.6732,
      "step": 5937
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36258452009729825,
      "learning_rate": 1.8969476652941656e-05,
      "loss": 0.6623,
      "step": 5938
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.359348690468356,
      "learning_rate": 1.8963692368245413e-05,
      "loss": 0.655,
      "step": 5939
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3642006500784841,
      "learning_rate": 1.8957908170463045e-05,
      "loss": 0.6487,
      "step": 5940
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34219295219892454,
      "learning_rate": 1.895212406007965e-05,
      "loss": 0.6355,
      "step": 5941
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38003396296900427,
      "learning_rate": 1.8946340037580353e-05,
      "loss": 0.6626,
      "step": 5942
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4476383398874988,
      "learning_rate": 1.8940556103450244e-05,
      "loss": 0.6935,
      "step": 5943
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43291373776410036,
      "learning_rate": 1.8934772258174406e-05,
      "loss": 0.6998,
      "step": 5944
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36310734621056406,
      "learning_rate": 1.892898850223794e-05,
      "loss": 0.6267,
      "step": 5945
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35711759698688517,
      "learning_rate": 1.8923204836125914e-05,
      "loss": 0.6617,
      "step": 5946
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.40196895189969123,
      "learning_rate": 1.89174212603234e-05,
      "loss": 0.6534,
      "step": 5947
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3838663001439704,
      "learning_rate": 1.891163777531545e-05,
      "loss": 0.7006,
      "step": 5948
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41526109280415,
      "learning_rate": 1.8905854381587132e-05,
      "loss": 0.691,
      "step": 5949
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.40089964587755295,
      "learning_rate": 1.8900071079623482e-05,
      "loss": 0.6838,
      "step": 5950
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4099286598125118,
      "learning_rate": 1.8894287869909543e-05,
      "loss": 0.6366,
      "step": 5951
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35184266865165176,
      "learning_rate": 1.8888504752930343e-05,
      "loss": 0.6546,
      "step": 5952
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3894953109437389,
      "learning_rate": 1.8882721729170913e-05,
      "loss": 0.6883,
      "step": 5953
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3638499644399089,
      "learning_rate": 1.8876938799116256e-05,
      "loss": 0.6542,
      "step": 5954
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41491767318360817,
      "learning_rate": 1.8871155963251394e-05,
      "loss": 0.683,
      "step": 5955
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35483124652945514,
      "learning_rate": 1.8865373222061315e-05,
      "loss": 0.6449,
      "step": 5956
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3329597550428592,
      "learning_rate": 1.8859590576031008e-05,
      "loss": 0.6554,
      "step": 5957
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38151455661391165,
      "learning_rate": 1.8853808025645458e-05,
      "loss": 0.6402,
      "step": 5958
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4108298023586774,
      "learning_rate": 1.8848025571389654e-05,
      "loss": 0.702,
      "step": 5959
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3648623718552771,
      "learning_rate": 1.8842243213748556e-05,
      "loss": 0.6344,
      "step": 5960
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35138489536877926,
      "learning_rate": 1.8836460953207117e-05,
      "loss": 0.6413,
      "step": 5961
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3876656971399804,
      "learning_rate": 1.8830678790250293e-05,
      "loss": 0.6539,
      "step": 5962
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36205186631192576,
      "learning_rate": 1.882489672536303e-05,
      "loss": 0.6501,
      "step": 5963
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35316440197994764,
      "learning_rate": 1.8819114759030255e-05,
      "loss": 0.6551,
      "step": 5964
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3897578218539838,
      "learning_rate": 1.8813332891736896e-05,
      "loss": 0.6717,
      "step": 5965
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4007525845919926,
      "learning_rate": 1.8807551123967878e-05,
      "loss": 0.6696,
      "step": 5966
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4007174800722288,
      "learning_rate": 1.8801769456208112e-05,
      "loss": 0.6649,
      "step": 5967
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3658163878629648,
      "learning_rate": 1.879598788894249e-05,
      "loss": 0.6656,
      "step": 5968
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3843568031625587,
      "learning_rate": 1.879020642265591e-05,
      "loss": 0.631,
      "step": 5969
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3674756643482892,
      "learning_rate": 1.878442505783326e-05,
      "loss": 0.6561,
      "step": 5970
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3872805164534559,
      "learning_rate": 1.8778643794959406e-05,
      "loss": 0.6416,
      "step": 5971
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38164484447919106,
      "learning_rate": 1.877286263451922e-05,
      "loss": 0.6567,
      "step": 5972
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42021531912725013,
      "learning_rate": 1.8767081576997566e-05,
      "loss": 0.7002,
      "step": 5973
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3811369296125639,
      "learning_rate": 1.8761300622879297e-05,
      "loss": 0.6869,
      "step": 5974
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38050324296066473,
      "learning_rate": 1.8755519772649244e-05,
      "loss": 0.6883,
      "step": 5975
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36249769537368515,
      "learning_rate": 1.8749739026792246e-05,
      "loss": 0.7092,
      "step": 5976
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3466124113186821,
      "learning_rate": 1.874395838579313e-05,
      "loss": 0.6787,
      "step": 5977
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3679294053174802,
      "learning_rate": 1.8738177850136693e-05,
      "loss": 0.6862,
      "step": 5978
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34606730788360973,
      "learning_rate": 1.8732397420307763e-05,
      "loss": 0.6649,
      "step": 5979
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45452394670234764,
      "learning_rate": 1.8726617096791137e-05,
      "loss": 0.6513,
      "step": 5980
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4415710889267077,
      "learning_rate": 1.8720836880071593e-05,
      "loss": 0.649,
      "step": 5981
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34802964322902946,
      "learning_rate": 1.8715056770633916e-05,
      "loss": 0.6814,
      "step": 5982
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3724794652001299,
      "learning_rate": 1.8709276768962878e-05,
      "loss": 0.6127,
      "step": 5983
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38603810506271385,
      "learning_rate": 1.870349687554323e-05,
      "loss": 0.6652,
      "step": 5984
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3966880573966789,
      "learning_rate": 1.8697717090859736e-05,
      "loss": 0.6598,
      "step": 5985
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.418844207076819,
      "learning_rate": 1.869193741539714e-05,
      "loss": 0.6701,
      "step": 5986
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39653325897469827,
      "learning_rate": 1.8686157849640175e-05,
      "loss": 0.6718,
      "step": 5987
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41411889009792696,
      "learning_rate": 1.868037839407356e-05,
      "loss": 0.6654,
      "step": 5988
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3675391997341278,
      "learning_rate": 1.867459904918202e-05,
      "loss": 0.6328,
      "step": 5989
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4124770771752708,
      "learning_rate": 1.8668819815450255e-05,
      "loss": 0.6562,
      "step": 5990
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3606416702283165,
      "learning_rate": 1.8663040693362963e-05,
      "loss": 0.6487,
      "step": 5991
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41385726250828453,
      "learning_rate": 1.865726168340483e-05,
      "loss": 0.6732,
      "step": 5992
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4201560488651027,
      "learning_rate": 1.865148278606054e-05,
      "loss": 0.6759,
      "step": 5993
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40133404365199343,
      "learning_rate": 1.8645704001814766e-05,
      "loss": 0.6761,
      "step": 5994
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4285666769938302,
      "learning_rate": 1.8639925331152158e-05,
      "loss": 0.6745,
      "step": 5995
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42092364579760294,
      "learning_rate": 1.863414677455737e-05,
      "loss": 0.6873,
      "step": 5996
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41904239362892665,
      "learning_rate": 1.8628368332515047e-05,
      "loss": 0.6343,
      "step": 5997
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3937717916720544,
      "learning_rate": 1.862259000550981e-05,
      "loss": 0.6947,
      "step": 5998
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3577638824324315,
      "learning_rate": 1.861681179402628e-05,
      "loss": 0.6938,
      "step": 5999
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.436947237981185,
      "learning_rate": 1.8611033698549083e-05,
      "loss": 0.632,
      "step": 6000
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3946652261323811,
      "learning_rate": 1.8605255719562813e-05,
      "loss": 0.6507,
      "step": 6001
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3997002254656954,
      "learning_rate": 1.8599477857552057e-05,
      "loss": 0.6495,
      "step": 6002
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3808894026035726,
      "learning_rate": 1.8593700113001406e-05,
      "loss": 0.6925,
      "step": 6003
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.45078491134016124,
      "learning_rate": 1.8587922486395428e-05,
      "loss": 0.7,
      "step": 6004
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4286814464800725,
      "learning_rate": 1.8582144978218683e-05,
      "loss": 0.6731,
      "step": 6005
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38834528630396814,
      "learning_rate": 1.857636758895572e-05,
      "loss": 0.6493,
      "step": 6006
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3276697031549784,
      "learning_rate": 1.8570590319091095e-05,
      "loss": 0.6663,
      "step": 6007
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37788059842444816,
      "learning_rate": 1.8564813169109336e-05,
      "loss": 0.6783,
      "step": 6008
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3803680998099298,
      "learning_rate": 1.855903613949496e-05,
      "loss": 0.688,
      "step": 6009
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.348811682127278,
      "learning_rate": 1.8553259230732485e-05,
      "loss": 0.6306,
      "step": 6010
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.401425770699158,
      "learning_rate": 1.8547482443306408e-05,
      "loss": 0.6497,
      "step": 6011
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37907561843733334,
      "learning_rate": 1.854170577770123e-05,
      "loss": 0.6713,
      "step": 6012
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4019089087775945,
      "learning_rate": 1.8535929234401412e-05,
      "loss": 0.6513,
      "step": 6013
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4029441095223109,
      "learning_rate": 1.8530152813891455e-05,
      "loss": 0.6598,
      "step": 6014
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.43289530694906053,
      "learning_rate": 1.8524376516655803e-05,
      "loss": 0.6978,
      "step": 6015
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3964233513314269,
      "learning_rate": 1.851860034317891e-05,
      "loss": 0.6564,
      "step": 6016
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37116323788390865,
      "learning_rate": 1.8512824293945223e-05,
      "loss": 0.6327,
      "step": 6017
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37081793494113835,
      "learning_rate": 1.850704836943916e-05,
      "loss": 0.6692,
      "step": 6018
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34585345111401045,
      "learning_rate": 1.850127257014515e-05,
      "loss": 0.6573,
      "step": 6019
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3985679736248795,
      "learning_rate": 1.8495496896547595e-05,
      "loss": 0.6714,
      "step": 6020
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40578711853382515,
      "learning_rate": 1.8489721349130902e-05,
      "loss": 0.6572,
      "step": 6021
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3664625885918712,
      "learning_rate": 1.848394592837945e-05,
      "loss": 0.6284,
      "step": 6022
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37708015535254863,
      "learning_rate": 1.847817063477763e-05,
      "loss": 0.6534,
      "step": 6023
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36780495623757836,
      "learning_rate": 1.84723954688098e-05,
      "loss": 0.6829,
      "step": 6024
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39947411672482613,
      "learning_rate": 1.8466620430960316e-05,
      "loss": 0.6793,
      "step": 6025
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4148588806770042,
      "learning_rate": 1.8460845521713523e-05,
      "loss": 0.6563,
      "step": 6026
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3894403669305272,
      "learning_rate": 1.8455070741553754e-05,
      "loss": 0.6764,
      "step": 6027
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3902343197883742,
      "learning_rate": 1.8449296090965344e-05,
      "loss": 0.6128,
      "step": 6028
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3669939215633856,
      "learning_rate": 1.844352157043259e-05,
      "loss": 0.6367,
      "step": 6029
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3950576398478754,
      "learning_rate": 1.843774718043981e-05,
      "loss": 0.6973,
      "step": 6030
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41437133155751227,
      "learning_rate": 1.8431972921471283e-05,
      "loss": 0.6654,
      "step": 6031
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3588579640488357,
      "learning_rate": 1.8426198794011295e-05,
      "loss": 0.6747,
      "step": 6032
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3722118668816425,
      "learning_rate": 1.8420424798544112e-05,
      "loss": 0.6251,
      "step": 6033
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37174710709272674,
      "learning_rate": 1.8414650935553987e-05,
      "loss": 0.6645,
      "step": 6034
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4138850740128858,
      "learning_rate": 1.8408877205525185e-05,
      "loss": 0.6876,
      "step": 6035
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4227197905738399,
      "learning_rate": 1.8403103608941925e-05,
      "loss": 0.666,
      "step": 6036
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3657717278882233,
      "learning_rate": 1.8397330146288437e-05,
      "loss": 0.6523,
      "step": 6037
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3942683631259372,
      "learning_rate": 1.8391556818048938e-05,
      "loss": 0.6862,
      "step": 6038
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3791297831179583,
      "learning_rate": 1.838578362470762e-05,
      "loss": 0.6286,
      "step": 6039
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3561452075290325,
      "learning_rate": 1.8380010566748685e-05,
      "loss": 0.6733,
      "step": 6040
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36230582806024164,
      "learning_rate": 1.8374237644656297e-05,
      "loss": 0.6902,
      "step": 6041
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3891724159433369,
      "learning_rate": 1.8368464858914645e-05,
      "loss": 0.6879,
      "step": 6042
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38416471677017583,
      "learning_rate": 1.836269221000787e-05,
      "loss": 0.639,
      "step": 6043
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3610684325167674,
      "learning_rate": 1.8356919698420125e-05,
      "loss": 0.667,
      "step": 6044
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3837789895236173,
      "learning_rate": 1.8351147324635538e-05,
      "loss": 0.6727,
      "step": 6045
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39683530859887695,
      "learning_rate": 1.8345375089138238e-05,
      "loss": 0.65,
      "step": 6046
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38897176063375954,
      "learning_rate": 1.833960299241233e-05,
      "loss": 0.6784,
      "step": 6047
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42759106213518344,
      "learning_rate": 1.8333831034941907e-05,
      "loss": 0.7294,
      "step": 6048
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34513670144348674,
      "learning_rate": 1.8328059217211065e-05,
      "loss": 0.6331,
      "step": 6049
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3534395224593825,
      "learning_rate": 1.832228753970388e-05,
      "loss": 0.6554,
      "step": 6050
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37728810841816474,
      "learning_rate": 1.8316516002904415e-05,
      "loss": 0.6944,
      "step": 6051
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3609673448546752,
      "learning_rate": 1.831074460729672e-05,
      "loss": 0.6602,
      "step": 6052
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37369442988805707,
      "learning_rate": 1.8304973353364833e-05,
      "loss": 0.6878,
      "step": 6053
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3912966532153092,
      "learning_rate": 1.8299202241592785e-05,
      "loss": 0.6809,
      "step": 6054
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4053796982924857,
      "learning_rate": 1.829343127246458e-05,
      "loss": 0.6881,
      "step": 6055
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3766723561592657,
      "learning_rate": 1.828766044646424e-05,
      "loss": 0.6977,
      "step": 6056
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3633421053011105,
      "learning_rate": 1.828188976407576e-05,
      "loss": 0.64,
      "step": 6057
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36064648792458587,
      "learning_rate": 1.82761192257831e-05,
      "loss": 0.6359,
      "step": 6058
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40230627728255236,
      "learning_rate": 1.827034883207024e-05,
      "loss": 0.6848,
      "step": 6059
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37165761178538226,
      "learning_rate": 1.826457858342114e-05,
      "loss": 0.7056,
      "step": 6060
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3500915942240148,
      "learning_rate": 1.825880848031973e-05,
      "loss": 0.6521,
      "step": 6061
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38935110183270677,
      "learning_rate": 1.8253038523249955e-05,
      "loss": 0.6902,
      "step": 6062
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33203660867391666,
      "learning_rate": 1.8247268712695727e-05,
      "loss": 0.6297,
      "step": 6063
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3784179118149076,
      "learning_rate": 1.824149904914096e-05,
      "loss": 0.6752,
      "step": 6064
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33411684176188067,
      "learning_rate": 1.8235729533069545e-05,
      "loss": 0.6221,
      "step": 6065
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32421019129388956,
      "learning_rate": 1.8229960164965356e-05,
      "loss": 0.6565,
      "step": 6066
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38400764636853435,
      "learning_rate": 1.8224190945312276e-05,
      "loss": 0.6869,
      "step": 6067
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36727112641645565,
      "learning_rate": 1.8218421874594148e-05,
      "loss": 0.6705,
      "step": 6068
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4086195045501173,
      "learning_rate": 1.8212652953294834e-05,
      "loss": 0.6627,
      "step": 6069
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35897727840451404,
      "learning_rate": 1.8206884181898153e-05,
      "loss": 0.6247,
      "step": 6070
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.38990077829996456,
      "learning_rate": 1.820111556088794e-05,
      "loss": 0.6872,
      "step": 6071
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37543449125879813,
      "learning_rate": 1.819534709074798e-05,
      "loss": 0.6349,
      "step": 6072
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3655640974289323,
      "learning_rate": 1.818957877196209e-05,
      "loss": 0.6905,
      "step": 6073
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39996894039366543,
      "learning_rate": 1.818381060501404e-05,
      "loss": 0.6447,
      "step": 6074
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39484953832059183,
      "learning_rate": 1.817804259038759e-05,
      "loss": 0.6585,
      "step": 6075
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3981493602840331,
      "learning_rate": 1.817227472856652e-05,
      "loss": 0.6304,
      "step": 6076
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4742639879559945,
      "learning_rate": 1.8166507020034557e-05,
      "loss": 0.6871,
      "step": 6077
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3485674655145832,
      "learning_rate": 1.8160739465275442e-05,
      "loss": 0.6302,
      "step": 6078
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33708311048156087,
      "learning_rate": 1.815497206477288e-05,
      "loss": 0.6596,
      "step": 6079
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3981461812109724,
      "learning_rate": 1.8149204819010587e-05,
      "loss": 0.669,
      "step": 6080
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3922330030938326,
      "learning_rate": 1.814343772847225e-05,
      "loss": 0.6898,
      "step": 6081
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3933465374900603,
      "learning_rate": 1.813767079364154e-05,
      "loss": 0.6846,
      "step": 6082
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3873765462698859,
      "learning_rate": 1.8131904015002135e-05,
      "loss": 0.6496,
      "step": 6083
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.43607166969637967,
      "learning_rate": 1.8126137393037687e-05,
      "loss": 0.6912,
      "step": 6084
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3830007997687013,
      "learning_rate": 1.8120370928231838e-05,
      "loss": 0.7069,
      "step": 6085
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3952427574405216,
      "learning_rate": 1.81146046210682e-05,
      "loss": 0.6702,
      "step": 6086
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42907484738926693,
      "learning_rate": 1.8108838472030403e-05,
      "loss": 0.666,
      "step": 6087
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3970005499595577,
      "learning_rate": 1.810307248160204e-05,
      "loss": 0.665,
      "step": 6088
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40850643264236586,
      "learning_rate": 1.8097306650266685e-05,
      "loss": 0.6622,
      "step": 6089
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35513056726844566,
      "learning_rate": 1.809154097850793e-05,
      "loss": 0.6779,
      "step": 6090
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3784938014002365,
      "learning_rate": 1.808577546680933e-05,
      "loss": 0.6508,
      "step": 6091
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34795817869004253,
      "learning_rate": 1.8080010115654435e-05,
      "loss": 0.6453,
      "step": 6092
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4149312152021158,
      "learning_rate": 1.8074244925526766e-05,
      "loss": 0.6823,
      "step": 6093
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3701101236279939,
      "learning_rate": 1.8068479896909853e-05,
      "loss": 0.6574,
      "step": 6094
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.37755381180394515,
      "learning_rate": 1.8062715030287198e-05,
      "loss": 0.6394,
      "step": 6095
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3488022912584285,
      "learning_rate": 1.8056950326142284e-05,
      "loss": 0.6446,
      "step": 6096
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33013572046938156,
      "learning_rate": 1.805118578495861e-05,
      "loss": 0.6516,
      "step": 6097
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37436556028227824,
      "learning_rate": 1.8045421407219627e-05,
      "loss": 0.644,
      "step": 6098
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3484793478913938,
      "learning_rate": 1.8039657193408788e-05,
      "loss": 0.6495,
      "step": 6099
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38487715121355714,
      "learning_rate": 1.803389314400954e-05,
      "loss": 0.6907,
      "step": 6100
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33801046491000797,
      "learning_rate": 1.8028129259505295e-05,
      "loss": 0.65,
      "step": 6101
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41720535293839356,
      "learning_rate": 1.8022365540379468e-05,
      "loss": 0.6911,
      "step": 6102
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3632563190920977,
      "learning_rate": 1.8016601987115448e-05,
      "loss": 0.6677,
      "step": 6103
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4152715704583654,
      "learning_rate": 1.801083860019663e-05,
      "loss": 0.6587,
      "step": 6104
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34851230392631766,
      "learning_rate": 1.8005075380106384e-05,
      "loss": 0.6624,
      "step": 6105
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3801005851741569,
      "learning_rate": 1.7999312327328048e-05,
      "loss": 0.6924,
      "step": 6106
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33383832224825544,
      "learning_rate": 1.7993549442344973e-05,
      "loss": 0.649,
      "step": 6107
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3660019245622059,
      "learning_rate": 1.7987786725640484e-05,
      "loss": 0.6758,
      "step": 6108
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36220072525358604,
      "learning_rate": 1.798202417769789e-05,
      "loss": 0.6553,
      "step": 6109
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3656473943387404,
      "learning_rate": 1.7976261799000486e-05,
      "loss": 0.67,
      "step": 6110
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38106120970458124,
      "learning_rate": 1.7970499590031568e-05,
      "loss": 0.6714,
      "step": 6111
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36724114600323116,
      "learning_rate": 1.79647375512744e-05,
      "loss": 0.6655,
      "step": 6112
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4090498596859173,
      "learning_rate": 1.795897568321223e-05,
      "loss": 0.6773,
      "step": 6113
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3419712939210981,
      "learning_rate": 1.795321398632831e-05,
      "loss": 0.6416,
      "step": 6114
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.366944086846295,
      "learning_rate": 1.794745246110586e-05,
      "loss": 0.6665,
      "step": 6115
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3443298020601596,
      "learning_rate": 1.794169110802809e-05,
      "loss": 0.6458,
      "step": 6116
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34345812956091193,
      "learning_rate": 1.7935929927578196e-05,
      "loss": 0.661,
      "step": 6117
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38915573672131676,
      "learning_rate": 1.7930168920239375e-05,
      "loss": 0.6669,
      "step": 6118
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40343816110608055,
      "learning_rate": 1.7924408086494786e-05,
      "loss": 0.717,
      "step": 6119
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38721168132890316,
      "learning_rate": 1.7918647426827583e-05,
      "loss": 0.6637,
      "step": 6120
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37445298983952907,
      "learning_rate": 1.7912886941720908e-05,
      "loss": 0.6391,
      "step": 6121
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38477099021792044,
      "learning_rate": 1.7907126631657884e-05,
      "loss": 0.6778,
      "step": 6122
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36110436519162825,
      "learning_rate": 1.790136649712162e-05,
      "loss": 0.6412,
      "step": 6123
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3501527385314313,
      "learning_rate": 1.789560653859521e-05,
      "loss": 0.6586,
      "step": 6124
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4410568164395181,
      "learning_rate": 1.7889846756561743e-05,
      "loss": 0.7308,
      "step": 6125
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3720713888739694,
      "learning_rate": 1.7884087151504282e-05,
      "loss": 0.671,
      "step": 6126
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3469535961677806,
      "learning_rate": 1.787832772390587e-05,
      "loss": 0.6062,
      "step": 6127
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4162622187875775,
      "learning_rate": 1.7872568474249557e-05,
      "loss": 0.6833,
      "step": 6128
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3603268638750935,
      "learning_rate": 1.7866809403018352e-05,
      "loss": 0.6397,
      "step": 6129
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36390782163351176,
      "learning_rate": 1.7861050510695267e-05,
      "loss": 0.6923,
      "step": 6130
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3698776631060797,
      "learning_rate": 1.7855291797763287e-05,
      "loss": 0.6442,
      "step": 6131
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3616372964456043,
      "learning_rate": 1.7849533264705402e-05,
      "loss": 0.6657,
      "step": 6132
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.397073312196676,
      "learning_rate": 1.784377491200456e-05,
      "loss": 0.6725,
      "step": 6133
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.35679756571131227,
      "learning_rate": 1.7838016740143716e-05,
      "loss": 0.6594,
      "step": 6134
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37437906915824903,
      "learning_rate": 1.78322587496058e-05,
      "loss": 0.6512,
      "step": 6135
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39523273292671457,
      "learning_rate": 1.7826500940873718e-05,
      "loss": 0.6752,
      "step": 6136
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4036181306806692,
      "learning_rate": 1.782074331443038e-05,
      "loss": 0.6424,
      "step": 6137
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3534019107408713,
      "learning_rate": 1.7814985870758662e-05,
      "loss": 0.631,
      "step": 6138
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4132218784456245,
      "learning_rate": 1.780922861034145e-05,
      "loss": 0.666,
      "step": 6139
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3787057903420683,
      "learning_rate": 1.780347153366158e-05,
      "loss": 0.6914,
      "step": 6140
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3884796239554781,
      "learning_rate": 1.7797714641201907e-05,
      "loss": 0.6574,
      "step": 6141
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3682341239449366,
      "learning_rate": 1.7791957933445247e-05,
      "loss": 0.6421,
      "step": 6142
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39374502547937035,
      "learning_rate": 1.7786201410874404e-05,
      "loss": 0.7075,
      "step": 6143
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4066489974323301,
      "learning_rate": 1.778044507397217e-05,
      "loss": 0.6692,
      "step": 6144
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.407110211128882,
      "learning_rate": 1.7774688923221336e-05,
      "loss": 0.6832,
      "step": 6145
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40473735342265377,
      "learning_rate": 1.7768932959104655e-05,
      "loss": 0.6677,
      "step": 6146
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4158426825697275,
      "learning_rate": 1.776317718210487e-05,
      "loss": 0.6667,
      "step": 6147
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7496223181178325,
      "learning_rate": 1.7757421592704716e-05,
      "loss": 0.4056,
      "step": 6148
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3668563611923282,
      "learning_rate": 1.7751666191386903e-05,
      "loss": 0.6381,
      "step": 6149
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3514509434303357,
      "learning_rate": 1.7745910978634128e-05,
      "loss": 0.6846,
      "step": 6150
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3645304348676136,
      "learning_rate": 1.7740155954929075e-05,
      "loss": 0.6775,
      "step": 6151
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36425472486329347,
      "learning_rate": 1.773440112075442e-05,
      "loss": 0.6416,
      "step": 6152
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3671454729665971,
      "learning_rate": 1.7728646476592802e-05,
      "loss": 0.6709,
      "step": 6153
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.374642773696499,
      "learning_rate": 1.7722892022926862e-05,
      "loss": 0.6511,
      "step": 6154
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3583316044933307,
      "learning_rate": 1.7717137760239224e-05,
      "loss": 0.6494,
      "step": 6155
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39448885272271134,
      "learning_rate": 1.7711383689012474e-05,
      "loss": 0.6611,
      "step": 6156
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39029146488489336,
      "learning_rate": 1.770562980972922e-05,
      "loss": 0.6906,
      "step": 6157
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4002916155881407,
      "learning_rate": 1.7699876122872013e-05,
      "loss": 0.6606,
      "step": 6158
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34795269091093783,
      "learning_rate": 1.7694122628923428e-05,
      "loss": 0.6497,
      "step": 6159
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36774583509948017,
      "learning_rate": 1.7688369328365986e-05,
      "loss": 0.6347,
      "step": 6160
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36322620850530485,
      "learning_rate": 1.768261622168222e-05,
      "loss": 0.6342,
      "step": 6161
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3693088680779243,
      "learning_rate": 1.7676863309354636e-05,
      "loss": 0.6504,
      "step": 6162
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3384294032186127,
      "learning_rate": 1.7671110591865713e-05,
      "loss": 0.666,
      "step": 6163
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4047892431024787,
      "learning_rate": 1.766535806969794e-05,
      "loss": 0.6859,
      "step": 6164
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4176139978310527,
      "learning_rate": 1.7659605743333756e-05,
      "loss": 0.6793,
      "step": 6165
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3703746544033456,
      "learning_rate": 1.765385361325562e-05,
      "loss": 0.6556,
      "step": 6166
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3622588160497515,
      "learning_rate": 1.7648101679945945e-05,
      "loss": 0.6346,
      "step": 6167
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37654547905328134,
      "learning_rate": 1.7642349943887146e-05,
      "loss": 0.6453,
      "step": 6168
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3749055567188797,
      "learning_rate": 1.763659840556161e-05,
      "loss": 0.6622,
      "step": 6169
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3737128228472733,
      "learning_rate": 1.7630847065451705e-05,
      "loss": 0.6597,
      "step": 6170
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3521986411587532,
      "learning_rate": 1.76250959240398e-05,
      "loss": 0.6388,
      "step": 6171
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34677535017296496,
      "learning_rate": 1.7619344981808225e-05,
      "loss": 0.6676,
      "step": 6172
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3584173688448504,
      "learning_rate": 1.7613594239239317e-05,
      "loss": 0.6278,
      "step": 6173
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37875554565345493,
      "learning_rate": 1.7607843696815376e-05,
      "loss": 0.6967,
      "step": 6174
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.375805926522617,
      "learning_rate": 1.76020933550187e-05,
      "loss": 0.6707,
      "step": 6175
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3939148057136871,
      "learning_rate": 1.7596343214331557e-05,
      "loss": 0.7174,
      "step": 6176
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41672860596631295,
      "learning_rate": 1.7590593275236202e-05,
      "loss": 0.6803,
      "step": 6177
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3907140781478671,
      "learning_rate": 1.7584843538214885e-05,
      "loss": 0.6883,
      "step": 6178
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3262630230469612,
      "learning_rate": 1.7579094003749814e-05,
      "loss": 0.6046,
      "step": 6179
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3425358861812898,
      "learning_rate": 1.757334467232321e-05,
      "loss": 0.6456,
      "step": 6180
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3643161368162634,
      "learning_rate": 1.7567595544417258e-05,
      "loss": 0.6794,
      "step": 6181
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4351047096330197,
      "learning_rate": 1.7561846620514132e-05,
      "loss": 0.6963,
      "step": 6182
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42912764998208786,
      "learning_rate": 1.755609790109599e-05,
      "loss": 0.7054,
      "step": 6183
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3835160625147012,
      "learning_rate": 1.7550349386644954e-05,
      "loss": 0.6943,
      "step": 6184
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3499464763352683,
      "learning_rate": 1.7544601077643167e-05,
      "loss": 0.6417,
      "step": 6185
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42297892300386175,
      "learning_rate": 1.7538852974572712e-05,
      "loss": 0.6737,
      "step": 6186
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39885051217639356,
      "learning_rate": 1.7533105077915695e-05,
      "loss": 0.6573,
      "step": 6187
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4061721725075757,
      "learning_rate": 1.7527357388154166e-05,
      "loss": 0.7028,
      "step": 6188
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6988784997529006,
      "learning_rate": 1.7521609905770196e-05,
      "loss": 0.4364,
      "step": 6189
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.35321244946219377,
      "learning_rate": 1.7515862631245807e-05,
      "loss": 0.6502,
      "step": 6190
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38618276824556474,
      "learning_rate": 1.751011556506302e-05,
      "loss": 0.665,
      "step": 6191
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3810719346882352,
      "learning_rate": 1.7504368707703834e-05,
      "loss": 0.69,
      "step": 6192
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31238640347137675,
      "learning_rate": 1.7498622059650224e-05,
      "loss": 0.6344,
      "step": 6193
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.390891127919515,
      "learning_rate": 1.749287562138416e-05,
      "loss": 0.7034,
      "step": 6194
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42949670031586856,
      "learning_rate": 1.74871293933876e-05,
      "loss": 0.661,
      "step": 6195
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.348906197337217,
      "learning_rate": 1.748138337614246e-05,
      "loss": 0.6374,
      "step": 6196
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37732326030413216,
      "learning_rate": 1.7475637570130654e-05,
      "loss": 0.6403,
      "step": 6197
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3561199299328831,
      "learning_rate": 1.7469891975834077e-05,
      "loss": 0.6504,
      "step": 6198
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38359024843736694,
      "learning_rate": 1.7464146593734606e-05,
      "loss": 0.675,
      "step": 6199
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40594652798755065,
      "learning_rate": 1.7458401424314085e-05,
      "loss": 0.6819,
      "step": 6200
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3579124852152324,
      "learning_rate": 1.7452656468054377e-05,
      "loss": 0.6214,
      "step": 6201
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41862065358183564,
      "learning_rate": 1.7446911725437296e-05,
      "loss": 0.6848,
      "step": 6202
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3866038566225923,
      "learning_rate": 1.7441167196944645e-05,
      "loss": 0.6554,
      "step": 6203
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3821517442504272,
      "learning_rate": 1.743542288305821e-05,
      "loss": 0.6906,
      "step": 6204
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41255992813437,
      "learning_rate": 1.742967878425976e-05,
      "loss": 0.6815,
      "step": 6205
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3397379781808885,
      "learning_rate": 1.7423934901031048e-05,
      "loss": 0.6657,
      "step": 6206
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3452597430509131,
      "learning_rate": 1.7418191233853797e-05,
      "loss": 0.6557,
      "step": 6207
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3258167944959985,
      "learning_rate": 1.741244778320973e-05,
      "loss": 0.6571,
      "step": 6208
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37159440599254057,
      "learning_rate": 1.7406704549580555e-05,
      "loss": 0.6758,
      "step": 6209
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3913561042016895,
      "learning_rate": 1.7400961533447928e-05,
      "loss": 0.6831,
      "step": 6210
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.40833231610829995,
      "learning_rate": 1.739521873529352e-05,
      "loss": 0.6177,
      "step": 6211
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3889248963642968,
      "learning_rate": 1.7389476155598974e-05,
      "loss": 0.6258,
      "step": 6212
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3707706771866988,
      "learning_rate": 1.7383733794845904e-05,
      "loss": 0.6251,
      "step": 6213
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36965002152740395,
      "learning_rate": 1.7377991653515925e-05,
      "loss": 0.6584,
      "step": 6214
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35052586123526314,
      "learning_rate": 1.7372249732090617e-05,
      "loss": 0.6425,
      "step": 6215
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3627284809227986,
      "learning_rate": 1.7366508031051553e-05,
      "loss": 0.6562,
      "step": 6216
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37978421927127654,
      "learning_rate": 1.7360766550880276e-05,
      "loss": 0.6712,
      "step": 6217
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34968997712341304,
      "learning_rate": 1.7355025292058326e-05,
      "loss": 0.6857,
      "step": 6218
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37283676830343015,
      "learning_rate": 1.734928425506721e-05,
      "loss": 0.6513,
      "step": 6219
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36694883117495714,
      "learning_rate": 1.7343543440388423e-05,
      "loss": 0.6565,
      "step": 6220
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37777892274422253,
      "learning_rate": 1.7337802848503428e-05,
      "loss": 0.6979,
      "step": 6221
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39652753960829873,
      "learning_rate": 1.7332062479893705e-05,
      "loss": 0.6739,
      "step": 6222
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.38105264661484234,
      "learning_rate": 1.732632233504068e-05,
      "loss": 0.6534,
      "step": 6223
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37598363031086607,
      "learning_rate": 1.7320582414425766e-05,
      "loss": 0.6835,
      "step": 6224
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37703246085356235,
      "learning_rate": 1.7314842718530376e-05,
      "loss": 0.6582,
      "step": 6225
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35847408266141373,
      "learning_rate": 1.7309103247835882e-05,
      "loss": 0.6554,
      "step": 6226
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3604511729696484,
      "learning_rate": 1.730336400282364e-05,
      "loss": 0.6391,
      "step": 6227
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3818580306847463,
      "learning_rate": 1.7297624983975008e-05,
      "loss": 0.7101,
      "step": 6228
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36687116707275214,
      "learning_rate": 1.729188619177131e-05,
      "loss": 0.6626,
      "step": 6229
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35782345460407966,
      "learning_rate": 1.7286147626693845e-05,
      "loss": 0.6639,
      "step": 6230
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3757447453394121,
      "learning_rate": 1.7280409289223895e-05,
      "loss": 0.6781,
      "step": 6231
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37535863764004107,
      "learning_rate": 1.727467117984274e-05,
      "loss": 0.6923,
      "step": 6232
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4205252245746486,
      "learning_rate": 1.7268933299031623e-05,
      "loss": 0.6799,
      "step": 6233
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7437051348287913,
      "learning_rate": 1.726319564727176e-05,
      "loss": 0.4228,
      "step": 6234
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3543665049556188,
      "learning_rate": 1.725745822504438e-05,
      "loss": 0.6716,
      "step": 6235
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3822450468329461,
      "learning_rate": 1.7251721032830672e-05,
      "loss": 0.6755,
      "step": 6236
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3968930736355358,
      "learning_rate": 1.7245984071111797e-05,
      "loss": 0.6831,
      "step": 6237
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42922543445267336,
      "learning_rate": 1.724024734036891e-05,
      "loss": 0.6993,
      "step": 6238
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4021489983944351,
      "learning_rate": 1.723451084108315e-05,
      "loss": 0.6719,
      "step": 6239
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37467213274131644,
      "learning_rate": 1.7228774573735628e-05,
      "loss": 0.6706,
      "step": 6240
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34548977323041136,
      "learning_rate": 1.7223038538807424e-05,
      "loss": 0.6254,
      "step": 6241
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4170124826489186,
      "learning_rate": 1.721730273677963e-05,
      "loss": 0.7071,
      "step": 6242
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3751234073868669,
      "learning_rate": 1.72115671681333e-05,
      "loss": 0.6571,
      "step": 6243
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36138276672180675,
      "learning_rate": 1.720583183334946e-05,
      "loss": 0.6281,
      "step": 6244
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3417193926407258,
      "learning_rate": 1.720009673290913e-05,
      "loss": 0.6549,
      "step": 6245
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.385782072760506,
      "learning_rate": 1.7194361867293307e-05,
      "loss": 0.6585,
      "step": 6246
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.43712434910810843,
      "learning_rate": 1.7188627236982963e-05,
      "loss": 0.6809,
      "step": 6247
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4209690090468845,
      "learning_rate": 1.7182892842459057e-05,
      "loss": 0.6641,
      "step": 6248
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3821904741983179,
      "learning_rate": 1.7177158684202526e-05,
      "loss": 0.6414,
      "step": 6249
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9130509742510539,
      "learning_rate": 1.7171424762694292e-05,
      "loss": 0.4288,
      "step": 6250
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3645361725181334,
      "learning_rate": 1.716569107841524e-05,
      "loss": 0.6737,
      "step": 6251
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3734646285862689,
      "learning_rate": 1.7159957631846264e-05,
      "loss": 0.6447,
      "step": 6252
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3732908475351696,
      "learning_rate": 1.7154224423468208e-05,
      "loss": 0.6626,
      "step": 6253
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39066782843984416,
      "learning_rate": 1.7148491453761905e-05,
      "loss": 0.6677,
      "step": 6254
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3719332824165103,
      "learning_rate": 1.7142758723208182e-05,
      "loss": 0.6522,
      "step": 6255
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.386109922039753,
      "learning_rate": 1.713702623228784e-05,
      "loss": 0.6233,
      "step": 6256
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37835654744476965,
      "learning_rate": 1.713129398148165e-05,
      "loss": 0.6041,
      "step": 6257
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3584175430524925,
      "learning_rate": 1.7125561971270367e-05,
      "loss": 0.6375,
      "step": 6258
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39790361903019256,
      "learning_rate": 1.711983020213473e-05,
      "loss": 0.6671,
      "step": 6259
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3717818411188811,
      "learning_rate": 1.711409867455546e-05,
      "loss": 0.6697,
      "step": 6260
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3639014323101536,
      "learning_rate": 1.7108367389013244e-05,
      "loss": 0.657,
      "step": 6261
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36341893918754065,
      "learning_rate": 1.7102636345988757e-05,
      "loss": 0.6709,
      "step": 6262
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39545229312489555,
      "learning_rate": 1.7096905545962668e-05,
      "loss": 0.6391,
      "step": 6263
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37550882530619223,
      "learning_rate": 1.7091174989415607e-05,
      "loss": 0.651,
      "step": 6264
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.41211506308017765,
      "learning_rate": 1.7085444676828183e-05,
      "loss": 0.6937,
      "step": 6265
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3679738400465241,
      "learning_rate": 1.7079714608680996e-05,
      "loss": 0.6641,
      "step": 6266
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4035880815260543,
      "learning_rate": 1.707398478545462e-05,
      "loss": 0.7072,
      "step": 6267
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4055340557635398,
      "learning_rate": 1.70682552076296e-05,
      "loss": 0.6936,
      "step": 6268
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37018346390499096,
      "learning_rate": 1.7062525875686472e-05,
      "loss": 0.6283,
      "step": 6269
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.43418832777220445,
      "learning_rate": 1.705679679010576e-05,
      "loss": 0.6979,
      "step": 6270
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35632153622748475,
      "learning_rate": 1.7051067951367945e-05,
      "loss": 0.6626,
      "step": 6271
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.41068233482704625,
      "learning_rate": 1.70453393599535e-05,
      "loss": 0.733,
      "step": 6272
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37799060179712146,
      "learning_rate": 1.7039611016342873e-05,
      "loss": 0.6795,
      "step": 6273
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37073122184595453,
      "learning_rate": 1.7033882921016495e-05,
      "loss": 0.6682,
      "step": 6274
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36216332474411045,
      "learning_rate": 1.7028155074454776e-05,
      "loss": 0.6447,
      "step": 6275
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42745360429462215,
      "learning_rate": 1.7022427477138093e-05,
      "loss": 0.6689,
      "step": 6276
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.41019200735471384,
      "learning_rate": 1.701670012954683e-05,
      "loss": 0.7117,
      "step": 6277
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4534891096200368,
      "learning_rate": 1.7010973032161322e-05,
      "loss": 0.6703,
      "step": 6278
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3854586045681494,
      "learning_rate": 1.70052461854619e-05,
      "loss": 0.6368,
      "step": 6279
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3768333419387794,
      "learning_rate": 1.699951958992886e-05,
      "loss": 0.6585,
      "step": 6280
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.397248227888796,
      "learning_rate": 1.699379324604249e-05,
      "loss": 0.6772,
      "step": 6281
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37628726719487565,
      "learning_rate": 1.698806715428305e-05,
      "loss": 0.649,
      "step": 6282
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3550002335940869,
      "learning_rate": 1.6982341315130772e-05,
      "loss": 0.6381,
      "step": 6283
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.38752691523343574,
      "learning_rate": 1.6976615729065893e-05,
      "loss": 0.6627,
      "step": 6284
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3699211607491397,
      "learning_rate": 1.6970890396568598e-05,
      "loss": 0.6679,
      "step": 6285
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8348823811643804,
      "learning_rate": 1.696516531811907e-05,
      "loss": 0.4292,
      "step": 6286
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36427717941975024,
      "learning_rate": 1.695944049419746e-05,
      "loss": 0.6811,
      "step": 6287
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4180967682823252,
      "learning_rate": 1.6953715925283902e-05,
      "loss": 0.6919,
      "step": 6288
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3943502430812148,
      "learning_rate": 1.6947991611858515e-05,
      "loss": 0.6702,
      "step": 6289
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.377664433701754,
      "learning_rate": 1.694226755440138e-05,
      "loss": 0.6558,
      "step": 6290
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35522354843371734,
      "learning_rate": 1.6936543753392575e-05,
      "loss": 0.6479,
      "step": 6291
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3568131099292656,
      "learning_rate": 1.6930820209312143e-05,
      "loss": 0.6323,
      "step": 6292
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3927877507422774,
      "learning_rate": 1.6925096922640115e-05,
      "loss": 0.636,
      "step": 6293
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.43122470944545865,
      "learning_rate": 1.69193738938565e-05,
      "loss": 0.6931,
      "step": 6294
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3616440550977476,
      "learning_rate": 1.691365112344127e-05,
      "loss": 0.6563,
      "step": 6295
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37144242628856683,
      "learning_rate": 1.6907928611874396e-05,
      "loss": 0.6637,
      "step": 6296
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4001374702090097,
      "learning_rate": 1.6902206359635805e-05,
      "loss": 0.7043,
      "step": 6297
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3447738902809339,
      "learning_rate": 1.6896484367205438e-05,
      "loss": 0.6539,
      "step": 6298
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4231978016213923,
      "learning_rate": 1.6890762635063173e-05,
      "loss": 0.6823,
      "step": 6299
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3803617408915129,
      "learning_rate": 1.6885041163688894e-05,
      "loss": 0.6969,
      "step": 6300
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4244528800710599,
      "learning_rate": 1.6879319953562453e-05,
      "loss": 0.6727,
      "step": 6301
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.354408425802889,
      "learning_rate": 1.6873599005163673e-05,
      "loss": 0.6785,
      "step": 6302
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39391765966767867,
      "learning_rate": 1.6867878318972372e-05,
      "loss": 0.6427,
      "step": 6303
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3797924253923245,
      "learning_rate": 1.6862157895468326e-05,
      "loss": 0.6594,
      "step": 6304
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39378403095757164,
      "learning_rate": 1.6856437735131317e-05,
      "loss": 0.6592,
      "step": 6305
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3961795499325621,
      "learning_rate": 1.6850717838441076e-05,
      "loss": 0.6969,
      "step": 6306
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4016149654278798,
      "learning_rate": 1.6844998205877328e-05,
      "loss": 0.6252,
      "step": 6307
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3856089986024519,
      "learning_rate": 1.6839278837919767e-05,
      "loss": 0.7068,
      "step": 6308
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39372833935243273,
      "learning_rate": 1.683355973504808e-05,
      "loss": 0.6859,
      "step": 6309
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4292837682074259,
      "learning_rate": 1.6827840897741907e-05,
      "loss": 0.6852,
      "step": 6310
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36667133217509845,
      "learning_rate": 1.6822122326480884e-05,
      "loss": 0.634,
      "step": 6311
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4340488710766237,
      "learning_rate": 1.6816404021744625e-05,
      "loss": 0.6607,
      "step": 6312
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4109721091430278,
      "learning_rate": 1.681068598401272e-05,
      "loss": 0.6693,
      "step": 6313
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4220858305548624,
      "learning_rate": 1.680496821376473e-05,
      "loss": 0.6489,
      "step": 6314
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.399059303509175,
      "learning_rate": 1.6799250711480195e-05,
      "loss": 0.6704,
      "step": 6315
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3970637920866404,
      "learning_rate": 1.6793533477638638e-05,
      "loss": 0.6646,
      "step": 6316
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3468363867012377,
      "learning_rate": 1.6787816512719552e-05,
      "loss": 0.671,
      "step": 6317
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.40361282392992315,
      "learning_rate": 1.6782099817202423e-05,
      "loss": 0.6572,
      "step": 6318
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.382570342957086,
      "learning_rate": 1.677638339156669e-05,
      "loss": 0.7042,
      "step": 6319
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4140191032808364,
      "learning_rate": 1.6770667236291792e-05,
      "loss": 0.7048,
      "step": 6320
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.41128114508340624,
      "learning_rate": 1.6764951351857136e-05,
      "loss": 0.6632,
      "step": 6321
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3977736524874915,
      "learning_rate": 1.6759235738742095e-05,
      "loss": 0.6843,
      "step": 6322
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33331762499930123,
      "learning_rate": 1.675352039742605e-05,
      "loss": 0.6259,
      "step": 6323
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3930201489916108,
      "learning_rate": 1.6747805328388314e-05,
      "loss": 0.6688,
      "step": 6324
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40695682884611356,
      "learning_rate": 1.6742090532108228e-05,
      "loss": 0.6895,
      "step": 6325
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39721415182424474,
      "learning_rate": 1.673637600906507e-05,
      "loss": 0.6603,
      "step": 6326
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3927988927024034,
      "learning_rate": 1.6730661759738117e-05,
      "loss": 0.6632,
      "step": 6327
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3900803452399001,
      "learning_rate": 1.6724947784606617e-05,
      "loss": 0.6695,
      "step": 6328
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40472699943432505,
      "learning_rate": 1.6719234084149786e-05,
      "loss": 0.6793,
      "step": 6329
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39455836695352076,
      "learning_rate": 1.6713520658846834e-05,
      "loss": 0.6508,
      "step": 6330
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34537555305467477,
      "learning_rate": 1.6707807509176924e-05,
      "loss": 0.6858,
      "step": 6331
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34997168004534807,
      "learning_rate": 1.6702094635619233e-05,
      "loss": 0.6786,
      "step": 6332
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39214498753288796,
      "learning_rate": 1.6696382038652877e-05,
      "loss": 0.6754,
      "step": 6333
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4300619869649855,
      "learning_rate": 1.6690669718756978e-05,
      "loss": 0.6843,
      "step": 6334
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36085161265389676,
      "learning_rate": 1.6684957676410604e-05,
      "loss": 0.6354,
      "step": 6335
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39725572007110743,
      "learning_rate": 1.6679245912092834e-05,
      "loss": 0.6793,
      "step": 6336
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3439183704569091,
      "learning_rate": 1.6673534426282695e-05,
      "loss": 0.634,
      "step": 6337
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35362637684937853,
      "learning_rate": 1.6667823219459202e-05,
      "loss": 0.6109,
      "step": 6338
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3782187604431425,
      "learning_rate": 1.6662112292101352e-05,
      "loss": 0.6395,
      "step": 6339
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4084043626614145,
      "learning_rate": 1.6656401644688118e-05,
      "loss": 0.6296,
      "step": 6340
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.41928259778379223,
      "learning_rate": 1.6650691277698442e-05,
      "loss": 0.6861,
      "step": 6341
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33539689902307407,
      "learning_rate": 1.664498119161124e-05,
      "loss": 0.6813,
      "step": 6342
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32992541656481583,
      "learning_rate": 1.6639271386905415e-05,
      "loss": 0.6366,
      "step": 6343
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38612525778604256,
      "learning_rate": 1.663356186405984e-05,
      "loss": 0.6516,
      "step": 6344
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40844506070116277,
      "learning_rate": 1.662785262355336e-05,
      "loss": 0.6702,
      "step": 6345
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38701358063747104,
      "learning_rate": 1.662214366586481e-05,
      "loss": 0.6225,
      "step": 6346
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3984968702431914,
      "learning_rate": 1.6616434991472995e-05,
      "loss": 0.6793,
      "step": 6347
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37789216426161026,
      "learning_rate": 1.6610726600856693e-05,
      "loss": 0.6883,
      "step": 6348
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3993861567626901,
      "learning_rate": 1.660501849449465e-05,
      "loss": 0.6416,
      "step": 6349
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37119051861051305,
      "learning_rate": 1.6599310672865615e-05,
      "loss": 0.6704,
      "step": 6350
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34160889029346747,
      "learning_rate": 1.6593603136448285e-05,
      "loss": 0.6796,
      "step": 6351
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46623691776154297,
      "learning_rate": 1.6587895885721334e-05,
      "loss": 0.7232,
      "step": 6352
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3604335404875464,
      "learning_rate": 1.6582188921163444e-05,
      "loss": 0.6789,
      "step": 6353
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3901625106016861,
      "learning_rate": 1.6576482243253243e-05,
      "loss": 0.6745,
      "step": 6354
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3361401037726267,
      "learning_rate": 1.657077585246934e-05,
      "loss": 0.6354,
      "step": 6355
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4060130973197448,
      "learning_rate": 1.6565069749290323e-05,
      "loss": 0.6941,
      "step": 6356
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3505247985310733,
      "learning_rate": 1.6559363934194763e-05,
      "loss": 0.6204,
      "step": 6357
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38378767741649694,
      "learning_rate": 1.655365840766119e-05,
      "loss": 0.647,
      "step": 6358
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36945946706923355,
      "learning_rate": 1.654795317016812e-05,
      "loss": 0.6306,
      "step": 6359
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38316801613753765,
      "learning_rate": 1.6542248222194052e-05,
      "loss": 0.6374,
      "step": 6360
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.43623419920675444,
      "learning_rate": 1.6536543564217456e-05,
      "loss": 0.7097,
      "step": 6361
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.424432581600455,
      "learning_rate": 1.6530839196716762e-05,
      "loss": 0.6832,
      "step": 6362
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4094587595180036,
      "learning_rate": 1.65251351201704e-05,
      "loss": 0.632,
      "step": 6363
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37693683181768256,
      "learning_rate": 1.6519431335056755e-05,
      "loss": 0.6544,
      "step": 6364
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38449600926988386,
      "learning_rate": 1.65137278418542e-05,
      "loss": 0.7195,
      "step": 6365
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42070062098776234,
      "learning_rate": 1.6508024641041078e-05,
      "loss": 0.7044,
      "step": 6366
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3599558183780246,
      "learning_rate": 1.6502321733095713e-05,
      "loss": 0.6646,
      "step": 6367
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3601217072130176,
      "learning_rate": 1.6496619118496405e-05,
      "loss": 0.6442,
      "step": 6368
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42049701561576847,
      "learning_rate": 1.6490916797721415e-05,
      "loss": 0.6623,
      "step": 6369
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3678030403111523,
      "learning_rate": 1.6485214771249002e-05,
      "loss": 0.6658,
      "step": 6370
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40102644956787614,
      "learning_rate": 1.6479513039557383e-05,
      "loss": 0.6633,
      "step": 6371
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3655077631928436,
      "learning_rate": 1.6473811603124746e-05,
      "loss": 0.6341,
      "step": 6372
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3677395161185553,
      "learning_rate": 1.646811046242927e-05,
      "loss": 0.6605,
      "step": 6373
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3612538678500238,
      "learning_rate": 1.646240961794911e-05,
      "loss": 0.6546,
      "step": 6374
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40371683286084603,
      "learning_rate": 1.645670907016238e-05,
      "loss": 0.6635,
      "step": 6375
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38171216207969255,
      "learning_rate": 1.6451008819547183e-05,
      "loss": 0.6637,
      "step": 6376
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38110036951328324,
      "learning_rate": 1.6445308866581592e-05,
      "loss": 0.6526,
      "step": 6377
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39587966505260763,
      "learning_rate": 1.6439609211743653e-05,
      "loss": 0.6367,
      "step": 6378
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4041998825467321,
      "learning_rate": 1.6433909855511384e-05,
      "loss": 0.6893,
      "step": 6379
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.443407316729816,
      "learning_rate": 1.6428210798362787e-05,
      "loss": 0.7324,
      "step": 6380
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4024605535177139,
      "learning_rate": 1.642251204077584e-05,
      "loss": 0.6562,
      "step": 6381
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4244450390889948,
      "learning_rate": 1.6416813583228494e-05,
      "loss": 0.7088,
      "step": 6382
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39747111875881996,
      "learning_rate": 1.641111542619866e-05,
      "loss": 0.6776,
      "step": 6383
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38632999616872266,
      "learning_rate": 1.6405417570164244e-05,
      "loss": 0.6935,
      "step": 6384
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3937054827743806,
      "learning_rate": 1.6399720015603112e-05,
      "loss": 0.6586,
      "step": 6385
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42268720108004054,
      "learning_rate": 1.6394022762993116e-05,
      "loss": 0.6409,
      "step": 6386
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3757851718025145,
      "learning_rate": 1.6388325812812066e-05,
      "loss": 0.6588,
      "step": 6387
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4107886408739257,
      "learning_rate": 1.638262916553778e-05,
      "loss": 0.6609,
      "step": 6388
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4050563623389178,
      "learning_rate": 1.637693282164802e-05,
      "loss": 0.6915,
      "step": 6389
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34898874997085194,
      "learning_rate": 1.637123678162052e-05,
      "loss": 0.6267,
      "step": 6390
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3459379955220398,
      "learning_rate": 1.6365541045933014e-05,
      "loss": 0.6547,
      "step": 6391
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40520676566210845,
      "learning_rate": 1.6359845615063188e-05,
      "loss": 0.6572,
      "step": 6392
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39328626665462413,
      "learning_rate": 1.635415048948872e-05,
      "loss": 0.6437,
      "step": 6393
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34541826102707984,
      "learning_rate": 1.6348455669687235e-05,
      "loss": 0.6624,
      "step": 6394
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3547852154283883,
      "learning_rate": 1.634276115613637e-05,
      "loss": 0.6391,
      "step": 6395
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4401208767180593,
      "learning_rate": 1.633706694931371e-05,
      "loss": 0.6575,
      "step": 6396
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3786327239875456,
      "learning_rate": 1.6331373049696823e-05,
      "loss": 0.6868,
      "step": 6397
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42391369946043056,
      "learning_rate": 1.632567945776325e-05,
      "loss": 0.6351,
      "step": 6398
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34658530647219415,
      "learning_rate": 1.63199861739905e-05,
      "loss": 0.6538,
      "step": 6399
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.43191750421658637,
      "learning_rate": 1.6314293198856056e-05,
      "loss": 0.6706,
      "step": 6400
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3790885375844184,
      "learning_rate": 1.6308600532837404e-05,
      "loss": 0.6778,
      "step": 6401
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38107406822391193,
      "learning_rate": 1.6302908176411963e-05,
      "loss": 0.6572,
      "step": 6402
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3953963463370588,
      "learning_rate": 1.6297216130057146e-05,
      "loss": 0.667,
      "step": 6403
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.41280430757858255,
      "learning_rate": 1.6291524394250347e-05,
      "loss": 0.6925,
      "step": 6404
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38369121647927507,
      "learning_rate": 1.6285832969468917e-05,
      "loss": 0.6698,
      "step": 6405
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3701586759728273,
      "learning_rate": 1.6280141856190182e-05,
      "loss": 0.6622,
      "step": 6406
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38965134010302627,
      "learning_rate": 1.6274451054891456e-05,
      "loss": 0.6871,
      "step": 6407
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3851234811967809,
      "learning_rate": 1.6268760566050028e-05,
      "loss": 0.6985,
      "step": 6408
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36934408757726384,
      "learning_rate": 1.6263070390143147e-05,
      "loss": 0.6318,
      "step": 6409
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33881377907247867,
      "learning_rate": 1.6257380527648034e-05,
      "loss": 0.6485,
      "step": 6410
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35595673986173215,
      "learning_rate": 1.6251690979041902e-05,
      "loss": 0.6352,
      "step": 6411
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39684223336537167,
      "learning_rate": 1.6246001744801918e-05,
      "loss": 0.7134,
      "step": 6412
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38354542048467744,
      "learning_rate": 1.6240312825405232e-05,
      "loss": 0.6725,
      "step": 6413
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3667267237579898,
      "learning_rate": 1.623462422132896e-05,
      "loss": 0.6302,
      "step": 6414
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36193155686549483,
      "learning_rate": 1.6228935933050223e-05,
      "loss": 0.6498,
      "step": 6415
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38736951602356756,
      "learning_rate": 1.622324796104607e-05,
      "loss": 0.6822,
      "step": 6416
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4428525883716987,
      "learning_rate": 1.6217560305793547e-05,
      "loss": 0.6549,
      "step": 6417
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3928433761538122,
      "learning_rate": 1.6211872967769677e-05,
      "loss": 0.7052,
      "step": 6418
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3192806080364808,
      "learning_rate": 1.6206185947451447e-05,
      "loss": 0.6628,
      "step": 6419
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38946951922622225,
      "learning_rate": 1.6200499245315814e-05,
      "loss": 0.6311,
      "step": 6420
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3914986140613158,
      "learning_rate": 1.6194812861839716e-05,
      "loss": 0.6786,
      "step": 6421
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38842328830299955,
      "learning_rate": 1.618912679750008e-05,
      "loss": 0.6866,
      "step": 6422
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38692609995772,
      "learning_rate": 1.6183441052773775e-05,
      "loss": 0.663,
      "step": 6423
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39981559198298483,
      "learning_rate": 1.6177755628137655e-05,
      "loss": 0.6814,
      "step": 6424
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39629303033464947,
      "learning_rate": 1.6172070524068556e-05,
      "loss": 0.6857,
      "step": 6425
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.395402085196888,
      "learning_rate": 1.6166385741043278e-05,
      "loss": 0.6722,
      "step": 6426
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4008785590258158,
      "learning_rate": 1.61607012795386e-05,
      "loss": 0.6712,
      "step": 6427
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35427400368240075,
      "learning_rate": 1.6155017140031263e-05,
      "loss": 0.6482,
      "step": 6428
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3753496509613136,
      "learning_rate": 1.6149333322998005e-05,
      "loss": 0.648,
      "step": 6429
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35998497723926154,
      "learning_rate": 1.6143649828915502e-05,
      "loss": 0.676,
      "step": 6430
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4191593717865704,
      "learning_rate": 1.6137966658260438e-05,
      "loss": 0.6367,
      "step": 6431
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3981601067448385,
      "learning_rate": 1.6132283811509447e-05,
      "loss": 0.6849,
      "step": 6432
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37091708405236623,
      "learning_rate": 1.6126601289139133e-05,
      "loss": 0.6513,
      "step": 6433
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7361041474723661,
      "learning_rate": 1.61209190916261e-05,
      "loss": 0.3699,
      "step": 6434
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37888404854314733,
      "learning_rate": 1.6115237219446886e-05,
      "loss": 0.6815,
      "step": 6435
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39831819737910873,
      "learning_rate": 1.6109555673078044e-05,
      "loss": 0.6948,
      "step": 6436
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4249210439628663,
      "learning_rate": 1.6103874452996065e-05,
      "loss": 0.6637,
      "step": 6437
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3301347305373287,
      "learning_rate": 1.609819355967744e-05,
      "loss": 0.6335,
      "step": 6438
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.681146417317818,
      "learning_rate": 1.60925129935986e-05,
      "loss": 0.3936,
      "step": 6439
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43838294695902646,
      "learning_rate": 1.6086832755235976e-05,
      "loss": 0.6785,
      "step": 6440
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4950516703628113,
      "learning_rate": 1.6081152845065967e-05,
      "loss": 0.633,
      "step": 6441
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40191224047601376,
      "learning_rate": 1.607547326356493e-05,
      "loss": 0.6763,
      "step": 6442
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3753319516730143,
      "learning_rate": 1.6069794011209217e-05,
      "loss": 0.6216,
      "step": 6443
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.42961426290415167,
      "learning_rate": 1.606411508847513e-05,
      "loss": 0.6968,
      "step": 6444
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4525475170394052,
      "learning_rate": 1.605843649583896e-05,
      "loss": 0.698,
      "step": 6445
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4491477551056361,
      "learning_rate": 1.6052758233776965e-05,
      "loss": 0.6562,
      "step": 6446
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4351380873052887,
      "learning_rate": 1.604708030276536e-05,
      "loss": 0.6757,
      "step": 6447
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.352227552139575,
      "learning_rate": 1.6041402703280364e-05,
      "loss": 0.6501,
      "step": 6448
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4157367830110458,
      "learning_rate": 1.6035725435798134e-05,
      "loss": 0.6905,
      "step": 6449
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4497046921569664,
      "learning_rate": 1.6030048500794832e-05,
      "loss": 0.6062,
      "step": 6450
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43855719569403234,
      "learning_rate": 1.602437189874656e-05,
      "loss": 0.6288,
      "step": 6451
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3507222166969037,
      "learning_rate": 1.6018695630129425e-05,
      "loss": 0.6671,
      "step": 6452
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3700378302534637,
      "learning_rate": 1.601301969541947e-05,
      "loss": 0.6948,
      "step": 6453
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44869992670062125,
      "learning_rate": 1.600734409509275e-05,
      "loss": 0.6748,
      "step": 6454
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4331559636875994,
      "learning_rate": 1.6001668829625253e-05,
      "loss": 0.659,
      "step": 6455
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36238869278138885,
      "learning_rate": 1.5995993899492955e-05,
      "loss": 0.6722,
      "step": 6456
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34468446480492265,
      "learning_rate": 1.599031930517182e-05,
      "loss": 0.6433,
      "step": 6457
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36676417771010733,
      "learning_rate": 1.5984645047137762e-05,
      "loss": 0.6636,
      "step": 6458
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43270693459475895,
      "learning_rate": 1.5978971125866675e-05,
      "loss": 0.6674,
      "step": 6459
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3922712611905469,
      "learning_rate": 1.597329754183442e-05,
      "loss": 0.6326,
      "step": 6460
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35921748105970613,
      "learning_rate": 1.5967624295516844e-05,
      "loss": 0.6603,
      "step": 6461
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37132988629982966,
      "learning_rate": 1.5961951387389746e-05,
      "loss": 0.646,
      "step": 6462
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4241176755995649,
      "learning_rate": 1.59562788179289e-05,
      "loss": 0.6775,
      "step": 6463
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4263315685544682,
      "learning_rate": 1.5950606587610072e-05,
      "loss": 0.6871,
      "step": 6464
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40707237244159156,
      "learning_rate": 1.594493469690898e-05,
      "loss": 0.6509,
      "step": 6465
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4210526358201399,
      "learning_rate": 1.593926314630132e-05,
      "loss": 0.7205,
      "step": 6466
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36518263719141486,
      "learning_rate": 1.5933591936262754e-05,
      "loss": 0.662,
      "step": 6467
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3987313685775587,
      "learning_rate": 1.5927921067268922e-05,
      "loss": 0.6741,
      "step": 6468
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37994461109142574,
      "learning_rate": 1.592225053979543e-05,
      "loss": 0.6277,
      "step": 6469
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3892866430647403,
      "learning_rate": 1.5916580354317857e-05,
      "loss": 0.6302,
      "step": 6470
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3388914273472078,
      "learning_rate": 1.591091051131176e-05,
      "loss": 0.6425,
      "step": 6471
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3549833532269894,
      "learning_rate": 1.5905241011252667e-05,
      "loss": 0.6437,
      "step": 6472
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38380085726804425,
      "learning_rate": 1.5899571854616064e-05,
      "loss": 0.6771,
      "step": 6473
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3711632300810791,
      "learning_rate": 1.5893903041877413e-05,
      "loss": 0.669,
      "step": 6474
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3404144793211765,
      "learning_rate": 1.588823457351216e-05,
      "loss": 0.6115,
      "step": 6475
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32654605480221943,
      "learning_rate": 1.5882566449995702e-05,
      "loss": 0.6604,
      "step": 6476
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3321218870542612,
      "learning_rate": 1.587689867180342e-05,
      "loss": 0.6403,
      "step": 6477
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41661162617600983,
      "learning_rate": 1.587123123941067e-05,
      "loss": 0.6681,
      "step": 6478
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3786261429400147,
      "learning_rate": 1.5865564153292774e-05,
      "loss": 0.6621,
      "step": 6479
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38035416168836833,
      "learning_rate": 1.5859897413925017e-05,
      "loss": 0.6699,
      "step": 6480
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.42328995445403855,
      "learning_rate": 1.5854231021782665e-05,
      "loss": 0.6428,
      "step": 6481
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39689414520400423,
      "learning_rate": 1.584856497734095e-05,
      "loss": 0.6705,
      "step": 6482
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0618298556257937,
      "learning_rate": 1.584289928107507e-05,
      "loss": 0.3946,
      "step": 6483
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3722815316331392,
      "learning_rate": 1.5837233933460217e-05,
      "loss": 0.6504,
      "step": 6484
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41954837387649296,
      "learning_rate": 1.583156893497152e-05,
      "loss": 0.6586,
      "step": 6485
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4272330074547714,
      "learning_rate": 1.5825904286084105e-05,
      "loss": 0.6889,
      "step": 6486
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38330179465675124,
      "learning_rate": 1.5820239987273056e-05,
      "loss": 0.6637,
      "step": 6487
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3972046096507086,
      "learning_rate": 1.5814576039013433e-05,
      "loss": 0.7155,
      "step": 6488
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3835055535924871,
      "learning_rate": 1.5808912441780266e-05,
      "loss": 0.684,
      "step": 6489
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3885283598167176,
      "learning_rate": 1.580324919604854e-05,
      "loss": 0.6576,
      "step": 6490
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33284417296107693,
      "learning_rate": 1.579758630229324e-05,
      "loss": 0.6519,
      "step": 6491
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38592235346143644,
      "learning_rate": 1.5791923760989304e-05,
      "loss": 0.6905,
      "step": 6492
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3680836125479079,
      "learning_rate": 1.578626157261164e-05,
      "loss": 0.6581,
      "step": 6493
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37521198677220347,
      "learning_rate": 1.578059973763513e-05,
      "loss": 0.705,
      "step": 6494
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.316747133779362,
      "learning_rate": 1.5774938256534625e-05,
      "loss": 0.657,
      "step": 6495
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36660562113675993,
      "learning_rate": 1.5769277129784945e-05,
      "loss": 0.6196,
      "step": 6496
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35016287637044835,
      "learning_rate": 1.5763616357860876e-05,
      "loss": 0.6395,
      "step": 6497
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4180583476098815,
      "learning_rate": 1.575795594123719e-05,
      "loss": 0.6865,
      "step": 6498
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3555273392237722,
      "learning_rate": 1.575229588038862e-05,
      "loss": 0.6663,
      "step": 6499
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3916753392704066,
      "learning_rate": 1.5746636175789868e-05,
      "loss": 0.6706,
      "step": 6500
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36931722109610093,
      "learning_rate": 1.5740976827915595e-05,
      "loss": 0.6242,
      "step": 6501
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3781189443078932,
      "learning_rate": 1.573531783724046e-05,
      "loss": 0.6169,
      "step": 6502
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3988872375311005,
      "learning_rate": 1.572965920423906e-05,
      "loss": 0.6756,
      "step": 6503
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4288314390721685,
      "learning_rate": 1.5724000929385984e-05,
      "loss": 0.6554,
      "step": 6504
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39251184542013057,
      "learning_rate": 1.5718343013155786e-05,
      "loss": 0.6469,
      "step": 6505
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3946859408828728,
      "learning_rate": 1.5712685456022995e-05,
      "loss": 0.6549,
      "step": 6506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.973168239562627,
      "learning_rate": 1.57070282584621e-05,
      "loss": 0.4388,
      "step": 6507
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4207716208863915,
      "learning_rate": 1.570137142094755e-05,
      "loss": 0.7254,
      "step": 6508
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.47689867927175605,
      "learning_rate": 1.5695714943953796e-05,
      "loss": 0.623,
      "step": 6509
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34239952633928783,
      "learning_rate": 1.5690058827955227e-05,
      "loss": 0.676,
      "step": 6510
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3899153828759465,
      "learning_rate": 1.5684403073426217e-05,
      "loss": 0.6732,
      "step": 6511
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40739136978100154,
      "learning_rate": 1.567874768084111e-05,
      "loss": 0.6519,
      "step": 6512
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.350116378962567,
      "learning_rate": 1.567309265067422e-05,
      "loss": 0.6436,
      "step": 6513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35216372386466727,
      "learning_rate": 1.566743798339982e-05,
      "loss": 0.6558,
      "step": 6514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3608558487885429,
      "learning_rate": 1.5661783679492168e-05,
      "loss": 0.655,
      "step": 6515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4336477036109507,
      "learning_rate": 1.565612973942548e-05,
      "loss": 0.6365,
      "step": 6516
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43293156417290923,
      "learning_rate": 1.565047616367394e-05,
      "loss": 0.6971,
      "step": 6517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37587858103038124,
      "learning_rate": 1.564482295271171e-05,
      "loss": 0.6882,
      "step": 6518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3889278955280302,
      "learning_rate": 1.5639170107012922e-05,
      "loss": 0.6568,
      "step": 6519
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.372854262536517,
      "learning_rate": 1.5633517627051672e-05,
      "loss": 0.6523,
      "step": 6520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4494391529734613,
      "learning_rate": 1.5627865513302024e-05,
      "loss": 0.7092,
      "step": 6521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3956860371260834,
      "learning_rate": 1.5622213766238018e-05,
      "loss": 0.6914,
      "step": 6522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36511339140654775,
      "learning_rate": 1.5616562386333657e-05,
      "loss": 0.6805,
      "step": 6523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37320421699954387,
      "learning_rate": 1.561091137406291e-05,
      "loss": 0.6486,
      "step": 6524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39132920026707646,
      "learning_rate": 1.560526072989972e-05,
      "loss": 0.6351,
      "step": 6525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7072554186008859,
      "learning_rate": 1.5599610454318017e-05,
      "loss": 0.419,
      "step": 6526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35995291796097384,
      "learning_rate": 1.5593960547791668e-05,
      "loss": 0.6245,
      "step": 6527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3553290450811332,
      "learning_rate": 1.558831101079452e-05,
      "loss": 0.6618,
      "step": 6528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34531484650345806,
      "learning_rate": 1.5582661843800408e-05,
      "loss": 0.6643,
      "step": 6529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3168867119575308,
      "learning_rate": 1.5577013047283113e-05,
      "loss": 0.6504,
      "step": 6530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3458510338755475,
      "learning_rate": 1.5571364621716386e-05,
      "loss": 0.6593,
      "step": 6531
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40111690492267815,
      "learning_rate": 1.5565716567573955e-05,
      "loss": 0.6479,
      "step": 6532
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3749202868177191,
      "learning_rate": 1.556006888532953e-05,
      "loss": 0.6432,
      "step": 6533
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36720041689366084,
      "learning_rate": 1.5554421575456765e-05,
      "loss": 0.6604,
      "step": 6534
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3773201470842027,
      "learning_rate": 1.5548774638429286e-05,
      "loss": 0.6345,
      "step": 6535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39793082576581096,
      "learning_rate": 1.554312807472071e-05,
      "loss": 0.6741,
      "step": 6536
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37731855753043997,
      "learning_rate": 1.5537481884804602e-05,
      "loss": 0.6647,
      "step": 6537
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3735810362915109,
      "learning_rate": 1.553183606915449e-05,
      "loss": 0.7265,
      "step": 6538
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3742396356614337,
      "learning_rate": 1.552619062824389e-05,
      "loss": 0.685,
      "step": 6539
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3615622972107379,
      "learning_rate": 1.5520545562546284e-05,
      "loss": 0.6698,
      "step": 6540
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38809236430044664,
      "learning_rate": 1.5514900872535117e-05,
      "loss": 0.6897,
      "step": 6541
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3769629895097341,
      "learning_rate": 1.550925655868379e-05,
      "loss": 0.6563,
      "step": 6542
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4094049818443495,
      "learning_rate": 1.5503612621465696e-05,
      "loss": 0.6688,
      "step": 6543
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4092203812354076,
      "learning_rate": 1.5497969061354176e-05,
      "loss": 0.6592,
      "step": 6544
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38409981260494586,
      "learning_rate": 1.5492325878822557e-05,
      "loss": 0.6487,
      "step": 6545
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3450963095831095,
      "learning_rate": 1.548668307434412e-05,
      "loss": 0.6316,
      "step": 6546
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36706552351513316,
      "learning_rate": 1.5481040648392126e-05,
      "loss": 0.6831,
      "step": 6547
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3705896406740443,
      "learning_rate": 1.5475398601439792e-05,
      "loss": 0.6449,
      "step": 6548
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3922343631277957,
      "learning_rate": 1.5469756933960322e-05,
      "loss": 0.6898,
      "step": 6549
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3677963098003038,
      "learning_rate": 1.546411564642686e-05,
      "loss": 0.6594,
      "step": 6550
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3882349901886048,
      "learning_rate": 1.545847473931254e-05,
      "loss": 0.7207,
      "step": 6551
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3718481317557781,
      "learning_rate": 1.5452834213090462e-05,
      "loss": 0.6526,
      "step": 6552
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3657759471188909,
      "learning_rate": 1.5447194068233678e-05,
      "loss": 0.6797,
      "step": 6553
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37500103524385314,
      "learning_rate": 1.544155430521524e-05,
      "loss": 0.7015,
      "step": 6554
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3809854324285571,
      "learning_rate": 1.543591492450813e-05,
      "loss": 0.653,
      "step": 6555
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3586600938980735,
      "learning_rate": 1.5430275926585328e-05,
      "loss": 0.652,
      "step": 6556
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4256472051190124,
      "learning_rate": 1.5424637311919767e-05,
      "loss": 0.6923,
      "step": 6557
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36088619859355253,
      "learning_rate": 1.5418999080984342e-05,
      "loss": 0.6683,
      "step": 6558
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39970450318567613,
      "learning_rate": 1.541336123425194e-05,
      "loss": 0.6374,
      "step": 6559
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37482183096607485,
      "learning_rate": 1.540772377219538e-05,
      "loss": 0.7033,
      "step": 6560
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3504414754797319,
      "learning_rate": 1.5402086695287494e-05,
      "loss": 0.6198,
      "step": 6561
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37271625841700823,
      "learning_rate": 1.5396450004001032e-05,
      "loss": 0.6893,
      "step": 6562
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3713615792796298,
      "learning_rate": 1.539081369880876e-05,
      "loss": 0.6662,
      "step": 6563
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37952918974786365,
      "learning_rate": 1.5385177780183372e-05,
      "loss": 0.6467,
      "step": 6564
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3235453686618035,
      "learning_rate": 1.537954224859755e-05,
      "loss": 0.6326,
      "step": 6565
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3491570216027613,
      "learning_rate": 1.5373907104523935e-05,
      "loss": 0.6487,
      "step": 6566
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36737273152866645,
      "learning_rate": 1.5368272348435148e-05,
      "loss": 0.6621,
      "step": 6567
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3801507361204886,
      "learning_rate": 1.536263798080377e-05,
      "loss": 0.6834,
      "step": 6568
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3762771857250546,
      "learning_rate": 1.535700400210234e-05,
      "loss": 0.6629,
      "step": 6569
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36930301112541936,
      "learning_rate": 1.535137041280338e-05,
      "loss": 0.6123,
      "step": 6570
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39755350513801846,
      "learning_rate": 1.5345737213379362e-05,
      "loss": 0.6683,
      "step": 6571
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3907157028425302,
      "learning_rate": 1.5340104404302748e-05,
      "loss": 0.664,
      "step": 6572
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4058956544293698,
      "learning_rate": 1.5334471986045946e-05,
      "loss": 0.7001,
      "step": 6573
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3924988952022943,
      "learning_rate": 1.5328839959081347e-05,
      "loss": 0.6453,
      "step": 6574
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41377568894666156,
      "learning_rate": 1.53232083238813e-05,
      "loss": 0.6748,
      "step": 6575
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41210061702266415,
      "learning_rate": 1.5317577080918126e-05,
      "loss": 0.6838,
      "step": 6576
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36701902960920085,
      "learning_rate": 1.5311946230664104e-05,
      "loss": 0.619,
      "step": 6577
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3978703175459812,
      "learning_rate": 1.5306315773591488e-05,
      "loss": 0.6777,
      "step": 6578
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.331168228206264,
      "learning_rate": 1.5300685710172504e-05,
      "loss": 0.6469,
      "step": 6579
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42159239769711565,
      "learning_rate": 1.5295056040879323e-05,
      "loss": 0.6561,
      "step": 6580
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.34043737926024553,
      "learning_rate": 1.528942676618412e-05,
      "loss": 0.6596,
      "step": 6581
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39550860643556796,
      "learning_rate": 1.5283797886559004e-05,
      "loss": 0.7004,
      "step": 6582
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38339277686216383,
      "learning_rate": 1.5278169402476064e-05,
      "loss": 0.6522,
      "step": 6583
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4176590900788083,
      "learning_rate": 1.5272541314407353e-05,
      "loss": 0.6565,
      "step": 6584
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3815524414111862,
      "learning_rate": 1.526691362282489e-05,
      "loss": 0.6356,
      "step": 6585
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4091911699715864,
      "learning_rate": 1.526128632820067e-05,
      "loss": 0.6428,
      "step": 6586
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3790389015355689,
      "learning_rate": 1.5255659431006635e-05,
      "loss": 0.69,
      "step": 6587
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3822936725400608,
      "learning_rate": 1.5250032931714723e-05,
      "loss": 0.696,
      "step": 6588
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35225765641239165,
      "learning_rate": 1.5244406830796805e-05,
      "loss": 0.6678,
      "step": 6589
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42346198399697377,
      "learning_rate": 1.523878112872475e-05,
      "loss": 0.706,
      "step": 6590
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6674910555567962,
      "learning_rate": 1.5233155825970369e-05,
      "loss": 0.3703,
      "step": 6591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3946585149736605,
      "learning_rate": 1.522753092300545e-05,
      "loss": 0.6879,
      "step": 6592
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4050098728321342,
      "learning_rate": 1.522190642030175e-05,
      "loss": 0.6573,
      "step": 6593
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39171812513523596,
      "learning_rate": 1.5216282318330983e-05,
      "loss": 0.6903,
      "step": 6594
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4032700605272319,
      "learning_rate": 1.5210658617564846e-05,
      "loss": 0.686,
      "step": 6595
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.636332809843573,
      "learning_rate": 1.5205035318474986e-05,
      "loss": 0.374,
      "step": 6596
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39984565473521805,
      "learning_rate": 1.5199412421533026e-05,
      "loss": 0.646,
      "step": 6597
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4009679539948833,
      "learning_rate": 1.5193789927210547e-05,
      "loss": 0.6643,
      "step": 6598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3484135042668901,
      "learning_rate": 1.5188167835979099e-05,
      "loss": 0.6436,
      "step": 6599
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32314258795774436,
      "learning_rate": 1.5182546148310209e-05,
      "loss": 0.6707,
      "step": 6600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32837289619223703,
      "learning_rate": 1.5176924864675343e-05,
      "loss": 0.6752,
      "step": 6601
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40452150570942114,
      "learning_rate": 1.5171303985545978e-05,
      "loss": 0.6604,
      "step": 6602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7968944830388781,
      "learning_rate": 1.5165683511393507e-05,
      "loss": 0.4113,
      "step": 6603
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.375190135147312,
      "learning_rate": 1.5160063442689328e-05,
      "loss": 0.6432,
      "step": 6604
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3685520768613683,
      "learning_rate": 1.515444377990478e-05,
      "loss": 0.6355,
      "step": 6605
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3558774221265008,
      "learning_rate": 1.5148824523511182e-05,
      "loss": 0.6726,
      "step": 6606
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36315226960843416,
      "learning_rate": 1.5143205673979813e-05,
      "loss": 0.6563,
      "step": 6607
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3474839553879419,
      "learning_rate": 1.513758723178191e-05,
      "loss": 0.6267,
      "step": 6608
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3707026949860718,
      "learning_rate": 1.5131969197388698e-05,
      "loss": 0.6696,
      "step": 6609
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3579673162946958,
      "learning_rate": 1.5126351571271357e-05,
      "loss": 0.6722,
      "step": 6610
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4270706190942728,
      "learning_rate": 1.5120734353901021e-05,
      "loss": 0.6985,
      "step": 6611
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3842048172470233,
      "learning_rate": 1.5115117545748797e-05,
      "loss": 0.6436,
      "step": 6612
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3625517919144654,
      "learning_rate": 1.5109501147285773e-05,
      "loss": 0.6886,
      "step": 6613
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31709901628247594,
      "learning_rate": 1.5103885158982978e-05,
      "loss": 0.6746,
      "step": 6614
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35780478679005684,
      "learning_rate": 1.5098269581311416e-05,
      "loss": 0.6467,
      "step": 6615
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3390802297837821,
      "learning_rate": 1.5092654414742066e-05,
      "loss": 0.6252,
      "step": 6616
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37485851872684084,
      "learning_rate": 1.5087039659745868e-05,
      "loss": 0.6384,
      "step": 6617
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3415413439752288,
      "learning_rate": 1.5081425316793722e-05,
      "loss": 0.6451,
      "step": 6618
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38220049437993237,
      "learning_rate": 1.507581138635649e-05,
      "loss": 0.6693,
      "step": 6619
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43937013528711844,
      "learning_rate": 1.5070197868905014e-05,
      "loss": 0.7133,
      "step": 6620
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35161519554508947,
      "learning_rate": 1.5064584764910089e-05,
      "loss": 0.6746,
      "step": 6621
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3529654255295701,
      "learning_rate": 1.505897207484247e-05,
      "loss": 0.6481,
      "step": 6622
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33594909345887425,
      "learning_rate": 1.5053359799172902e-05,
      "loss": 0.6525,
      "step": 6623
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3603719863853356,
      "learning_rate": 1.5047747938372078e-05,
      "loss": 0.6491,
      "step": 6624
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3815291328488509,
      "learning_rate": 1.504213649291065e-05,
      "loss": 0.6517,
      "step": 6625
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4361789504730209,
      "learning_rate": 1.5036525463259247e-05,
      "loss": 0.6892,
      "step": 6626
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7497158463500639,
      "learning_rate": 1.503091484988846e-05,
      "loss": 0.3831,
      "step": 6627
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3859059904917893,
      "learning_rate": 1.5025304653268842e-05,
      "loss": 0.686,
      "step": 6628
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36954891485085634,
      "learning_rate": 1.5019694873870912e-05,
      "loss": 0.6535,
      "step": 6629
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3423735630387423,
      "learning_rate": 1.501408551216516e-05,
      "loss": 0.6321,
      "step": 6630
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3919976541617079,
      "learning_rate": 1.5008476568622038e-05,
      "loss": 0.6772,
      "step": 6631
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38875242690256173,
      "learning_rate": 1.5002868043711956e-05,
      "loss": 0.7008,
      "step": 6632
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.376462653247241,
      "learning_rate": 1.4997259937905297e-05,
      "loss": 0.6526,
      "step": 6633
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4020945826237358,
      "learning_rate": 1.4991652251672406e-05,
      "loss": 0.6584,
      "step": 6634
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32714043268496357,
      "learning_rate": 1.498604498548359e-05,
      "loss": 0.6036,
      "step": 6635
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4296445192689355,
      "learning_rate": 1.4980438139809119e-05,
      "loss": 0.6898,
      "step": 6636
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3900441894898729,
      "learning_rate": 1.4974831715119249e-05,
      "loss": 0.6556,
      "step": 6637
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3835733945537995,
      "learning_rate": 1.4969225711884174e-05,
      "loss": 0.6805,
      "step": 6638
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36230308420442486,
      "learning_rate": 1.4963620130574059e-05,
      "loss": 0.6198,
      "step": 6639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7306017118128993,
      "learning_rate": 1.4958014971659046e-05,
      "loss": 0.4022,
      "step": 6640
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3686751985644678,
      "learning_rate": 1.4952410235609228e-05,
      "loss": 0.678,
      "step": 6641
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37539868170625573,
      "learning_rate": 1.494680592289466e-05,
      "loss": 0.6603,
      "step": 6642
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3786072853515829,
      "learning_rate": 1.4941202033985373e-05,
      "loss": 0.6471,
      "step": 6643
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40969788229206977,
      "learning_rate": 1.4935598569351376e-05,
      "loss": 0.6671,
      "step": 6644
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43582185586553446,
      "learning_rate": 1.4929995529462607e-05,
      "loss": 0.6553,
      "step": 6645
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38735249577776476,
      "learning_rate": 1.4924392914788985e-05,
      "loss": 0.6799,
      "step": 6646
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3675633363699539,
      "learning_rate": 1.4918790725800402e-05,
      "loss": 0.6514,
      "step": 6647
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40762811629310575,
      "learning_rate": 1.491318896296671e-05,
      "loss": 0.6574,
      "step": 6648
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3699559405069331,
      "learning_rate": 1.4907587626757707e-05,
      "loss": 0.6716,
      "step": 6649
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36840996517175656,
      "learning_rate": 1.4901986717643178e-05,
      "loss": 0.669,
      "step": 6650
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3621529433267245,
      "learning_rate": 1.4896386236092872e-05,
      "loss": 0.6723,
      "step": 6651
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4118800363375353,
      "learning_rate": 1.4890786182576493e-05,
      "loss": 0.6854,
      "step": 6652
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.380757386026812,
      "learning_rate": 1.4885186557563698e-05,
      "loss": 0.6804,
      "step": 6653
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3621570042900328,
      "learning_rate": 1.4879587361524136e-05,
      "loss": 0.6383,
      "step": 6654
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34355220729393865,
      "learning_rate": 1.4873988594927398e-05,
      "loss": 0.6293,
      "step": 6655
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4029581959595728,
      "learning_rate": 1.4868390258243035e-05,
      "loss": 0.6614,
      "step": 6656
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38411196828209354,
      "learning_rate": 1.4862792351940591e-05,
      "loss": 0.6774,
      "step": 6657
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38833985982221747,
      "learning_rate": 1.485719487648955e-05,
      "loss": 0.7101,
      "step": 6658
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3798804794364572,
      "learning_rate": 1.4851597832359366e-05,
      "loss": 0.6758,
      "step": 6659
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35468159931547694,
      "learning_rate": 1.484600122001945e-05,
      "loss": 0.6368,
      "step": 6660
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3467411007377212,
      "learning_rate": 1.4840405039939189e-05,
      "loss": 0.6503,
      "step": 6661
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3595184020663615,
      "learning_rate": 1.4834809292587923e-05,
      "loss": 0.6698,
      "step": 6662
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6607638013406574,
      "learning_rate": 1.4829213978434965e-05,
      "loss": 0.3963,
      "step": 6663
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3578152910729226,
      "learning_rate": 1.4823619097949584e-05,
      "loss": 0.6415,
      "step": 6664
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3690814677356581,
      "learning_rate": 1.4818024651601022e-05,
      "loss": 0.6538,
      "step": 6665
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3919079002331539,
      "learning_rate": 1.4812430639858471e-05,
      "loss": 0.6604,
      "step": 6666
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3984956909150552,
      "learning_rate": 1.4806837063191102e-05,
      "loss": 0.6706,
      "step": 6667
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4519515954717804,
      "learning_rate": 1.4801243922068036e-05,
      "loss": 0.7209,
      "step": 6668
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38413371432665516,
      "learning_rate": 1.479565121695836e-05,
      "loss": 0.6306,
      "step": 6669
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38919188020330875,
      "learning_rate": 1.4790058948331124e-05,
      "loss": 0.6782,
      "step": 6670
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3416304117627373,
      "learning_rate": 1.4784467116655365e-05,
      "loss": 0.6635,
      "step": 6671
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35454010054208634,
      "learning_rate": 1.4778875722400048e-05,
      "loss": 0.671,
      "step": 6672
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36978122320474344,
      "learning_rate": 1.4773284766034115e-05,
      "loss": 0.6445,
      "step": 6673
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.33742777950693625,
      "learning_rate": 1.4767694248026478e-05,
      "loss": 0.6284,
      "step": 6674
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3565639038076416,
      "learning_rate": 1.476210416884601e-05,
      "loss": 0.6283,
      "step": 6675
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3945766232008895,
      "learning_rate": 1.4756514528961532e-05,
      "loss": 0.6524,
      "step": 6676
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3823449030731492,
      "learning_rate": 1.4750925328841846e-05,
      "loss": 0.669,
      "step": 6677
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3303654285139958,
      "learning_rate": 1.474533656895572e-05,
      "loss": 0.6509,
      "step": 6678
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37936820992065445,
      "learning_rate": 1.4739748249771868e-05,
      "loss": 0.636,
      "step": 6679
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3542285346719276,
      "learning_rate": 1.4734160371758979e-05,
      "loss": 0.6436,
      "step": 6680
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37805015155200106,
      "learning_rate": 1.4728572935385699e-05,
      "loss": 0.6986,
      "step": 6681
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3881353813148153,
      "learning_rate": 1.4722985941120644e-05,
      "loss": 0.6996,
      "step": 6682
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3502061029124956,
      "learning_rate": 1.4717399389432379e-05,
      "loss": 0.6434,
      "step": 6683
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41653975566449397,
      "learning_rate": 1.4711813280789443e-05,
      "loss": 0.6549,
      "step": 6684
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36894020785285603,
      "learning_rate": 1.4706227615660348e-05,
      "loss": 0.6373,
      "step": 6685
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3735792686324412,
      "learning_rate": 1.4700642394513553e-05,
      "loss": 0.6779,
      "step": 6686
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.370074889850592,
      "learning_rate": 1.4695057617817475e-05,
      "loss": 0.6707,
      "step": 6687
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3655974747094493,
      "learning_rate": 1.4689473286040512e-05,
      "loss": 0.6695,
      "step": 6688
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3599570396157657,
      "learning_rate": 1.4683889399651005e-05,
      "loss": 0.6736,
      "step": 6689
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3499174905364315,
      "learning_rate": 1.467830595911728e-05,
      "loss": 0.6473,
      "step": 6690
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34396155859127664,
      "learning_rate": 1.46727229649076e-05,
      "loss": 0.6368,
      "step": 6691
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3375492897944913,
      "learning_rate": 1.4667140417490217e-05,
      "loss": 0.639,
      "step": 6692
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38431326917267744,
      "learning_rate": 1.4661558317333322e-05,
      "loss": 0.6622,
      "step": 6693
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39214908047646513,
      "learning_rate": 1.465597666490509e-05,
      "loss": 0.648,
      "step": 6694
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4281220532497497,
      "learning_rate": 1.4650395460673638e-05,
      "loss": 0.669,
      "step": 6695
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39656259760450024,
      "learning_rate": 1.4644814705107058e-05,
      "loss": 0.7001,
      "step": 6696
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3878862486761333,
      "learning_rate": 1.4639234398673404e-05,
      "loss": 0.6841,
      "step": 6697
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3662372769013589,
      "learning_rate": 1.4633654541840679e-05,
      "loss": 0.6321,
      "step": 6698
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38388948895154806,
      "learning_rate": 1.4628075135076874e-05,
      "loss": 0.656,
      "step": 6699
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37160480899260384,
      "learning_rate": 1.4622496178849916e-05,
      "loss": 0.6511,
      "step": 6700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4043088920732799,
      "learning_rate": 1.4616917673627712e-05,
      "loss": 0.6328,
      "step": 6701
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3992584748403595,
      "learning_rate": 1.4611339619878125e-05,
      "loss": 0.6552,
      "step": 6702
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6300623253070786,
      "learning_rate": 1.4605762018068967e-05,
      "loss": 0.3612,
      "step": 6703
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41550984518089895,
      "learning_rate": 1.4600184868668041e-05,
      "loss": 0.6797,
      "step": 6704
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3778737850133864,
      "learning_rate": 1.4594608172143081e-05,
      "loss": 0.6742,
      "step": 6705
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36516258249997036,
      "learning_rate": 1.4589031928961813e-05,
      "loss": 0.6668,
      "step": 6706
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3521340773835436,
      "learning_rate": 1.4583456139591899e-05,
      "loss": 0.6144,
      "step": 6707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3671844080591086,
      "learning_rate": 1.4577880804500982e-05,
      "loss": 0.6723,
      "step": 6708
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44270950837098527,
      "learning_rate": 1.457230592415665e-05,
      "loss": 0.6605,
      "step": 6709
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3979223213010094,
      "learning_rate": 1.4566731499026463e-05,
      "loss": 0.6908,
      "step": 6710
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.595825837383706,
      "learning_rate": 1.456115752957795e-05,
      "loss": 0.4009,
      "step": 6711
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38923723445764824,
      "learning_rate": 1.4555584016278575e-05,
      "loss": 0.6395,
      "step": 6712
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38837124550811636,
      "learning_rate": 1.4550010959595801e-05,
      "loss": 0.6828,
      "step": 6713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41299850573446195,
      "learning_rate": 1.4544438359997024e-05,
      "loss": 0.6585,
      "step": 6714
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.394974603554839,
      "learning_rate": 1.4538866217949617e-05,
      "loss": 0.6599,
      "step": 6715
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.43394457693005756,
      "learning_rate": 1.4533294533920903e-05,
      "loss": 0.7427,
      "step": 6716
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40742346609897206,
      "learning_rate": 1.4527723308378172e-05,
      "loss": 0.6976,
      "step": 6717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3906668661979183,
      "learning_rate": 1.4522152541788684e-05,
      "loss": 0.6424,
      "step": 6718
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3954874531007802,
      "learning_rate": 1.4516582234619636e-05,
      "loss": 0.6918,
      "step": 6719
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39790187333243654,
      "learning_rate": 1.4511012387338221e-05,
      "loss": 0.663,
      "step": 6720
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.417626522255501,
      "learning_rate": 1.4505443000411568e-05,
      "loss": 0.6371,
      "step": 6721
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.387099682766446,
      "learning_rate": 1.4499874074306774e-05,
      "loss": 0.6552,
      "step": 6722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3021490136705355,
      "learning_rate": 1.44943056094909e-05,
      "loss": 0.6359,
      "step": 6723
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3662278207795116,
      "learning_rate": 1.4488737606430965e-05,
      "loss": 0.6731,
      "step": 6724
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6404949656394954,
      "learning_rate": 1.4483170065593954e-05,
      "loss": 0.4011,
      "step": 6725
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38715679676782044,
      "learning_rate": 1.4477602987446797e-05,
      "loss": 0.6637,
      "step": 6726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4019493279167293,
      "learning_rate": 1.4472036372456413e-05,
      "loss": 0.6877,
      "step": 6727
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36718165086390586,
      "learning_rate": 1.4466470221089666e-05,
      "loss": 0.6832,
      "step": 6728
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35912761712319957,
      "learning_rate": 1.4460904533813379e-05,
      "loss": 0.6775,
      "step": 6729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3871882262590158,
      "learning_rate": 1.4455339311094334e-05,
      "loss": 0.6825,
      "step": 6730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35522772567143357,
      "learning_rate": 1.4449774553399292e-05,
      "loss": 0.6352,
      "step": 6731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37248327747501064,
      "learning_rate": 1.4444210261194953e-05,
      "loss": 0.6953,
      "step": 6732
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3654130939283171,
      "learning_rate": 1.4438646434947981e-05,
      "loss": 0.6433,
      "step": 6733
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38534851213542876,
      "learning_rate": 1.443308307512502e-05,
      "loss": 0.6554,
      "step": 6734
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35243035719204274,
      "learning_rate": 1.4427520182192662e-05,
      "loss": 0.6327,
      "step": 6735
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34426637700036794,
      "learning_rate": 1.442195775661746e-05,
      "loss": 0.6647,
      "step": 6736
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3171747129428197,
      "learning_rate": 1.4416395798865917e-05,
      "loss": 0.6246,
      "step": 6737
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3582194681288791,
      "learning_rate": 1.441083430940452e-05,
      "loss": 0.6579,
      "step": 6738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3495227844036008,
      "learning_rate": 1.4405273288699691e-05,
      "loss": 0.6355,
      "step": 6739
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3912866831556337,
      "learning_rate": 1.4399712737217843e-05,
      "loss": 0.7055,
      "step": 6740
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40876538267722046,
      "learning_rate": 1.439415265542532e-05,
      "loss": 0.6617,
      "step": 6741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36640634200755234,
      "learning_rate": 1.4388593043788448e-05,
      "loss": 0.647,
      "step": 6742
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4107147073381436,
      "learning_rate": 1.4383033902773502e-05,
      "loss": 0.6535,
      "step": 6743
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3706266051074597,
      "learning_rate": 1.4377475232846714e-05,
      "loss": 0.666,
      "step": 6744
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37603550962409305,
      "learning_rate": 1.4371917034474292e-05,
      "loss": 0.6417,
      "step": 6745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37413196199421817,
      "learning_rate": 1.4366359308122381e-05,
      "loss": 0.6609,
      "step": 6746
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3719506300961426,
      "learning_rate": 1.4360802054257121e-05,
      "loss": 0.6281,
      "step": 6747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4103416920361377,
      "learning_rate": 1.435524527334458e-05,
      "loss": 0.6837,
      "step": 6748
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3936628083146454,
      "learning_rate": 1.43496889658508e-05,
      "loss": 0.6667,
      "step": 6749
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34905087185981215,
      "learning_rate": 1.4344133132241785e-05,
      "loss": 0.648,
      "step": 6750
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42055944560305053,
      "learning_rate": 1.4338577772983493e-05,
      "loss": 0.6633,
      "step": 6751
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38587381654112085,
      "learning_rate": 1.4333022888541848e-05,
      "loss": 0.6712,
      "step": 6752
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3650846443476519,
      "learning_rate": 1.432746847938272e-05,
      "loss": 0.6206,
      "step": 6753
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4074820698673332,
      "learning_rate": 1.4321914545971965e-05,
      "loss": 0.6774,
      "step": 6754
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41623676607432497,
      "learning_rate": 1.4316361088775384e-05,
      "loss": 0.6471,
      "step": 6755
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37409159073103065,
      "learning_rate": 1.4310808108258731e-05,
      "loss": 0.6825,
      "step": 6756
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37872979698134995,
      "learning_rate": 1.430525560488773e-05,
      "loss": 0.6552,
      "step": 6757
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3559531267934472,
      "learning_rate": 1.4299703579128069e-05,
      "loss": 0.6288,
      "step": 6758
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37634556359416976,
      "learning_rate": 1.4294152031445383e-05,
      "loss": 0.6461,
      "step": 6759
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.377919493460524,
      "learning_rate": 1.4288600962305265e-05,
      "loss": 0.6533,
      "step": 6760
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39234310365141134,
      "learning_rate": 1.4283050372173292e-05,
      "loss": 0.6651,
      "step": 6761
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3850843796235422,
      "learning_rate": 1.4277500261514987e-05,
      "loss": 0.6416,
      "step": 6762
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3586932181597116,
      "learning_rate": 1.4271950630795824e-05,
      "loss": 0.6498,
      "step": 6763
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3424219329309957,
      "learning_rate": 1.4266401480481236e-05,
      "loss": 0.6359,
      "step": 6764
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36585627145792177,
      "learning_rate": 1.4260852811036637e-05,
      "loss": 0.7112,
      "step": 6765
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4089590692245139,
      "learning_rate": 1.4255304622927382e-05,
      "loss": 0.6865,
      "step": 6766
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3504855304443536,
      "learning_rate": 1.4249756916618782e-05,
      "loss": 0.6417,
      "step": 6767
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.421859515162206,
      "learning_rate": 1.424420969257613e-05,
      "loss": 0.691,
      "step": 6768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4476271832074248,
      "learning_rate": 1.4238662951264664e-05,
      "loss": 0.6666,
      "step": 6769
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3810894367843521,
      "learning_rate": 1.4233116693149578e-05,
      "loss": 0.6827,
      "step": 6770
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3504875822828294,
      "learning_rate": 1.4227570918696026e-05,
      "loss": 0.654,
      "step": 6771
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39536393810352044,
      "learning_rate": 1.422202562836913e-05,
      "loss": 0.6596,
      "step": 6772
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39962904069978245,
      "learning_rate": 1.421648082263397e-05,
      "loss": 0.6504,
      "step": 6773
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39420465122429027,
      "learning_rate": 1.4210936501955572e-05,
      "loss": 0.6386,
      "step": 6774
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3816388393148949,
      "learning_rate": 1.420539266679894e-05,
      "loss": 0.684,
      "step": 6775
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3486487791979551,
      "learning_rate": 1.4199849317629034e-05,
      "loss": 0.6369,
      "step": 6776
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44202194811915957,
      "learning_rate": 1.4194306454910757e-05,
      "loss": 0.6647,
      "step": 6777
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39080091339899675,
      "learning_rate": 1.4188764079108985e-05,
      "loss": 0.716,
      "step": 6778
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3354351882766707,
      "learning_rate": 1.4183222190688555e-05,
      "loss": 0.6721,
      "step": 6779
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36019055400822214,
      "learning_rate": 1.4177680790114248e-05,
      "loss": 0.658,
      "step": 6780
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4067574198489496,
      "learning_rate": 1.417213987785082e-05,
      "loss": 0.6575,
      "step": 6781
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35983881467518153,
      "learning_rate": 1.416659945436299e-05,
      "loss": 0.6371,
      "step": 6782
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37764871141259687,
      "learning_rate": 1.416105952011542e-05,
      "loss": 0.6551,
      "step": 6783
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38096327101259864,
      "learning_rate": 1.415552007557273e-05,
      "loss": 0.6585,
      "step": 6784
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4339215179826359,
      "learning_rate": 1.4149981121199518e-05,
      "loss": 0.655,
      "step": 6785
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.42651343090005744,
      "learning_rate": 1.4144442657460324e-05,
      "loss": 0.6957,
      "step": 6786
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35731371124033784,
      "learning_rate": 1.413890468481965e-05,
      "loss": 0.6821,
      "step": 6787
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36671454746005994,
      "learning_rate": 1.4133367203741957e-05,
      "loss": 0.6455,
      "step": 6788
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3836228721768411,
      "learning_rate": 1.4127830214691677e-05,
      "loss": 0.671,
      "step": 6789
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4413485242232214,
      "learning_rate": 1.4122293718133189e-05,
      "loss": 0.6599,
      "step": 6790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3664706736100499,
      "learning_rate": 1.4116757714530822e-05,
      "loss": 0.662,
      "step": 6791
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33116158643189647,
      "learning_rate": 1.4111222204348884e-05,
      "loss": 0.6345,
      "step": 6792
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3588606902620312,
      "learning_rate": 1.4105687188051632e-05,
      "loss": 0.634,
      "step": 6793
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4186981122925569,
      "learning_rate": 1.4100152666103269e-05,
      "loss": 0.6775,
      "step": 6794
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3969794572626506,
      "learning_rate": 1.4094618638967972e-05,
      "loss": 0.6897,
      "step": 6795
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3798565841261676,
      "learning_rate": 1.4089085107109887e-05,
      "loss": 0.6784,
      "step": 6796
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6834248982307664,
      "learning_rate": 1.4083552070993096e-05,
      "loss": 0.4338,
      "step": 6797
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41246221088410495,
      "learning_rate": 1.4078019531081645e-05,
      "loss": 0.6386,
      "step": 6798
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3751615696682723,
      "learning_rate": 1.4072487487839546e-05,
      "loss": 0.6475,
      "step": 6799
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3806492473923906,
      "learning_rate": 1.4066955941730763e-05,
      "loss": 0.6496,
      "step": 6800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3957193503361286,
      "learning_rate": 1.4061424893219217e-05,
      "loss": 0.6721,
      "step": 6801
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3612295611198104,
      "learning_rate": 1.4055894342768788e-05,
      "loss": 0.6565,
      "step": 6802
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3564193057756643,
      "learning_rate": 1.405036429084333e-05,
      "loss": 0.6534,
      "step": 6803
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3973209371530019,
      "learning_rate": 1.4044834737906634e-05,
      "loss": 0.6406,
      "step": 6804
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4264462951556027,
      "learning_rate": 1.4039305684422454e-05,
      "loss": 0.7044,
      "step": 6805
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3367830310896987,
      "learning_rate": 1.4033777130854506e-05,
      "loss": 0.6348,
      "step": 6806
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3507356877018016,
      "learning_rate": 1.4028249077666467e-05,
      "loss": 0.649,
      "step": 6807
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.365549824445764,
      "learning_rate": 1.4022721525321968e-05,
      "loss": 0.6413,
      "step": 6808
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34417652483592504,
      "learning_rate": 1.4017194474284585e-05,
      "loss": 0.665,
      "step": 6809
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34475794062213955,
      "learning_rate": 1.4011667925017886e-05,
      "loss": 0.6494,
      "step": 6810
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37874552136597933,
      "learning_rate": 1.4006141877985364e-05,
      "loss": 0.681,
      "step": 6811
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3970380320371924,
      "learning_rate": 1.4000616333650486e-05,
      "loss": 0.6767,
      "step": 6812
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.31137021978715274,
      "learning_rate": 1.399509129247667e-05,
      "loss": 0.6371,
      "step": 6813
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3933878684394202,
      "learning_rate": 1.398956675492729e-05,
      "loss": 0.6682,
      "step": 6814
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3991379654886455,
      "learning_rate": 1.3984042721465697e-05,
      "loss": 0.6265,
      "step": 6815
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.324065583152694,
      "learning_rate": 1.3978519192555165e-05,
      "loss": 0.5999,
      "step": 6816
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4196465175755481,
      "learning_rate": 1.3972996168658962e-05,
      "loss": 0.6769,
      "step": 6817
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7527117465621368,
      "learning_rate": 1.396747365024029e-05,
      "loss": 0.4089,
      "step": 6818
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4119600584368631,
      "learning_rate": 1.396195163776232e-05,
      "loss": 0.6728,
      "step": 6819
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3496217214113135,
      "learning_rate": 1.3956430131688175e-05,
      "loss": 0.6335,
      "step": 6820
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3431044463119188,
      "learning_rate": 1.395090913248093e-05,
      "loss": 0.6507,
      "step": 6821
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4138701247653241,
      "learning_rate": 1.3945388640603628e-05,
      "loss": 0.7092,
      "step": 6822
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3630668513446897,
      "learning_rate": 1.3939868656519275e-05,
      "loss": 0.6472,
      "step": 6823
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32247802645706436,
      "learning_rate": 1.3934349180690822e-05,
      "loss": 0.6145,
      "step": 6824
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3563251478526718,
      "learning_rate": 1.3928830213581172e-05,
      "loss": 0.6569,
      "step": 6825
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36087273998812996,
      "learning_rate": 1.3923311755653206e-05,
      "loss": 0.6539,
      "step": 6826
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3825543477969204,
      "learning_rate": 1.3917793807369742e-05,
      "loss": 0.6645,
      "step": 6827
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38800784607814176,
      "learning_rate": 1.3912276369193562e-05,
      "loss": 0.7004,
      "step": 6828
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40537867261055516,
      "learning_rate": 1.3906759441587408e-05,
      "loss": 0.6199,
      "step": 6829
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34416514015982425,
      "learning_rate": 1.3901243025013988e-05,
      "loss": 0.6614,
      "step": 6830
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4168653200167947,
      "learning_rate": 1.389572711993595e-05,
      "loss": 0.7013,
      "step": 6831
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38737043122024484,
      "learning_rate": 1.38902117268159e-05,
      "loss": 0.6552,
      "step": 6832
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4099954868575676,
      "learning_rate": 1.3884696846116424e-05,
      "loss": 0.6573,
      "step": 6833
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3480978830130539,
      "learning_rate": 1.3879182478300029e-05,
      "loss": 0.6546,
      "step": 6834
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3585706573516925,
      "learning_rate": 1.3873668623829213e-05,
      "loss": 0.6494,
      "step": 6835
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3782965131441553,
      "learning_rate": 1.3868155283166403e-05,
      "loss": 0.6844,
      "step": 6836
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4045889443146263,
      "learning_rate": 1.3862642456774011e-05,
      "loss": 0.6401,
      "step": 6837
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41917675445410557,
      "learning_rate": 1.3857130145114381e-05,
      "loss": 0.6495,
      "step": 6838
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35505122573980485,
      "learning_rate": 1.385161834864983e-05,
      "loss": 0.6568,
      "step": 6839
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40759560381166204,
      "learning_rate": 1.3846107067842625e-05,
      "loss": 0.6996,
      "step": 6840
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4304810878370861,
      "learning_rate": 1.3840596303154982e-05,
      "loss": 0.6586,
      "step": 6841
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40633208098434326,
      "learning_rate": 1.3835086055049094e-05,
      "loss": 0.6646,
      "step": 6842
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3694071295258274,
      "learning_rate": 1.3829576323987083e-05,
      "loss": 0.6776,
      "step": 6843
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3919115494572393,
      "learning_rate": 1.3824067110431063e-05,
      "loss": 0.6076,
      "step": 6844
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38878623401166224,
      "learning_rate": 1.3818558414843073e-05,
      "loss": 0.633,
      "step": 6845
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3564262808647381,
      "learning_rate": 1.3813050237685126e-05,
      "loss": 0.6743,
      "step": 6846
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3543962028808962,
      "learning_rate": 1.3807542579419184e-05,
      "loss": 0.6327,
      "step": 6847
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39477205201522914,
      "learning_rate": 1.3802035440507165e-05,
      "loss": 0.6964,
      "step": 6848
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34242979905641774,
      "learning_rate": 1.379652882141095e-05,
      "loss": 0.6428,
      "step": 6849
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.409794654549452,
      "learning_rate": 1.3791022722592361e-05,
      "loss": 0.6904,
      "step": 6850
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36015487497300014,
      "learning_rate": 1.378551714451321e-05,
      "loss": 0.6559,
      "step": 6851
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3672574561043366,
      "learning_rate": 1.3780012087635224e-05,
      "loss": 0.6666,
      "step": 6852
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3844375633032232,
      "learning_rate": 1.3774507552420118e-05,
      "loss": 0.7014,
      "step": 6853
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40430252626219293,
      "learning_rate": 1.3769003539329547e-05,
      "loss": 0.7008,
      "step": 6854
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34695947145295325,
      "learning_rate": 1.3763500048825117e-05,
      "loss": 0.6584,
      "step": 6855
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35513193347139616,
      "learning_rate": 1.375799708136841e-05,
      "loss": 0.6556,
      "step": 6856
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38756554398349363,
      "learning_rate": 1.375249463742094e-05,
      "loss": 0.7001,
      "step": 6857
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38705409665026763,
      "learning_rate": 1.374699271744421e-05,
      "loss": 0.6641,
      "step": 6858
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3574370285036465,
      "learning_rate": 1.3741491321899645e-05,
      "loss": 0.6415,
      "step": 6859
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39508210426362844,
      "learning_rate": 1.3735990451248649e-05,
      "loss": 0.6609,
      "step": 6860
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38777111694495386,
      "learning_rate": 1.3730490105952567e-05,
      "loss": 0.6682,
      "step": 6861
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35740467843471924,
      "learning_rate": 1.3724990286472707e-05,
      "loss": 0.6407,
      "step": 6862
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37956931579051906,
      "learning_rate": 1.3719490993270335e-05,
      "loss": 0.6469,
      "step": 6863
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37547516595340735,
      "learning_rate": 1.371399222680666e-05,
      "loss": 0.6446,
      "step": 6864
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40462598402394373,
      "learning_rate": 1.3708493987542877e-05,
      "loss": 0.6869,
      "step": 6865
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39252134213717343,
      "learning_rate": 1.3702996275940097e-05,
      "loss": 0.7115,
      "step": 6866
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35399784118484984,
      "learning_rate": 1.3697499092459419e-05,
      "loss": 0.6513,
      "step": 6867
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40849907168835886,
      "learning_rate": 1.369200243756188e-05,
      "loss": 0.6912,
      "step": 6868
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3816985388164921,
      "learning_rate": 1.3686506311708479e-05,
      "loss": 0.6406,
      "step": 6869
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36114513575496004,
      "learning_rate": 1.368101071536017e-05,
      "loss": 0.6683,
      "step": 6870
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7444701464985362,
      "learning_rate": 1.3675515648977853e-05,
      "loss": 0.4158,
      "step": 6871
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37167224117550113,
      "learning_rate": 1.3670021113022402e-05,
      "loss": 0.6631,
      "step": 6872
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36560149426324584,
      "learning_rate": 1.3664527107954642e-05,
      "loss": 0.6551,
      "step": 6873
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35668285255786863,
      "learning_rate": 1.3659033634235343e-05,
      "loss": 0.687,
      "step": 6874
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40831532222971173,
      "learning_rate": 1.3653540692325229e-05,
      "loss": 0.6815,
      "step": 6875
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3627835519070063,
      "learning_rate": 1.3648048282685e-05,
      "loss": 0.641,
      "step": 6876
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41399574463420086,
      "learning_rate": 1.3642556405775287e-05,
      "loss": 0.6507,
      "step": 6877
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37864040338145155,
      "learning_rate": 1.3637065062056683e-05,
      "loss": 0.6232,
      "step": 6878
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5686476029293557,
      "learning_rate": 1.3631574251989753e-05,
      "loss": 0.4059,
      "step": 6879
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.43297917537444724,
      "learning_rate": 1.3626083976035003e-05,
      "loss": 0.6604,
      "step": 6880
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40681781901049513,
      "learning_rate": 1.362059423465289e-05,
      "loss": 0.6517,
      "step": 6881
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.39285455829705623,
      "learning_rate": 1.3615105028303833e-05,
      "loss": 0.6791,
      "step": 6882
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37403941160459164,
      "learning_rate": 1.360961635744821e-05,
      "loss": 0.6532,
      "step": 6883
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40222803597595697,
      "learning_rate": 1.3604128222546344e-05,
      "loss": 0.6625,
      "step": 6884
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4067257665988949,
      "learning_rate": 1.3598640624058512e-05,
      "loss": 0.6415,
      "step": 6885
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3413807789451829,
      "learning_rate": 1.3593153562444963e-05,
      "loss": 0.6181,
      "step": 6886
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4076470800551287,
      "learning_rate": 1.3587667038165892e-05,
      "loss": 0.6901,
      "step": 6887
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3856891388061515,
      "learning_rate": 1.3582181051681444e-05,
      "loss": 0.663,
      "step": 6888
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4233577644162888,
      "learning_rate": 1.3576695603451718e-05,
      "loss": 0.643,
      "step": 6889
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4000522847897913,
      "learning_rate": 1.3571210693936774e-05,
      "loss": 0.6861,
      "step": 6890
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3365168596360846,
      "learning_rate": 1.3565726323596622e-05,
      "loss": 0.6303,
      "step": 6891
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3756354334463335,
      "learning_rate": 1.3560242492891234e-05,
      "loss": 0.6184,
      "step": 6892
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4140263080397678,
      "learning_rate": 1.355475920228053e-05,
      "loss": 0.6461,
      "step": 6893
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6416894685295836,
      "learning_rate": 1.354927645222439e-05,
      "loss": 0.4342,
      "step": 6894
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38826244563042367,
      "learning_rate": 1.3543794243182643e-05,
      "loss": 0.6803,
      "step": 6895
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42815665139597514,
      "learning_rate": 1.3538312575615078e-05,
      "loss": 0.6968,
      "step": 6896
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3061588945753031,
      "learning_rate": 1.3532831449981435e-05,
      "loss": 0.6142,
      "step": 6897
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3617029703337822,
      "learning_rate": 1.3527350866741402e-05,
      "loss": 0.6011,
      "step": 6898
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38065914720909205,
      "learning_rate": 1.3521870826354633e-05,
      "loss": 0.6577,
      "step": 6899
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40439094223420735,
      "learning_rate": 1.351639132928074e-05,
      "loss": 0.6981,
      "step": 6900
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3299849988639282,
      "learning_rate": 1.3510912375979274e-05,
      "loss": 0.6516,
      "step": 6901
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34685444027840373,
      "learning_rate": 1.350543396690975e-05,
      "loss": 0.6087,
      "step": 6902
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4066703368368352,
      "learning_rate": 1.349995610253164e-05,
      "loss": 0.6658,
      "step": 6903
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3873508016178074,
      "learning_rate": 1.3494478783304361e-05,
      "loss": 0.6035,
      "step": 6904
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3918261990532257,
      "learning_rate": 1.348900200968728e-05,
      "loss": 0.6569,
      "step": 6905
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3580163005963298,
      "learning_rate": 1.3483525782139743e-05,
      "loss": 0.6289,
      "step": 6906
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3614286024378373,
      "learning_rate": 1.347805010112103e-05,
      "loss": 0.6565,
      "step": 6907
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3454879570835759,
      "learning_rate": 1.3472574967090382e-05,
      "loss": 0.6349,
      "step": 6908
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37993694378521886,
      "learning_rate": 1.3467100380506982e-05,
      "loss": 0.64,
      "step": 6909
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3885388593504384,
      "learning_rate": 1.3461626341829988e-05,
      "loss": 0.6761,
      "step": 6910
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.333457637448722,
      "learning_rate": 1.3456152851518499e-05,
      "loss": 0.6396,
      "step": 6911
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3847306402491253,
      "learning_rate": 1.3450679910031554e-05,
      "loss": 0.687,
      "step": 6912
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3714459970375766,
      "learning_rate": 1.3445207517828182e-05,
      "loss": 0.6535,
      "step": 6913
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3945218471330601,
      "learning_rate": 1.3439735675367345e-05,
      "loss": 0.6599,
      "step": 6914
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40753226612922056,
      "learning_rate": 1.3434264383107953e-05,
      "loss": 0.6601,
      "step": 6915
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.39574282223688245,
      "learning_rate": 1.3428793641508873e-05,
      "loss": 0.6837,
      "step": 6916
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38161353813298676,
      "learning_rate": 1.3423323451028938e-05,
      "loss": 0.664,
      "step": 6917
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36852016863095277,
      "learning_rate": 1.3417853812126925e-05,
      "loss": 0.678,
      "step": 6918
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3757110361089484,
      "learning_rate": 1.3412384725261553e-05,
      "loss": 0.6458,
      "step": 6919
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.439040372818604,
      "learning_rate": 1.3406916190891522e-05,
      "loss": 0.7054,
      "step": 6920
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37221935498317243,
      "learning_rate": 1.3401448209475474e-05,
      "loss": 0.6544,
      "step": 6921
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4118988248527222,
      "learning_rate": 1.3395980781471996e-05,
      "loss": 0.6812,
      "step": 6922
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35592830800541786,
      "learning_rate": 1.3390513907339627e-05,
      "loss": 0.6592,
      "step": 6923
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3794932406389272,
      "learning_rate": 1.3385047587536879e-05,
      "loss": 0.6706,
      "step": 6924
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3941169265581349,
      "learning_rate": 1.3379581822522198e-05,
      "loss": 0.6823,
      "step": 6925
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36683139888880184,
      "learning_rate": 1.3374116612753993e-05,
      "loss": 0.6117,
      "step": 6926
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.39155239754967425,
      "learning_rate": 1.3368651958690628e-05,
      "loss": 0.6745,
      "step": 6927
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3907183417070046,
      "learning_rate": 1.3363187860790418e-05,
      "loss": 0.6792,
      "step": 6928
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3858254375406018,
      "learning_rate": 1.335772431951162e-05,
      "loss": 0.6574,
      "step": 6929
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3420511330090024,
      "learning_rate": 1.335226133531247e-05,
      "loss": 0.6485,
      "step": 6930
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3667403839417643,
      "learning_rate": 1.334679890865113e-05,
      "loss": 0.6782,
      "step": 6931
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3784692168144186,
      "learning_rate": 1.3341337039985729e-05,
      "loss": 0.6636,
      "step": 6932
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38469708589548773,
      "learning_rate": 1.3335875729774341e-05,
      "loss": 0.6511,
      "step": 6933
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32526318198242565,
      "learning_rate": 1.3330414978475019e-05,
      "loss": 0.6755,
      "step": 6934
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4094404400826596,
      "learning_rate": 1.3324954786545736e-05,
      "loss": 0.6932,
      "step": 6935
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3668636228590166,
      "learning_rate": 1.331949515444443e-05,
      "loss": 0.6533,
      "step": 6936
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33671806332605003,
      "learning_rate": 1.3314036082629003e-05,
      "loss": 0.6329,
      "step": 6937
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33468463572527085,
      "learning_rate": 1.3308577571557293e-05,
      "loss": 0.6668,
      "step": 6938
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3758375050234979,
      "learning_rate": 1.3303119621687098e-05,
      "loss": 0.6574,
      "step": 6939
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40044386373797825,
      "learning_rate": 1.3297662233476167e-05,
      "loss": 0.6932,
      "step": 6940
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.39336666835547013,
      "learning_rate": 1.3292205407382222e-05,
      "loss": 0.6597,
      "step": 6941
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.41304247046882764,
      "learning_rate": 1.3286749143862903e-05,
      "loss": 0.6522,
      "step": 6942
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37034990815409113,
      "learning_rate": 1.3281293443375827e-05,
      "loss": 0.6786,
      "step": 6943
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3962727477945113,
      "learning_rate": 1.3275838306378558e-05,
      "loss": 0.6928,
      "step": 6944
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4224633626145415,
      "learning_rate": 1.3270383733328607e-05,
      "loss": 0.7066,
      "step": 6945
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3616032605682271,
      "learning_rate": 1.3264929724683443e-05,
      "loss": 0.6457,
      "step": 6946
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38389597048964186,
      "learning_rate": 1.3259476280900483e-05,
      "loss": 0.6605,
      "step": 6947
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3732555312440347,
      "learning_rate": 1.325402340243712e-05,
      "loss": 0.6147,
      "step": 6948
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3492276186047787,
      "learning_rate": 1.3248571089750662e-05,
      "loss": 0.6579,
      "step": 6949
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.41252755426585114,
      "learning_rate": 1.3243119343298392e-05,
      "loss": 0.6502,
      "step": 6950
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37683644221666757,
      "learning_rate": 1.3237668163537545e-05,
      "loss": 0.6338,
      "step": 6951
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3801508807223988,
      "learning_rate": 1.3232217550925298e-05,
      "loss": 0.6341,
      "step": 6952
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34867433864050307,
      "learning_rate": 1.3226767505918796e-05,
      "loss": 0.6548,
      "step": 6953
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37269890121736965,
      "learning_rate": 1.3221318028975116e-05,
      "loss": 0.6808,
      "step": 6954
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.367401896876148,
      "learning_rate": 1.3215869120551315e-05,
      "loss": 0.7139,
      "step": 6955
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3784680713240907,
      "learning_rate": 1.3210420781104371e-05,
      "loss": 0.6626,
      "step": 6956
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42035326763183545,
      "learning_rate": 1.3204973011091242e-05,
      "loss": 0.6728,
      "step": 6957
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36448568342144344,
      "learning_rate": 1.319952581096882e-05,
      "loss": 0.6071,
      "step": 6958
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37008241262478486,
      "learning_rate": 1.3194079181193953e-05,
      "loss": 0.6506,
      "step": 6959
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3706686687844343,
      "learning_rate": 1.3188633122223449e-05,
      "loss": 0.6755,
      "step": 6960
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7254175570494951,
      "learning_rate": 1.318318763451405e-05,
      "loss": 0.3988,
      "step": 6961
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3860682455843109,
      "learning_rate": 1.317774271852248e-05,
      "loss": 0.6553,
      "step": 6962
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4035314836206492,
      "learning_rate": 1.317229837470538e-05,
      "loss": 0.6493,
      "step": 6963
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38936858937402075,
      "learning_rate": 1.3166854603519381e-05,
      "loss": 0.6615,
      "step": 6964
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.383996036886315,
      "learning_rate": 1.316141140542103e-05,
      "loss": 0.678,
      "step": 6965
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4117611623656412,
      "learning_rate": 1.3155968780866843e-05,
      "loss": 0.6817,
      "step": 6966
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3960135898667482,
      "learning_rate": 1.3150526730313293e-05,
      "loss": 0.6682,
      "step": 6967
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36400821356444174,
      "learning_rate": 1.3145085254216784e-05,
      "loss": 0.6447,
      "step": 6968
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3654909710384413,
      "learning_rate": 1.3139644353033707e-05,
      "loss": 0.6665,
      "step": 6969
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37507835736638245,
      "learning_rate": 1.3134204027220367e-05,
      "loss": 0.5998,
      "step": 6970
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3880842553053863,
      "learning_rate": 1.312876427723305e-05,
      "loss": 0.6877,
      "step": 6971
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.41103349229481423,
      "learning_rate": 1.312332510352797e-05,
      "loss": 0.6329,
      "step": 6972
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33594479150255685,
      "learning_rate": 1.3117886506561309e-05,
      "loss": 0.6119,
      "step": 6973
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.39032257165671164,
      "learning_rate": 1.3112448486789197e-05,
      "loss": 0.6716,
      "step": 6974
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3470659227115671,
      "learning_rate": 1.3107011044667705e-05,
      "loss": 0.6313,
      "step": 6975
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40266100687758716,
      "learning_rate": 1.310157418065288e-05,
      "loss": 0.6687,
      "step": 6976
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.343005209594206,
      "learning_rate": 1.3096137895200693e-05,
      "loss": 0.6464,
      "step": 6977
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34787292845235646,
      "learning_rate": 1.3090702188767088e-05,
      "loss": 0.6271,
      "step": 6978
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3581238213057354,
      "learning_rate": 1.3085267061807945e-05,
      "loss": 0.6687,
      "step": 6979
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4620683251650599,
      "learning_rate": 1.3079832514779099e-05,
      "loss": 0.7317,
      "step": 6980
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32825133364189485,
      "learning_rate": 1.3074398548136346e-05,
      "loss": 0.6566,
      "step": 6981
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4075870255838404,
      "learning_rate": 1.3068965162335417e-05,
      "loss": 0.7128,
      "step": 6982
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3837664783844843,
      "learning_rate": 1.3063532357832017e-05,
      "loss": 0.688,
      "step": 6983
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3494948790852925,
      "learning_rate": 1.3058100135081775e-05,
      "loss": 0.6339,
      "step": 6984
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.338185346708509,
      "learning_rate": 1.3052668494540295e-05,
      "loss": 0.663,
      "step": 6985
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3607607583208377,
      "learning_rate": 1.3047237436663113e-05,
      "loss": 0.6769,
      "step": 6986
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42071719608087443,
      "learning_rate": 1.3041806961905739e-05,
      "loss": 0.6793,
      "step": 6987
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3595729554255157,
      "learning_rate": 1.3036377070723607e-05,
      "loss": 0.6553,
      "step": 6988
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3675741085549042,
      "learning_rate": 1.3030947763572112e-05,
      "loss": 0.6397,
      "step": 6989
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4139278787570495,
      "learning_rate": 1.3025519040906617e-05,
      "loss": 0.6558,
      "step": 6990
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7654643774930155,
      "learning_rate": 1.3020090903182423e-05,
      "loss": 0.3895,
      "step": 6991
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3749414809189833,
      "learning_rate": 1.3014663350854775e-05,
      "loss": 0.6533,
      "step": 6992
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38607218483443456,
      "learning_rate": 1.3009236384378872e-05,
      "loss": 0.6536,
      "step": 6993
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42103999525498104,
      "learning_rate": 1.3003810004209872e-05,
      "loss": 0.6373,
      "step": 6994
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4131901422644114,
      "learning_rate": 1.299838421080287e-05,
      "loss": 0.6493,
      "step": 6995
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37036792961296894,
      "learning_rate": 1.299295900461294e-05,
      "loss": 0.6282,
      "step": 6996
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36940109241270214,
      "learning_rate": 1.2987534386095066e-05,
      "loss": 0.6379,
      "step": 6997
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3714947764381304,
      "learning_rate": 1.2982110355704223e-05,
      "loss": 0.6697,
      "step": 6998
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3854018370235404,
      "learning_rate": 1.2976686913895306e-05,
      "loss": 0.66,
      "step": 6999
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36967388265022644,
      "learning_rate": 1.2971264061123174e-05,
      "loss": 0.659,
      "step": 7000
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40272518324367035,
      "learning_rate": 1.296584179784264e-05,
      "loss": 0.6653,
      "step": 7001
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36092640581190205,
      "learning_rate": 1.2960420124508451e-05,
      "loss": 0.6656,
      "step": 7002
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39268298353333375,
      "learning_rate": 1.2954999041575331e-05,
      "loss": 0.625,
      "step": 7003
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38895276304892473,
      "learning_rate": 1.294957854949793e-05,
      "loss": 0.6779,
      "step": 7004
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37943042168918734,
      "learning_rate": 1.2944158648730866e-05,
      "loss": 0.6582,
      "step": 7005
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40399624241210014,
      "learning_rate": 1.2938739339728695e-05,
      "loss": 0.6648,
      "step": 7006
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.625879195084284,
      "learning_rate": 1.2933320622945924e-05,
      "loss": 0.3963,
      "step": 7007
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39285526556768297,
      "learning_rate": 1.2927902498837021e-05,
      "loss": 0.7005,
      "step": 7008
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36830609585748536,
      "learning_rate": 1.2922484967856388e-05,
      "loss": 0.6325,
      "step": 7009
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3602463052875375,
      "learning_rate": 1.29170680304584e-05,
      "loss": 0.6331,
      "step": 7010
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.41704008703323,
      "learning_rate": 1.2911651687097358e-05,
      "loss": 0.6654,
      "step": 7011
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32515659878804387,
      "learning_rate": 1.2906235938227534e-05,
      "loss": 0.6304,
      "step": 7012
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4028221893885498,
      "learning_rate": 1.290082078430313e-05,
      "loss": 0.6858,
      "step": 7013
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3695561185556252,
      "learning_rate": 1.2895406225778316e-05,
      "loss": 0.62,
      "step": 7014
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3586481809205545,
      "learning_rate": 1.28899922631072e-05,
      "loss": 0.6694,
      "step": 7015
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3558925512276084,
      "learning_rate": 1.2884578896743842e-05,
      "loss": 0.6454,
      "step": 7016
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35473273888168266,
      "learning_rate": 1.2879166127142262e-05,
      "loss": 0.668,
      "step": 7017
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4268551606615157,
      "learning_rate": 1.287375395475642e-05,
      "loss": 0.7304,
      "step": 7018
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32185170202993973,
      "learning_rate": 1.286834238004023e-05,
      "loss": 0.5989,
      "step": 7019
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3537112566119191,
      "learning_rate": 1.2862931403447549e-05,
      "loss": 0.6272,
      "step": 7020
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38761529166037034,
      "learning_rate": 1.2857521025432194e-05,
      "loss": 0.6212,
      "step": 7021
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.390742754930072,
      "learning_rate": 1.2852111246447928e-05,
      "loss": 0.6488,
      "step": 7022
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40301327595513137,
      "learning_rate": 1.2846702066948446e-05,
      "loss": 0.6481,
      "step": 7023
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3825762511631756,
      "learning_rate": 1.284129348738743e-05,
      "loss": 0.6655,
      "step": 7024
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3852204196023819,
      "learning_rate": 1.2835885508218487e-05,
      "loss": 0.6336,
      "step": 7025
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3846618993306521,
      "learning_rate": 1.2830478129895174e-05,
      "loss": 0.6714,
      "step": 7026
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6640087582259332,
      "learning_rate": 1.2825071352870998e-05,
      "loss": 0.3795,
      "step": 7027
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32531223883914534,
      "learning_rate": 1.2819665177599425e-05,
      "loss": 0.6352,
      "step": 7028
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3976024084069077,
      "learning_rate": 1.2814259604533864e-05,
      "loss": 0.6668,
      "step": 7029
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4328064354368938,
      "learning_rate": 1.2808854634127663e-05,
      "loss": 0.6721,
      "step": 7030
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3972566630574191,
      "learning_rate": 1.2803450266834144e-05,
      "loss": 0.6846,
      "step": 7031
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4386009967001194,
      "learning_rate": 1.279804650310656e-05,
      "loss": 0.679,
      "step": 7032
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37947639248954096,
      "learning_rate": 1.2792643343398124e-05,
      "loss": 0.6209,
      "step": 7033
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4043414860317462,
      "learning_rate": 1.2787240788161978e-05,
      "loss": 0.647,
      "step": 7034
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3863050136046476,
      "learning_rate": 1.2781838837851241e-05,
      "loss": 0.6699,
      "step": 7035
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4114461055955723,
      "learning_rate": 1.2776437492918965e-05,
      "loss": 0.6825,
      "step": 7036
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3813371060085545,
      "learning_rate": 1.2771036753818144e-05,
      "loss": 0.6793,
      "step": 7037
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38351788594148045,
      "learning_rate": 1.2765636621001743e-05,
      "loss": 0.6676,
      "step": 7038
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38758075001867714,
      "learning_rate": 1.2760237094922667e-05,
      "loss": 0.6758,
      "step": 7039
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37387564811137786,
      "learning_rate": 1.2754838176033761e-05,
      "loss": 0.6616,
      "step": 7040
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39761609831844774,
      "learning_rate": 1.2749439864787827e-05,
      "loss": 0.6582,
      "step": 7041
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4144047912805623,
      "learning_rate": 1.2744042161637615e-05,
      "loss": 0.6763,
      "step": 7042
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4484129121357413,
      "learning_rate": 1.2738645067035822e-05,
      "loss": 0.6289,
      "step": 7043
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38543159143782835,
      "learning_rate": 1.2733248581435096e-05,
      "loss": 0.6627,
      "step": 7044
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32991158083935246,
      "learning_rate": 1.2727852705288037e-05,
      "loss": 0.6789,
      "step": 7045
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39056321378244624,
      "learning_rate": 1.2722457439047192e-05,
      "loss": 0.6543,
      "step": 7046
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36429785348092014,
      "learning_rate": 1.271706278316505e-05,
      "loss": 0.6399,
      "step": 7047
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3890906835956002,
      "learning_rate": 1.2711668738094058e-05,
      "loss": 0.6701,
      "step": 7048
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36821870255115285,
      "learning_rate": 1.270627530428661e-05,
      "loss": 0.6463,
      "step": 7049
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38479429440491036,
      "learning_rate": 1.2700882482195038e-05,
      "loss": 0.6297,
      "step": 7050
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36703558504666467,
      "learning_rate": 1.2695490272271633e-05,
      "loss": 0.6728,
      "step": 7051
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4644642441632504,
      "learning_rate": 1.2690098674968646e-05,
      "loss": 0.7095,
      "step": 7052
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4025250299586788,
      "learning_rate": 1.2684707690738253e-05,
      "loss": 0.6543,
      "step": 7053
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7291834235218045,
      "learning_rate": 1.2679317320032592e-05,
      "loss": 0.3679,
      "step": 7054
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6347479929329166,
      "learning_rate": 1.2673927563303745e-05,
      "loss": 0.3952,
      "step": 7055
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35346285643909076,
      "learning_rate": 1.2668538421003752e-05,
      "loss": 0.6479,
      "step": 7056
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3819187497316271,
      "learning_rate": 1.2663149893584577e-05,
      "loss": 0.6575,
      "step": 7057
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3481737879313125,
      "learning_rate": 1.2657761981498158e-05,
      "loss": 0.6372,
      "step": 7058
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.404213413684479,
      "learning_rate": 1.2652374685196383e-05,
      "loss": 0.7412,
      "step": 7059
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3713874207944895,
      "learning_rate": 1.264698800513107e-05,
      "loss": 0.695,
      "step": 7060
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.34790327915040115,
      "learning_rate": 1.2641601941753984e-05,
      "loss": 0.6483,
      "step": 7061
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3368242024649939,
      "learning_rate": 1.2636216495516866e-05,
      "loss": 0.6532,
      "step": 7062
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35768075873163296,
      "learning_rate": 1.2630831666871375e-05,
      "loss": 0.6924,
      "step": 7063
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7747069088521352,
      "learning_rate": 1.2625447456269132e-05,
      "loss": 0.3913,
      "step": 7064
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39608379557303214,
      "learning_rate": 1.2620063864161695e-05,
      "loss": 0.6502,
      "step": 7065
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3572218295036481,
      "learning_rate": 1.2614680891000601e-05,
      "loss": 0.6372,
      "step": 7066
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4107379698206744,
      "learning_rate": 1.26092985372373e-05,
      "loss": 0.6614,
      "step": 7067
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3719267101676062,
      "learning_rate": 1.2603916803323201e-05,
      "loss": 0.6777,
      "step": 7068
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33215241421640135,
      "learning_rate": 1.259853568970967e-05,
      "loss": 0.6624,
      "step": 7069
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3706348464127473,
      "learning_rate": 1.2593155196848011e-05,
      "loss": 0.6723,
      "step": 7070
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4166631042817325,
      "learning_rate": 1.2587775325189484e-05,
      "loss": 0.6847,
      "step": 7071
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3324369365251248,
      "learning_rate": 1.258239607518528e-05,
      "loss": 0.6259,
      "step": 7072
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3933853041711182,
      "learning_rate": 1.2577017447286566e-05,
      "loss": 0.6763,
      "step": 7073
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3464324226212296,
      "learning_rate": 1.257163944194443e-05,
      "loss": 0.646,
      "step": 7074
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3483689516274047,
      "learning_rate": 1.2566262059609928e-05,
      "loss": 0.6386,
      "step": 7075
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3272098800828924,
      "learning_rate": 1.256088530073405e-05,
      "loss": 0.6383,
      "step": 7076
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3099727934824809,
      "learning_rate": 1.2555509165767732e-05,
      "loss": 0.6208,
      "step": 7077
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36971471334932204,
      "learning_rate": 1.2550133655161867e-05,
      "loss": 0.6508,
      "step": 7078
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4103326582214985,
      "learning_rate": 1.2544758769367303e-05,
      "loss": 0.6705,
      "step": 7079
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39487914502342897,
      "learning_rate": 1.2539384508834816e-05,
      "loss": 0.7058,
      "step": 7080
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39131477618887645,
      "learning_rate": 1.2534010874015134e-05,
      "loss": 0.6067,
      "step": 7081
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3542712615096632,
      "learning_rate": 1.2528637865358952e-05,
      "loss": 0.6742,
      "step": 7082
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39883449658256154,
      "learning_rate": 1.2523265483316886e-05,
      "loss": 0.6681,
      "step": 7083
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3312772867736211,
      "learning_rate": 1.2517893728339515e-05,
      "loss": 0.6143,
      "step": 7084
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36918316630689324,
      "learning_rate": 1.251252260087735e-05,
      "loss": 0.6315,
      "step": 7085
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37455670543964925,
      "learning_rate": 1.2507152101380884e-05,
      "loss": 0.6393,
      "step": 7086
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3890636036006312,
      "learning_rate": 1.2501782230300523e-05,
      "loss": 0.6499,
      "step": 7087
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.43537503604135336,
      "learning_rate": 1.2496412988086626e-05,
      "loss": 0.6812,
      "step": 7088
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35263955468481917,
      "learning_rate": 1.2491044375189514e-05,
      "loss": 0.6439,
      "step": 7089
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3673531360827571,
      "learning_rate": 1.2485676392059438e-05,
      "loss": 0.6269,
      "step": 7090
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4116504059408823,
      "learning_rate": 1.248030903914661e-05,
      "loss": 0.6859,
      "step": 7091
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44278376640814093,
      "learning_rate": 1.2474942316901171e-05,
      "loss": 0.6783,
      "step": 7092
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3595312091218712,
      "learning_rate": 1.2469576225773245e-05,
      "loss": 0.6206,
      "step": 7093
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36946236119429743,
      "learning_rate": 1.2464210766212866e-05,
      "loss": 0.6611,
      "step": 7094
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35360385670434047,
      "learning_rate": 1.2458845938670021e-05,
      "loss": 0.6777,
      "step": 7095
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36371305638719925,
      "learning_rate": 1.245348174359467e-05,
      "loss": 0.6612,
      "step": 7096
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3989230283338681,
      "learning_rate": 1.2448118181436683e-05,
      "loss": 0.6902,
      "step": 7097
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3634016210783603,
      "learning_rate": 1.2442755252645905e-05,
      "loss": 0.6504,
      "step": 7098
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3770066650342382,
      "learning_rate": 1.2437392957672115e-05,
      "loss": 0.6197,
      "step": 7099
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38017779966899523,
      "learning_rate": 1.243203129696505e-05,
      "loss": 0.655,
      "step": 7100
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.34107218426757224,
      "learning_rate": 1.2426670270974378e-05,
      "loss": 0.6171,
      "step": 7101
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37468301992791925,
      "learning_rate": 1.2421309880149723e-05,
      "loss": 0.6489,
      "step": 7102
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35629271875817053,
      "learning_rate": 1.241595012494066e-05,
      "loss": 0.6202,
      "step": 7103
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39647032793187953,
      "learning_rate": 1.2410591005796695e-05,
      "loss": 0.7016,
      "step": 7104
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3805117198293867,
      "learning_rate": 1.2405232523167303e-05,
      "loss": 0.6781,
      "step": 7105
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3772550366874699,
      "learning_rate": 1.2399874677501878e-05,
      "loss": 0.6199,
      "step": 7106
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41611382627397014,
      "learning_rate": 1.2394517469249791e-05,
      "loss": 0.676,
      "step": 7107
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4043881273479861,
      "learning_rate": 1.2389160898860336e-05,
      "loss": 0.6325,
      "step": 7108
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38555529713861164,
      "learning_rate": 1.2383804966782771e-05,
      "loss": 0.6499,
      "step": 7109
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7941415493873449,
      "learning_rate": 1.2378449673466281e-05,
      "loss": 0.4126,
      "step": 7110
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3604834296561234,
      "learning_rate": 1.2373095019360014e-05,
      "loss": 0.6412,
      "step": 7111
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4510606993248681,
      "learning_rate": 1.2367741004913056e-05,
      "loss": 0.6755,
      "step": 7112
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37158342277876055,
      "learning_rate": 1.2362387630574438e-05,
      "loss": 0.6714,
      "step": 7113
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3717496436736187,
      "learning_rate": 1.235703489679315e-05,
      "loss": 0.6953,
      "step": 7114
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3983302440407542,
      "learning_rate": 1.2351682804018115e-05,
      "loss": 0.668,
      "step": 7115
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43352093905515143,
      "learning_rate": 1.2346331352698206e-05,
      "loss": 0.6743,
      "step": 7116
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36102013946872025,
      "learning_rate": 1.2340980543282246e-05,
      "loss": 0.6532,
      "step": 7117
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3871739399060525,
      "learning_rate": 1.2335630376218992e-05,
      "loss": 0.6476,
      "step": 7118
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41566751754867637,
      "learning_rate": 1.2330280851957168e-05,
      "loss": 0.6272,
      "step": 7119
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40138673527766394,
      "learning_rate": 1.232493197094542e-05,
      "loss": 0.6446,
      "step": 7120
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43840968630439187,
      "learning_rate": 1.2319583733632363e-05,
      "loss": 0.6727,
      "step": 7121
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4274982522424711,
      "learning_rate": 1.231423614046654e-05,
      "loss": 0.6838,
      "step": 7122
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40464473699388154,
      "learning_rate": 1.2308889191896454e-05,
      "loss": 0.6705,
      "step": 7123
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37113451609014697,
      "learning_rate": 1.230354288837054e-05,
      "loss": 0.6524,
      "step": 7124
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41066315023835925,
      "learning_rate": 1.229819723033719e-05,
      "loss": 0.6595,
      "step": 7125
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38183106949883966,
      "learning_rate": 1.229285221824474e-05,
      "loss": 0.6484,
      "step": 7126
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35970460413658367,
      "learning_rate": 1.2287507852541456e-05,
      "loss": 0.6498,
      "step": 7127
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44540265337460944,
      "learning_rate": 1.2282164133675583e-05,
      "loss": 0.6864,
      "step": 7128
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44758705622973066,
      "learning_rate": 1.227682106209528e-05,
      "loss": 0.6906,
      "step": 7129
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4173146023267157,
      "learning_rate": 1.2271478638248671e-05,
      "loss": 0.6276,
      "step": 7130
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.390183339195612,
      "learning_rate": 1.2266136862583811e-05,
      "loss": 0.6394,
      "step": 7131
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4092539380528033,
      "learning_rate": 1.2260795735548716e-05,
      "loss": 0.7062,
      "step": 7132
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40875422415370305,
      "learning_rate": 1.2255455257591335e-05,
      "loss": 0.6523,
      "step": 7133
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44029032492143233,
      "learning_rate": 1.2250115429159562e-05,
      "loss": 0.6866,
      "step": 7134
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4055803674238316,
      "learning_rate": 1.2244776250701252e-05,
      "loss": 0.6605,
      "step": 7135
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3931961366435444,
      "learning_rate": 1.2239437722664195e-05,
      "loss": 0.6577,
      "step": 7136
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4255381859797202,
      "learning_rate": 1.2234099845496125e-05,
      "loss": 0.6436,
      "step": 7137
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36518967109098827,
      "learning_rate": 1.2228762619644718e-05,
      "loss": 0.6496,
      "step": 7138
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34560662375837825,
      "learning_rate": 1.222342604555761e-05,
      "loss": 0.6361,
      "step": 7139
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4191078610323384,
      "learning_rate": 1.2218090123682367e-05,
      "loss": 0.6703,
      "step": 7140
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4335311810347461,
      "learning_rate": 1.2212754854466495e-05,
      "loss": 0.6669,
      "step": 7141
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4157647489405416,
      "learning_rate": 1.2207420238357478e-05,
      "loss": 0.6121,
      "step": 7142
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3353279447036604,
      "learning_rate": 1.2202086275802718e-05,
      "loss": 0.6711,
      "step": 7143
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3700798956513911,
      "learning_rate": 1.2196752967249564e-05,
      "loss": 0.6421,
      "step": 7144
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3488060643601359,
      "learning_rate": 1.2191420313145309e-05,
      "loss": 0.6344,
      "step": 7145
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40975971212060247,
      "learning_rate": 1.2186088313937208e-05,
      "loss": 0.6327,
      "step": 7146
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36020351149785024,
      "learning_rate": 1.2180756970072443e-05,
      "loss": 0.6701,
      "step": 7147
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35728839625181014,
      "learning_rate": 1.2175426281998137e-05,
      "loss": 0.6547,
      "step": 7148
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39643813207767886,
      "learning_rate": 1.2170096250161386e-05,
      "loss": 0.6764,
      "step": 7149
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3740893529582047,
      "learning_rate": 1.2164766875009208e-05,
      "loss": 0.6408,
      "step": 7150
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6208876378658454,
      "learning_rate": 1.2159438156988573e-05,
      "loss": 0.3996,
      "step": 7151
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36713826420476303,
      "learning_rate": 1.2154110096546386e-05,
      "loss": 0.6268,
      "step": 7152
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3729030097998883,
      "learning_rate": 1.2148782694129513e-05,
      "loss": 0.6508,
      "step": 7153
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37404502507434323,
      "learning_rate": 1.2143455950184752e-05,
      "loss": 0.6767,
      "step": 7154
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3575326503800695,
      "learning_rate": 1.2138129865158845e-05,
      "loss": 0.6673,
      "step": 7155
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40685097288624494,
      "learning_rate": 1.2132804439498496e-05,
      "loss": 0.662,
      "step": 7156
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38824247163807357,
      "learning_rate": 1.2127479673650342e-05,
      "loss": 0.6646,
      "step": 7157
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3519252916302765,
      "learning_rate": 1.212215556806096e-05,
      "loss": 0.6112,
      "step": 7158
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4032220570049776,
      "learning_rate": 1.211683212317687e-05,
      "loss": 0.6295,
      "step": 7159
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3407667420994673,
      "learning_rate": 1.2111509339444556e-05,
      "loss": 0.6363,
      "step": 7160
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3941294131282422,
      "learning_rate": 1.210618721731042e-05,
      "loss": 0.6704,
      "step": 7161
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3827897314729407,
      "learning_rate": 1.2100865757220835e-05,
      "loss": 0.67,
      "step": 7162
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4204793760418244,
      "learning_rate": 1.2095544959622095e-05,
      "loss": 0.6996,
      "step": 7163
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3611201660611975,
      "learning_rate": 1.2090224824960455e-05,
      "loss": 0.6499,
      "step": 7164
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3548129089387144,
      "learning_rate": 1.2084905353682106e-05,
      "loss": 0.6393,
      "step": 7165
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35578901716908135,
      "learning_rate": 1.2079586546233188e-05,
      "loss": 0.6557,
      "step": 7166
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38262025727434557,
      "learning_rate": 1.207426840305978e-05,
      "loss": 0.6396,
      "step": 7167
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4028240767825606,
      "learning_rate": 1.2068950924607905e-05,
      "loss": 0.6684,
      "step": 7168
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37451599022899534,
      "learning_rate": 1.2063634111323538e-05,
      "loss": 0.6509,
      "step": 7169
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3929488176956665,
      "learning_rate": 1.2058317963652599e-05,
      "loss": 0.6505,
      "step": 7170
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.355661647843933,
      "learning_rate": 1.2053002482040941e-05,
      "loss": 0.6421,
      "step": 7171
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4225533709107706,
      "learning_rate": 1.2047687666934365e-05,
      "loss": 0.6963,
      "step": 7172
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3818545872834764,
      "learning_rate": 1.204237351877862e-05,
      "loss": 0.6555,
      "step": 7173
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4197882526429991,
      "learning_rate": 1.2037060038019402e-05,
      "loss": 0.6536,
      "step": 7174
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4132972344481147,
      "learning_rate": 1.2031747225102332e-05,
      "loss": 0.6678,
      "step": 7175
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39864449596515034,
      "learning_rate": 1.2026435080473005e-05,
      "loss": 0.6504,
      "step": 7176
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4015456292041813,
      "learning_rate": 1.202112360457694e-05,
      "loss": 0.689,
      "step": 7177
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33640359213916915,
      "learning_rate": 1.20158127978596e-05,
      "loss": 0.6511,
      "step": 7178
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39586456761026334,
      "learning_rate": 1.20105026607664e-05,
      "loss": 0.6858,
      "step": 7179
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40235446434537114,
      "learning_rate": 1.2005193193742692e-05,
      "loss": 0.6614,
      "step": 7180
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3581283616066295,
      "learning_rate": 1.1999884397233776e-05,
      "loss": 0.6576,
      "step": 7181
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37384477951059325,
      "learning_rate": 1.1994576271684887e-05,
      "loss": 0.634,
      "step": 7182
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3999529867377945,
      "learning_rate": 1.1989268817541225e-05,
      "loss": 0.6845,
      "step": 7183
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36130908776793036,
      "learning_rate": 1.1983962035247911e-05,
      "loss": 0.6711,
      "step": 7184
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4393498161242087,
      "learning_rate": 1.1978655925250024e-05,
      "loss": 0.6712,
      "step": 7185
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3483583712984945,
      "learning_rate": 1.1973350487992572e-05,
      "loss": 0.6332,
      "step": 7186
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3749933131521836,
      "learning_rate": 1.1968045723920528e-05,
      "loss": 0.6827,
      "step": 7187
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41758680353948124,
      "learning_rate": 1.1962741633478779e-05,
      "loss": 0.6882,
      "step": 7188
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34574129229127604,
      "learning_rate": 1.1957438217112187e-05,
      "loss": 0.6276,
      "step": 7189
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34416350645887833,
      "learning_rate": 1.1952135475265538e-05,
      "loss": 0.6343,
      "step": 7190
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34682119857378196,
      "learning_rate": 1.1946833408383572e-05,
      "loss": 0.6415,
      "step": 7191
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3355275861970978,
      "learning_rate": 1.1941532016910955e-05,
      "loss": 0.6263,
      "step": 7192
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3899234839777463,
      "learning_rate": 1.1936231301292322e-05,
      "loss": 0.6372,
      "step": 7193
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3911382745573858,
      "learning_rate": 1.1930931261972231e-05,
      "loss": 0.657,
      "step": 7194
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38406206120411707,
      "learning_rate": 1.1925631899395187e-05,
      "loss": 0.664,
      "step": 7195
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39198390635611724,
      "learning_rate": 1.1920333214005639e-05,
      "loss": 0.6636,
      "step": 7196
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40284776093321667,
      "learning_rate": 1.1915035206247994e-05,
      "loss": 0.6256,
      "step": 7197
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3573772191106524,
      "learning_rate": 1.1909737876566582e-05,
      "loss": 0.6744,
      "step": 7198
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37357310727465565,
      "learning_rate": 1.1904441225405684e-05,
      "loss": 0.6494,
      "step": 7199
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3564568062353889,
      "learning_rate": 1.1899145253209521e-05,
      "loss": 0.6535,
      "step": 7200
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35551558621447593,
      "learning_rate": 1.1893849960422265e-05,
      "loss": 0.6425,
      "step": 7201
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4139137342508386,
      "learning_rate": 1.188855534748802e-05,
      "loss": 0.6644,
      "step": 7202
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43290705092685655,
      "learning_rate": 1.1883261414850837e-05,
      "loss": 0.7135,
      "step": 7203
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41290024297013417,
      "learning_rate": 1.1877968162954727e-05,
      "loss": 0.6416,
      "step": 7204
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.385500411271751,
      "learning_rate": 1.1872675592243615e-05,
      "loss": 0.723,
      "step": 7205
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38271031232803887,
      "learning_rate": 1.1867383703161383e-05,
      "loss": 0.6462,
      "step": 7206
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.42684142775282863,
      "learning_rate": 1.1862092496151862e-05,
      "loss": 0.6815,
      "step": 7207
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3314511526849946,
      "learning_rate": 1.1856801971658815e-05,
      "loss": 0.6316,
      "step": 7208
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35622629044426857,
      "learning_rate": 1.1851512130125951e-05,
      "loss": 0.6563,
      "step": 7209
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43787943924552797,
      "learning_rate": 1.1846222971996916e-05,
      "loss": 0.7239,
      "step": 7210
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36306212138642807,
      "learning_rate": 1.1840934497715322e-05,
      "loss": 0.6384,
      "step": 7211
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3789773394197976,
      "learning_rate": 1.1835646707724698e-05,
      "loss": 0.6549,
      "step": 7212
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36645960160815183,
      "learning_rate": 1.1830359602468522e-05,
      "loss": 0.662,
      "step": 7213
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3589364183240015,
      "learning_rate": 1.1825073182390221e-05,
      "loss": 0.6429,
      "step": 7214
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.383893242842123,
      "learning_rate": 1.181978744793316e-05,
      "loss": 0.6283,
      "step": 7215
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3900851995591668,
      "learning_rate": 1.1814502399540646e-05,
      "loss": 0.6549,
      "step": 7216
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3695136590701065,
      "learning_rate": 1.1809218037655923e-05,
      "loss": 0.6532,
      "step": 7217
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3989342391502995,
      "learning_rate": 1.1803934362722199e-05,
      "loss": 0.6677,
      "step": 7218
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39033785487412437,
      "learning_rate": 1.1798651375182604e-05,
      "loss": 0.6447,
      "step": 7219
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4107575050787189,
      "learning_rate": 1.1793369075480205e-05,
      "loss": 0.6323,
      "step": 7220
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40856012929142227,
      "learning_rate": 1.178808746405804e-05,
      "loss": 0.6621,
      "step": 7221
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36664141198347944,
      "learning_rate": 1.1782806541359055e-05,
      "loss": 0.6822,
      "step": 7222
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3993424172114376,
      "learning_rate": 1.1777526307826166e-05,
      "loss": 0.6932,
      "step": 7223
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36461467702774203,
      "learning_rate": 1.1772246763902206e-05,
      "loss": 0.677,
      "step": 7224
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40132708711228304,
      "learning_rate": 1.1766967910029985e-05,
      "loss": 0.6802,
      "step": 7225
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.362393150704086,
      "learning_rate": 1.1761689746652216e-05,
      "loss": 0.6579,
      "step": 7226
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39321711004797033,
      "learning_rate": 1.1756412274211581e-05,
      "loss": 0.6811,
      "step": 7227
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41859722562027657,
      "learning_rate": 1.1751135493150696e-05,
      "loss": 0.6412,
      "step": 7228
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36885740993124455,
      "learning_rate": 1.1745859403912108e-05,
      "loss": 0.6527,
      "step": 7229
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3959743954006356,
      "learning_rate": 1.174058400693833e-05,
      "loss": 0.6496,
      "step": 7230
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3640709953460773,
      "learning_rate": 1.1735309302671786e-05,
      "loss": 0.6343,
      "step": 7231
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35104252670851926,
      "learning_rate": 1.173003529155488e-05,
      "loss": 0.6491,
      "step": 7232
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4199431723555471,
      "learning_rate": 1.1724761974029918e-05,
      "loss": 0.7019,
      "step": 7233
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37008469715231507,
      "learning_rate": 1.1719489350539185e-05,
      "loss": 0.6261,
      "step": 7234
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3821776438549922,
      "learning_rate": 1.1714217421524878e-05,
      "loss": 0.6434,
      "step": 7235
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4188109970288183,
      "learning_rate": 1.1708946187429145e-05,
      "loss": 0.6828,
      "step": 7236
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3821216997970765,
      "learning_rate": 1.1703675648694083e-05,
      "loss": 0.642,
      "step": 7237
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39182859706282325,
      "learning_rate": 1.1698405805761723e-05,
      "loss": 0.7053,
      "step": 7238
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40769604394398984,
      "learning_rate": 1.1693136659074048e-05,
      "loss": 0.6683,
      "step": 7239
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35666465090240423,
      "learning_rate": 1.1687868209072964e-05,
      "loss": 0.6654,
      "step": 7240
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3884280775572331,
      "learning_rate": 1.1682600456200337e-05,
      "loss": 0.657,
      "step": 7241
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4128784404922895,
      "learning_rate": 1.1677333400897968e-05,
      "loss": 0.687,
      "step": 7242
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3768177239012927,
      "learning_rate": 1.1672067043607593e-05,
      "loss": 0.6396,
      "step": 7243
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3700212467612773,
      "learning_rate": 1.1666801384770889e-05,
      "loss": 0.6674,
      "step": 7244
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.332254228523482,
      "learning_rate": 1.1661536424829493e-05,
      "loss": 0.6534,
      "step": 7245
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39306345755049343,
      "learning_rate": 1.1656272164224978e-05,
      "loss": 0.6516,
      "step": 7246
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4210932841830402,
      "learning_rate": 1.165100860339883e-05,
      "loss": 0.6419,
      "step": 7247
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36420735205155247,
      "learning_rate": 1.164574574279251e-05,
      "loss": 0.6484,
      "step": 7248
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39191358129633413,
      "learning_rate": 1.1640483582847408e-05,
      "loss": 0.6642,
      "step": 7249
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36196160382002657,
      "learning_rate": 1.1635222124004848e-05,
      "loss": 0.6457,
      "step": 7250
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3675645609915014,
      "learning_rate": 1.1629961366706106e-05,
      "loss": 0.616,
      "step": 7251
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37959397655625193,
      "learning_rate": 1.1624701311392395e-05,
      "loss": 0.6394,
      "step": 7252
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42606357806918344,
      "learning_rate": 1.1619441958504871e-05,
      "loss": 0.7058,
      "step": 7253
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4078899147980299,
      "learning_rate": 1.1614183308484637e-05,
      "loss": 0.6725,
      "step": 7254
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6083869348632464,
      "learning_rate": 1.1608925361772712e-05,
      "loss": 0.399,
      "step": 7255
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3689654173895416,
      "learning_rate": 1.1603668118810084e-05,
      "loss": 0.6439,
      "step": 7256
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38952664218671557,
      "learning_rate": 1.1598411580037677e-05,
      "loss": 0.651,
      "step": 7257
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39198797449154826,
      "learning_rate": 1.159315574589633e-05,
      "loss": 0.6853,
      "step": 7258
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31984509735536826,
      "learning_rate": 1.1587900616826864e-05,
      "loss": 0.6561,
      "step": 7259
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.371610831974684,
      "learning_rate": 1.1582646193270024e-05,
      "loss": 0.661,
      "step": 7260
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4005429476526965,
      "learning_rate": 1.1577392475666473e-05,
      "loss": 0.6332,
      "step": 7261
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.44104381122976255,
      "learning_rate": 1.1572139464456846e-05,
      "loss": 0.6469,
      "step": 7262
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38462106919949396,
      "learning_rate": 1.1566887160081708e-05,
      "loss": 0.661,
      "step": 7263
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4119864229524075,
      "learning_rate": 1.1561635562981555e-05,
      "loss": 0.6668,
      "step": 7264
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.448747740659616,
      "learning_rate": 1.1556384673596832e-05,
      "loss": 0.7205,
      "step": 7265
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35580102815334647,
      "learning_rate": 1.1551134492367933e-05,
      "loss": 0.6837,
      "step": 7266
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4166455756354626,
      "learning_rate": 1.1545885019735176e-05,
      "loss": 0.648,
      "step": 7267
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37931876879103865,
      "learning_rate": 1.1540636256138837e-05,
      "loss": 0.6707,
      "step": 7268
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36245151111182533,
      "learning_rate": 1.1535388202019125e-05,
      "loss": 0.6806,
      "step": 7269
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39250668237831426,
      "learning_rate": 1.1530140857816172e-05,
      "loss": 0.6746,
      "step": 7270
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3953686162057185,
      "learning_rate": 1.1524894223970077e-05,
      "loss": 0.5994,
      "step": 7271
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3660017202490788,
      "learning_rate": 1.1519648300920872e-05,
      "loss": 0.6779,
      "step": 7272
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33615311063401043,
      "learning_rate": 1.151440308910852e-05,
      "loss": 0.6484,
      "step": 7273
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38173945456502134,
      "learning_rate": 1.150915858897294e-05,
      "loss": 0.6486,
      "step": 7274
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37386254562243887,
      "learning_rate": 1.1503914800953969e-05,
      "loss": 0.6762,
      "step": 7275
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3634330738740288,
      "learning_rate": 1.14986717254914e-05,
      "loss": 0.6392,
      "step": 7276
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6181362884249523,
      "learning_rate": 1.1493429363024975e-05,
      "loss": 0.4109,
      "step": 7277
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3845847075105887,
      "learning_rate": 1.148818771399435e-05,
      "loss": 0.7063,
      "step": 7278
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39117550423624936,
      "learning_rate": 1.148294677883913e-05,
      "loss": 0.6795,
      "step": 7279
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4213033327971479,
      "learning_rate": 1.1477706557998895e-05,
      "loss": 0.6587,
      "step": 7280
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37904549927751685,
      "learning_rate": 1.1472467051913112e-05,
      "loss": 0.6507,
      "step": 7281
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3397777965715359,
      "learning_rate": 1.146722826102122e-05,
      "loss": 0.6562,
      "step": 7282
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39796322747368706,
      "learning_rate": 1.1461990185762595e-05,
      "loss": 0.6582,
      "step": 7283
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3624812510746516,
      "learning_rate": 1.1456752826576535e-05,
      "loss": 0.6933,
      "step": 7284
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37120320262885326,
      "learning_rate": 1.1451516183902295e-05,
      "loss": 0.6303,
      "step": 7285
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36929551086161627,
      "learning_rate": 1.144628025817907e-05,
      "loss": 0.6557,
      "step": 7286
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5824917865480262,
      "learning_rate": 1.144104504984599e-05,
      "loss": 0.375,
      "step": 7287
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33041231639663887,
      "learning_rate": 1.1435810559342136e-05,
      "loss": 0.6356,
      "step": 7288
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42946207679019965,
      "learning_rate": 1.1430576787106492e-05,
      "loss": 0.6876,
      "step": 7289
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3965469627146834,
      "learning_rate": 1.1425343733578029e-05,
      "loss": 0.6149,
      "step": 7290
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36073634564521134,
      "learning_rate": 1.1420111399195633e-05,
      "loss": 0.6587,
      "step": 7291
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40011586081299205,
      "learning_rate": 1.1414879784398136e-05,
      "loss": 0.6694,
      "step": 7292
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3884067084523315,
      "learning_rate": 1.140964888962429e-05,
      "loss": 0.7008,
      "step": 7293
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37463195417835493,
      "learning_rate": 1.1404418715312832e-05,
      "loss": 0.6902,
      "step": 7294
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3840435258661708,
      "learning_rate": 1.1399189261902387e-05,
      "loss": 0.6522,
      "step": 7295
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3758690345680278,
      "learning_rate": 1.1393960529831553e-05,
      "loss": 0.6479,
      "step": 7296
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41075157553303393,
      "learning_rate": 1.138873251953886e-05,
      "loss": 0.6993,
      "step": 7297
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37300273102446563,
      "learning_rate": 1.1383505231462768e-05,
      "loss": 0.6567,
      "step": 7298
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37494315356572505,
      "learning_rate": 1.1378278666041684e-05,
      "loss": 0.6385,
      "step": 7299
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3841934853401548,
      "learning_rate": 1.1373052823713957e-05,
      "loss": 0.6243,
      "step": 7300
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3833979933468831,
      "learning_rate": 1.1367827704917868e-05,
      "loss": 0.5778,
      "step": 7301
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.407919485894522,
      "learning_rate": 1.1362603310091647e-05,
      "loss": 0.6083,
      "step": 7302
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3616567732465642,
      "learning_rate": 1.1357379639673463e-05,
      "loss": 0.643,
      "step": 7303
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3670705059130043,
      "learning_rate": 1.13521566941014e-05,
      "loss": 0.6546,
      "step": 7304
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39311200445219113,
      "learning_rate": 1.1346934473813516e-05,
      "loss": 0.6461,
      "step": 7305
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37771358927663395,
      "learning_rate": 1.134171297924779e-05,
      "loss": 0.6678,
      "step": 7306
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33879409786686177,
      "learning_rate": 1.1336492210842127e-05,
      "loss": 0.6442,
      "step": 7307
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3772706279014245,
      "learning_rate": 1.1331272169034415e-05,
      "loss": 0.6585,
      "step": 7308
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39337448480682724,
      "learning_rate": 1.132605285426243e-05,
      "loss": 0.6664,
      "step": 7309
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3417549879037791,
      "learning_rate": 1.1320834266963914e-05,
      "loss": 0.6482,
      "step": 7310
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40044648843038366,
      "learning_rate": 1.1315616407576555e-05,
      "loss": 0.6556,
      "step": 7311
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36956843904284403,
      "learning_rate": 1.1310399276537956e-05,
      "loss": 0.6295,
      "step": 7312
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3803131530145928,
      "learning_rate": 1.1305182874285669e-05,
      "loss": 0.5924,
      "step": 7313
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36301555286860204,
      "learning_rate": 1.1299967201257198e-05,
      "loss": 0.6446,
      "step": 7314
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36394418453623595,
      "learning_rate": 1.129475225788997e-05,
      "loss": 0.6382,
      "step": 7315
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3616910244446297,
      "learning_rate": 1.1289538044621355e-05,
      "loss": 0.6391,
      "step": 7316
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3627807709511747,
      "learning_rate": 1.1284324561888672e-05,
      "loss": 0.6001,
      "step": 7317
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39510995390670756,
      "learning_rate": 1.127911181012916e-05,
      "loss": 0.6898,
      "step": 7318
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3511400436919915,
      "learning_rate": 1.1273899789780003e-05,
      "loss": 0.647,
      "step": 7319
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36723329380429054,
      "learning_rate": 1.126868850127834e-05,
      "loss": 0.6719,
      "step": 7320
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3940443159813259,
      "learning_rate": 1.1263477945061214e-05,
      "loss": 0.6506,
      "step": 7321
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3661365924720159,
      "learning_rate": 1.1258268121565656e-05,
      "loss": 0.6344,
      "step": 7322
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4522209081072032,
      "learning_rate": 1.1253059031228589e-05,
      "loss": 0.7147,
      "step": 7323
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40240588667329785,
      "learning_rate": 1.1247850674486894e-05,
      "loss": 0.6925,
      "step": 7324
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3677125120142994,
      "learning_rate": 1.1242643051777393e-05,
      "loss": 0.6713,
      "step": 7325
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4185239789075223,
      "learning_rate": 1.1237436163536853e-05,
      "loss": 0.6875,
      "step": 7326
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5985668711826418,
      "learning_rate": 1.123223001020195e-05,
      "loss": 0.4041,
      "step": 7327
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.39592568898149083,
      "learning_rate": 1.1227024592209326e-05,
      "loss": 0.6495,
      "step": 7328
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3797684948354436,
      "learning_rate": 1.1221819909995558e-05,
      "loss": 0.6598,
      "step": 7329
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4045418662286048,
      "learning_rate": 1.1216615963997148e-05,
      "loss": 0.6544,
      "step": 7330
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3769418149710266,
      "learning_rate": 1.1211412754650559e-05,
      "loss": 0.664,
      "step": 7331
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4230139047299097,
      "learning_rate": 1.1206210282392162e-05,
      "loss": 0.7221,
      "step": 7332
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4148380115020535,
      "learning_rate": 1.1201008547658287e-05,
      "loss": 0.6915,
      "step": 7333
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3741792251545366,
      "learning_rate": 1.11958075508852e-05,
      "loss": 0.6216,
      "step": 7334
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3969042135838683,
      "learning_rate": 1.1190607292509101e-05,
      "loss": 0.6152,
      "step": 7335
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3944009611823756,
      "learning_rate": 1.1185407772966129e-05,
      "loss": 0.6441,
      "step": 7336
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4148497900908925,
      "learning_rate": 1.118020899269237e-05,
      "loss": 0.6653,
      "step": 7337
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3634229414320191,
      "learning_rate": 1.1175010952123821e-05,
      "loss": 0.6481,
      "step": 7338
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3922144850401135,
      "learning_rate": 1.1169813651696448e-05,
      "loss": 0.6633,
      "step": 7339
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3822364911574785,
      "learning_rate": 1.1164617091846149e-05,
      "loss": 0.6214,
      "step": 7340
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4216101605054428,
      "learning_rate": 1.1159421273008723e-05,
      "loss": 0.683,
      "step": 7341
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42243288977689375,
      "learning_rate": 1.1154226195619979e-05,
      "loss": 0.6629,
      "step": 7342
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4088123324830311,
      "learning_rate": 1.114903186011559e-05,
      "loss": 0.6531,
      "step": 7343
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3625088405450067,
      "learning_rate": 1.114383826693121e-05,
      "loss": 0.6268,
      "step": 7344
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3856850915076508,
      "learning_rate": 1.113864541650243e-05,
      "loss": 0.6574,
      "step": 7345
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35173319030313055,
      "learning_rate": 1.1133453309264745e-05,
      "loss": 0.6477,
      "step": 7346
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42103866784457855,
      "learning_rate": 1.1128261945653627e-05,
      "loss": 0.6938,
      "step": 7347
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3960647492348902,
      "learning_rate": 1.1123071326104462e-05,
      "loss": 0.6376,
      "step": 7348
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40702147002092953,
      "learning_rate": 1.1117881451052583e-05,
      "loss": 0.6641,
      "step": 7349
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.34357061859970645,
      "learning_rate": 1.1112692320933265e-05,
      "loss": 0.6722,
      "step": 7350
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37561224734478926,
      "learning_rate": 1.1107503936181715e-05,
      "loss": 0.6313,
      "step": 7351
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35479615705366824,
      "learning_rate": 1.110231629723306e-05,
      "loss": 0.6736,
      "step": 7352
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3738028453737473,
      "learning_rate": 1.1097129404522394e-05,
      "loss": 0.6445,
      "step": 7353
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.39022671400228415,
      "learning_rate": 1.1091943258484739e-05,
      "loss": 0.6719,
      "step": 7354
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4083462300532891,
      "learning_rate": 1.1086757859555028e-05,
      "loss": 0.6576,
      "step": 7355
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.48922326136655975,
      "learning_rate": 1.1081573208168181e-05,
      "loss": 0.6427,
      "step": 7356
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.41707629193046203,
      "learning_rate": 1.1076389304759028e-05,
      "loss": 0.6801,
      "step": 7357
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3978730685711791,
      "learning_rate": 1.1071206149762315e-05,
      "loss": 0.6455,
      "step": 7358
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3370654251621542,
      "learning_rate": 1.1066023743612759e-05,
      "loss": 0.6646,
      "step": 7359
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4037245310843265,
      "learning_rate": 1.1060842086745014e-05,
      "loss": 0.6523,
      "step": 7360
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32792940588925557,
      "learning_rate": 1.1055661179593638e-05,
      "loss": 0.6199,
      "step": 7361
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4399361068885234,
      "learning_rate": 1.1050481022593154e-05,
      "loss": 0.6923,
      "step": 7362
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3519538658956133,
      "learning_rate": 1.104530161617802e-05,
      "loss": 0.6311,
      "step": 7363
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37049633700853535,
      "learning_rate": 1.1040122960782626e-05,
      "loss": 0.6392,
      "step": 7364
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40568008653786,
      "learning_rate": 1.1034945056841306e-05,
      "loss": 0.6639,
      "step": 7365
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.44468842106468764,
      "learning_rate": 1.102976790478831e-05,
      "loss": 0.6517,
      "step": 7366
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3690180221902822,
      "learning_rate": 1.1024591505057845e-05,
      "loss": 0.6248,
      "step": 7367
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36206199798128097,
      "learning_rate": 1.101941585808406e-05,
      "loss": 0.6226,
      "step": 7368
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.41595406409220165,
      "learning_rate": 1.1014240964301001e-05,
      "loss": 0.6905,
      "step": 7369
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3821866829125698,
      "learning_rate": 1.1009066824142713e-05,
      "loss": 0.6487,
      "step": 7370
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3382147267017413,
      "learning_rate": 1.1003893438043138e-05,
      "loss": 0.6667,
      "step": 7371
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3686601334630612,
      "learning_rate": 1.0998720806436149e-05,
      "loss": 0.6509,
      "step": 7372
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37555967893705394,
      "learning_rate": 1.0993548929755576e-05,
      "loss": 0.6769,
      "step": 7373
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37171233971285,
      "learning_rate": 1.0988377808435184e-05,
      "loss": 0.6778,
      "step": 7374
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.38871021379675186,
      "learning_rate": 1.0983207442908654e-05,
      "loss": 0.6631,
      "step": 7375
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3725767938116196,
      "learning_rate": 1.0978037833609627e-05,
      "loss": 0.6456,
      "step": 7376
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42398078787249077,
      "learning_rate": 1.0972868980971669e-05,
      "loss": 0.6385,
      "step": 7377
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37225323592547904,
      "learning_rate": 1.096770088542829e-05,
      "loss": 0.6252,
      "step": 7378
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42945946347446845,
      "learning_rate": 1.0962533547412937e-05,
      "loss": 0.6426,
      "step": 7379
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3837172666498463,
      "learning_rate": 1.0957366967358971e-05,
      "loss": 0.6837,
      "step": 7380
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3672813861573344,
      "learning_rate": 1.0952201145699719e-05,
      "loss": 0.6342,
      "step": 7381
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3824631573594955,
      "learning_rate": 1.0947036082868427e-05,
      "loss": 0.6449,
      "step": 7382
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3903543739818308,
      "learning_rate": 1.0941871779298289e-05,
      "loss": 0.6863,
      "step": 7383
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3607991692554497,
      "learning_rate": 1.0936708235422425e-05,
      "loss": 0.6387,
      "step": 7384
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4047740220185367,
      "learning_rate": 1.0931545451673902e-05,
      "loss": 0.6711,
      "step": 7385
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33949289041030273,
      "learning_rate": 1.0926383428485704e-05,
      "loss": 0.6517,
      "step": 7386
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.39851198618856365,
      "learning_rate": 1.0921222166290768e-05,
      "loss": 0.6467,
      "step": 7387
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.39264897733318993,
      "learning_rate": 1.0916061665521971e-05,
      "loss": 0.6235,
      "step": 7388
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4157587130548841,
      "learning_rate": 1.0910901926612105e-05,
      "loss": 0.6498,
      "step": 7389
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40830153895324955,
      "learning_rate": 1.090574294999391e-05,
      "loss": 0.6844,
      "step": 7390
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3514017353015487,
      "learning_rate": 1.0900584736100086e-05,
      "loss": 0.6258,
      "step": 7391
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3866446304777923,
      "learning_rate": 1.0895427285363224e-05,
      "loss": 0.6626,
      "step": 7392
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.29200035822201326,
      "learning_rate": 1.089027059821588e-05,
      "loss": 0.5895,
      "step": 7393
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37412218562594135,
      "learning_rate": 1.0885114675090546e-05,
      "loss": 0.6974,
      "step": 7394
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42459829323082826,
      "learning_rate": 1.0879959516419629e-05,
      "loss": 0.6618,
      "step": 7395
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33180064172106455,
      "learning_rate": 1.0874805122635492e-05,
      "loss": 0.6168,
      "step": 7396
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3934725486469213,
      "learning_rate": 1.086965149417043e-05,
      "loss": 0.67,
      "step": 7397
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37266855768395757,
      "learning_rate": 1.0864498631456672e-05,
      "loss": 0.6643,
      "step": 7398
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32279759230833827,
      "learning_rate": 1.0859346534926387e-05,
      "loss": 0.6319,
      "step": 7399
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.41232410861226765,
      "learning_rate": 1.0854195205011662e-05,
      "loss": 0.6427,
      "step": 7400
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3688095899241346,
      "learning_rate": 1.084904464214454e-05,
      "loss": 0.6165,
      "step": 7401
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.38462902335636207,
      "learning_rate": 1.0843894846756999e-05,
      "loss": 0.6711,
      "step": 7402
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40402768179374826,
      "learning_rate": 1.083874581928093e-05,
      "loss": 0.674,
      "step": 7403
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3515704983303314,
      "learning_rate": 1.0833597560148181e-05,
      "loss": 0.669,
      "step": 7404
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37191464292242604,
      "learning_rate": 1.0828450069790546e-05,
      "loss": 0.658,
      "step": 7405
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43156642510998794,
      "learning_rate": 1.0823303348639725e-05,
      "loss": 0.6965,
      "step": 7406
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4185146997638674,
      "learning_rate": 1.0818157397127365e-05,
      "loss": 0.6668,
      "step": 7407
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3746646522796371,
      "learning_rate": 1.0813012215685064e-05,
      "loss": 0.6535,
      "step": 7408
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3774670790266697,
      "learning_rate": 1.0807867804744326e-05,
      "loss": 0.6519,
      "step": 7409
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37725726091577655,
      "learning_rate": 1.0802724164736614e-05,
      "loss": 0.6684,
      "step": 7410
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37854994121979957,
      "learning_rate": 1.0797581296093317e-05,
      "loss": 0.6528,
      "step": 7411
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3864365148749639,
      "learning_rate": 1.0792439199245766e-05,
      "loss": 0.6386,
      "step": 7412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3943882171837861,
      "learning_rate": 1.0787297874625216e-05,
      "loss": 0.6304,
      "step": 7413
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3298691780027351,
      "learning_rate": 1.0782157322662879e-05,
      "loss": 0.6449,
      "step": 7414
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.38975569241873453,
      "learning_rate": 1.0777017543789863e-05,
      "loss": 0.6497,
      "step": 7415
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4479834002656451,
      "learning_rate": 1.0771878538437251e-05,
      "loss": 0.6927,
      "step": 7416
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.618644241021425,
      "learning_rate": 1.0766740307036039e-05,
      "loss": 0.3936,
      "step": 7417
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36768902580744106,
      "learning_rate": 1.0761602850017167e-05,
      "loss": 0.6283,
      "step": 7418
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.362181132230673,
      "learning_rate": 1.0756466167811514e-05,
      "loss": 0.6642,
      "step": 7419
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35530595032513995,
      "learning_rate": 1.0751330260849873e-05,
      "loss": 0.6497,
      "step": 7420
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.332561738840002,
      "learning_rate": 1.0746195129562993e-05,
      "loss": 0.6288,
      "step": 7421
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5681915020836983,
      "learning_rate": 1.074106077438156e-05,
      "loss": 0.3666,
      "step": 7422
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37666937223272706,
      "learning_rate": 1.0735927195736169e-05,
      "loss": 0.6635,
      "step": 7423
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6240996594731607,
      "learning_rate": 1.0730794394057372e-05,
      "loss": 0.4048,
      "step": 7424
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3679802697576497,
      "learning_rate": 1.0725662369775666e-05,
      "loss": 0.6464,
      "step": 7425
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42410651821240214,
      "learning_rate": 1.072053112332145e-05,
      "loss": 0.6673,
      "step": 7426
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.400525588260957,
      "learning_rate": 1.071540065512508e-05,
      "loss": 0.6241,
      "step": 7427
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4054006182669078,
      "learning_rate": 1.0710270965616856e-05,
      "loss": 0.6509,
      "step": 7428
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32685257314962163,
      "learning_rate": 1.0705142055226979e-05,
      "loss": 0.631,
      "step": 7429
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3792922960449371,
      "learning_rate": 1.0700013924385609e-05,
      "loss": 0.6581,
      "step": 7430
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35404732370515046,
      "learning_rate": 1.069488657352284e-05,
      "loss": 0.6456,
      "step": 7431
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3322727924081505,
      "learning_rate": 1.0689760003068698e-05,
      "loss": 0.6755,
      "step": 7432
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35594583776082017,
      "learning_rate": 1.0684634213453149e-05,
      "loss": 0.6239,
      "step": 7433
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35289111126679873,
      "learning_rate": 1.0679509205106068e-05,
      "loss": 0.625,
      "step": 7434
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4009232652999035,
      "learning_rate": 1.0674384978457297e-05,
      "loss": 0.6508,
      "step": 7435
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3384867099512607,
      "learning_rate": 1.0669261533936601e-05,
      "loss": 0.6243,
      "step": 7436
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4148440250675845,
      "learning_rate": 1.0664138871973664e-05,
      "loss": 0.717,
      "step": 7437
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40681339684074475,
      "learning_rate": 1.0659016992998116e-05,
      "loss": 0.6322,
      "step": 7438
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36412097490891926,
      "learning_rate": 1.065389589743955e-05,
      "loss": 0.6689,
      "step": 7439
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42334708484698735,
      "learning_rate": 1.0648775585727437e-05,
      "loss": 0.6732,
      "step": 7440
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3142661806341109,
      "learning_rate": 1.0643656058291226e-05,
      "loss": 0.6322,
      "step": 7441
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5799233954171662,
      "learning_rate": 1.0638537315560286e-05,
      "loss": 0.3809,
      "step": 7442
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3829949893646865,
      "learning_rate": 1.063341935796391e-05,
      "loss": 0.6802,
      "step": 7443
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35363867241230584,
      "learning_rate": 1.0628302185931338e-05,
      "loss": 0.6416,
      "step": 7444
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4049504663751377,
      "learning_rate": 1.0623185799891746e-05,
      "loss": 0.6829,
      "step": 7445
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35358330685426154,
      "learning_rate": 1.0618070200274237e-05,
      "loss": 0.6042,
      "step": 7446
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3673084663004475,
      "learning_rate": 1.0612955387507848e-05,
      "loss": 0.6605,
      "step": 7447
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.565402379700766,
      "learning_rate": 1.0607841362021562e-05,
      "loss": 0.3875,
      "step": 7448
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3841551896002205,
      "learning_rate": 1.0602728124244273e-05,
      "loss": 0.6424,
      "step": 7449
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3387241138958466,
      "learning_rate": 1.0597615674604821e-05,
      "loss": 0.6484,
      "step": 7450
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3659949892256244,
      "learning_rate": 1.0592504013532002e-05,
      "loss": 0.6567,
      "step": 7451
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3842829720039259,
      "learning_rate": 1.0587393141454489e-05,
      "loss": 0.6537,
      "step": 7452
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3585341884272895,
      "learning_rate": 1.0582283058800967e-05,
      "loss": 0.6339,
      "step": 7453
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3978629269872138,
      "learning_rate": 1.0577173765999977e-05,
      "loss": 0.6607,
      "step": 7454
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.38748081891342795,
      "learning_rate": 1.0572065263480046e-05,
      "loss": 0.638,
      "step": 7455
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36134440890009517,
      "learning_rate": 1.0566957551669622e-05,
      "loss": 0.6735,
      "step": 7456
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34377015247497755,
      "learning_rate": 1.0561850630997073e-05,
      "loss": 0.6506,
      "step": 7457
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3510636019786201,
      "learning_rate": 1.0556744501890708e-05,
      "loss": 0.6344,
      "step": 7458
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41515020070363856,
      "learning_rate": 1.055163916477878e-05,
      "loss": 0.6419,
      "step": 7459
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4177118705798604,
      "learning_rate": 1.0546534620089464e-05,
      "loss": 0.6683,
      "step": 7460
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3769878726127236,
      "learning_rate": 1.0541430868250871e-05,
      "loss": 0.6529,
      "step": 7461
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4114935860760994,
      "learning_rate": 1.0536327909691057e-05,
      "loss": 0.664,
      "step": 7462
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41437657527342603,
      "learning_rate": 1.0531225744837984e-05,
      "loss": 0.6769,
      "step": 7463
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34146492010143614,
      "learning_rate": 1.052612437411957e-05,
      "loss": 0.6111,
      "step": 7464
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4298800346761534,
      "learning_rate": 1.0521023797963674e-05,
      "loss": 0.675,
      "step": 7465
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3609188620127475,
      "learning_rate": 1.0515924016798048e-05,
      "loss": 0.611,
      "step": 7466
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37917677504253705,
      "learning_rate": 1.0510825031050435e-05,
      "loss": 0.6721,
      "step": 7467
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3994150482088532,
      "learning_rate": 1.050572684114846e-05,
      "loss": 0.6696,
      "step": 7468
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4262749880690067,
      "learning_rate": 1.050062944751971e-05,
      "loss": 0.6578,
      "step": 7469
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39730815065474234,
      "learning_rate": 1.0495532850591696e-05,
      "loss": 0.6322,
      "step": 7470
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35341144662215623,
      "learning_rate": 1.0490437050791868e-05,
      "loss": 0.6396,
      "step": 7471
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41173312733645645,
      "learning_rate": 1.0485342048547596e-05,
      "loss": 0.7,
      "step": 7472
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4167539917851391,
      "learning_rate": 1.0480247844286192e-05,
      "loss": 0.6436,
      "step": 7473
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42856103056656014,
      "learning_rate": 1.0475154438434908e-05,
      "loss": 0.6597,
      "step": 7474
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3524598873858037,
      "learning_rate": 1.0470061831420915e-05,
      "loss": 0.64,
      "step": 7475
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3324083197539225,
      "learning_rate": 1.0464970023671335e-05,
      "loss": 0.6307,
      "step": 7476
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3906165343440663,
      "learning_rate": 1.0459879015613196e-05,
      "loss": 0.6294,
      "step": 7477
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4393274425196587,
      "learning_rate": 1.0454788807673484e-05,
      "loss": 0.6481,
      "step": 7478
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3602863425083007,
      "learning_rate": 1.0449699400279114e-05,
      "loss": 0.6274,
      "step": 7479
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40349696753950354,
      "learning_rate": 1.0444610793856906e-05,
      "loss": 0.6313,
      "step": 7480
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4343769472995396,
      "learning_rate": 1.0439522988833656e-05,
      "loss": 0.6959,
      "step": 7481
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3863540425671207,
      "learning_rate": 1.043443598563608e-05,
      "loss": 0.6702,
      "step": 7482
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.38304107830418027,
      "learning_rate": 1.0429349784690795e-05,
      "loss": 0.6734,
      "step": 7483
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3624683881702185,
      "learning_rate": 1.0424264386424383e-05,
      "loss": 0.6324,
      "step": 7484
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3363610895808478,
      "learning_rate": 1.0419179791263362e-05,
      "loss": 0.6616,
      "step": 7485
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37230235376277154,
      "learning_rate": 1.0414095999634153e-05,
      "loss": 0.6714,
      "step": 7486
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3366888965277665,
      "learning_rate": 1.0409013011963136e-05,
      "loss": 0.6337,
      "step": 7487
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39251687558255965,
      "learning_rate": 1.0403930828676616e-05,
      "loss": 0.6988,
      "step": 7488
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4123489809142593,
      "learning_rate": 1.0398849450200827e-05,
      "loss": 0.6418,
      "step": 7489
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3949377237430749,
      "learning_rate": 1.0393768876961948e-05,
      "loss": 0.675,
      "step": 7490
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3672537336663184,
      "learning_rate": 1.0388689109386063e-05,
      "loss": 0.6597,
      "step": 7491
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37155967064802786,
      "learning_rate": 1.0383610147899214e-05,
      "loss": 0.6546,
      "step": 7492
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3733207245153475,
      "learning_rate": 1.0378531992927378e-05,
      "loss": 0.6381,
      "step": 7493
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4241336052198225,
      "learning_rate": 1.0373454644896428e-05,
      "loss": 0.6589,
      "step": 7494
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4320418944013328,
      "learning_rate": 1.0368378104232217e-05,
      "loss": 0.6845,
      "step": 7495
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39565677100639474,
      "learning_rate": 1.0363302371360513e-05,
      "loss": 0.6541,
      "step": 7496
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35629186855993683,
      "learning_rate": 1.035822744670699e-05,
      "loss": 0.6293,
      "step": 7497
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3926068380328891,
      "learning_rate": 1.035315333069729e-05,
      "loss": 0.6762,
      "step": 7498
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4072848324342751,
      "learning_rate": 1.0348080023756974e-05,
      "loss": 0.6395,
      "step": 7499
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4123623648872649,
      "learning_rate": 1.0343007526311518e-05,
      "loss": 0.6702,
      "step": 7500
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.401417061003871,
      "learning_rate": 1.0337935838786372e-05,
      "loss": 0.6592,
      "step": 7501
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3703867292136563,
      "learning_rate": 1.033286496160687e-05,
      "loss": 0.6422,
      "step": 7502
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4219318475265085,
      "learning_rate": 1.032779489519831e-05,
      "loss": 0.6874,
      "step": 7503
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40935103585870136,
      "learning_rate": 1.0322725639985908e-05,
      "loss": 0.6739,
      "step": 7504
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4235375922498512,
      "learning_rate": 1.0317657196394833e-05,
      "loss": 0.6563,
      "step": 7505
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3727589043259438,
      "learning_rate": 1.0312589564850142e-05,
      "loss": 0.6524,
      "step": 7506
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3973024765501498,
      "learning_rate": 1.0307522745776868e-05,
      "loss": 0.658,
      "step": 7507
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4035099643697816,
      "learning_rate": 1.0302456739599953e-05,
      "loss": 0.6449,
      "step": 7508
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4265163964125569,
      "learning_rate": 1.0297391546744281e-05,
      "loss": 0.7094,
      "step": 7509
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.38187183106259714,
      "learning_rate": 1.0292327167634673e-05,
      "loss": 0.6272,
      "step": 7510
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3569667151783373,
      "learning_rate": 1.0287263602695852e-05,
      "loss": 0.624,
      "step": 7511
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39597835317047303,
      "learning_rate": 1.02822008523525e-05,
      "loss": 0.6678,
      "step": 7512
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3648038941932822,
      "learning_rate": 1.0277138917029235e-05,
      "loss": 0.6898,
      "step": 7513
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4333912545119961,
      "learning_rate": 1.0272077797150572e-05,
      "loss": 0.665,
      "step": 7514
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4400219757425655,
      "learning_rate": 1.0267017493141005e-05,
      "loss": 0.6739,
      "step": 7515
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43501692812545556,
      "learning_rate": 1.0261958005424932e-05,
      "loss": 0.6702,
      "step": 7516
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3936800819579928,
      "learning_rate": 1.0256899334426675e-05,
      "loss": 0.6543,
      "step": 7517
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.382388803676576,
      "learning_rate": 1.0251841480570502e-05,
      "loss": 0.649,
      "step": 7518
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3594472884726732,
      "learning_rate": 1.024678444428062e-05,
      "loss": 0.6506,
      "step": 7519
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3832184759131738,
      "learning_rate": 1.0241728225981139e-05,
      "loss": 0.6303,
      "step": 7520
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4138323084774874,
      "learning_rate": 1.0236672826096127e-05,
      "loss": 0.6726,
      "step": 7521
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3816479154393022,
      "learning_rate": 1.0231618245049576e-05,
      "loss": 0.6537,
      "step": 7522
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3963019311263643,
      "learning_rate": 1.0226564483265405e-05,
      "loss": 0.6668,
      "step": 7523
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.38825719287269955,
      "learning_rate": 1.0221511541167473e-05,
      "loss": 0.6869,
      "step": 7524
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3985408868903484,
      "learning_rate": 1.0216459419179552e-05,
      "loss": 0.6576,
      "step": 7525
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43180485463402535,
      "learning_rate": 1.0211408117725362e-05,
      "loss": 0.6648,
      "step": 7526
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36659348120211904,
      "learning_rate": 1.0206357637228555e-05,
      "loss": 0.6534,
      "step": 7527
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44411898609484385,
      "learning_rate": 1.0201307978112704e-05,
      "loss": 0.6489,
      "step": 7528
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35011766284347945,
      "learning_rate": 1.0196259140801317e-05,
      "loss": 0.6148,
      "step": 7529
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39284610385124896,
      "learning_rate": 1.0191211125717846e-05,
      "loss": 0.6832,
      "step": 7530
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33172336769659094,
      "learning_rate": 1.0186163933285644e-05,
      "loss": 0.6331,
      "step": 7531
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3640056307927643,
      "learning_rate": 1.018111756392802e-05,
      "loss": 0.6532,
      "step": 7532
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39185116965514705,
      "learning_rate": 1.0176072018068216e-05,
      "loss": 0.6786,
      "step": 7533
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3892005893814947,
      "learning_rate": 1.0171027296129381e-05,
      "loss": 0.6915,
      "step": 7534
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37943483896637176,
      "learning_rate": 1.0165983398534606e-05,
      "loss": 0.6394,
      "step": 7535
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3356169291980147,
      "learning_rate": 1.0160940325706946e-05,
      "loss": 0.6429,
      "step": 7536
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.353301431226542,
      "learning_rate": 1.0155898078069326e-05,
      "loss": 0.5955,
      "step": 7537
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3516660016453198,
      "learning_rate": 1.015085665604465e-05,
      "loss": 0.6662,
      "step": 7538
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3540739898866982,
      "learning_rate": 1.0145816060055736e-05,
      "loss": 0.6461,
      "step": 7539
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3712059972398169,
      "learning_rate": 1.014077629052532e-05,
      "loss": 0.6168,
      "step": 7540
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3947496770469976,
      "learning_rate": 1.013573734787609e-05,
      "loss": 0.6727,
      "step": 7541
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3483415117409612,
      "learning_rate": 1.013069923253066e-05,
      "loss": 0.6205,
      "step": 7542
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37152874240670697,
      "learning_rate": 1.0125661944911562e-05,
      "loss": 0.6564,
      "step": 7543
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.660471637273447,
      "learning_rate": 1.0120625485441278e-05,
      "loss": 0.3985,
      "step": 7544
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40022253843514705,
      "learning_rate": 1.0115589854542201e-05,
      "loss": 0.6542,
      "step": 7545
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3504447146224605,
      "learning_rate": 1.011055505263666e-05,
      "loss": 0.6671,
      "step": 7546
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3566507955416982,
      "learning_rate": 1.0105521080146934e-05,
      "loss": 0.6153,
      "step": 7547
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3707395428880963,
      "learning_rate": 1.0100487937495197e-05,
      "loss": 0.6486,
      "step": 7548
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35597964291571493,
      "learning_rate": 1.0095455625103571e-05,
      "loss": 0.6303,
      "step": 7549
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3490907174875903,
      "learning_rate": 1.009042414339414e-05,
      "loss": 0.6622,
      "step": 7550
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4026692731657083,
      "learning_rate": 1.0085393492788857e-05,
      "loss": 0.6322,
      "step": 7551
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4962563088038233,
      "learning_rate": 1.008036367370965e-05,
      "loss": 0.7268,
      "step": 7552
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34934884414141704,
      "learning_rate": 1.007533468657837e-05,
      "loss": 0.6298,
      "step": 7553
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39090659853069415,
      "learning_rate": 1.0070306531816775e-05,
      "loss": 0.6709,
      "step": 7554
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3503929161734929,
      "learning_rate": 1.006527920984658e-05,
      "loss": 0.6501,
      "step": 7555
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3797424763474792,
      "learning_rate": 1.0060252721089419e-05,
      "loss": 0.6822,
      "step": 7556
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3321884763848922,
      "learning_rate": 1.0055227065966857e-05,
      "loss": 0.6242,
      "step": 7557
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3447452475986493,
      "learning_rate": 1.00502022449004e-05,
      "loss": 0.6328,
      "step": 7558
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.424389361248158,
      "learning_rate": 1.0045178258311457e-05,
      "loss": 0.6984,
      "step": 7559
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.38256414793797444,
      "learning_rate": 1.0040155106621388e-05,
      "loss": 0.661,
      "step": 7560
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37838676805587107,
      "learning_rate": 1.0035132790251485e-05,
      "loss": 0.6589,
      "step": 7561
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.33976317404947853,
      "learning_rate": 1.0030111309622967e-05,
      "loss": 0.6261,
      "step": 7562
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3254167851395278,
      "learning_rate": 1.0025090665156958e-05,
      "loss": 0.6342,
      "step": 7563
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3806727466827316,
      "learning_rate": 1.0020070857274562e-05,
      "loss": 0.6452,
      "step": 7564
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4158595552993751,
      "learning_rate": 1.0015051886396765e-05,
      "loss": 0.6637,
      "step": 7565
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.41332094403518566,
      "learning_rate": 1.0010033752944506e-05,
      "loss": 0.6592,
      "step": 7566
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3953252104675211,
      "learning_rate": 1.0005016457338659e-05,
      "loss": 0.6165,
      "step": 7567
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40904779907992306,
      "learning_rate": 1.0000000000000006e-05,
      "loss": 0.653,
      "step": 7568
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34719112990965195,
      "learning_rate": 9.994984381349272e-06,
      "loss": 0.6609,
      "step": 7569
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4214996455689858,
      "learning_rate": 9.989969601807117e-06,
      "loss": 0.6672,
      "step": 7570
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4111746348792906,
      "learning_rate": 9.984955661794123e-06,
      "loss": 0.6981,
      "step": 7571
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4061767782973408,
      "learning_rate": 9.979942561730803e-06,
      "loss": 0.6361,
      "step": 7572
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3324098448244689,
      "learning_rate": 9.974930302037604e-06,
      "loss": 0.6662,
      "step": 7573
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35857597247668205,
      "learning_rate": 9.969918883134886e-06,
      "loss": 0.6564,
      "step": 7574
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3275488600179223,
      "learning_rate": 9.96490830544296e-06,
      "loss": 0.592,
      "step": 7575
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3790564844847263,
      "learning_rate": 9.959898569382063e-06,
      "loss": 0.6466,
      "step": 7576
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35776075894178516,
      "learning_rate": 9.95488967537233e-06,
      "loss": 0.6709,
      "step": 7577
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3923753654804498,
      "learning_rate": 9.949881623833886e-06,
      "loss": 0.6502,
      "step": 7578
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3517712297730752,
      "learning_rate": 9.944874415186722e-06,
      "loss": 0.6466,
      "step": 7579
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39490659422912266,
      "learning_rate": 9.9398680498508e-06,
      "loss": 0.6429,
      "step": 7580
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37676088783655026,
      "learning_rate": 9.934862528246004e-06,
      "loss": 0.6364,
      "step": 7581
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3717094715443563,
      "learning_rate": 9.929857850792122e-06,
      "loss": 0.6686,
      "step": 7582
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4017897078908569,
      "learning_rate": 9.924854017908893e-06,
      "loss": 0.6772,
      "step": 7583
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3780876478771069,
      "learning_rate": 9.919851030016006e-06,
      "loss": 0.6275,
      "step": 7584
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3958180546731125,
      "learning_rate": 9.914848887533031e-06,
      "loss": 0.6918,
      "step": 7585
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.38338954493144406,
      "learning_rate": 9.9098475908795e-06,
      "loss": 0.6419,
      "step": 7586
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34901854574676117,
      "learning_rate": 9.904847140474874e-06,
      "loss": 0.6218,
      "step": 7587
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3966638269478712,
      "learning_rate": 9.899847536738519e-06,
      "loss": 0.7061,
      "step": 7588
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.38813011829896243,
      "learning_rate": 9.894848780089752e-06,
      "loss": 0.6644,
      "step": 7589
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37392831944392463,
      "learning_rate": 9.889850870947814e-06,
      "loss": 0.6487,
      "step": 7590
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3583496769475122,
      "learning_rate": 9.884853809731875e-06,
      "loss": 0.5991,
      "step": 7591
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3923707657369173,
      "learning_rate": 9.879857596861032e-06,
      "loss": 0.672,
      "step": 7592
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3961883338657734,
      "learning_rate": 9.874862232754314e-06,
      "loss": 0.6849,
      "step": 7593
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36455133871059947,
      "learning_rate": 9.869867717830668e-06,
      "loss": 0.658,
      "step": 7594
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35525658201590565,
      "learning_rate": 9.864874052508982e-06,
      "loss": 0.636,
      "step": 7595
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.310705461382214,
      "learning_rate": 9.859881237208075e-06,
      "loss": 0.6523,
      "step": 7596
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40030636817401927,
      "learning_rate": 9.854889272346669e-06,
      "loss": 0.6986,
      "step": 7597
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34799514131245857,
      "learning_rate": 9.84989815834346e-06,
      "loss": 0.6396,
      "step": 7598
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5729135889379611,
      "learning_rate": 9.844907895617028e-06,
      "loss": 0.3718,
      "step": 7599
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3651240871710727,
      "learning_rate": 9.839918484585909e-06,
      "loss": 0.6689,
      "step": 7600
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.60300165738776,
      "learning_rate": 9.834929925668561e-06,
      "loss": 0.3827,
      "step": 7601
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3795938571434617,
      "learning_rate": 9.829942219283358e-06,
      "loss": 0.6768,
      "step": 7602
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32197672188357324,
      "learning_rate": 9.82495536584862e-06,
      "loss": 0.6282,
      "step": 7603
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.38832195293947375,
      "learning_rate": 9.819969365782582e-06,
      "loss": 0.6725,
      "step": 7604
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4044042024792946,
      "learning_rate": 9.814984219503423e-06,
      "loss": 0.6642,
      "step": 7605
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.384183595743195,
      "learning_rate": 9.809999927429237e-06,
      "loss": 0.6418,
      "step": 7606
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36397800763706145,
      "learning_rate": 9.805016489978057e-06,
      "loss": 0.6575,
      "step": 7607
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.45069106741936893,
      "learning_rate": 9.800033907567825e-06,
      "loss": 0.6676,
      "step": 7608
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37162194846366753,
      "learning_rate": 9.795052180616431e-06,
      "loss": 0.6519,
      "step": 7609
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3827459656418621,
      "learning_rate": 9.790071309541692e-06,
      "loss": 0.6637,
      "step": 7610
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40730959115527193,
      "learning_rate": 9.785091294761329e-06,
      "loss": 0.6412,
      "step": 7611
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4086134726059586,
      "learning_rate": 9.780112136693036e-06,
      "loss": 0.6567,
      "step": 7612
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3649548946840366,
      "learning_rate": 9.77513383575439e-06,
      "loss": 0.662,
      "step": 7613
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4481028712359468,
      "learning_rate": 9.770156392362917e-06,
      "loss": 0.6741,
      "step": 7614
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4073554227148017,
      "learning_rate": 9.765179806936083e-06,
      "loss": 0.6542,
      "step": 7615
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4640479033104814,
      "learning_rate": 9.76020407989125e-06,
      "loss": 0.6819,
      "step": 7616
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34812433008634186,
      "learning_rate": 9.755229211645734e-06,
      "loss": 0.625,
      "step": 7617
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3634032217232418,
      "learning_rate": 9.750255202616772e-06,
      "loss": 0.6506,
      "step": 7618
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3620589920268963,
      "learning_rate": 9.745282053221527e-06,
      "loss": 0.6211,
      "step": 7619
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.360854918979405,
      "learning_rate": 9.740309763877092e-06,
      "loss": 0.6365,
      "step": 7620
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4160196421790295,
      "learning_rate": 9.735338335000494e-06,
      "loss": 0.6486,
      "step": 7621
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39079213509252486,
      "learning_rate": 9.730367767008667e-06,
      "loss": 0.6352,
      "step": 7622
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4278183506631072,
      "learning_rate": 9.725398060318491e-06,
      "loss": 0.6675,
      "step": 7623
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4159376197451218,
      "learning_rate": 9.72042921534678e-06,
      "loss": 0.6685,
      "step": 7624
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35644822206748505,
      "learning_rate": 9.71546123251024e-06,
      "loss": 0.6362,
      "step": 7625
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35642861178965324,
      "learning_rate": 9.710494112225555e-06,
      "loss": 0.6276,
      "step": 7626
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3674963054464154,
      "learning_rate": 9.705527854909306e-06,
      "loss": 0.6573,
      "step": 7627
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40378591051567114,
      "learning_rate": 9.700562460978001e-06,
      "loss": 0.6342,
      "step": 7628
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3393557523421136,
      "learning_rate": 9.69559793084808e-06,
      "loss": 0.6307,
      "step": 7629
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.41789892550478797,
      "learning_rate": 9.690634264935925e-06,
      "loss": 0.6581,
      "step": 7630
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40000059877573835,
      "learning_rate": 9.685671463657816e-06,
      "loss": 0.6299,
      "step": 7631
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40616690852449927,
      "learning_rate": 9.680709527429983e-06,
      "loss": 0.6682,
      "step": 7632
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3902947677787791,
      "learning_rate": 9.675748456668582e-06,
      "loss": 0.6916,
      "step": 7633
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3790407675083745,
      "learning_rate": 9.670788251789686e-06,
      "loss": 0.614,
      "step": 7634
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4237493399104006,
      "learning_rate": 9.665828913209317e-06,
      "loss": 0.6484,
      "step": 7635
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4065807020507622,
      "learning_rate": 9.660870441343387e-06,
      "loss": 0.6197,
      "step": 7636
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39783148404266394,
      "learning_rate": 9.655912836607765e-06,
      "loss": 0.6243,
      "step": 7637
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42727552411187564,
      "learning_rate": 9.65095609941825e-06,
      "loss": 0.6848,
      "step": 7638
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3779341608795306,
      "learning_rate": 9.646000230190533e-06,
      "loss": 0.638,
      "step": 7639
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37097253855321277,
      "learning_rate": 9.64104522934028e-06,
      "loss": 0.6589,
      "step": 7640
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3572576794087264,
      "learning_rate": 9.63609109728306e-06,
      "loss": 0.6605,
      "step": 7641
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3411466819528957,
      "learning_rate": 9.631137834434357e-06,
      "loss": 0.6308,
      "step": 7642
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43017916094131936,
      "learning_rate": 9.6261854412096e-06,
      "loss": 0.6232,
      "step": 7643
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8463177807985238,
      "learning_rate": 9.621233918024148e-06,
      "loss": 0.4105,
      "step": 7644
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5276487572442582,
      "learning_rate": 9.616283265293269e-06,
      "loss": 0.6884,
      "step": 7645
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39517137243027733,
      "learning_rate": 9.61133348343217e-06,
      "loss": 0.6568,
      "step": 7646
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3620364611532948,
      "learning_rate": 9.606384572855985e-06,
      "loss": 0.6846,
      "step": 7647
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.359355499093785,
      "learning_rate": 9.601436533979776e-06,
      "loss": 0.6719,
      "step": 7648
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3980725646871753,
      "learning_rate": 9.596489367218524e-06,
      "loss": 0.6445,
      "step": 7649
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3893432678517697,
      "learning_rate": 9.591543072987152e-06,
      "loss": 0.6838,
      "step": 7650
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42029595764130134,
      "learning_rate": 9.586597651700486e-06,
      "loss": 0.6608,
      "step": 7651
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3370578800749508,
      "learning_rate": 9.581653103773298e-06,
      "loss": 0.6353,
      "step": 7652
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39200921892488966,
      "learning_rate": 9.576709429620283e-06,
      "loss": 0.6591,
      "step": 7653
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40750379760015104,
      "learning_rate": 9.571766629656059e-06,
      "loss": 0.6624,
      "step": 7654
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36539795433827665,
      "learning_rate": 9.566824704295182e-06,
      "loss": 0.6473,
      "step": 7655
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3667102402006322,
      "learning_rate": 9.561883653952108e-06,
      "loss": 0.6487,
      "step": 7656
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.346873241055038,
      "learning_rate": 9.556943479041248e-06,
      "loss": 0.6501,
      "step": 7657
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39723920098085597,
      "learning_rate": 9.552004179976932e-06,
      "loss": 0.6363,
      "step": 7658
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39307751478934005,
      "learning_rate": 9.547065757173402e-06,
      "loss": 0.646,
      "step": 7659
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37682749391145876,
      "learning_rate": 9.542128211044835e-06,
      "loss": 0.6488,
      "step": 7660
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35270710692043367,
      "learning_rate": 9.537191542005357e-06,
      "loss": 0.6364,
      "step": 7661
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38717802662030176,
      "learning_rate": 9.532255750468984e-06,
      "loss": 0.6246,
      "step": 7662
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41642667700106567,
      "learning_rate": 9.527320836849679e-06,
      "loss": 0.6786,
      "step": 7663
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.42095598352117086,
      "learning_rate": 9.522386801561334e-06,
      "loss": 0.691,
      "step": 7664
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4175801382786169,
      "learning_rate": 9.517453645017749e-06,
      "loss": 0.67,
      "step": 7665
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3553158750589532,
      "learning_rate": 9.512521367632665e-06,
      "loss": 0.6292,
      "step": 7666
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4222478268576393,
      "learning_rate": 9.507589969819748e-06,
      "loss": 0.6545,
      "step": 7667
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3606116473300498,
      "learning_rate": 9.50265945199259e-06,
      "loss": 0.6424,
      "step": 7668
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35038967270619814,
      "learning_rate": 9.497729814564714e-06,
      "loss": 0.6393,
      "step": 7669
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4353669956000703,
      "learning_rate": 9.492801057949545e-06,
      "loss": 0.6582,
      "step": 7670
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.383774511706616,
      "learning_rate": 9.487873182560463e-06,
      "loss": 0.625,
      "step": 7671
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35113830796124856,
      "learning_rate": 9.48294618881077e-06,
      "loss": 0.6634,
      "step": 7672
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3562181856574731,
      "learning_rate": 9.478020077113664e-06,
      "loss": 0.6535,
      "step": 7673
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46111726447253304,
      "learning_rate": 9.473094847882314e-06,
      "loss": 0.655,
      "step": 7674
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40397510011739535,
      "learning_rate": 9.468170501529792e-06,
      "loss": 0.6871,
      "step": 7675
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.31131333812403267,
      "learning_rate": 9.463247038469083e-06,
      "loss": 0.5874,
      "step": 7676
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38050084651429167,
      "learning_rate": 9.458324459113123e-06,
      "loss": 0.6518,
      "step": 7677
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6779595342282365,
      "learning_rate": 9.453402763874763e-06,
      "loss": 0.384,
      "step": 7678
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4097764522448972,
      "learning_rate": 9.448481953166773e-06,
      "loss": 0.6511,
      "step": 7679
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3811210153624033,
      "learning_rate": 9.443562027401856e-06,
      "loss": 0.6606,
      "step": 7680
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4042733429803781,
      "learning_rate": 9.438642986992641e-06,
      "loss": 0.678,
      "step": 7681
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40744683447282254,
      "learning_rate": 9.433724832351688e-06,
      "loss": 0.6371,
      "step": 7682
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41495102493539765,
      "learning_rate": 9.428807563891469e-06,
      "loss": 0.6726,
      "step": 7683
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4355381499801228,
      "learning_rate": 9.4238911820244e-06,
      "loss": 0.6837,
      "step": 7684
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3556788242591238,
      "learning_rate": 9.4189756871628e-06,
      "loss": 0.6347,
      "step": 7685
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3739337987027743,
      "learning_rate": 9.41406107971893e-06,
      "loss": 0.6329,
      "step": 7686
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3880595617098307,
      "learning_rate": 9.409147360104971e-06,
      "loss": 0.6287,
      "step": 7687
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3515875870027908,
      "learning_rate": 9.404234528733035e-06,
      "loss": 0.6279,
      "step": 7688
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36155294569810226,
      "learning_rate": 9.399322586015159e-06,
      "loss": 0.6526,
      "step": 7689
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39786272157077257,
      "learning_rate": 9.394411532363285e-06,
      "loss": 0.705,
      "step": 7690
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.376910281020477,
      "learning_rate": 9.389501368189311e-06,
      "loss": 0.6133,
      "step": 7691
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36457023451903514,
      "learning_rate": 9.384592093905049e-06,
      "loss": 0.6362,
      "step": 7692
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34425846998770276,
      "learning_rate": 9.379683709922221e-06,
      "loss": 0.6777,
      "step": 7693
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3460668710327661,
      "learning_rate": 9.374776216652486e-06,
      "loss": 0.6464,
      "step": 7694
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39400671526003683,
      "learning_rate": 9.369869614507455e-06,
      "loss": 0.6333,
      "step": 7695
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39913774214282793,
      "learning_rate": 9.364963903898612e-06,
      "loss": 0.685,
      "step": 7696
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3493200833987311,
      "learning_rate": 9.360059085237401e-06,
      "loss": 0.6689,
      "step": 7697
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3982548128208006,
      "learning_rate": 9.355155158935194e-06,
      "loss": 0.6822,
      "step": 7698
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40575116477260803,
      "learning_rate": 9.350252125403261e-06,
      "loss": 0.653,
      "step": 7699
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39783292269725556,
      "learning_rate": 9.345349985052821e-06,
      "loss": 0.6397,
      "step": 7700
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41287911583712605,
      "learning_rate": 9.340448738295009e-06,
      "loss": 0.6957,
      "step": 7701
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3289085904336584,
      "learning_rate": 9.335548385540887e-06,
      "loss": 0.6295,
      "step": 7702
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4370753726643895,
      "learning_rate": 9.330648927201451e-06,
      "loss": 0.6735,
      "step": 7703
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40855242170887845,
      "learning_rate": 9.325750363687599e-06,
      "loss": 0.6564,
      "step": 7704
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43303136843768075,
      "learning_rate": 9.32085269541017e-06,
      "loss": 0.6221,
      "step": 7705
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41206063485091105,
      "learning_rate": 9.31595592277993e-06,
      "loss": 0.6623,
      "step": 7706
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3766767966682483,
      "learning_rate": 9.311060046207567e-06,
      "loss": 0.6672,
      "step": 7707
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39060175820135656,
      "learning_rate": 9.306165066103678e-06,
      "loss": 0.671,
      "step": 7708
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36382607786581095,
      "learning_rate": 9.301270982878826e-06,
      "loss": 0.6466,
      "step": 7709
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40799093042074874,
      "learning_rate": 9.296377796943446e-06,
      "loss": 0.6715,
      "step": 7710
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3965352433003578,
      "learning_rate": 9.291485508707932e-06,
      "loss": 0.6748,
      "step": 7711
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39103056001115316,
      "learning_rate": 9.286594118582606e-06,
      "loss": 0.6764,
      "step": 7712
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3949331893446896,
      "learning_rate": 9.281703626977685e-06,
      "loss": 0.6295,
      "step": 7713
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3851296708237774,
      "learning_rate": 9.276814034303334e-06,
      "loss": 0.6553,
      "step": 7714
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37932115086177565,
      "learning_rate": 9.271925340969644e-06,
      "loss": 0.6665,
      "step": 7715
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38223775408662597,
      "learning_rate": 9.267037547386615e-06,
      "loss": 0.6428,
      "step": 7716
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3964660472033747,
      "learning_rate": 9.262150653964187e-06,
      "loss": 0.6834,
      "step": 7717
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43477799853040944,
      "learning_rate": 9.257264661112222e-06,
      "loss": 0.6525,
      "step": 7718
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35807782608719435,
      "learning_rate": 9.25237956924049e-06,
      "loss": 0.652,
      "step": 7719
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.406647978213579,
      "learning_rate": 9.247495378758703e-06,
      "loss": 0.6575,
      "step": 7720
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38244751420189466,
      "learning_rate": 9.242612090076499e-06,
      "loss": 0.7076,
      "step": 7721
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38313326743289994,
      "learning_rate": 9.237729703603416e-06,
      "loss": 0.6769,
      "step": 7722
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37652829185781395,
      "learning_rate": 9.232848219748956e-06,
      "loss": 0.6561,
      "step": 7723
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3872361837803539,
      "learning_rate": 9.227967638922505e-06,
      "loss": 0.6862,
      "step": 7724
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38760587030800725,
      "learning_rate": 9.223087961533402e-06,
      "loss": 0.6787,
      "step": 7725
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40466793152234404,
      "learning_rate": 9.218209187990903e-06,
      "loss": 0.6434,
      "step": 7726
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37697064356529797,
      "learning_rate": 9.213331318704171e-06,
      "loss": 0.6421,
      "step": 7727
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3687170986167417,
      "learning_rate": 9.208454354082312e-06,
      "loss": 0.6372,
      "step": 7728
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3941323654675696,
      "learning_rate": 9.203578294534355e-06,
      "loss": 0.652,
      "step": 7729
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36275463768774846,
      "learning_rate": 9.198703140469247e-06,
      "loss": 0.6631,
      "step": 7730
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40072274507493305,
      "learning_rate": 9.193828892295864e-06,
      "loss": 0.6568,
      "step": 7731
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6759003069059906,
      "learning_rate": 9.188955550423007e-06,
      "loss": 0.3861,
      "step": 7732
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4292630517636464,
      "learning_rate": 9.184083115259384e-06,
      "loss": 0.6592,
      "step": 7733
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4127448341707592,
      "learning_rate": 9.179211587213648e-06,
      "loss": 0.6359,
      "step": 7734
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3633908546528776,
      "learning_rate": 9.174340966694375e-06,
      "loss": 0.6624,
      "step": 7735
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.32629298670087675,
      "learning_rate": 9.169471254110038e-06,
      "loss": 0.6143,
      "step": 7736
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39712297763462123,
      "learning_rate": 9.16460244986908e-06,
      "loss": 0.6523,
      "step": 7737
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40058094566622987,
      "learning_rate": 9.159734554379823e-06,
      "loss": 0.6803,
      "step": 7738
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4267290529207137,
      "learning_rate": 9.154867568050534e-06,
      "loss": 0.6881,
      "step": 7739
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4201933026837913,
      "learning_rate": 9.150001491289406e-06,
      "loss": 0.6754,
      "step": 7740
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.401391778578089,
      "learning_rate": 9.14513632450456e-06,
      "loss": 0.6861,
      "step": 7741
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35010216976340774,
      "learning_rate": 9.14027206810401e-06,
      "loss": 0.6122,
      "step": 7742
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41368015791249574,
      "learning_rate": 9.135408722495727e-06,
      "loss": 0.6822,
      "step": 7743
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38094557746756563,
      "learning_rate": 9.130546288087594e-06,
      "loss": 0.6482,
      "step": 7744
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4108813254259386,
      "learning_rate": 9.125684765287416e-06,
      "loss": 0.6471,
      "step": 7745
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3594391703821467,
      "learning_rate": 9.120824154502933e-06,
      "loss": 0.6511,
      "step": 7746
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41595204276124326,
      "learning_rate": 9.115964456141781e-06,
      "loss": 0.6837,
      "step": 7747
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.45276608975633037,
      "learning_rate": 9.111105670611545e-06,
      "loss": 0.7213,
      "step": 7748
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7559375040083267,
      "learning_rate": 9.106247798319735e-06,
      "loss": 0.4178,
      "step": 7749
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.42657829033729505,
      "learning_rate": 9.101390839673749e-06,
      "loss": 0.6559,
      "step": 7750
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40899584748570345,
      "learning_rate": 9.09653479508096e-06,
      "loss": 0.6746,
      "step": 7751
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4290013852540321,
      "learning_rate": 9.091679664948637e-06,
      "loss": 0.6441,
      "step": 7752
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38878633056854317,
      "learning_rate": 9.086825449683958e-06,
      "loss": 0.6599,
      "step": 7753
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3810862626603407,
      "learning_rate": 9.08197214969405e-06,
      "loss": 0.6032,
      "step": 7754
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3613636345163916,
      "learning_rate": 9.077119765385958e-06,
      "loss": 0.6529,
      "step": 7755
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4578800746337158,
      "learning_rate": 9.072268297166626e-06,
      "loss": 0.6574,
      "step": 7756
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40908608615636965,
      "learning_rate": 9.067417745442969e-06,
      "loss": 0.6494,
      "step": 7757
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41306731797872204,
      "learning_rate": 9.062568110621774e-06,
      "loss": 0.6658,
      "step": 7758
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4094244355887518,
      "learning_rate": 9.057719393109785e-06,
      "loss": 0.6715,
      "step": 7759
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37804122980350474,
      "learning_rate": 9.05287159331366e-06,
      "loss": 0.6503,
      "step": 7760
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40132469707972507,
      "learning_rate": 9.048024711639968e-06,
      "loss": 0.6836,
      "step": 7761
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37161525684114205,
      "learning_rate": 9.043178748495217e-06,
      "loss": 0.6443,
      "step": 7762
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3908976910753457,
      "learning_rate": 9.038333704285834e-06,
      "loss": 0.6776,
      "step": 7763
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34202091427646275,
      "learning_rate": 9.033489579418162e-06,
      "loss": 0.6186,
      "step": 7764
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36694865493071127,
      "learning_rate": 9.028646374298475e-06,
      "loss": 0.6736,
      "step": 7765
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3835614019365457,
      "learning_rate": 9.023804089332974e-06,
      "loss": 0.6486,
      "step": 7766
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4013228552925798,
      "learning_rate": 9.018962724927763e-06,
      "loss": 0.6422,
      "step": 7767
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3855954157571046,
      "learning_rate": 9.014122281488883e-06,
      "loss": 0.6467,
      "step": 7768
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.396688820162499,
      "learning_rate": 9.009282759422308e-06,
      "loss": 0.6375,
      "step": 7769
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37157269420241357,
      "learning_rate": 9.004444159133903e-06,
      "loss": 0.6449,
      "step": 7770
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39003378239376213,
      "learning_rate": 8.99960648102949e-06,
      "loss": 0.658,
      "step": 7771
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3854871556193348,
      "learning_rate": 8.994769725514806e-06,
      "loss": 0.6548,
      "step": 7772
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3798478865257914,
      "learning_rate": 8.989933892995486e-06,
      "loss": 0.6695,
      "step": 7773
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3807928247225659,
      "learning_rate": 8.985098983877114e-06,
      "loss": 0.6428,
      "step": 7774
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4587872626871995,
      "learning_rate": 8.980264998565196e-06,
      "loss": 0.7123,
      "step": 7775
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37998466552659377,
      "learning_rate": 8.975431937465136e-06,
      "loss": 0.6749,
      "step": 7776
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38271368019205254,
      "learning_rate": 8.970599800982284e-06,
      "loss": 0.6502,
      "step": 7777
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38376256337869485,
      "learning_rate": 8.965768589521908e-06,
      "loss": 0.6779,
      "step": 7778
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3906607646990577,
      "learning_rate": 8.960938303489193e-06,
      "loss": 0.6442,
      "step": 7779
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37935985955853097,
      "learning_rate": 8.956108943289259e-06,
      "loss": 0.6314,
      "step": 7780
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3610100356732147,
      "learning_rate": 8.951280509327122e-06,
      "loss": 0.6493,
      "step": 7781
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3932664245059382,
      "learning_rate": 8.946453002007747e-06,
      "loss": 0.6138,
      "step": 7782
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4079410286951423,
      "learning_rate": 8.941626421736016e-06,
      "loss": 0.6485,
      "step": 7783
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6640144888818886,
      "learning_rate": 8.936800768916706e-06,
      "loss": 0.3662,
      "step": 7784
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3385427014067833,
      "learning_rate": 8.931976043954564e-06,
      "loss": 0.6309,
      "step": 7785
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4381115803655169,
      "learning_rate": 8.92715224725423e-06,
      "loss": 0.6527,
      "step": 7786
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4495928694247019,
      "learning_rate": 8.922329379220257e-06,
      "loss": 0.6625,
      "step": 7787
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4320709053427458,
      "learning_rate": 8.917507440257143e-06,
      "loss": 0.625,
      "step": 7788
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4112491681036847,
      "learning_rate": 8.9126864307693e-06,
      "loss": 0.6801,
      "step": 7789
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4037646353545434,
      "learning_rate": 8.90786635116105e-06,
      "loss": 0.6594,
      "step": 7790
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4292740527058236,
      "learning_rate": 8.903047201836655e-06,
      "loss": 0.6849,
      "step": 7791
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.46893688307474557,
      "learning_rate": 8.898228983200286e-06,
      "loss": 0.6969,
      "step": 7792
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38176423403972215,
      "learning_rate": 8.893411695656047e-06,
      "loss": 0.6724,
      "step": 7793
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3855042566796135,
      "learning_rate": 8.888595339607961e-06,
      "loss": 0.6827,
      "step": 7794
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3374838945550721,
      "learning_rate": 8.883779915459957e-06,
      "loss": 0.6378,
      "step": 7795
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42165643197677477,
      "learning_rate": 8.87896542361591e-06,
      "loss": 0.6742,
      "step": 7796
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3754899537189898,
      "learning_rate": 8.874151864479601e-06,
      "loss": 0.6186,
      "step": 7797
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5834831955386519,
      "learning_rate": 8.869339238454737e-06,
      "loss": 0.3526,
      "step": 7798
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4084706576345041,
      "learning_rate": 8.864527545944949e-06,
      "loss": 0.6251,
      "step": 7799
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5859851051599368,
      "learning_rate": 8.859716787353796e-06,
      "loss": 0.364,
      "step": 7800
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3507665773729728,
      "learning_rate": 8.854906963084733e-06,
      "loss": 0.6719,
      "step": 7801
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38256140790018384,
      "learning_rate": 8.850098073541165e-06,
      "loss": 0.6432,
      "step": 7802
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4006580254716692,
      "learning_rate": 8.845290119126411e-06,
      "loss": 0.6566,
      "step": 7803
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3754801904856243,
      "learning_rate": 8.840483100243698e-06,
      "loss": 0.5998,
      "step": 7804
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3288772343828508,
      "learning_rate": 8.83567701729618e-06,
      "loss": 0.6685,
      "step": 7805
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.43272157528235927,
      "learning_rate": 8.830871870686965e-06,
      "loss": 0.6881,
      "step": 7806
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38322134956331083,
      "learning_rate": 8.826067660819028e-06,
      "loss": 0.6588,
      "step": 7807
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3586038118302693,
      "learning_rate": 8.821264388095304e-06,
      "loss": 0.6676,
      "step": 7808
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3805866137766948,
      "learning_rate": 8.816462052918641e-06,
      "loss": 0.6055,
      "step": 7809
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3519341577526149,
      "learning_rate": 8.811660655691792e-06,
      "loss": 0.6631,
      "step": 7810
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36743700042062993,
      "learning_rate": 8.806860196817452e-06,
      "loss": 0.6613,
      "step": 7811
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35748922382979226,
      "learning_rate": 8.80206067669823e-06,
      "loss": 0.6258,
      "step": 7812
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38891555787129106,
      "learning_rate": 8.797262095736653e-06,
      "loss": 0.661,
      "step": 7813
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3932453314165813,
      "learning_rate": 8.792464454335184e-06,
      "loss": 0.665,
      "step": 7814
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38456892292586786,
      "learning_rate": 8.787667752896179e-06,
      "loss": 0.6519,
      "step": 7815
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3761589715405768,
      "learning_rate": 8.782871991821938e-06,
      "loss": 0.6359,
      "step": 7816
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3485191810500503,
      "learning_rate": 8.778077171514684e-06,
      "loss": 0.6458,
      "step": 7817
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3395774584047836,
      "learning_rate": 8.773283292376535e-06,
      "loss": 0.624,
      "step": 7818
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4001517726961656,
      "learning_rate": 8.76849035480955e-06,
      "loss": 0.6068,
      "step": 7819
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36450188543933437,
      "learning_rate": 8.763698359215733e-06,
      "loss": 0.6618,
      "step": 7820
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.65952017580375,
      "learning_rate": 8.758907305996957e-06,
      "loss": 0.3672,
      "step": 7821
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36675094238156536,
      "learning_rate": 8.754117195555047e-06,
      "loss": 0.6649,
      "step": 7822
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3758108005431391,
      "learning_rate": 8.749328028291757e-06,
      "loss": 0.6351,
      "step": 7823
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.371773188570461,
      "learning_rate": 8.744539804608727e-06,
      "loss": 0.6452,
      "step": 7824
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3829717741081965,
      "learning_rate": 8.739752524907552e-06,
      "loss": 0.6514,
      "step": 7825
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3268219566894255,
      "learning_rate": 8.734966189589733e-06,
      "loss": 0.6157,
      "step": 7826
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4405666748124489,
      "learning_rate": 8.730180799056698e-06,
      "loss": 0.6636,
      "step": 7827
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.39092260801557843,
      "learning_rate": 8.725396353709785e-06,
      "loss": 0.6798,
      "step": 7828
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.389365077689043,
      "learning_rate": 8.720612853950273e-06,
      "loss": 0.6574,
      "step": 7829
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3922570885931402,
      "learning_rate": 8.715830300179331e-06,
      "loss": 0.6263,
      "step": 7830
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36906937624759284,
      "learning_rate": 8.711048692798075e-06,
      "loss": 0.6834,
      "step": 7831
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3756347040971336,
      "learning_rate": 8.706268032207536e-06,
      "loss": 0.629,
      "step": 7832
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3891450109613006,
      "learning_rate": 8.701488318808646e-06,
      "loss": 0.6613,
      "step": 7833
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33597533083557113,
      "learning_rate": 8.696709553002297e-06,
      "loss": 0.6234,
      "step": 7834
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37427679149921744,
      "learning_rate": 8.691931735189262e-06,
      "loss": 0.6698,
      "step": 7835
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.39570016655189305,
      "learning_rate": 8.687154865770256e-06,
      "loss": 0.6693,
      "step": 7836
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35876532231807434,
      "learning_rate": 8.682378945145915e-06,
      "loss": 0.6441,
      "step": 7837
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6354130715847972,
      "learning_rate": 8.677603973716777e-06,
      "loss": 0.3789,
      "step": 7838
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41282915079243543,
      "learning_rate": 8.672829951883312e-06,
      "loss": 0.6218,
      "step": 7839
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37489718331620464,
      "learning_rate": 8.668056880045935e-06,
      "loss": 0.6458,
      "step": 7840
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37384898688012386,
      "learning_rate": 8.663284758604934e-06,
      "loss": 0.6322,
      "step": 7841
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42573235824375094,
      "learning_rate": 8.658513587960549e-06,
      "loss": 0.6953,
      "step": 7842
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40242781669112615,
      "learning_rate": 8.65374336851294e-06,
      "loss": 0.6681,
      "step": 7843
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36733394388084306,
      "learning_rate": 8.648974100662165e-06,
      "loss": 0.6322,
      "step": 7844
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40523674332237813,
      "learning_rate": 8.644205784808221e-06,
      "loss": 0.6662,
      "step": 7845
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40631812691736785,
      "learning_rate": 8.639438421351025e-06,
      "loss": 0.6666,
      "step": 7846
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5741349091360537,
      "learning_rate": 8.634672010690409e-06,
      "loss": 0.379,
      "step": 7847
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3342955355538247,
      "learning_rate": 8.629906553226132e-06,
      "loss": 0.6144,
      "step": 7848
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38658576087178476,
      "learning_rate": 8.625142049357853e-06,
      "loss": 0.6132,
      "step": 7849
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3714322236870626,
      "learning_rate": 8.620378499485176e-06,
      "loss": 0.6706,
      "step": 7850
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3732352296955538,
      "learning_rate": 8.615615904007607e-06,
      "loss": 0.648,
      "step": 7851
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37133707734504756,
      "learning_rate": 8.610854263324593e-06,
      "loss": 0.6684,
      "step": 7852
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35162471655504646,
      "learning_rate": 8.606093577835465e-06,
      "loss": 0.6627,
      "step": 7853
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.366024376267161,
      "learning_rate": 8.601333847939519e-06,
      "loss": 0.6004,
      "step": 7854
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3807067307926991,
      "learning_rate": 8.596575074035931e-06,
      "loss": 0.6374,
      "step": 7855
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36576253592679836,
      "learning_rate": 8.59181725652382e-06,
      "loss": 0.6375,
      "step": 7856
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3611384012421839,
      "learning_rate": 8.587060395802225e-06,
      "loss": 0.6178,
      "step": 7857
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3974729654468784,
      "learning_rate": 8.582304492270084e-06,
      "loss": 0.6356,
      "step": 7858
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42740534018870857,
      "learning_rate": 8.577549546326276e-06,
      "loss": 0.6051,
      "step": 7859
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3887488088397861,
      "learning_rate": 8.572795558369594e-06,
      "loss": 0.6219,
      "step": 7860
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3733238773914748,
      "learning_rate": 8.568042528798745e-06,
      "loss": 0.663,
      "step": 7861
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0854867495830132,
      "learning_rate": 8.563290458012366e-06,
      "loss": 0.3829,
      "step": 7862
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.45732376326509544,
      "learning_rate": 8.558539346409009e-06,
      "loss": 0.7118,
      "step": 7863
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40955041407159093,
      "learning_rate": 8.553789194387134e-06,
      "loss": 0.6359,
      "step": 7864
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34763093662735595,
      "learning_rate": 8.549040002345134e-06,
      "loss": 0.622,
      "step": 7865
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.355359448230471,
      "learning_rate": 8.54429177068133e-06,
      "loss": 0.6325,
      "step": 7866
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.391968030655878,
      "learning_rate": 8.539544499793924e-06,
      "loss": 0.6422,
      "step": 7867
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40367643828987115,
      "learning_rate": 8.534798190081093e-06,
      "loss": 0.6622,
      "step": 7868
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3144806088931552,
      "learning_rate": 8.530052841940885e-06,
      "loss": 0.5985,
      "step": 7869
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.355331810455646,
      "learning_rate": 8.525308455771293e-06,
      "loss": 0.6667,
      "step": 7870
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3762689622104312,
      "learning_rate": 8.520565031970232e-06,
      "loss": 0.6459,
      "step": 7871
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3921965064559968,
      "learning_rate": 8.515822570935508e-06,
      "loss": 0.6478,
      "step": 7872
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35288442560816036,
      "learning_rate": 8.511081073064878e-06,
      "loss": 0.6566,
      "step": 7873
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3699645670917092,
      "learning_rate": 8.506340538756002e-06,
      "loss": 0.6315,
      "step": 7874
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.39864163049661255,
      "learning_rate": 8.501600968406465e-06,
      "loss": 0.6124,
      "step": 7875
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38095698737726663,
      "learning_rate": 8.496862362413768e-06,
      "loss": 0.6834,
      "step": 7876
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44907070268171306,
      "learning_rate": 8.492124721175339e-06,
      "loss": 0.6764,
      "step": 7877
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4209699257590067,
      "learning_rate": 8.487388045088502e-06,
      "loss": 0.6241,
      "step": 7878
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3958409620038691,
      "learning_rate": 8.48265233455053e-06,
      "loss": 0.646,
      "step": 7879
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.386423747624354,
      "learning_rate": 8.477917589958604e-06,
      "loss": 0.6538,
      "step": 7880
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3778344508027301,
      "learning_rate": 8.473183811709797e-06,
      "loss": 0.6647,
      "step": 7881
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4067437648366826,
      "learning_rate": 8.468451000201161e-06,
      "loss": 0.6351,
      "step": 7882
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36470134904044316,
      "learning_rate": 8.463719155829604e-06,
      "loss": 0.6161,
      "step": 7883
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37471806707628363,
      "learning_rate": 8.45898827899199e-06,
      "loss": 0.6445,
      "step": 7884
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40322118444994587,
      "learning_rate": 8.454258370085091e-06,
      "loss": 0.6753,
      "step": 7885
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35858680271983706,
      "learning_rate": 8.449529429505605e-06,
      "loss": 0.6362,
      "step": 7886
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41474638836808536,
      "learning_rate": 8.44480145765013e-06,
      "loss": 0.6789,
      "step": 7887
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40468386573375903,
      "learning_rate": 8.440074454915202e-06,
      "loss": 0.67,
      "step": 7888
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.379008777502422,
      "learning_rate": 8.435348421697267e-06,
      "loss": 0.6213,
      "step": 7889
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3561725887477223,
      "learning_rate": 8.430623358392693e-06,
      "loss": 0.6269,
      "step": 7890
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37455308839056545,
      "learning_rate": 8.425899265397775e-06,
      "loss": 0.6442,
      "step": 7891
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3623477150293226,
      "learning_rate": 8.421176143108698e-06,
      "loss": 0.6348,
      "step": 7892
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3766449369318264,
      "learning_rate": 8.416453991921597e-06,
      "loss": 0.6539,
      "step": 7893
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38049827827745225,
      "learning_rate": 8.411732812232516e-06,
      "loss": 0.6294,
      "step": 7894
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45133957167830235,
      "learning_rate": 8.407012604437394e-06,
      "loss": 0.7087,
      "step": 7895
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.42673039584033884,
      "learning_rate": 8.40229336893213e-06,
      "loss": 0.6399,
      "step": 7896
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3985282970143197,
      "learning_rate": 8.397575106112526e-06,
      "loss": 0.6518,
      "step": 7897
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4332123801359489,
      "learning_rate": 8.392857816374276e-06,
      "loss": 0.6693,
      "step": 7898
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3870317585194247,
      "learning_rate": 8.388141500113023e-06,
      "loss": 0.6676,
      "step": 7899
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39576908525586246,
      "learning_rate": 8.383426157724326e-06,
      "loss": 0.682,
      "step": 7900
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37098250215807876,
      "learning_rate": 8.378711789603642e-06,
      "loss": 0.6362,
      "step": 7901
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.416213003015496,
      "learning_rate": 8.373998396146363e-06,
      "loss": 0.6222,
      "step": 7902
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3224066472098334,
      "learning_rate": 8.3692859777478e-06,
      "loss": 0.6406,
      "step": 7903
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3845504276250178,
      "learning_rate": 8.364574534803174e-06,
      "loss": 0.6819,
      "step": 7904
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33265561820758144,
      "learning_rate": 8.359864067707639e-06,
      "loss": 0.6222,
      "step": 7905
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3524898062872627,
      "learning_rate": 8.355154576856237e-06,
      "loss": 0.6094,
      "step": 7906
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3961520573428367,
      "learning_rate": 8.35044606264396e-06,
      "loss": 0.6666,
      "step": 7907
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44236269001692335,
      "learning_rate": 8.3457385254657e-06,
      "loss": 0.6426,
      "step": 7908
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36257979173489036,
      "learning_rate": 8.341031965716278e-06,
      "loss": 0.6455,
      "step": 7909
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3811653129108927,
      "learning_rate": 8.336326383790423e-06,
      "loss": 0.6617,
      "step": 7910
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38512108185241206,
      "learning_rate": 8.331621780082795e-06,
      "loss": 0.6125,
      "step": 7911
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36227208145821055,
      "learning_rate": 8.326918154987948e-06,
      "loss": 0.614,
      "step": 7912
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35949495713257895,
      "learning_rate": 8.32221550890038e-06,
      "loss": 0.6098,
      "step": 7913
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6360979135684459,
      "learning_rate": 8.317513842214502e-06,
      "loss": 0.3803,
      "step": 7914
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3866634921866569,
      "learning_rate": 8.31281315532462e-06,
      "loss": 0.6524,
      "step": 7915
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39766975608481986,
      "learning_rate": 8.308113448624976e-06,
      "loss": 0.6314,
      "step": 7916
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30401005630898165,
      "learning_rate": 8.303414722509754e-06,
      "loss": 0.649,
      "step": 7917
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39434901923558524,
      "learning_rate": 8.298716977373004e-06,
      "loss": 0.6887,
      "step": 7918
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3801796495037899,
      "learning_rate": 8.294020213608731e-06,
      "loss": 0.6586,
      "step": 7919
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38770441245366405,
      "learning_rate": 8.289324431610852e-06,
      "loss": 0.6678,
      "step": 7920
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4174425958125841,
      "learning_rate": 8.284629631773183e-06,
      "loss": 0.6327,
      "step": 7921
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41245715376641234,
      "learning_rate": 8.279935814489476e-06,
      "loss": 0.6883,
      "step": 7922
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35180932815866545,
      "learning_rate": 8.275242980153402e-06,
      "loss": 0.6373,
      "step": 7923
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39498685168952724,
      "learning_rate": 8.270551129158538e-06,
      "loss": 0.6613,
      "step": 7924
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40288599764023647,
      "learning_rate": 8.26586026189839e-06,
      "loss": 0.6662,
      "step": 7925
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38324054722694234,
      "learning_rate": 8.261170378766365e-06,
      "loss": 0.6539,
      "step": 7926
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3760417441982562,
      "learning_rate": 8.256481480155804e-06,
      "loss": 0.648,
      "step": 7927
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39780286865150255,
      "learning_rate": 8.251793566459967e-06,
      "loss": 0.6704,
      "step": 7928
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.34817902985983457,
      "learning_rate": 8.247106638071999e-06,
      "loss": 0.6351,
      "step": 7929
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.401876759154591,
      "learning_rate": 8.242420695385012e-06,
      "loss": 0.6717,
      "step": 7930
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3974631064525725,
      "learning_rate": 8.237735738792008e-06,
      "loss": 0.6448,
      "step": 7931
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40358736481101937,
      "learning_rate": 8.233051768685893e-06,
      "loss": 0.6548,
      "step": 7932
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3736221544851422,
      "learning_rate": 8.228368785459517e-06,
      "loss": 0.6428,
      "step": 7933
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3585816075683066,
      "learning_rate": 8.223686789505643e-06,
      "loss": 0.5895,
      "step": 7934
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3709982159743263,
      "learning_rate": 8.219005781216926e-06,
      "loss": 0.6488,
      "step": 7935
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4064225097138253,
      "learning_rate": 8.214325760985969e-06,
      "loss": 0.6525,
      "step": 7936
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39465306687916807,
      "learning_rate": 8.209646729205276e-06,
      "loss": 0.6468,
      "step": 7937
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3703821879201368,
      "learning_rate": 8.204968686267272e-06,
      "loss": 0.664,
      "step": 7938
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3937241812868365,
      "learning_rate": 8.200291632564307e-06,
      "loss": 0.6748,
      "step": 7939
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38067746715817474,
      "learning_rate": 8.195615568488627e-06,
      "loss": 0.6407,
      "step": 7940
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3972324896586241,
      "learning_rate": 8.19094049443241e-06,
      "loss": 0.6481,
      "step": 7941
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38270351436578126,
      "learning_rate": 8.186266410787754e-06,
      "loss": 0.69,
      "step": 7942
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39934117616784454,
      "learning_rate": 8.181593317946665e-06,
      "loss": 0.6917,
      "step": 7943
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4300177735462058,
      "learning_rate": 8.176921216301075e-06,
      "loss": 0.683,
      "step": 7944
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3366156661840615,
      "learning_rate": 8.172250106242827e-06,
      "loss": 0.6495,
      "step": 7945
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3319729819296055,
      "learning_rate": 8.167579988163675e-06,
      "loss": 0.6403,
      "step": 7946
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39251698396402235,
      "learning_rate": 8.162910862455296e-06,
      "loss": 0.6855,
      "step": 7947
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3674055025323252,
      "learning_rate": 8.158242729509295e-06,
      "loss": 0.6936,
      "step": 7948
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36272821004186867,
      "learning_rate": 8.153575589717168e-06,
      "loss": 0.6541,
      "step": 7949
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36918538249793686,
      "learning_rate": 8.148909443470343e-06,
      "loss": 0.6476,
      "step": 7950
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3623155074285083,
      "learning_rate": 8.144244291160186e-06,
      "loss": 0.6414,
      "step": 7951
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.601041910863602,
      "learning_rate": 8.139580133177933e-06,
      "loss": 0.3833,
      "step": 7952
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3976407359764906,
      "learning_rate": 8.134916969914773e-06,
      "loss": 0.6396,
      "step": 7953
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.610925786704964,
      "learning_rate": 8.1302548017618e-06,
      "loss": 0.3717,
      "step": 7954
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.34492842510079275,
      "learning_rate": 8.12559362911002e-06,
      "loss": 0.6408,
      "step": 7955
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4114148698877109,
      "learning_rate": 8.120933452350358e-06,
      "loss": 0.6473,
      "step": 7956
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3307309559724524,
      "learning_rate": 8.116274271873664e-06,
      "loss": 0.6231,
      "step": 7957
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3416289221295381,
      "learning_rate": 8.11161608807069e-06,
      "loss": 0.6024,
      "step": 7958
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3663365320850245,
      "learning_rate": 8.106958901332129e-06,
      "loss": 0.6363,
      "step": 7959
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.34938179518508383,
      "learning_rate": 8.102302712048553e-06,
      "loss": 0.6518,
      "step": 7960
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3765650456598689,
      "learning_rate": 8.097647520610478e-06,
      "loss": 0.6679,
      "step": 7961
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3946328696680534,
      "learning_rate": 8.09299332740834e-06,
      "loss": 0.6889,
      "step": 7962
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3576722990198514,
      "learning_rate": 8.088340132832467e-06,
      "loss": 0.6409,
      "step": 7963
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38269139128506596,
      "learning_rate": 8.08368793727311e-06,
      "loss": 0.6246,
      "step": 7964
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3475538126922574,
      "learning_rate": 8.079036741120471e-06,
      "loss": 0.6106,
      "step": 7965
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.374503584848179,
      "learning_rate": 8.074386544764617e-06,
      "loss": 0.6334,
      "step": 7966
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38659278569997413,
      "learning_rate": 8.069737348595559e-06,
      "loss": 0.652,
      "step": 7967
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4086440951741846,
      "learning_rate": 8.065089153003229e-06,
      "loss": 0.6816,
      "step": 7968
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3397322082927636,
      "learning_rate": 8.060441958377451e-06,
      "loss": 0.6459,
      "step": 7969
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.47492930817926005,
      "learning_rate": 8.055795765107988e-06,
      "loss": 0.6493,
      "step": 7970
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3503312126011516,
      "learning_rate": 8.051150573584508e-06,
      "loss": 0.6441,
      "step": 7971
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.327549935613,
      "learning_rate": 8.046506384196599e-06,
      "loss": 0.6355,
      "step": 7972
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4232038073428005,
      "learning_rate": 8.041863197333764e-06,
      "loss": 0.6286,
      "step": 7973
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3678809973978965,
      "learning_rate": 8.037221013385425e-06,
      "loss": 0.6315,
      "step": 7974
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4144959236786745,
      "learning_rate": 8.03257983274091e-06,
      "loss": 0.6867,
      "step": 7975
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3941711100660563,
      "learning_rate": 8.027939655789467e-06,
      "loss": 0.604,
      "step": 7976
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3755122860945654,
      "learning_rate": 8.023300482920278e-06,
      "loss": 0.6482,
      "step": 7977
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38291126257011565,
      "learning_rate": 8.018662314522396e-06,
      "loss": 0.631,
      "step": 7978
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36248398738289433,
      "learning_rate": 8.01402515098485e-06,
      "loss": 0.696,
      "step": 7979
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4073027939461656,
      "learning_rate": 8.009388992696534e-06,
      "loss": 0.6497,
      "step": 7980
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4568125849994383,
      "learning_rate": 8.00475384004628e-06,
      "loss": 0.7071,
      "step": 7981
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4117696297401613,
      "learning_rate": 8.000119693422843e-06,
      "loss": 0.6604,
      "step": 7982
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3717853878350845,
      "learning_rate": 7.995486553214868e-06,
      "loss": 0.6311,
      "step": 7983
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4306259673621188,
      "learning_rate": 7.990854419810939e-06,
      "loss": 0.6564,
      "step": 7984
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39395253637828026,
      "learning_rate": 7.986223293599544e-06,
      "loss": 0.6616,
      "step": 7985
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40652341622860766,
      "learning_rate": 7.981593174969096e-06,
      "loss": 0.645,
      "step": 7986
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3803674203274929,
      "learning_rate": 7.97696406430791e-06,
      "loss": 0.6481,
      "step": 7987
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4403546503213205,
      "learning_rate": 7.972335962004237e-06,
      "loss": 0.6742,
      "step": 7988
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7263989829283433,
      "learning_rate": 7.967708868446213e-06,
      "loss": 0.4058,
      "step": 7989
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33816608260359293,
      "learning_rate": 7.963082784021916e-06,
      "loss": 0.6179,
      "step": 7990
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3860629573993039,
      "learning_rate": 7.958457709119337e-06,
      "loss": 0.6427,
      "step": 7991
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3869869668965141,
      "learning_rate": 7.953833644126349e-06,
      "loss": 0.6273,
      "step": 7992
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3903870746195781,
      "learning_rate": 7.949210589430805e-06,
      "loss": 0.6764,
      "step": 7993
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3764555597052667,
      "learning_rate": 7.944588545420404e-06,
      "loss": 0.6572,
      "step": 7994
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35576789875741743,
      "learning_rate": 7.939967512482804e-06,
      "loss": 0.6381,
      "step": 7995
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39354507689963575,
      "learning_rate": 7.935347491005569e-06,
      "loss": 0.6482,
      "step": 7996
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33155600162644183,
      "learning_rate": 7.930728481376166e-06,
      "loss": 0.6135,
      "step": 7997
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3979861892469978,
      "learning_rate": 7.926110483981988e-06,
      "loss": 0.6785,
      "step": 7998
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3518675546344672,
      "learning_rate": 7.92149349921034e-06,
      "loss": 0.6359,
      "step": 7999
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3383765196638535,
      "learning_rate": 7.916877527448447e-06,
      "loss": 0.6072,
      "step": 8000
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39279239199248367,
      "learning_rate": 7.912262569083444e-06,
      "loss": 0.6594,
      "step": 8001
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3976193004400414,
      "learning_rate": 7.907648624502386e-06,
      "loss": 0.6618,
      "step": 8002
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.41054337060673046,
      "learning_rate": 7.903035694092227e-06,
      "loss": 0.6532,
      "step": 8003
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37689002298334456,
      "learning_rate": 7.898423778239857e-06,
      "loss": 0.6482,
      "step": 8004
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4259348818785242,
      "learning_rate": 7.893812877332075e-06,
      "loss": 0.6765,
      "step": 8005
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36096177424281195,
      "learning_rate": 7.889202991755573e-06,
      "loss": 0.6555,
      "step": 8006
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37448389366952467,
      "learning_rate": 7.884594121896995e-06,
      "loss": 0.6238,
      "step": 8007
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3779206139280553,
      "learning_rate": 7.879986268142888e-06,
      "loss": 0.6149,
      "step": 8008
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5854303645140463,
      "learning_rate": 7.875379430879684e-06,
      "loss": 0.3725,
      "step": 8009
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37489850776721056,
      "learning_rate": 7.870773610493764e-06,
      "loss": 0.6603,
      "step": 8010
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3869062653586214,
      "learning_rate": 7.866168807371424e-06,
      "loss": 0.6724,
      "step": 8011
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3628892158736783,
      "learning_rate": 7.861565021898836e-06,
      "loss": 0.6679,
      "step": 8012
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34875787774608286,
      "learning_rate": 7.856962254462144e-06,
      "loss": 0.6381,
      "step": 8013
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3479050779403151,
      "learning_rate": 7.852360505447353e-06,
      "loss": 0.6397,
      "step": 8014
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3844741139082617,
      "learning_rate": 7.84775977524042e-06,
      "loss": 0.678,
      "step": 8015
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3883541573822523,
      "learning_rate": 7.843160064227204e-06,
      "loss": 0.6117,
      "step": 8016
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3664933290418997,
      "learning_rate": 7.838561372793466e-06,
      "loss": 0.6052,
      "step": 8017
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3526811285805841,
      "learning_rate": 7.833963701324897e-06,
      "loss": 0.5956,
      "step": 8018
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37192443300244316,
      "learning_rate": 7.829367050207097e-06,
      "loss": 0.6391,
      "step": 8019
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36927174699640086,
      "learning_rate": 7.824771419825588e-06,
      "loss": 0.6443,
      "step": 8020
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3834231850322172,
      "learning_rate": 7.820176810565793e-06,
      "loss": 0.659,
      "step": 8021
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39419170147299726,
      "learning_rate": 7.815583222813067e-06,
      "loss": 0.6633,
      "step": 8022
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39626380421700647,
      "learning_rate": 7.810990656952657e-06,
      "loss": 0.5916,
      "step": 8023
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.391230742342482,
      "learning_rate": 7.806399113369738e-06,
      "loss": 0.6401,
      "step": 8024
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4040319793614508,
      "learning_rate": 7.801808592449405e-06,
      "loss": 0.6153,
      "step": 8025
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3715671533372542,
      "learning_rate": 7.79721909457664e-06,
      "loss": 0.6657,
      "step": 8026
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38275096811631176,
      "learning_rate": 7.792630620136388e-06,
      "loss": 0.662,
      "step": 8027
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3351998542479529,
      "learning_rate": 7.788043169513454e-06,
      "loss": 0.5718,
      "step": 8028
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3816260325579834,
      "learning_rate": 7.783456743092595e-06,
      "loss": 0.6209,
      "step": 8029
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3941249848571516,
      "learning_rate": 7.77887134125846e-06,
      "loss": 0.594,
      "step": 8030
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3917503600388991,
      "learning_rate": 7.774286964395636e-06,
      "loss": 0.6353,
      "step": 8031
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35339565961900593,
      "learning_rate": 7.769703612888593e-06,
      "loss": 0.6093,
      "step": 8032
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37286147338718006,
      "learning_rate": 7.765121287121737e-06,
      "loss": 0.6512,
      "step": 8033
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.376145576847113,
      "learning_rate": 7.760539987479383e-06,
      "loss": 0.6746,
      "step": 8034
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34622249195534194,
      "learning_rate": 7.755959714345757e-06,
      "loss": 0.6309,
      "step": 8035
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5508627257911753,
      "learning_rate": 7.75138046810501e-06,
      "loss": 0.3593,
      "step": 8036
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40186173009315684,
      "learning_rate": 7.746802249141186e-06,
      "loss": 0.6511,
      "step": 8037
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3820365313514589,
      "learning_rate": 7.742225057838258e-06,
      "loss": 0.6676,
      "step": 8038
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36734964790263586,
      "learning_rate": 7.737648894580117e-06,
      "loss": 0.6142,
      "step": 8039
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6068172406178309,
      "learning_rate": 7.733073759750537e-06,
      "loss": 0.3563,
      "step": 8040
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5831584847154241,
      "learning_rate": 7.728499653733259e-06,
      "loss": 0.4209,
      "step": 8041
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3744505471817746,
      "learning_rate": 7.723926576911898e-06,
      "loss": 0.6466,
      "step": 8042
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40357262944776084,
      "learning_rate": 7.719354529669985e-06,
      "loss": 0.6382,
      "step": 8043
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40181821364642056,
      "learning_rate": 7.714783512390973e-06,
      "loss": 0.6881,
      "step": 8044
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40704736677743086,
      "learning_rate": 7.71021352545824e-06,
      "loss": 0.6459,
      "step": 8045
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4051670996784528,
      "learning_rate": 7.705644569255048e-06,
      "loss": 0.6309,
      "step": 8046
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4159968705418176,
      "learning_rate": 7.7010766441646e-06,
      "loss": 0.6338,
      "step": 8047
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4002207929371665,
      "learning_rate": 7.69650975057e-06,
      "loss": 0.6445,
      "step": 8048
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4026350727904939,
      "learning_rate": 7.69194388885427e-06,
      "loss": 0.6641,
      "step": 8049
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44083134463477,
      "learning_rate": 7.687379059400347e-06,
      "loss": 0.6087,
      "step": 8050
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3849892111960571,
      "learning_rate": 7.682815262591064e-06,
      "loss": 0.6604,
      "step": 8051
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3724472883550453,
      "learning_rate": 7.678252498809191e-06,
      "loss": 0.6434,
      "step": 8052
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6232315636964325,
      "learning_rate": 7.673690768437406e-06,
      "loss": 0.3646,
      "step": 8053
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42366260873305983,
      "learning_rate": 7.669130071858277e-06,
      "loss": 0.662,
      "step": 8054
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39845711845606263,
      "learning_rate": 7.664570409454323e-06,
      "loss": 0.6729,
      "step": 8055
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4037614485444794,
      "learning_rate": 7.66001178160796e-06,
      "loss": 0.6467,
      "step": 8056
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37821686052824305,
      "learning_rate": 7.655454188701496e-06,
      "loss": 0.6546,
      "step": 8057
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.45148059524554335,
      "learning_rate": 7.650897631117185e-06,
      "loss": 0.6226,
      "step": 8058
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38084093990410045,
      "learning_rate": 7.646342109237179e-06,
      "loss": 0.6777,
      "step": 8059
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3775471369058361,
      "learning_rate": 7.641787623443535e-06,
      "loss": 0.654,
      "step": 8060
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3807894726199406,
      "learning_rate": 7.637234174118237e-06,
      "loss": 0.6517,
      "step": 8061
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3560301326543443,
      "learning_rate": 7.632681761643182e-06,
      "loss": 0.5745,
      "step": 8062
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42270284784724743,
      "learning_rate": 7.628130386400168e-06,
      "loss": 0.6781,
      "step": 8063
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3776608180307289,
      "learning_rate": 7.62358004877092e-06,
      "loss": 0.6445,
      "step": 8064
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3816101898728454,
      "learning_rate": 7.619030749137071e-06,
      "loss": 0.6679,
      "step": 8065
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37987338646136176,
      "learning_rate": 7.614482487880155e-06,
      "loss": 0.616,
      "step": 8066
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36566188301574304,
      "learning_rate": 7.609935265381634e-06,
      "loss": 0.6717,
      "step": 8067
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40845484636026486,
      "learning_rate": 7.605389082022878e-06,
      "loss": 0.6847,
      "step": 8068
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39886144710861365,
      "learning_rate": 7.600843938185172e-06,
      "loss": 0.6206,
      "step": 8069
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35761806839714166,
      "learning_rate": 7.5962998342497185e-06,
      "loss": 0.6454,
      "step": 8070
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4125132850645186,
      "learning_rate": 7.591756770597609e-06,
      "loss": 0.6689,
      "step": 8071
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3572289184362762,
      "learning_rate": 7.587214747609875e-06,
      "loss": 0.6343,
      "step": 8072
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3247816845160264,
      "learning_rate": 7.582673765667457e-06,
      "loss": 0.6187,
      "step": 8073
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3531329475897983,
      "learning_rate": 7.578133825151186e-06,
      "loss": 0.6366,
      "step": 8074
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34168339286237637,
      "learning_rate": 7.573594926441821e-06,
      "loss": 0.6373,
      "step": 8075
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39683938911359545,
      "learning_rate": 7.569057069920058e-06,
      "loss": 0.6217,
      "step": 8076
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3950987662700686,
      "learning_rate": 7.564520255966459e-06,
      "loss": 0.6498,
      "step": 8077
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3982400914516018,
      "learning_rate": 7.559984484961527e-06,
      "loss": 0.6571,
      "step": 8078
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36042790945177383,
      "learning_rate": 7.555449757285678e-06,
      "loss": 0.6348,
      "step": 8079
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38182113720689187,
      "learning_rate": 7.550916073319223e-06,
      "loss": 0.6461,
      "step": 8080
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36142972109114524,
      "learning_rate": 7.546383433442403e-06,
      "loss": 0.6837,
      "step": 8081
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4162230519424224,
      "learning_rate": 7.541851838035363e-06,
      "loss": 0.7085,
      "step": 8082
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40760514107832435,
      "learning_rate": 7.537321287478165e-06,
      "loss": 0.626,
      "step": 8083
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3175951125661171,
      "learning_rate": 7.532791782150788e-06,
      "loss": 0.6259,
      "step": 8084
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.393855810695903,
      "learning_rate": 7.528263322433098e-06,
      "loss": 0.6556,
      "step": 8085
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.661230180218676,
      "learning_rate": 7.523735908704901e-06,
      "loss": 0.3689,
      "step": 8086
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.402353225874624,
      "learning_rate": 7.519209541345907e-06,
      "loss": 0.6164,
      "step": 8087
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4245414327233141,
      "learning_rate": 7.5146842207357415e-06,
      "loss": 0.6496,
      "step": 8088
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35440024312455465,
      "learning_rate": 7.510159947253919e-06,
      "loss": 0.6271,
      "step": 8089
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34263130776332223,
      "learning_rate": 7.505636721279912e-06,
      "loss": 0.6141,
      "step": 8090
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4354801800060924,
      "learning_rate": 7.5011145431930574e-06,
      "loss": 0.685,
      "step": 8091
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35156937719937736,
      "learning_rate": 7.49659341337263e-06,
      "loss": 0.6426,
      "step": 8092
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38940088858411526,
      "learning_rate": 7.49207333219782e-06,
      "loss": 0.6607,
      "step": 8093
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44838811300568854,
      "learning_rate": 7.487554300047706e-06,
      "loss": 0.6813,
      "step": 8094
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3354518870471337,
      "learning_rate": 7.483036317301295e-06,
      "loss": 0.6396,
      "step": 8095
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3909692240721195,
      "learning_rate": 7.478519384337528e-06,
      "loss": 0.6328,
      "step": 8096
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33820107785827996,
      "learning_rate": 7.474003501535208e-06,
      "loss": 0.6349,
      "step": 8097
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39218886785242996,
      "learning_rate": 7.469488669273088e-06,
      "loss": 0.6801,
      "step": 8098
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.387344461906718,
      "learning_rate": 7.464974887929828e-06,
      "loss": 0.6427,
      "step": 8099
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4217363899386411,
      "learning_rate": 7.460462157883976e-06,
      "loss": 0.6814,
      "step": 8100
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3965752702653183,
      "learning_rate": 7.455950479514023e-06,
      "loss": 0.6563,
      "step": 8101
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3200191076294959,
      "learning_rate": 7.45143985319835e-06,
      "loss": 0.662,
      "step": 8102
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3616870795360135,
      "learning_rate": 7.446930279315263e-06,
      "loss": 0.6433,
      "step": 8103
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3888580828439993,
      "learning_rate": 7.442421758242981e-06,
      "loss": 0.6509,
      "step": 8104
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34146179883838407,
      "learning_rate": 7.437914290359613e-06,
      "loss": 0.6068,
      "step": 8105
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4390653432340362,
      "learning_rate": 7.433407876043201e-06,
      "loss": 0.6943,
      "step": 8106
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36697358836875893,
      "learning_rate": 7.428902515671703e-06,
      "loss": 0.6578,
      "step": 8107
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3787671603447021,
      "learning_rate": 7.424398209622961e-06,
      "loss": 0.6767,
      "step": 8108
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4062720899091138,
      "learning_rate": 7.4198949582747474e-06,
      "loss": 0.6972,
      "step": 8109
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41417205013396524,
      "learning_rate": 7.415392762004763e-06,
      "loss": 0.6714,
      "step": 8110
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4361701377482034,
      "learning_rate": 7.410891621190584e-06,
      "loss": 0.6701,
      "step": 8111
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.347122129612128,
      "learning_rate": 7.406391536209721e-06,
      "loss": 0.6659,
      "step": 8112
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4320631994687574,
      "learning_rate": 7.401892507439597e-06,
      "loss": 0.675,
      "step": 8113
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3479928316474306,
      "learning_rate": 7.397394535257525e-06,
      "loss": 0.6142,
      "step": 8114
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3602665976118137,
      "learning_rate": 7.392897620040755e-06,
      "loss": 0.6806,
      "step": 8115
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38226300869402974,
      "learning_rate": 7.388401762166437e-06,
      "loss": 0.6192,
      "step": 8116
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36447581222623954,
      "learning_rate": 7.383906962011631e-06,
      "loss": 0.6178,
      "step": 8117
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3463125275613387,
      "learning_rate": 7.37941321995332e-06,
      "loss": 0.6387,
      "step": 8118
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.339821594231107,
      "learning_rate": 7.374920536368371e-06,
      "loss": 0.6326,
      "step": 8119
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3912437655397039,
      "learning_rate": 7.370428911633592e-06,
      "loss": 0.6433,
      "step": 8120
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36125616405307825,
      "learning_rate": 7.365938346125685e-06,
      "loss": 0.6276,
      "step": 8121
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36656092126322165,
      "learning_rate": 7.361448840221279e-06,
      "loss": 0.6359,
      "step": 8122
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.39879433334847897,
      "learning_rate": 7.356960394296882e-06,
      "loss": 0.6223,
      "step": 8123
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37053448851507065,
      "learning_rate": 7.352473008728962e-06,
      "loss": 0.6587,
      "step": 8124
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3475224973004984,
      "learning_rate": 7.347986683893848e-06,
      "loss": 0.6206,
      "step": 8125
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37548600074103455,
      "learning_rate": 7.343501420167811e-06,
      "loss": 0.6528,
      "step": 8126
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41878027171052956,
      "learning_rate": 7.3390172179270314e-06,
      "loss": 0.6659,
      "step": 8127
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33442340598330833,
      "learning_rate": 7.334534077547582e-06,
      "loss": 0.6271,
      "step": 8128
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3967884990500279,
      "learning_rate": 7.330051999405463e-06,
      "loss": 0.6799,
      "step": 8129
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.48987803855952294,
      "learning_rate": 7.32557098387658e-06,
      "loss": 0.6799,
      "step": 8130
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38123811993365875,
      "learning_rate": 7.3210910313367535e-06,
      "loss": 0.6844,
      "step": 8131
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43210811002128663,
      "learning_rate": 7.31661214216171e-06,
      "loss": 0.6576,
      "step": 8132
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38881179301749375,
      "learning_rate": 7.312134316727093e-06,
      "loss": 0.6672,
      "step": 8133
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3878402281028553,
      "learning_rate": 7.3076575554084425e-06,
      "loss": 0.651,
      "step": 8134
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3714761305524285,
      "learning_rate": 7.303181858581225e-06,
      "loss": 0.6359,
      "step": 8135
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3715290292502861,
      "learning_rate": 7.298707226620818e-06,
      "loss": 0.6355,
      "step": 8136
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38415778207453605,
      "learning_rate": 7.294233659902481e-06,
      "loss": 0.6314,
      "step": 8137
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.39231491307801863,
      "learning_rate": 7.289761158801438e-06,
      "loss": 0.6735,
      "step": 8138
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41697989354199355,
      "learning_rate": 7.2852897236927675e-06,
      "loss": 0.6822,
      "step": 8139
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33361979194676905,
      "learning_rate": 7.280819354951496e-06,
      "loss": 0.647,
      "step": 8140
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36209332325143845,
      "learning_rate": 7.276350052952548e-06,
      "loss": 0.6298,
      "step": 8141
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3698347976455967,
      "learning_rate": 7.27188181807075e-06,
      "loss": 0.642,
      "step": 8142
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3741219535904208,
      "learning_rate": 7.267414650680851e-06,
      "loss": 0.6731,
      "step": 8143
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4043033869878581,
      "learning_rate": 7.2629485511575075e-06,
      "loss": 0.6275,
      "step": 8144
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35898002043935734,
      "learning_rate": 7.2584835198752875e-06,
      "loss": 0.6061,
      "step": 8145
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3421877181297941,
      "learning_rate": 7.254019557208667e-06,
      "loss": 0.6474,
      "step": 8146
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37237568755403905,
      "learning_rate": 7.249556663532038e-06,
      "loss": 0.641,
      "step": 8147
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37021570122928243,
      "learning_rate": 7.245094839219688e-06,
      "loss": 0.6692,
      "step": 8148
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6068370602565044,
      "learning_rate": 7.24063408464583e-06,
      "loss": 0.3858,
      "step": 8149
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37960075885892147,
      "learning_rate": 7.236174400184588e-06,
      "loss": 0.6535,
      "step": 8150
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37625587199854327,
      "learning_rate": 7.231715786209972e-06,
      "loss": 0.6498,
      "step": 8151
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43292678820790237,
      "learning_rate": 7.227258243095938e-06,
      "loss": 0.6542,
      "step": 8152
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.380297363424522,
      "learning_rate": 7.222801771216337e-06,
      "loss": 0.6441,
      "step": 8153
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.39768934525868,
      "learning_rate": 7.218346370944913e-06,
      "loss": 0.6534,
      "step": 8154
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.381744066777726,
      "learning_rate": 7.213892042655344e-06,
      "loss": 0.6047,
      "step": 8155
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38715343827170595,
      "learning_rate": 7.209438786721212e-06,
      "loss": 0.6411,
      "step": 8156
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.46635447495485743,
      "learning_rate": 7.2049866035159975e-06,
      "loss": 0.649,
      "step": 8157
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3925770558840692,
      "learning_rate": 7.200535493413104e-06,
      "loss": 0.6715,
      "step": 8158
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4240321329326816,
      "learning_rate": 7.196085456785839e-06,
      "loss": 0.6224,
      "step": 8159
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3945886377867638,
      "learning_rate": 7.1916364940074254e-06,
      "loss": 0.6316,
      "step": 8160
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.40080509106711554,
      "learning_rate": 7.187188605450996e-06,
      "loss": 0.6536,
      "step": 8161
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35638534013897316,
      "learning_rate": 7.182741791489578e-06,
      "loss": 0.6324,
      "step": 8162
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3558283821942239,
      "learning_rate": 7.1782960524961276e-06,
      "loss": 0.6319,
      "step": 8163
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3423084490836396,
      "learning_rate": 7.173851388843507e-06,
      "loss": 0.5932,
      "step": 8164
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.34638870612790895,
      "learning_rate": 7.169407800904468e-06,
      "loss": 0.5859,
      "step": 8165
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33363105393814335,
      "learning_rate": 7.164965289051709e-06,
      "loss": 0.6189,
      "step": 8166
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6274209507506022,
      "learning_rate": 7.160523853657817e-06,
      "loss": 0.3532,
      "step": 8167
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38721053976113623,
      "learning_rate": 7.156083495095276e-06,
      "loss": 0.6316,
      "step": 8168
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3964390862586294,
      "learning_rate": 7.151644213736499e-06,
      "loss": 0.685,
      "step": 8169
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3966163775028178,
      "learning_rate": 7.147206009953815e-06,
      "loss": 0.6726,
      "step": 8170
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36498808930811827,
      "learning_rate": 7.142768884119431e-06,
      "loss": 0.6611,
      "step": 8171
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3508685151021271,
      "learning_rate": 7.138332836605495e-06,
      "loss": 0.6034,
      "step": 8172
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3562015959108758,
      "learning_rate": 7.13389786778405e-06,
      "loss": 0.6565,
      "step": 8173
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3852493360293631,
      "learning_rate": 7.129463978027054e-06,
      "loss": 0.629,
      "step": 8174
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36720477659637313,
      "learning_rate": 7.125031167706375e-06,
      "loss": 0.6319,
      "step": 8175
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3278257720114058,
      "learning_rate": 7.120599437193778e-06,
      "loss": 0.6344,
      "step": 8176
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4081768226195052,
      "learning_rate": 7.116168786860951e-06,
      "loss": 0.6773,
      "step": 8177
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38341495251146696,
      "learning_rate": 7.1117392170794876e-06,
      "loss": 0.6684,
      "step": 8178
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35991318273232076,
      "learning_rate": 7.107310728220893e-06,
      "loss": 0.6061,
      "step": 8179
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4319422167247264,
      "learning_rate": 7.102883320656575e-06,
      "loss": 0.6512,
      "step": 8180
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4141178738888802,
      "learning_rate": 7.098456994757865e-06,
      "loss": 0.6451,
      "step": 8181
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3763272788170095,
      "learning_rate": 7.094031750895977e-06,
      "loss": 0.6602,
      "step": 8182
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.34262304866704374,
      "learning_rate": 7.0896075894420624e-06,
      "loss": 0.6456,
      "step": 8183
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4171855864902703,
      "learning_rate": 7.085184510767173e-06,
      "loss": 0.6702,
      "step": 8184
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.362662364961235,
      "learning_rate": 7.08076251524225e-06,
      "loss": 0.6291,
      "step": 8185
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36793233889098986,
      "learning_rate": 7.076341603238177e-06,
      "loss": 0.6268,
      "step": 8186
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35687141982846315,
      "learning_rate": 7.071921775125732e-06,
      "loss": 0.5933,
      "step": 8187
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3523599368533408,
      "learning_rate": 7.067503031275589e-06,
      "loss": 0.6165,
      "step": 8188
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37980375114873355,
      "learning_rate": 7.063085372058347e-06,
      "loss": 0.6233,
      "step": 8189
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.391150587483317,
      "learning_rate": 7.058668797844519e-06,
      "loss": 0.632,
      "step": 8190
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3832660874687364,
      "learning_rate": 7.054253309004504e-06,
      "loss": 0.6287,
      "step": 8191
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3968091473784325,
      "learning_rate": 7.049838905908628e-06,
      "loss": 0.6325,
      "step": 8192
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41505028667176125,
      "learning_rate": 7.045425588927122e-06,
      "loss": 0.7111,
      "step": 8193
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38024126259851365,
      "learning_rate": 7.041013358430129e-06,
      "loss": 0.6597,
      "step": 8194
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4037088419011305,
      "learning_rate": 7.0366022147877e-06,
      "loss": 0.6315,
      "step": 8195
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36278453769046326,
      "learning_rate": 7.0321921583697815e-06,
      "loss": 0.6336,
      "step": 8196
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3763315401836733,
      "learning_rate": 7.027783189546247e-06,
      "loss": 0.6246,
      "step": 8197
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33667448746014467,
      "learning_rate": 7.023375308686875e-06,
      "loss": 0.6058,
      "step": 8198
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3711961491335867,
      "learning_rate": 7.018968516161331e-06,
      "loss": 0.6066,
      "step": 8199
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.39974580097302637,
      "learning_rate": 7.014562812339227e-06,
      "loss": 0.6846,
      "step": 8200
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35772942123432844,
      "learning_rate": 7.0101581975900644e-06,
      "loss": 0.6206,
      "step": 8201
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36903315140776966,
      "learning_rate": 7.005754672283238e-06,
      "loss": 0.6786,
      "step": 8202
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.39302111054231387,
      "learning_rate": 7.001352236788077e-06,
      "loss": 0.6301,
      "step": 8203
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.40524549845701857,
      "learning_rate": 6.996950891473811e-06,
      "loss": 0.6717,
      "step": 8204
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38648366004312623,
      "learning_rate": 6.992550636709561e-06,
      "loss": 0.6348,
      "step": 8205
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3971468230874585,
      "learning_rate": 6.9881514728643815e-06,
      "loss": 0.6744,
      "step": 8206
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3992776312746348,
      "learning_rate": 6.983753400307225e-06,
      "loss": 0.6467,
      "step": 8207
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38386934831057234,
      "learning_rate": 6.979356419406951e-06,
      "loss": 0.6086,
      "step": 8208
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3720326331362659,
      "learning_rate": 6.974960530532329e-06,
      "loss": 0.6661,
      "step": 8209
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3841202464003951,
      "learning_rate": 6.970565734052042e-06,
      "loss": 0.6366,
      "step": 8210
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36191288950746364,
      "learning_rate": 6.966172030334666e-06,
      "loss": 0.6604,
      "step": 8211
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3916342706055615,
      "learning_rate": 6.9617794197487e-06,
      "loss": 0.6548,
      "step": 8212
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3407179251809534,
      "learning_rate": 6.957387902662549e-06,
      "loss": 0.6664,
      "step": 8213
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3370379499667735,
      "learning_rate": 6.952997479444523e-06,
      "loss": 0.603,
      "step": 8214
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.46412040493455575,
      "learning_rate": 6.9486081504628476e-06,
      "loss": 0.6726,
      "step": 8215
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43102199853422957,
      "learning_rate": 6.944219916085637e-06,
      "loss": 0.7241,
      "step": 8216
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35868888314628533,
      "learning_rate": 6.939832776680937e-06,
      "loss": 0.5844,
      "step": 8217
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4274934515201532,
      "learning_rate": 6.935446732616695e-06,
      "loss": 0.7024,
      "step": 8218
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.40376641216699183,
      "learning_rate": 6.9310617842607505e-06,
      "loss": 0.6858,
      "step": 8219
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3602226035483309,
      "learning_rate": 6.926677931980866e-06,
      "loss": 0.6408,
      "step": 8220
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3918890599013596,
      "learning_rate": 6.922295176144724e-06,
      "loss": 0.6462,
      "step": 8221
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37915578345729845,
      "learning_rate": 6.91791351711989e-06,
      "loss": 0.6359,
      "step": 8222
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39038342901434653,
      "learning_rate": 6.913532955273848e-06,
      "loss": 0.652,
      "step": 8223
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4085372778480862,
      "learning_rate": 6.9091534909739984e-06,
      "loss": 0.6408,
      "step": 8224
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4014979941613326,
      "learning_rate": 6.904775124587629e-06,
      "loss": 0.6641,
      "step": 8225
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38540366104038715,
      "learning_rate": 6.900397856481955e-06,
      "loss": 0.6434,
      "step": 8226
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40196748599929427,
      "learning_rate": 6.896021687024092e-06,
      "loss": 0.6518,
      "step": 8227
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3658820498428344,
      "learning_rate": 6.891646616581067e-06,
      "loss": 0.6525,
      "step": 8228
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3599636480776641,
      "learning_rate": 6.8872726455198155e-06,
      "loss": 0.6263,
      "step": 8229
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35945811054128046,
      "learning_rate": 6.882899774207163e-06,
      "loss": 0.5798,
      "step": 8230
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36818810909052496,
      "learning_rate": 6.878528003009866e-06,
      "loss": 0.6892,
      "step": 8231
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42265753882129586,
      "learning_rate": 6.8741573322945845e-06,
      "loss": 0.6243,
      "step": 8232
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38340658365018854,
      "learning_rate": 6.869787762427869e-06,
      "loss": 0.5821,
      "step": 8233
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3631618271967299,
      "learning_rate": 6.86541929377619e-06,
      "loss": 0.6549,
      "step": 8234
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35764425993487337,
      "learning_rate": 6.861051926705946e-06,
      "loss": 0.6683,
      "step": 8235
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3905979318927695,
      "learning_rate": 6.856685661583402e-06,
      "loss": 0.6314,
      "step": 8236
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35626687365167997,
      "learning_rate": 6.852320498774758e-06,
      "loss": 0.6132,
      "step": 8237
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3928988648702639,
      "learning_rate": 6.847956438646124e-06,
      "loss": 0.6798,
      "step": 8238
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3681848922581865,
      "learning_rate": 6.843593481563491e-06,
      "loss": 0.6397,
      "step": 8239
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3510567674404767,
      "learning_rate": 6.839231627892784e-06,
      "loss": 0.6263,
      "step": 8240
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41874986059770897,
      "learning_rate": 6.834870877999829e-06,
      "loss": 0.6676,
      "step": 8241
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43519324482416955,
      "learning_rate": 6.83051123225035e-06,
      "loss": 0.6118,
      "step": 8242
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3995727071314704,
      "learning_rate": 6.82615269100999e-06,
      "loss": 0.6122,
      "step": 8243
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.359351336418022,
      "learning_rate": 6.821795254644304e-06,
      "loss": 0.6557,
      "step": 8244
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39620457799930076,
      "learning_rate": 6.817438923518725e-06,
      "loss": 0.6553,
      "step": 8245
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40622066194602807,
      "learning_rate": 6.8130836979986236e-06,
      "loss": 0.6411,
      "step": 8246
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37604646313143403,
      "learning_rate": 6.808729578449273e-06,
      "loss": 0.6068,
      "step": 8247
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40995669322468825,
      "learning_rate": 6.804376565235828e-06,
      "loss": 0.6913,
      "step": 8248
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39568227564089264,
      "learning_rate": 6.800024658723399e-06,
      "loss": 0.6544,
      "step": 8249
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3760801380712341,
      "learning_rate": 6.795673859276952e-06,
      "loss": 0.6696,
      "step": 8250
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3423169705031463,
      "learning_rate": 6.7913241672613925e-06,
      "loss": 0.5954,
      "step": 8251
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36557051773319416,
      "learning_rate": 6.7869755830415305e-06,
      "loss": 0.6241,
      "step": 8252
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38808082501586905,
      "learning_rate": 6.782628106982061e-06,
      "loss": 0.6119,
      "step": 8253
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4081201911993979,
      "learning_rate": 6.778281739447612e-06,
      "loss": 0.6248,
      "step": 8254
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4376940402917316,
      "learning_rate": 6.773936480802703e-06,
      "loss": 0.6396,
      "step": 8255
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.386437593541676,
      "learning_rate": 6.76959233141177e-06,
      "loss": 0.6447,
      "step": 8256
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38881053646131575,
      "learning_rate": 6.765249291639148e-06,
      "loss": 0.6633,
      "step": 8257
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40157786334151474,
      "learning_rate": 6.760907361849092e-06,
      "loss": 0.629,
      "step": 8258
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35269951082482737,
      "learning_rate": 6.7565665424057404e-06,
      "loss": 0.6123,
      "step": 8259
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3906947545572138,
      "learning_rate": 6.752226833673159e-06,
      "loss": 0.6443,
      "step": 8260
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3898499506363248,
      "learning_rate": 6.747888236015314e-06,
      "loss": 0.6583,
      "step": 8261
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.377101245698827,
      "learning_rate": 6.743550749796077e-06,
      "loss": 0.6693,
      "step": 8262
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39117308648655774,
      "learning_rate": 6.739214375379237e-06,
      "loss": 0.6496,
      "step": 8263
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3499381743240102,
      "learning_rate": 6.734879113128465e-06,
      "loss": 0.6493,
      "step": 8264
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4067331544981441,
      "learning_rate": 6.730544963407362e-06,
      "loss": 0.6171,
      "step": 8265
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38601904324518693,
      "learning_rate": 6.726211926579427e-06,
      "loss": 0.6731,
      "step": 8266
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36171766720440685,
      "learning_rate": 6.721880003008074e-06,
      "loss": 0.6341,
      "step": 8267
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33882346425148924,
      "learning_rate": 6.717549193056596e-06,
      "loss": 0.6284,
      "step": 8268
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6287057547741123,
      "learning_rate": 6.713219497088239e-06,
      "loss": 0.4229,
      "step": 8269
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3867611679771828,
      "learning_rate": 6.708890915466108e-06,
      "loss": 0.6454,
      "step": 8270
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4323373863599283,
      "learning_rate": 6.704563448553248e-06,
      "loss": 0.6839,
      "step": 8271
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.45401031297655126,
      "learning_rate": 6.700237096712598e-06,
      "loss": 0.6788,
      "step": 8272
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3694306364384412,
      "learning_rate": 6.695911860306996e-06,
      "loss": 0.6423,
      "step": 8273
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38432913634774263,
      "learning_rate": 6.691587739699199e-06,
      "loss": 0.6463,
      "step": 8274
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3570017007175063,
      "learning_rate": 6.6872647352518635e-06,
      "loss": 0.6307,
      "step": 8275
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.31998812940891996,
      "learning_rate": 6.6829428473275604e-06,
      "loss": 0.6389,
      "step": 8276
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42351248653153584,
      "learning_rate": 6.678622076288757e-06,
      "loss": 0.6862,
      "step": 8277
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40164442057026206,
      "learning_rate": 6.674302422497836e-06,
      "loss": 0.6238,
      "step": 8278
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3337499979766186,
      "learning_rate": 6.669983886317073e-06,
      "loss": 0.6096,
      "step": 8279
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.437336884604684,
      "learning_rate": 6.665666468108663e-06,
      "loss": 0.6983,
      "step": 8280
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4023622574341139,
      "learning_rate": 6.6613501682347085e-06,
      "loss": 0.5982,
      "step": 8281
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4017360893791086,
      "learning_rate": 6.657034987057192e-06,
      "loss": 0.6548,
      "step": 8282
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4003814052255883,
      "learning_rate": 6.652720924938052e-06,
      "loss": 0.665,
      "step": 8283
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35821469122964383,
      "learning_rate": 6.64840798223908e-06,
      "loss": 0.6654,
      "step": 8284
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41467867645798634,
      "learning_rate": 6.644096159322009e-06,
      "loss": 0.6493,
      "step": 8285
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3643560039880999,
      "learning_rate": 6.639785456548467e-06,
      "loss": 0.6424,
      "step": 8286
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3837832087169672,
      "learning_rate": 6.635475874279977e-06,
      "loss": 0.6384,
      "step": 8287
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3481073938891798,
      "learning_rate": 6.631167412877988e-06,
      "loss": 0.6352,
      "step": 8288
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41164806735638676,
      "learning_rate": 6.62686007270384e-06,
      "loss": 0.6462,
      "step": 8289
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.394184599111039,
      "learning_rate": 6.6225538541187875e-06,
      "loss": 0.6609,
      "step": 8290
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38488046572087886,
      "learning_rate": 6.61824875748399e-06,
      "loss": 0.622,
      "step": 8291
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3646323162478117,
      "learning_rate": 6.613944783160511e-06,
      "loss": 0.6349,
      "step": 8292
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.397637687750647,
      "learning_rate": 6.6096419315093115e-06,
      "loss": 0.6313,
      "step": 8293
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39389453116062195,
      "learning_rate": 6.605340202891273e-06,
      "loss": 0.671,
      "step": 8294
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36791786423600004,
      "learning_rate": 6.60103959766718e-06,
      "loss": 0.6774,
      "step": 8295
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3724987552768366,
      "learning_rate": 6.596740116197702e-06,
      "loss": 0.6216,
      "step": 8296
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41243043900397197,
      "learning_rate": 6.592441758843455e-06,
      "loss": 0.6371,
      "step": 8297
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4216335901719839,
      "learning_rate": 6.5881445259649214e-06,
      "loss": 0.6636,
      "step": 8298
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4013535072875014,
      "learning_rate": 6.583848417922507e-06,
      "loss": 0.6404,
      "step": 8299
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.46684759558744054,
      "learning_rate": 6.579553435076525e-06,
      "loss": 0.7033,
      "step": 8300
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34632900357671914,
      "learning_rate": 6.5752595777871915e-06,
      "loss": 0.6465,
      "step": 8301
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4382709564939071,
      "learning_rate": 6.5709668464146214e-06,
      "loss": 0.6728,
      "step": 8302
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40325893434175486,
      "learning_rate": 6.566675241318843e-06,
      "loss": 0.6663,
      "step": 8303
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4225472149213942,
      "learning_rate": 6.562384762859788e-06,
      "loss": 0.637,
      "step": 8304
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4021483239654833,
      "learning_rate": 6.5580954113972935e-06,
      "loss": 0.6829,
      "step": 8305
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3512028906628125,
      "learning_rate": 6.5538071872911106e-06,
      "loss": 0.6775,
      "step": 8306
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3607592373902661,
      "learning_rate": 6.549520090900874e-06,
      "loss": 0.6297,
      "step": 8307
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4095974534485234,
      "learning_rate": 6.545234122586144e-06,
      "loss": 0.6189,
      "step": 8308
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44873528333812707,
      "learning_rate": 6.540949282706383e-06,
      "loss": 0.6768,
      "step": 8309
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3539621169685123,
      "learning_rate": 6.53666557162094e-06,
      "loss": 0.6447,
      "step": 8310
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43037100816320134,
      "learning_rate": 6.532382989689103e-06,
      "loss": 0.6766,
      "step": 8311
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37763289710008824,
      "learning_rate": 6.528101537270044e-06,
      "loss": 0.6781,
      "step": 8312
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42168181075164707,
      "learning_rate": 6.523821214722834e-06,
      "loss": 0.6838,
      "step": 8313
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4016627078448681,
      "learning_rate": 6.519542022406464e-06,
      "loss": 0.6006,
      "step": 8314
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38861382313848775,
      "learning_rate": 6.515263960679832e-06,
      "loss": 0.6994,
      "step": 8315
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38019987750575684,
      "learning_rate": 6.510987029901716e-06,
      "loss": 0.6258,
      "step": 8316
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3763207826396553,
      "learning_rate": 6.506711230430831e-06,
      "loss": 0.6486,
      "step": 8317
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43563909606597495,
      "learning_rate": 6.502436562625778e-06,
      "loss": 0.6813,
      "step": 8318
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35386755889045896,
      "learning_rate": 6.498163026845068e-06,
      "loss": 0.6661,
      "step": 8319
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3670044684330195,
      "learning_rate": 6.493890623447128e-06,
      "loss": 0.6386,
      "step": 8320
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4618164653546924,
      "learning_rate": 6.489619352790262e-06,
      "loss": 0.6749,
      "step": 8321
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34809538955622626,
      "learning_rate": 6.4853492152327036e-06,
      "loss": 0.6349,
      "step": 8322
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3818236370569644,
      "learning_rate": 6.481080211132587e-06,
      "loss": 0.6289,
      "step": 8323
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3776686214641738,
      "learning_rate": 6.476812340847944e-06,
      "loss": 0.6335,
      "step": 8324
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4394356693956708,
      "learning_rate": 6.47254560473672e-06,
      "loss": 0.6835,
      "step": 8325
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36767527849821374,
      "learning_rate": 6.468280003156766e-06,
      "loss": 0.6247,
      "step": 8326
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36188777812203526,
      "learning_rate": 6.46401553646582e-06,
      "loss": 0.6521,
      "step": 8327
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44264776419620916,
      "learning_rate": 6.4597522050215414e-06,
      "loss": 0.6419,
      "step": 8328
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39739320145414003,
      "learning_rate": 6.4554900091815e-06,
      "loss": 0.68,
      "step": 8329
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3955260348663183,
      "learning_rate": 6.451228949303148e-06,
      "loss": 0.6474,
      "step": 8330
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3993475976003146,
      "learning_rate": 6.446969025743854e-06,
      "loss": 0.6444,
      "step": 8331
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40021136866093243,
      "learning_rate": 6.442710238860912e-06,
      "loss": 0.6319,
      "step": 8332
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3563995354013951,
      "learning_rate": 6.438452589011486e-06,
      "loss": 0.6565,
      "step": 8333
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36908646603405,
      "learning_rate": 6.434196076552661e-06,
      "loss": 0.6221,
      "step": 8334
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40574743091826093,
      "learning_rate": 6.4299407018414326e-06,
      "loss": 0.6606,
      "step": 8335
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4286704431283122,
      "learning_rate": 6.425686465234686e-06,
      "loss": 0.6343,
      "step": 8336
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38373550709046383,
      "learning_rate": 6.42143336708922e-06,
      "loss": 0.6607,
      "step": 8337
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3652076588548033,
      "learning_rate": 6.417181407761739e-06,
      "loss": 0.6509,
      "step": 8338
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42699160174911116,
      "learning_rate": 6.41293058760885e-06,
      "loss": 0.6905,
      "step": 8339
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3896515465925759,
      "learning_rate": 6.408680906987068e-06,
      "loss": 0.6934,
      "step": 8340
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39210055546038936,
      "learning_rate": 6.4044323662527994e-06,
      "loss": 0.6507,
      "step": 8341
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3979535547222069,
      "learning_rate": 6.400184965762368e-06,
      "loss": 0.6585,
      "step": 8342
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37907419554270777,
      "learning_rate": 6.395938705872007e-06,
      "loss": 0.6315,
      "step": 8343
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36403616357382745,
      "learning_rate": 6.39169358693783e-06,
      "loss": 0.6655,
      "step": 8344
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4131691979661021,
      "learning_rate": 6.387449609315872e-06,
      "loss": 0.6244,
      "step": 8345
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42164936682449694,
      "learning_rate": 6.383206773362087e-06,
      "loss": 0.6998,
      "step": 8346
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3730926612920406,
      "learning_rate": 6.378965079432298e-06,
      "loss": 0.6226,
      "step": 8347
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36639535465840134,
      "learning_rate": 6.374724527882259e-06,
      "loss": 0.6618,
      "step": 8348
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3615774348774872,
      "learning_rate": 6.370485119067624e-06,
      "loss": 0.6512,
      "step": 8349
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5517625272549924,
      "learning_rate": 6.366246853343936e-06,
      "loss": 0.34,
      "step": 8350
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3777252510806547,
      "learning_rate": 6.362009731066658e-06,
      "loss": 0.6158,
      "step": 8351
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4347797206889991,
      "learning_rate": 6.357773752591156e-06,
      "loss": 0.6924,
      "step": 8352
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3807189521405202,
      "learning_rate": 6.353538918272692e-06,
      "loss": 0.6284,
      "step": 8353
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3923080463748442,
      "learning_rate": 6.349305228466444e-06,
      "loss": 0.6657,
      "step": 8354
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3702067213185925,
      "learning_rate": 6.345072683527476e-06,
      "loss": 0.6565,
      "step": 8355
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37810279459390367,
      "learning_rate": 6.340841283810768e-06,
      "loss": 0.6542,
      "step": 8356
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42886339207747687,
      "learning_rate": 6.336611029671209e-06,
      "loss": 0.6788,
      "step": 8357
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3946583688164968,
      "learning_rate": 6.33238192146358e-06,
      "loss": 0.6205,
      "step": 8358
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38857411797779934,
      "learning_rate": 6.328153959542573e-06,
      "loss": 0.6544,
      "step": 8359
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38386136546494953,
      "learning_rate": 6.323927144262787e-06,
      "loss": 0.62,
      "step": 8360
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37965377184545396,
      "learning_rate": 6.31970147597871e-06,
      "loss": 0.6416,
      "step": 8361
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3865855642826104,
      "learning_rate": 6.315476955044751e-06,
      "loss": 0.6646,
      "step": 8362
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3885651063233069,
      "learning_rate": 6.311253581815215e-06,
      "loss": 0.6344,
      "step": 8363
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41791963652244274,
      "learning_rate": 6.307031356644304e-06,
      "loss": 0.6627,
      "step": 8364
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3952044759194078,
      "learning_rate": 6.302810279886131e-06,
      "loss": 0.6529,
      "step": 8365
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.380071497786828,
      "learning_rate": 6.2985903518947315e-06,
      "loss": 0.656,
      "step": 8366
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37467886447579396,
      "learning_rate": 6.294371573024003e-06,
      "loss": 0.5996,
      "step": 8367
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41425402521616517,
      "learning_rate": 6.290153943627782e-06,
      "loss": 0.6299,
      "step": 8368
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36528068559547744,
      "learning_rate": 6.285937464059794e-06,
      "loss": 0.634,
      "step": 8369
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38557915764565615,
      "learning_rate": 6.281722134673667e-06,
      "loss": 0.6672,
      "step": 8370
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3464078991027812,
      "learning_rate": 6.277507955822935e-06,
      "loss": 0.6455,
      "step": 8371
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3435829120827626,
      "learning_rate": 6.2732949278610374e-06,
      "loss": 0.6281,
      "step": 8372
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.359525811751563,
      "learning_rate": 6.269083051141318e-06,
      "loss": 0.629,
      "step": 8373
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4119340685778487,
      "learning_rate": 6.264872326017024e-06,
      "loss": 0.6375,
      "step": 8374
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3512242209736964,
      "learning_rate": 6.260662752841298e-06,
      "loss": 0.6745,
      "step": 8375
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40228804143582636,
      "learning_rate": 6.25645433196719e-06,
      "loss": 0.6821,
      "step": 8376
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39014907157696443,
      "learning_rate": 6.252247063747668e-06,
      "loss": 0.6489,
      "step": 8377
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39171095449049204,
      "learning_rate": 6.248040948535574e-06,
      "loss": 0.677,
      "step": 8378
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38733106626113517,
      "learning_rate": 6.243835986683673e-06,
      "loss": 0.6564,
      "step": 8379
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3499813921902184,
      "learning_rate": 6.2396321785446456e-06,
      "loss": 0.617,
      "step": 8380
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35780673082919917,
      "learning_rate": 6.235429524471044e-06,
      "loss": 0.6223,
      "step": 8381
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4107597888008911,
      "learning_rate": 6.2312280248153435e-06,
      "loss": 0.677,
      "step": 8382
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34475980590032795,
      "learning_rate": 6.227027679929927e-06,
      "loss": 0.6577,
      "step": 8383
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37127734976219706,
      "learning_rate": 6.222828490167059e-06,
      "loss": 0.6355,
      "step": 8384
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42257934908988365,
      "learning_rate": 6.218630455878926e-06,
      "loss": 0.6706,
      "step": 8385
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36820543364973496,
      "learning_rate": 6.214433577417614e-06,
      "loss": 0.6476,
      "step": 8386
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3806364192044869,
      "learning_rate": 6.210237855135109e-06,
      "loss": 0.6177,
      "step": 8387
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42073457922686985,
      "learning_rate": 6.206043289383301e-06,
      "loss": 0.6425,
      "step": 8388
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40214306079724943,
      "learning_rate": 6.201849880513991e-06,
      "loss": 0.688,
      "step": 8389
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40455090368316854,
      "learning_rate": 6.197657628878859e-06,
      "loss": 0.6524,
      "step": 8390
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4039106940343939,
      "learning_rate": 6.1934665348295155e-06,
      "loss": 0.6352,
      "step": 8391
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.364491674457229,
      "learning_rate": 6.189276598717464e-06,
      "loss": 0.6677,
      "step": 8392
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4102980735913743,
      "learning_rate": 6.185087820894091e-06,
      "loss": 0.6729,
      "step": 8393
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4031130626148364,
      "learning_rate": 6.180900201710733e-06,
      "loss": 0.6764,
      "step": 8394
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3480589285053677,
      "learning_rate": 6.176713741518581e-06,
      "loss": 0.6185,
      "step": 8395
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3996163414119778,
      "learning_rate": 6.17252844066875e-06,
      "loss": 0.6401,
      "step": 8396
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38063259154317175,
      "learning_rate": 6.1683442995122675e-06,
      "loss": 0.6098,
      "step": 8397
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3590500219529759,
      "learning_rate": 6.1641613184000374e-06,
      "loss": 0.6448,
      "step": 8398
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37926242720668074,
      "learning_rate": 6.159979497682889e-06,
      "loss": 0.6321,
      "step": 8399
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36534228674161806,
      "learning_rate": 6.1557988377115444e-06,
      "loss": 0.623,
      "step": 8400
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3591124439956431,
      "learning_rate": 6.1516193388366315e-06,
      "loss": 0.5998,
      "step": 8401
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40472990007695464,
      "learning_rate": 6.1474410014086805e-06,
      "loss": 0.6659,
      "step": 8402
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4318982529225075,
      "learning_rate": 6.14326382577813e-06,
      "loss": 0.6352,
      "step": 8403
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38477611332832595,
      "learning_rate": 6.139087812295302e-06,
      "loss": 0.6536,
      "step": 8404
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3707456944538274,
      "learning_rate": 6.134912961310438e-06,
      "loss": 0.6064,
      "step": 8405
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37009427907822096,
      "learning_rate": 6.130739273173685e-06,
      "loss": 0.6289,
      "step": 8406
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36033375747441954,
      "learning_rate": 6.126566748235068e-06,
      "loss": 0.6496,
      "step": 8407
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4084482117335584,
      "learning_rate": 6.122395386844555e-06,
      "loss": 0.6534,
      "step": 8408
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4121366016509993,
      "learning_rate": 6.118225189351974e-06,
      "loss": 0.6881,
      "step": 8409
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3781613618932412,
      "learning_rate": 6.114056156107082e-06,
      "loss": 0.698,
      "step": 8410
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40510836100707975,
      "learning_rate": 6.1098882874595375e-06,
      "loss": 0.6419,
      "step": 8411
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3892193817641419,
      "learning_rate": 6.105721583758879e-06,
      "loss": 0.6728,
      "step": 8412
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6192573298691729,
      "learning_rate": 6.101556045354573e-06,
      "loss": 0.3472,
      "step": 8413
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37598083475523675,
      "learning_rate": 6.0973916725959755e-06,
      "loss": 0.6175,
      "step": 8414
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43084518564669977,
      "learning_rate": 6.0932284658323486e-06,
      "loss": 0.6469,
      "step": 8415
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43383032960903317,
      "learning_rate": 6.089066425412855e-06,
      "loss": 0.654,
      "step": 8416
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3973191932376717,
      "learning_rate": 6.084905551686569e-06,
      "loss": 0.6488,
      "step": 8417
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3845949585083834,
      "learning_rate": 6.080745845002438e-06,
      "loss": 0.6482,
      "step": 8418
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34335246375229517,
      "learning_rate": 6.076587305709347e-06,
      "loss": 0.6411,
      "step": 8419
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3946028863019124,
      "learning_rate": 6.0724299341560676e-06,
      "loss": 0.6462,
      "step": 8420
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38846854615447396,
      "learning_rate": 6.068273730691256e-06,
      "loss": 0.5978,
      "step": 8421
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38899892898443234,
      "learning_rate": 6.06411869566351e-06,
      "loss": 0.6512,
      "step": 8422
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38936467876966113,
      "learning_rate": 6.059964829421303e-06,
      "loss": 0.6539,
      "step": 8423
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34736352975988105,
      "learning_rate": 6.055812132313006e-06,
      "loss": 0.6189,
      "step": 8424
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36262346343264407,
      "learning_rate": 6.051660604686905e-06,
      "loss": 0.6311,
      "step": 8425
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4169355482070757,
      "learning_rate": 6.047510246891188e-06,
      "loss": 0.6636,
      "step": 8426
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5972508839980312,
      "learning_rate": 6.043361059273931e-06,
      "loss": 0.3906,
      "step": 8427
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41683504244343095,
      "learning_rate": 6.039213042183127e-06,
      "loss": 0.6801,
      "step": 8428
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39112816355355,
      "learning_rate": 6.035066195966666e-06,
      "loss": 0.6435,
      "step": 8429
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.45106727731723356,
      "learning_rate": 6.030920520972336e-06,
      "loss": 0.7023,
      "step": 8430
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3466375223593925,
      "learning_rate": 6.026776017547837e-06,
      "loss": 0.6589,
      "step": 8431
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4030076364323028,
      "learning_rate": 6.022632686040755e-06,
      "loss": 0.6545,
      "step": 8432
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3541692328534118,
      "learning_rate": 6.018490526798586e-06,
      "loss": 0.635,
      "step": 8433
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4008205954755947,
      "learning_rate": 6.014349540168734e-06,
      "loss": 0.6533,
      "step": 8434
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3579158048170593,
      "learning_rate": 6.010209726498493e-06,
      "loss": 0.645,
      "step": 8435
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39071515057678596,
      "learning_rate": 6.006071086135068e-06,
      "loss": 0.6872,
      "step": 8436
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3962449494941694,
      "learning_rate": 6.0019336194255665e-06,
      "loss": 0.6452,
      "step": 8437
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42142153956318085,
      "learning_rate": 5.997797326716981e-06,
      "loss": 0.6709,
      "step": 8438
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35188349679539876,
      "learning_rate": 5.993662208356223e-06,
      "loss": 0.6167,
      "step": 8439
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3787819109816831,
      "learning_rate": 5.989528264690107e-06,
      "loss": 0.6498,
      "step": 8440
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3725161772664109,
      "learning_rate": 5.985395496065322e-06,
      "loss": 0.616,
      "step": 8441
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36078482362898767,
      "learning_rate": 5.9812639028285045e-06,
      "loss": 0.6216,
      "step": 8442
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4024425814456056,
      "learning_rate": 5.977133485326148e-06,
      "loss": 0.6676,
      "step": 8443
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3576625233474344,
      "learning_rate": 5.973004243904667e-06,
      "loss": 0.6561,
      "step": 8444
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4273111772154096,
      "learning_rate": 5.968876178910383e-06,
      "loss": 0.6863,
      "step": 8445
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43095559596888766,
      "learning_rate": 5.964749290689513e-06,
      "loss": 0.6342,
      "step": 8446
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4281082441501774,
      "learning_rate": 5.960623579588165e-06,
      "loss": 0.6674,
      "step": 8447
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3586633873669397,
      "learning_rate": 5.956499045952362e-06,
      "loss": 0.6502,
      "step": 8448
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3974986645398103,
      "learning_rate": 5.952375690128023e-06,
      "loss": 0.6274,
      "step": 8449
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4314984371047491,
      "learning_rate": 5.948253512460971e-06,
      "loss": 0.6665,
      "step": 8450
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37379330449943343,
      "learning_rate": 5.94413251329693e-06,
      "loss": 0.6275,
      "step": 8451
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42273490741865544,
      "learning_rate": 5.940012692981516e-06,
      "loss": 0.6558,
      "step": 8452
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3489482334607094,
      "learning_rate": 5.935894051860256e-06,
      "loss": 0.6081,
      "step": 8453
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37691726541673504,
      "learning_rate": 5.931776590278584e-06,
      "loss": 0.6345,
      "step": 8454
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4136684522705877,
      "learning_rate": 5.927660308581806e-06,
      "loss": 0.6608,
      "step": 8455
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3706066700995111,
      "learning_rate": 5.923545207115171e-06,
      "loss": 0.6664,
      "step": 8456
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3801689119379968,
      "learning_rate": 5.9194312862238045e-06,
      "loss": 0.6255,
      "step": 8457
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3945901682039432,
      "learning_rate": 5.915318546252724e-06,
      "loss": 0.6404,
      "step": 8458
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37182353380553773,
      "learning_rate": 5.911206987546867e-06,
      "loss": 0.6406,
      "step": 8459
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36884279449774243,
      "learning_rate": 5.907096610451073e-06,
      "loss": 0.6371,
      "step": 8460
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4089634553367813,
      "learning_rate": 5.902987415310057e-06,
      "loss": 0.6364,
      "step": 8461
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3830295795288276,
      "learning_rate": 5.898879402468465e-06,
      "loss": 0.6422,
      "step": 8462
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38007457456021565,
      "learning_rate": 5.894772572270826e-06,
      "loss": 0.653,
      "step": 8463
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3682142940545015,
      "learning_rate": 5.890666925061576e-06,
      "loss": 0.6335,
      "step": 8464
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42768350116777387,
      "learning_rate": 5.8865624611850595e-06,
      "loss": 0.708,
      "step": 8465
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.394508749978846,
      "learning_rate": 5.8824591809854955e-06,
      "loss": 0.656,
      "step": 8466
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4175650380317448,
      "learning_rate": 5.878357084807034e-06,
      "loss": 0.663,
      "step": 8467
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37309213899888605,
      "learning_rate": 5.874256172993713e-06,
      "loss": 0.6114,
      "step": 8468
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37082210766747403,
      "learning_rate": 5.8701564458894545e-06,
      "loss": 0.628,
      "step": 8469
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39783202627800107,
      "learning_rate": 5.866057903838116e-06,
      "loss": 0.629,
      "step": 8470
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3892565606461484,
      "learning_rate": 5.86196054718344e-06,
      "loss": 0.6585,
      "step": 8471
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38490667424508074,
      "learning_rate": 5.857864376269051e-06,
      "loss": 0.6091,
      "step": 8472
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3786145908726773,
      "learning_rate": 5.853769391438497e-06,
      "loss": 0.6591,
      "step": 8473
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32691747728834775,
      "learning_rate": 5.849675593035226e-06,
      "loss": 0.6196,
      "step": 8474
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4298492029855525,
      "learning_rate": 5.845582981402569e-06,
      "loss": 0.6228,
      "step": 8475
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36466576870068196,
      "learning_rate": 5.841491556883767e-06,
      "loss": 0.5833,
      "step": 8476
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3496309455465603,
      "learning_rate": 5.837401319821981e-06,
      "loss": 0.6305,
      "step": 8477
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37874966320429565,
      "learning_rate": 5.8333122705602364e-06,
      "loss": 0.646,
      "step": 8478
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3422379167607844,
      "learning_rate": 5.82922440944148e-06,
      "loss": 0.6173,
      "step": 8479
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38609755641224086,
      "learning_rate": 5.825137736808568e-06,
      "loss": 0.6547,
      "step": 8480
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3724110958615452,
      "learning_rate": 5.821052253004227e-06,
      "loss": 0.6749,
      "step": 8481
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38512817207860806,
      "learning_rate": 5.816967958371109e-06,
      "loss": 0.6687,
      "step": 8482
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40638704358472344,
      "learning_rate": 5.812884853251761e-06,
      "loss": 0.6568,
      "step": 8483
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40946161985506135,
      "learning_rate": 5.8088029379886265e-06,
      "loss": 0.6677,
      "step": 8484
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44366242625572166,
      "learning_rate": 5.804722212924056e-06,
      "loss": 0.6689,
      "step": 8485
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3653606590739518,
      "learning_rate": 5.800642678400286e-06,
      "loss": 0.6029,
      "step": 8486
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37934832454868683,
      "learning_rate": 5.796564334759464e-06,
      "loss": 0.639,
      "step": 8487
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40113195946828545,
      "learning_rate": 5.792487182343645e-06,
      "loss": 0.6479,
      "step": 8488
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3798186119324277,
      "learning_rate": 5.788411221494761e-06,
      "loss": 0.6611,
      "step": 8489
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3562705048747037,
      "learning_rate": 5.7843364525546595e-06,
      "loss": 0.6309,
      "step": 8490
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42044714348255835,
      "learning_rate": 5.780262875865103e-06,
      "loss": 0.6446,
      "step": 8491
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41206916616364725,
      "learning_rate": 5.7761904917677195e-06,
      "loss": 0.6594,
      "step": 8492
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37215976539455786,
      "learning_rate": 5.772119300604062e-06,
      "loss": 0.6462,
      "step": 8493
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37608792786730905,
      "learning_rate": 5.768049302715582e-06,
      "loss": 0.6375,
      "step": 8494
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41242155807811776,
      "learning_rate": 5.763980498443614e-06,
      "loss": 0.6811,
      "step": 8495
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37430388144947035,
      "learning_rate": 5.759912888129406e-06,
      "loss": 0.6336,
      "step": 8496
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3687876843170639,
      "learning_rate": 5.755846472114108e-06,
      "loss": 0.5974,
      "step": 8497
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3941481851841198,
      "learning_rate": 5.751781250738764e-06,
      "loss": 0.6447,
      "step": 8498
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.73546211370727,
      "learning_rate": 5.747717224344327e-06,
      "loss": 0.3494,
      "step": 8499
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3425235387875978,
      "learning_rate": 5.743654393271625e-06,
      "loss": 0.6079,
      "step": 8500
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3775203364219602,
      "learning_rate": 5.739592757861412e-06,
      "loss": 0.6639,
      "step": 8501
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3700062331254615,
      "learning_rate": 5.735532318454333e-06,
      "loss": 0.6469,
      "step": 8502
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43969221365753497,
      "learning_rate": 5.731473075390939e-06,
      "loss": 0.631,
      "step": 8503
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43008911817555895,
      "learning_rate": 5.7274150290116516e-06,
      "loss": 0.6803,
      "step": 8504
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4000795332678496,
      "learning_rate": 5.723358179656842e-06,
      "loss": 0.6693,
      "step": 8505
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39608437837640215,
      "learning_rate": 5.719302527666733e-06,
      "loss": 0.6575,
      "step": 8506
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3755584580695387,
      "learning_rate": 5.715248073381474e-06,
      "loss": 0.6586,
      "step": 8507
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39472822509749506,
      "learning_rate": 5.711194817141115e-06,
      "loss": 0.6378,
      "step": 8508
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3568061274923132,
      "learning_rate": 5.7071427592855845e-06,
      "loss": 0.661,
      "step": 8509
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3977413093549394,
      "learning_rate": 5.703091900154727e-06,
      "loss": 0.6344,
      "step": 8510
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3252122133700395,
      "learning_rate": 5.699042240088289e-06,
      "loss": 0.6145,
      "step": 8511
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3988106474299419,
      "learning_rate": 5.6949937794259035e-06,
      "loss": 0.6524,
      "step": 8512
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38297379640725193,
      "learning_rate": 5.690946518507115e-06,
      "loss": 0.6252,
      "step": 8513
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41919072589470646,
      "learning_rate": 5.686900457671367e-06,
      "loss": 0.6717,
      "step": 8514
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3738088151105167,
      "learning_rate": 5.682855597257986e-06,
      "loss": 0.6552,
      "step": 8515
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4283249856436553,
      "learning_rate": 5.6788119376062166e-06,
      "loss": 0.6343,
      "step": 8516
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3875620868714551,
      "learning_rate": 5.674769479055193e-06,
      "loss": 0.6333,
      "step": 8517
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3999592767117492,
      "learning_rate": 5.670728221943953e-06,
      "loss": 0.6532,
      "step": 8518
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3544701927107461,
      "learning_rate": 5.666688166611438e-06,
      "loss": 0.6666,
      "step": 8519
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39076423782530745,
      "learning_rate": 5.6626493133964684e-06,
      "loss": 0.6759,
      "step": 8520
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38973061112717317,
      "learning_rate": 5.6586116626377875e-06,
      "loss": 0.6047,
      "step": 8521
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5882792261206358,
      "learning_rate": 5.654575214674032e-06,
      "loss": 0.341,
      "step": 8522
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37943770134902005,
      "learning_rate": 5.650539969843723e-06,
      "loss": 0.6041,
      "step": 8523
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3728739508541006,
      "learning_rate": 5.6465059284852884e-06,
      "loss": 0.6348,
      "step": 8524
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41887741466446243,
      "learning_rate": 5.642473090937082e-06,
      "loss": 0.6543,
      "step": 8525
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41726936371920254,
      "learning_rate": 5.638441457537311e-06,
      "loss": 0.6533,
      "step": 8526
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34701127212945454,
      "learning_rate": 5.6344110286241095e-06,
      "loss": 0.6295,
      "step": 8527
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40814149744582856,
      "learning_rate": 5.630381804535514e-06,
      "loss": 0.6118,
      "step": 8528
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3649155165828563,
      "learning_rate": 5.6263537856094355e-06,
      "loss": 0.6422,
      "step": 8529
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3582907697958048,
      "learning_rate": 5.622326972183705e-06,
      "loss": 0.6366,
      "step": 8530
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3825295542712623,
      "learning_rate": 5.618301364596048e-06,
      "loss": 0.6403,
      "step": 8531
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3532207418810939,
      "learning_rate": 5.614276963184087e-06,
      "loss": 0.6111,
      "step": 8532
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3838635134081304,
      "learning_rate": 5.610253768285343e-06,
      "loss": 0.6669,
      "step": 8533
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39442831035927634,
      "learning_rate": 5.606231780237246e-06,
      "loss": 0.6503,
      "step": 8534
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3646202379786289,
      "learning_rate": 5.602210999377096e-06,
      "loss": 0.6426,
      "step": 8535
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40868738467488147,
      "learning_rate": 5.598191426042123e-06,
      "loss": 0.6295,
      "step": 8536
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3613562279424252,
      "learning_rate": 5.59417306056945e-06,
      "loss": 0.6551,
      "step": 8537
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41545397357765584,
      "learning_rate": 5.59015590329607e-06,
      "loss": 0.6371,
      "step": 8538
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3865222960958144,
      "learning_rate": 5.5861399545589245e-06,
      "loss": 0.6558,
      "step": 8539
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38045532727017334,
      "learning_rate": 5.582125214694809e-06,
      "loss": 0.6603,
      "step": 8540
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3669079654447416,
      "learning_rate": 5.578111684040437e-06,
      "loss": 0.6486,
      "step": 8541
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3695949790592402,
      "learning_rate": 5.574099362932428e-06,
      "loss": 0.6413,
      "step": 8542
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39236861719643373,
      "learning_rate": 5.570088251707278e-06,
      "loss": 0.6119,
      "step": 8543
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42213298598589727,
      "learning_rate": 5.566078350701398e-06,
      "loss": 0.6775,
      "step": 8544
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36175265731590217,
      "learning_rate": 5.562069660251097e-06,
      "loss": 0.6209,
      "step": 8545
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.350483146074315,
      "learning_rate": 5.558062180692574e-06,
      "loss": 0.6277,
      "step": 8546
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40268275742162185,
      "learning_rate": 5.554055912361937e-06,
      "loss": 0.6423,
      "step": 8547
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4125723793306073,
      "learning_rate": 5.550050855595189e-06,
      "loss": 0.6316,
      "step": 8548
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4004799093632565,
      "learning_rate": 5.54604701072822e-06,
      "loss": 0.6117,
      "step": 8549
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41859292613297305,
      "learning_rate": 5.542044378096831e-06,
      "loss": 0.6649,
      "step": 8550
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3458215483283052,
      "learning_rate": 5.538042958036724e-06,
      "loss": 0.5958,
      "step": 8551
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36747783165009756,
      "learning_rate": 5.534042750883479e-06,
      "loss": 0.6549,
      "step": 8552
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3887313895587194,
      "learning_rate": 5.530043756972609e-06,
      "loss": 0.6472,
      "step": 8553
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49228863113731036,
      "learning_rate": 5.526045976639487e-06,
      "loss": 0.7208,
      "step": 8554
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40351571376425327,
      "learning_rate": 5.52204941021941e-06,
      "loss": 0.5998,
      "step": 8555
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38825546509676895,
      "learning_rate": 5.518054058047571e-06,
      "loss": 0.6412,
      "step": 8556
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40728312835425273,
      "learning_rate": 5.514059920459041e-06,
      "loss": 0.6689,
      "step": 8557
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4154707633218057,
      "learning_rate": 5.51006699778881e-06,
      "loss": 0.6157,
      "step": 8558
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4095745679245367,
      "learning_rate": 5.5060752903717615e-06,
      "loss": 0.6009,
      "step": 8559
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.34024988868583234,
      "learning_rate": 5.502084798542675e-06,
      "loss": 0.6543,
      "step": 8560
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37857653463741026,
      "learning_rate": 5.498095522636226e-06,
      "loss": 0.626,
      "step": 8561
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33777525625565,
      "learning_rate": 5.494107462987e-06,
      "loss": 0.6141,
      "step": 8562
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3741423614712837,
      "learning_rate": 5.490120619929453e-06,
      "loss": 0.6068,
      "step": 8563
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43356084890237695,
      "learning_rate": 5.486134993797967e-06,
      "loss": 0.6682,
      "step": 8564
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3470078838837241,
      "learning_rate": 5.482150584926818e-06,
      "loss": 0.6431,
      "step": 8565
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4008878592215036,
      "learning_rate": 5.478167393650153e-06,
      "loss": 0.6921,
      "step": 8566
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3927009110360527,
      "learning_rate": 5.474185420302058e-06,
      "loss": 0.6834,
      "step": 8567
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41881353070686167,
      "learning_rate": 5.470204665216494e-06,
      "loss": 0.6563,
      "step": 8568
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39017731530010247,
      "learning_rate": 5.466225128727314e-06,
      "loss": 0.6605,
      "step": 8569
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3929493042697164,
      "learning_rate": 5.462246811168277e-06,
      "loss": 0.6645,
      "step": 8570
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3278692724576956,
      "learning_rate": 5.45826971287305e-06,
      "loss": 0.6525,
      "step": 8571
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37972120173680746,
      "learning_rate": 5.454293834175173e-06,
      "loss": 0.6327,
      "step": 8572
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.368852991986074,
      "learning_rate": 5.450319175408107e-06,
      "loss": 0.6216,
      "step": 8573
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37324670857980824,
      "learning_rate": 5.4463457369051984e-06,
      "loss": 0.6209,
      "step": 8574
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39013302088703716,
      "learning_rate": 5.4423735189996976e-06,
      "loss": 0.6873,
      "step": 8575
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.35862896065091954,
      "learning_rate": 5.438402522024755e-06,
      "loss": 0.5835,
      "step": 8576
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3509945926789561,
      "learning_rate": 5.4344327463134005e-06,
      "loss": 0.632,
      "step": 8577
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3817975935601193,
      "learning_rate": 5.4304641921985814e-06,
      "loss": 0.6533,
      "step": 8578
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.35303244611513024,
      "learning_rate": 5.42649686001314e-06,
      "loss": 0.6472,
      "step": 8579
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3890476387699357,
      "learning_rate": 5.422530750089794e-06,
      "loss": 0.6435,
      "step": 8580
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3908287934483434,
      "learning_rate": 5.418565862761196e-06,
      "loss": 0.6665,
      "step": 8581
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40035746788847104,
      "learning_rate": 5.414602198359875e-06,
      "loss": 0.6581,
      "step": 8582
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36866198262364225,
      "learning_rate": 5.410639757218248e-06,
      "loss": 0.6466,
      "step": 8583
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3630370381761335,
      "learning_rate": 5.406678539668644e-06,
      "loss": 0.6676,
      "step": 8584
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4074757009459034,
      "learning_rate": 5.402718546043293e-06,
      "loss": 0.6384,
      "step": 8585
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39569682208412343,
      "learning_rate": 5.3987597766743006e-06,
      "loss": 0.6665,
      "step": 8586
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44264153233263764,
      "learning_rate": 5.394802231893694e-06,
      "loss": 0.6871,
      "step": 8587
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39279062175796237,
      "learning_rate": 5.390845912033385e-06,
      "loss": 0.6686,
      "step": 8588
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36637306269054365,
      "learning_rate": 5.386890817425185e-06,
      "loss": 0.6519,
      "step": 8589
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4076680740352923,
      "learning_rate": 5.382936948400805e-06,
      "loss": 0.7006,
      "step": 8590
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3940534893424813,
      "learning_rate": 5.378984305291857e-06,
      "loss": 0.6694,
      "step": 8591
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4571258177435266,
      "learning_rate": 5.3750328884298295e-06,
      "loss": 0.6552,
      "step": 8592
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.376971338931741,
      "learning_rate": 5.3710826981461325e-06,
      "loss": 0.649,
      "step": 8593
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.343037827365174,
      "learning_rate": 5.367133734772063e-06,
      "loss": 0.6557,
      "step": 8594
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3875220841934707,
      "learning_rate": 5.363185998638814e-06,
      "loss": 0.644,
      "step": 8595
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41969657430968194,
      "learning_rate": 5.359239490077486e-06,
      "loss": 0.6232,
      "step": 8596
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3696412500243122,
      "learning_rate": 5.355294209419053e-06,
      "loss": 0.6403,
      "step": 8597
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3851491616027793,
      "learning_rate": 5.351350156994408e-06,
      "loss": 0.6616,
      "step": 8598
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3766472658431525,
      "learning_rate": 5.34740733313434e-06,
      "loss": 0.6229,
      "step": 8599
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3854591529056931,
      "learning_rate": 5.343465738169513e-06,
      "loss": 0.6195,
      "step": 8600
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39169477753381743,
      "learning_rate": 5.3395253724305185e-06,
      "loss": 0.6551,
      "step": 8601
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5654827840445618,
      "learning_rate": 5.3355862362478315e-06,
      "loss": 0.3599,
      "step": 8602
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5783583200095094,
      "learning_rate": 5.331648329951813e-06,
      "loss": 0.3902,
      "step": 8603
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4107667653438264,
      "learning_rate": 5.3277116538727316e-06,
      "loss": 0.649,
      "step": 8604
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3836944901163355,
      "learning_rate": 5.32377620834076e-06,
      "loss": 0.6421,
      "step": 8605
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8119154672380774,
      "learning_rate": 5.319841993685951e-06,
      "loss": 0.4236,
      "step": 8606
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42642127417382136,
      "learning_rate": 5.315909010238261e-06,
      "loss": 0.6572,
      "step": 8607
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.35399189991157803,
      "learning_rate": 5.31197725832755e-06,
      "loss": 0.6448,
      "step": 8608
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4310365406959397,
      "learning_rate": 5.308046738283568e-06,
      "loss": 0.6306,
      "step": 8609
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39199785167628814,
      "learning_rate": 5.304117450435966e-06,
      "loss": 0.6664,
      "step": 8610
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3939354831732652,
      "learning_rate": 5.300189395114281e-06,
      "loss": 0.6259,
      "step": 8611
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39977312747437,
      "learning_rate": 5.296262572647961e-06,
      "loss": 0.6463,
      "step": 8612
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36894281984999017,
      "learning_rate": 5.2923369833663444e-06,
      "loss": 0.6297,
      "step": 8613
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38204037993313444,
      "learning_rate": 5.288412627598652e-06,
      "loss": 0.6308,
      "step": 8614
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3801334810657433,
      "learning_rate": 5.284489505674031e-06,
      "loss": 0.6374,
      "step": 8615
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3471800140683173,
      "learning_rate": 5.280567617921508e-06,
      "loss": 0.6487,
      "step": 8616
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41418611434782726,
      "learning_rate": 5.276646964669998e-06,
      "loss": 0.6851,
      "step": 8617
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41088330708660414,
      "learning_rate": 5.2727275462483265e-06,
      "loss": 0.6655,
      "step": 8618
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37606064901499026,
      "learning_rate": 5.268809362985215e-06,
      "loss": 0.645,
      "step": 8619
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4055106532835569,
      "learning_rate": 5.264892415209264e-06,
      "loss": 0.6356,
      "step": 8620
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3397623488226809,
      "learning_rate": 5.260976703248992e-06,
      "loss": 0.6354,
      "step": 8621
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4196248019254704,
      "learning_rate": 5.257062227432803e-06,
      "loss": 0.6405,
      "step": 8622
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37014109631089454,
      "learning_rate": 5.253148988089001e-06,
      "loss": 0.6406,
      "step": 8623
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41807009530272565,
      "learning_rate": 5.249236985545783e-06,
      "loss": 0.6452,
      "step": 8624
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3427001953998964,
      "learning_rate": 5.2453262201312505e-06,
      "loss": 0.6551,
      "step": 8625
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3616319874342514,
      "learning_rate": 5.241416692173382e-06,
      "loss": 0.6005,
      "step": 8626
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.388838392450288,
      "learning_rate": 5.237508402000073e-06,
      "loss": 0.6483,
      "step": 8627
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3646197024526855,
      "learning_rate": 5.233601349939106e-06,
      "loss": 0.6239,
      "step": 8628
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3902352860140508,
      "learning_rate": 5.22969553631816e-06,
      "loss": 0.6563,
      "step": 8629
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42045045192454783,
      "learning_rate": 5.225790961464816e-06,
      "loss": 0.6608,
      "step": 8630
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40123940934114044,
      "learning_rate": 5.221887625706537e-06,
      "loss": 0.6699,
      "step": 8631
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.425933648401005,
      "learning_rate": 5.217985529370695e-06,
      "loss": 0.6838,
      "step": 8632
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3813965079012824,
      "learning_rate": 5.214084672784561e-06,
      "loss": 0.6303,
      "step": 8633
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4167414985311388,
      "learning_rate": 5.210185056275281e-06,
      "loss": 0.6762,
      "step": 8634
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3413748678931055,
      "learning_rate": 5.206286680169914e-06,
      "loss": 0.6485,
      "step": 8635
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3894821414991649,
      "learning_rate": 5.20238954479543e-06,
      "loss": 0.6648,
      "step": 8636
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7185515805658838,
      "learning_rate": 5.198493650478656e-06,
      "loss": 0.3982,
      "step": 8637
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39498444185242393,
      "learning_rate": 5.194598997546343e-06,
      "loss": 0.6646,
      "step": 8638
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.381308825959162,
      "learning_rate": 5.19070558632514e-06,
      "loss": 0.6246,
      "step": 8639
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39143576376069983,
      "learning_rate": 5.186813417141567e-06,
      "loss": 0.6345,
      "step": 8640
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41809974850277004,
      "learning_rate": 5.182922490322064e-06,
      "loss": 0.6424,
      "step": 8641
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3924112130204162,
      "learning_rate": 5.179032806192956e-06,
      "loss": 0.6226,
      "step": 8642
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3486268488273395,
      "learning_rate": 5.175144365080469e-06,
      "loss": 0.6233,
      "step": 8643
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3825314315341604,
      "learning_rate": 5.171257167310724e-06,
      "loss": 0.6339,
      "step": 8644
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40738588764324996,
      "learning_rate": 5.1673712132097266e-06,
      "loss": 0.6757,
      "step": 8645
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36047286138309204,
      "learning_rate": 5.163486503103394e-06,
      "loss": 0.6237,
      "step": 8646
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37331665187354784,
      "learning_rate": 5.1596030373175285e-06,
      "loss": 0.6398,
      "step": 8647
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3873076011780672,
      "learning_rate": 5.15572081617784e-06,
      "loss": 0.6389,
      "step": 8648
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6133491236242273,
      "learning_rate": 5.151839840009907e-06,
      "loss": 0.3638,
      "step": 8649
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3857104964223678,
      "learning_rate": 5.147960109139248e-06,
      "loss": 0.6597,
      "step": 8650
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3714623337099604,
      "learning_rate": 5.144081623891231e-06,
      "loss": 0.6371,
      "step": 8651
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3881596983415997,
      "learning_rate": 5.140204384591145e-06,
      "loss": 0.6528,
      "step": 8652
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3837598567870854,
      "learning_rate": 5.13632839156418e-06,
      "loss": 0.6353,
      "step": 8653
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3853640796022406,
      "learning_rate": 5.1324536451353955e-06,
      "loss": 0.6234,
      "step": 8654
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39524035824651355,
      "learning_rate": 5.128580145629767e-06,
      "loss": 0.654,
      "step": 8655
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4152226740402199,
      "learning_rate": 5.1247078933721625e-06,
      "loss": 0.6491,
      "step": 8656
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3972295934935937,
      "learning_rate": 5.120836888687342e-06,
      "loss": 0.6518,
      "step": 8657
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36967237947167997,
      "learning_rate": 5.116967131899961e-06,
      "loss": 0.6233,
      "step": 8658
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3821920490619642,
      "learning_rate": 5.1130986233345805e-06,
      "loss": 0.6565,
      "step": 8659
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3639799066086004,
      "learning_rate": 5.109231363315632e-06,
      "loss": 0.6684,
      "step": 8660
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41128031104816043,
      "learning_rate": 5.105365352167466e-06,
      "loss": 0.6401,
      "step": 8661
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4058276861733093,
      "learning_rate": 5.101500590214328e-06,
      "loss": 0.6627,
      "step": 8662
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38051648403512517,
      "learning_rate": 5.097637077780329e-06,
      "loss": 0.6292,
      "step": 8663
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3535521914658683,
      "learning_rate": 5.09377481518952e-06,
      "loss": 0.6179,
      "step": 8664
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4004780144860936,
      "learning_rate": 5.0899138027658136e-06,
      "loss": 0.6583,
      "step": 8665
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41709618025303763,
      "learning_rate": 5.0860540408330286e-06,
      "loss": 0.6562,
      "step": 8666
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40585559643875135,
      "learning_rate": 5.0821955297148835e-06,
      "loss": 0.6793,
      "step": 8667
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5461854038522741,
      "learning_rate": 5.078338269734979e-06,
      "loss": 0.3598,
      "step": 8668
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3438371929705231,
      "learning_rate": 5.074482261216822e-06,
      "loss": 0.621,
      "step": 8669
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37392020571261014,
      "learning_rate": 5.070627504483813e-06,
      "loss": 0.6392,
      "step": 8670
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34430104229054803,
      "learning_rate": 5.066773999859244e-06,
      "loss": 0.6421,
      "step": 8671
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4007431736076419,
      "learning_rate": 5.062921747666307e-06,
      "loss": 0.6404,
      "step": 8672
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4018353258581191,
      "learning_rate": 5.059070748228088e-06,
      "loss": 0.6815,
      "step": 8673
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42953749410840425,
      "learning_rate": 5.055221001867556e-06,
      "loss": 0.6433,
      "step": 8674
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4348277749791971,
      "learning_rate": 5.05137250890759e-06,
      "loss": 0.6549,
      "step": 8675
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42656097318061564,
      "learning_rate": 5.047525269670963e-06,
      "loss": 0.6097,
      "step": 8676
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.378560040470514,
      "learning_rate": 5.043679284480323e-06,
      "loss": 0.6319,
      "step": 8677
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4358850670111729,
      "learning_rate": 5.0398345536582495e-06,
      "loss": 0.6972,
      "step": 8678
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35364637226600537,
      "learning_rate": 5.03599107752718e-06,
      "loss": 0.6412,
      "step": 8679
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43998034050902135,
      "learning_rate": 5.032148856409467e-06,
      "loss": 0.6843,
      "step": 8680
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39202755794929106,
      "learning_rate": 5.028307890627351e-06,
      "loss": 0.6162,
      "step": 8681
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39184971897646514,
      "learning_rate": 5.024468180502979e-06,
      "loss": 0.6198,
      "step": 8682
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37080347656981766,
      "learning_rate": 5.020629726358368e-06,
      "loss": 0.6092,
      "step": 8683
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43149261855192567,
      "learning_rate": 5.016792528515451e-06,
      "loss": 0.627,
      "step": 8684
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3882419157691864,
      "learning_rate": 5.012956587296049e-06,
      "loss": 0.662,
      "step": 8685
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3955896235331086,
      "learning_rate": 5.009121903021881e-06,
      "loss": 0.6433,
      "step": 8686
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40635410128764976,
      "learning_rate": 5.005288476014558e-06,
      "loss": 0.642,
      "step": 8687
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33651870759511937,
      "learning_rate": 5.001456306595576e-06,
      "loss": 0.6114,
      "step": 8688
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3737289524355892,
      "learning_rate": 4.997625395086338e-06,
      "loss": 0.6638,
      "step": 8689
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4218825089324162,
      "learning_rate": 4.993795741808143e-06,
      "loss": 0.6466,
      "step": 8690
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36248484538154646,
      "learning_rate": 4.9899673470821765e-06,
      "loss": 0.6149,
      "step": 8691
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38452600421091276,
      "learning_rate": 4.986140211229518e-06,
      "loss": 0.6204,
      "step": 8692
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38872517556389347,
      "learning_rate": 4.982314334571156e-06,
      "loss": 0.6103,
      "step": 8693
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3788721878534021,
      "learning_rate": 4.978489717427948e-06,
      "loss": 0.6041,
      "step": 8694
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33414399674145334,
      "learning_rate": 4.974666360120666e-06,
      "loss": 0.6075,
      "step": 8695
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3788219036193552,
      "learning_rate": 4.970844262969974e-06,
      "loss": 0.6389,
      "step": 8696
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3980556171940249,
      "learning_rate": 4.967023426296413e-06,
      "loss": 0.6701,
      "step": 8697
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31218995887740597,
      "learning_rate": 4.963203850420455e-06,
      "loss": 0.6437,
      "step": 8698
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.382543368395524,
      "learning_rate": 4.959385535662422e-06,
      "loss": 0.6439,
      "step": 8699
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4227955313821126,
      "learning_rate": 4.955568482342561e-06,
      "loss": 0.6562,
      "step": 8700
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3853273516408991,
      "learning_rate": 4.951752690781009e-06,
      "loss": 0.6164,
      "step": 8701
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37005921882543635,
      "learning_rate": 4.947938161297778e-06,
      "loss": 0.6607,
      "step": 8702
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3827093225499516,
      "learning_rate": 4.944124894212796e-06,
      "loss": 0.6392,
      "step": 8703
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39208924032921216,
      "learning_rate": 4.9403128898458755e-06,
      "loss": 0.611,
      "step": 8704
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3485852756950204,
      "learning_rate": 4.936502148516724e-06,
      "loss": 0.6359,
      "step": 8705
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36632263521867253,
      "learning_rate": 4.932692670544947e-06,
      "loss": 0.6376,
      "step": 8706
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3760460163814622,
      "learning_rate": 4.928884456250045e-06,
      "loss": 0.6129,
      "step": 8707
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39915181624242835,
      "learning_rate": 4.925077505951396e-06,
      "loss": 0.6532,
      "step": 8708
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37710627470204905,
      "learning_rate": 4.921271819968292e-06,
      "loss": 0.6482,
      "step": 8709
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.392475586767602,
      "learning_rate": 4.917467398619915e-06,
      "loss": 0.6854,
      "step": 8710
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39732601330872813,
      "learning_rate": 4.91366424222532e-06,
      "loss": 0.6572,
      "step": 8711
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4136706905628372,
      "learning_rate": 4.909862351103492e-06,
      "loss": 0.6603,
      "step": 8712
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35067904897855623,
      "learning_rate": 4.9060617255732875e-06,
      "loss": 0.6518,
      "step": 8713
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.347209551720485,
      "learning_rate": 4.902262365953454e-06,
      "loss": 0.6321,
      "step": 8714
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39934358672429926,
      "learning_rate": 4.898464272562642e-06,
      "loss": 0.6355,
      "step": 8715
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3643551231110958,
      "learning_rate": 4.894667445719399e-06,
      "loss": 0.6521,
      "step": 8716
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4140943616600957,
      "learning_rate": 4.890871885742148e-06,
      "loss": 0.6409,
      "step": 8717
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43471251687891177,
      "learning_rate": 4.887077592949225e-06,
      "loss": 0.6941,
      "step": 8718
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42843498768219923,
      "learning_rate": 4.883284567658854e-06,
      "loss": 0.6932,
      "step": 8719
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36797683491929045,
      "learning_rate": 4.8794928101891505e-06,
      "loss": 0.6334,
      "step": 8720
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4063065824685716,
      "learning_rate": 4.875702320858127e-06,
      "loss": 0.6564,
      "step": 8721
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4274316675360034,
      "learning_rate": 4.871913099983681e-06,
      "loss": 0.6931,
      "step": 8722
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.47315473888452436,
      "learning_rate": 4.868125147883611e-06,
      "loss": 0.6445,
      "step": 8723
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43342704865681253,
      "learning_rate": 4.864338464875618e-06,
      "loss": 0.6455,
      "step": 8724
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3657849181107877,
      "learning_rate": 4.860553051277266e-06,
      "loss": 0.6446,
      "step": 8725
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3809495365449845,
      "learning_rate": 4.856768907406051e-06,
      "loss": 0.6742,
      "step": 8726
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3853359556515267,
      "learning_rate": 4.8529860335793456e-06,
      "loss": 0.6373,
      "step": 8727
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.45557805212950636,
      "learning_rate": 4.849204430114405e-06,
      "loss": 0.647,
      "step": 8728
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3235689512281716,
      "learning_rate": 4.845424097328389e-06,
      "loss": 0.6356,
      "step": 8729
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39702836510875855,
      "learning_rate": 4.841645035538358e-06,
      "loss": 0.6388,
      "step": 8730
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38801863099757194,
      "learning_rate": 4.837867245061245e-06,
      "loss": 0.6657,
      "step": 8731
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40872323789244225,
      "learning_rate": 4.834090726213896e-06,
      "loss": 0.6654,
      "step": 8732
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37498329381602863,
      "learning_rate": 4.8303154793130414e-06,
      "loss": 0.6322,
      "step": 8733
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34370949570814907,
      "learning_rate": 4.826541504675306e-06,
      "loss": 0.6369,
      "step": 8734
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3921595267361571,
      "learning_rate": 4.822768802617217e-06,
      "loss": 0.6222,
      "step": 8735
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35802848472781124,
      "learning_rate": 4.818997373455174e-06,
      "loss": 0.6399,
      "step": 8736
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4011404391799137,
      "learning_rate": 4.815227217505487e-06,
      "loss": 0.6223,
      "step": 8737
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35097854488411395,
      "learning_rate": 4.811458335084353e-06,
      "loss": 0.619,
      "step": 8738
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3841919687479473,
      "learning_rate": 4.807690726507868e-06,
      "loss": 0.669,
      "step": 8739
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3717056771691724,
      "learning_rate": 4.803924392092012e-06,
      "loss": 0.6531,
      "step": 8740
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.382489866431421,
      "learning_rate": 4.80015933215267e-06,
      "loss": 0.6785,
      "step": 8741
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37929426916678277,
      "learning_rate": 4.7963955470056035e-06,
      "loss": 0.6395,
      "step": 8742
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3857793960489497,
      "learning_rate": 4.792633036966481e-06,
      "loss": 0.6414,
      "step": 8743
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38334724754223276,
      "learning_rate": 4.788871802350867e-06,
      "loss": 0.6844,
      "step": 8744
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3840201437318604,
      "learning_rate": 4.785111843474197e-06,
      "loss": 0.6492,
      "step": 8745
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.41381718424075936,
      "learning_rate": 4.7813531606518156e-06,
      "loss": 0.6358,
      "step": 8746
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3654192121907108,
      "learning_rate": 4.7775957541989785e-06,
      "loss": 0.6601,
      "step": 8747
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3709304900035362,
      "learning_rate": 4.773839624430796e-06,
      "loss": 0.6471,
      "step": 8748
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40750036001701706,
      "learning_rate": 4.770084771662294e-06,
      "loss": 0.6551,
      "step": 8749
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34364184065100584,
      "learning_rate": 4.766331196208396e-06,
      "loss": 0.6284,
      "step": 8750
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.412724495729178,
      "learning_rate": 4.762578898383896e-06,
      "loss": 0.6513,
      "step": 8751
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3758767356275442,
      "learning_rate": 4.758827878503502e-06,
      "loss": 0.6688,
      "step": 8752
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39156561697989356,
      "learning_rate": 4.755078136881808e-06,
      "loss": 0.656,
      "step": 8753
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36673089687008337,
      "learning_rate": 4.751329673833298e-06,
      "loss": 0.5796,
      "step": 8754
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39701262363039147,
      "learning_rate": 4.74758248967236e-06,
      "loss": 0.692,
      "step": 8755
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38434557950552756,
      "learning_rate": 4.7438365847132525e-06,
      "loss": 0.6747,
      "step": 8756
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43726230694874313,
      "learning_rate": 4.740091959270146e-06,
      "loss": 0.6394,
      "step": 8757
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36026367186586367,
      "learning_rate": 4.7363486136571e-06,
      "loss": 0.6236,
      "step": 8758
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36047134188280955,
      "learning_rate": 4.732606548188059e-06,
      "loss": 0.5983,
      "step": 8759
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4411311712434885,
      "learning_rate": 4.728865763176862e-06,
      "loss": 0.6695,
      "step": 8760
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4026132278683794,
      "learning_rate": 4.725126258937262e-06,
      "loss": 0.6023,
      "step": 8761
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40719005570672073,
      "learning_rate": 4.721388035782872e-06,
      "loss": 0.6952,
      "step": 8762
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38133719241188185,
      "learning_rate": 4.717651094027214e-06,
      "loss": 0.6643,
      "step": 8763
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3512108780606712,
      "learning_rate": 4.713915433983707e-06,
      "loss": 0.6335,
      "step": 8764
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42015730747292596,
      "learning_rate": 4.710181055965646e-06,
      "loss": 0.6981,
      "step": 8765
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3707641352786461,
      "learning_rate": 4.706447960286236e-06,
      "loss": 0.6527,
      "step": 8766
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4128085049068147,
      "learning_rate": 4.702716147258564e-06,
      "loss": 0.6465,
      "step": 8767
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40222793617728164,
      "learning_rate": 4.698985617195615e-06,
      "loss": 0.6581,
      "step": 8768
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39224493283423933,
      "learning_rate": 4.695256370410262e-06,
      "loss": 0.6029,
      "step": 8769
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3835094867394265,
      "learning_rate": 4.691528407215282e-06,
      "loss": 0.6361,
      "step": 8770
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3488853485207181,
      "learning_rate": 4.687801727923318e-06,
      "loss": 0.6115,
      "step": 8771
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3515772838315621,
      "learning_rate": 4.684076332846932e-06,
      "loss": 0.6431,
      "step": 8772
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39567454884223796,
      "learning_rate": 4.680352222298568e-06,
      "loss": 0.6161,
      "step": 8773
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40266930091862574,
      "learning_rate": 4.676629396590562e-06,
      "loss": 0.6388,
      "step": 8774
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3702938999925647,
      "learning_rate": 4.672907856035147e-06,
      "loss": 0.6343,
      "step": 8775
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3262128776124832,
      "learning_rate": 4.669187600944433e-06,
      "loss": 0.6102,
      "step": 8776
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38323934317891556,
      "learning_rate": 4.6654686316304406e-06,
      "loss": 0.6191,
      "step": 8777
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3829258598755127,
      "learning_rate": 4.661750948405082e-06,
      "loss": 0.6782,
      "step": 8778
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4010033698739675,
      "learning_rate": 4.658034551580139e-06,
      "loss": 0.6808,
      "step": 8779
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4002037726621656,
      "learning_rate": 4.654319441467306e-06,
      "loss": 0.6348,
      "step": 8780
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3787429352098384,
      "learning_rate": 4.650605618378179e-06,
      "loss": 0.6291,
      "step": 8781
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43645964541432736,
      "learning_rate": 4.646893082624217e-06,
      "loss": 0.712,
      "step": 8782
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35263589058520783,
      "learning_rate": 4.64318183451679e-06,
      "loss": 0.6128,
      "step": 8783
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42962600925013555,
      "learning_rate": 4.6394718743671605e-06,
      "loss": 0.6441,
      "step": 8784
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3542875235257664,
      "learning_rate": 4.6357632024864695e-06,
      "loss": 0.6646,
      "step": 8785
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3543231377780949,
      "learning_rate": 4.632055819185764e-06,
      "loss": 0.6144,
      "step": 8786
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38893824416382605,
      "learning_rate": 4.628349724775976e-06,
      "loss": 0.5972,
      "step": 8787
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.37228131799247033,
      "learning_rate": 4.624644919567934e-06,
      "loss": 0.6859,
      "step": 8788
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42638358973139107,
      "learning_rate": 4.620941403872356e-06,
      "loss": 0.7028,
      "step": 8789
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3948442356872679,
      "learning_rate": 4.617239177999846e-06,
      "loss": 0.6634,
      "step": 8790
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.46093099952152966,
      "learning_rate": 4.6135382422609086e-06,
      "loss": 0.6545,
      "step": 8791
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.37108176481785937,
      "learning_rate": 4.609838596965943e-06,
      "loss": 0.6303,
      "step": 8792
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3690536089763618,
      "learning_rate": 4.6061402424252215e-06,
      "loss": 0.6306,
      "step": 8793
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4143063663764558,
      "learning_rate": 4.602443178948919e-06,
      "loss": 0.6534,
      "step": 8794
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4057413437463955,
      "learning_rate": 4.598747406847126e-06,
      "loss": 0.6772,
      "step": 8795
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3645278121489344,
      "learning_rate": 4.595052926429779e-06,
      "loss": 0.6265,
      "step": 8796
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36320086327345447,
      "learning_rate": 4.5913597380067424e-06,
      "loss": 0.5916,
      "step": 8797
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39638060862521857,
      "learning_rate": 4.58766784188776e-06,
      "loss": 0.6165,
      "step": 8798
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3498324612384455,
      "learning_rate": 4.583977238382455e-06,
      "loss": 0.6084,
      "step": 8799
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4275230796220715,
      "learning_rate": 4.580287927800362e-06,
      "loss": 0.6873,
      "step": 8800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3790939135990532,
      "learning_rate": 4.576599910450899e-06,
      "loss": 0.6276,
      "step": 8801
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40000567131689246,
      "learning_rate": 4.572913186643373e-06,
      "loss": 0.7029,
      "step": 8802
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3764652128321364,
      "learning_rate": 4.569227756686989e-06,
      "loss": 0.6438,
      "step": 8803
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42746693661104995,
      "learning_rate": 4.565543620890842e-06,
      "loss": 0.6619,
      "step": 8804
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38902212086572596,
      "learning_rate": 4.561860779563905e-06,
      "loss": 0.664,
      "step": 8805
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3867312901604982,
      "learning_rate": 4.558179233015061e-06,
      "loss": 0.6461,
      "step": 8806
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3804737707342548,
      "learning_rate": 4.554498981553082e-06,
      "loss": 0.615,
      "step": 8807
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4277463682860905,
      "learning_rate": 4.550820025486607e-06,
      "loss": 0.6382,
      "step": 8808
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3888397116618681,
      "learning_rate": 4.547142365124211e-06,
      "loss": 0.6549,
      "step": 8809
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3894190646796043,
      "learning_rate": 4.5434660007743144e-06,
      "loss": 0.6236,
      "step": 8810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3948961908001359,
      "learning_rate": 4.53979093274526e-06,
      "loss": 0.6093,
      "step": 8811
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3557726890355828,
      "learning_rate": 4.536117161345275e-06,
      "loss": 0.6228,
      "step": 8812
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.37799491871490337,
      "learning_rate": 4.532444686882462e-06,
      "loss": 0.6266,
      "step": 8813
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3660670740397674,
      "learning_rate": 4.528773509664832e-06,
      "loss": 0.6396,
      "step": 8814
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38457859437749015,
      "learning_rate": 4.5251036300002845e-06,
      "loss": 0.6459,
      "step": 8815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.41635604162210393,
      "learning_rate": 4.521435048196605e-06,
      "loss": 0.6336,
      "step": 8816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39705018256125596,
      "learning_rate": 4.517767764561476e-06,
      "loss": 0.6492,
      "step": 8817
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33902036828582427,
      "learning_rate": 4.514101779402471e-06,
      "loss": 0.6148,
      "step": 8818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4038940179115237,
      "learning_rate": 4.510437093027045e-06,
      "loss": 0.6605,
      "step": 8819
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40069346969604386,
      "learning_rate": 4.506773705742551e-06,
      "loss": 0.6395,
      "step": 8820
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3589295152398156,
      "learning_rate": 4.50311161785624e-06,
      "loss": 0.6528,
      "step": 8821
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38074636400357303,
      "learning_rate": 4.499450829675232e-06,
      "loss": 0.6667,
      "step": 8822
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3888377559111235,
      "learning_rate": 4.495791341506575e-06,
      "loss": 0.6503,
      "step": 8823
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39500505938767355,
      "learning_rate": 4.4921331536571655e-06,
      "loss": 0.6464,
      "step": 8824
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39852162016372944,
      "learning_rate": 4.4884762664338236e-06,
      "loss": 0.6686,
      "step": 8825
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.384030225861109,
      "learning_rate": 4.48482068014324e-06,
      "loss": 0.6558,
      "step": 8826
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36937135884081146,
      "learning_rate": 4.481166395092016e-06,
      "loss": 0.6239,
      "step": 8827
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4284269953317054,
      "learning_rate": 4.477513411586621e-06,
      "loss": 0.6439,
      "step": 8828
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3947743566983964,
      "learning_rate": 4.473861729933426e-06,
      "loss": 0.6461,
      "step": 8829
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42105612873979226,
      "learning_rate": 4.470211350438698e-06,
      "loss": 0.6745,
      "step": 8830
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4405511166341144,
      "learning_rate": 4.4665622734085905e-06,
      "loss": 0.6155,
      "step": 8831
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3896793940652241,
      "learning_rate": 4.462914499149149e-06,
      "loss": 0.6415,
      "step": 8832
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4231349822005197,
      "learning_rate": 4.459268027966299e-06,
      "loss": 0.6688,
      "step": 8833
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40709447076942645,
      "learning_rate": 4.455622860165872e-06,
      "loss": 0.6492,
      "step": 8834
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4136883541340995,
      "learning_rate": 4.451978996053588e-06,
      "loss": 0.6437,
      "step": 8835
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40355908758243075,
      "learning_rate": 4.448336435935037e-06,
      "loss": 0.6261,
      "step": 8836
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3927005257070015,
      "learning_rate": 4.444695180115734e-06,
      "loss": 0.6543,
      "step": 8837
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39280227368997894,
      "learning_rate": 4.4410552289010655e-06,
      "loss": 0.6177,
      "step": 8838
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34586107934796034,
      "learning_rate": 4.437416582596299e-06,
      "loss": 0.6242,
      "step": 8839
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36049716322563086,
      "learning_rate": 4.433779241506608e-06,
      "loss": 0.646,
      "step": 8840
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3772841569396355,
      "learning_rate": 4.430143205937058e-06,
      "loss": 0.6356,
      "step": 8841
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3520266851185404,
      "learning_rate": 4.4265084761925904e-06,
      "loss": 0.6093,
      "step": 8842
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35142960780975946,
      "learning_rate": 4.422875052578046e-06,
      "loss": 0.6241,
      "step": 8843
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5952174253351535,
      "learning_rate": 4.41924293539816e-06,
      "loss": 0.3805,
      "step": 8844
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3868476725284488,
      "learning_rate": 4.415612124957551e-06,
      "loss": 0.6471,
      "step": 8845
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3747212356030893,
      "learning_rate": 4.41198262156074e-06,
      "loss": 0.6217,
      "step": 8846
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34586411690345603,
      "learning_rate": 4.408354425512114e-06,
      "loss": 0.6132,
      "step": 8847
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.385823741369951,
      "learning_rate": 4.404727537115971e-06,
      "loss": 0.6643,
      "step": 8848
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39191888523645296,
      "learning_rate": 4.401101956676503e-06,
      "loss": 0.6059,
      "step": 8849
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4384155890001678,
      "learning_rate": 4.397477684497764e-06,
      "loss": 0.6575,
      "step": 8850
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3899493250142313,
      "learning_rate": 4.393854720883732e-06,
      "loss": 0.617,
      "step": 8851
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3835096182446417,
      "learning_rate": 4.390233066138266e-06,
      "loss": 0.665,
      "step": 8852
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.399529850337094,
      "learning_rate": 4.386612720565091e-06,
      "loss": 0.6681,
      "step": 8853
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40086858943836057,
      "learning_rate": 4.382993684467853e-06,
      "loss": 0.6735,
      "step": 8854
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.41876002815247937,
      "learning_rate": 4.379375958150081e-06,
      "loss": 0.6524,
      "step": 8855
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3832423418825845,
      "learning_rate": 4.37575954191517e-06,
      "loss": 0.6496,
      "step": 8856
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39390883919048775,
      "learning_rate": 4.372144436066448e-06,
      "loss": 0.6647,
      "step": 8857
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4441037932890089,
      "learning_rate": 4.3685306409070935e-06,
      "loss": 0.6141,
      "step": 8858
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4186507477777902,
      "learning_rate": 4.364918156740196e-06,
      "loss": 0.6379,
      "step": 8859
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3694022844423193,
      "learning_rate": 4.361306983868729e-06,
      "loss": 0.6468,
      "step": 8860
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3590177209935646,
      "learning_rate": 4.357697122595566e-06,
      "loss": 0.652,
      "step": 8861
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3810416951541152,
      "learning_rate": 4.354088573223448e-06,
      "loss": 0.6431,
      "step": 8862
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3741875364171314,
      "learning_rate": 4.350481336055026e-06,
      "loss": 0.612,
      "step": 8863
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38132942609927506,
      "learning_rate": 4.346875411392835e-06,
      "loss": 0.6142,
      "step": 8864
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3700923176685014,
      "learning_rate": 4.343270799539297e-06,
      "loss": 0.6388,
      "step": 8865
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38852746474284466,
      "learning_rate": 4.339667500796736e-06,
      "loss": 0.6304,
      "step": 8866
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5532784254983524,
      "learning_rate": 4.336065515467342e-06,
      "loss": 0.3849,
      "step": 8867
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3925238698996477,
      "learning_rate": 4.332464843853217e-06,
      "loss": 0.6435,
      "step": 8868
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40881650464239694,
      "learning_rate": 4.328865486256348e-06,
      "loss": 0.6428,
      "step": 8869
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39615787813899994,
      "learning_rate": 4.325267442978596e-06,
      "loss": 0.6643,
      "step": 8870
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3551153643921966,
      "learning_rate": 4.321670714321738e-06,
      "loss": 0.6207,
      "step": 8871
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3873510071702563,
      "learning_rate": 4.3180753005874275e-06,
      "loss": 0.6528,
      "step": 8872
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.41943378400075826,
      "learning_rate": 4.314481202077196e-06,
      "loss": 0.6647,
      "step": 8873
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4041260157020232,
      "learning_rate": 4.310888419092485e-06,
      "loss": 0.6513,
      "step": 8874
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4018552742316672,
      "learning_rate": 4.307296951934619e-06,
      "loss": 0.6379,
      "step": 8875
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3704784341715659,
      "learning_rate": 4.303706800904799e-06,
      "loss": 0.6229,
      "step": 8876
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3503341452327363,
      "learning_rate": 4.3001179663041355e-06,
      "loss": 0.6778,
      "step": 8877
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39724486248563073,
      "learning_rate": 4.296530448433615e-06,
      "loss": 0.6356,
      "step": 8878
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38993818336436914,
      "learning_rate": 4.292944247594124e-06,
      "loss": 0.6859,
      "step": 8879
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4093192357624669,
      "learning_rate": 4.289359364086434e-06,
      "loss": 0.6254,
      "step": 8880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43321704152005075,
      "learning_rate": 4.2857757982111935e-06,
      "loss": 0.6982,
      "step": 8881
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3740876818137387,
      "learning_rate": 4.282193550268961e-06,
      "loss": 0.6036,
      "step": 8882
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35238132346734374,
      "learning_rate": 4.27861262056017e-06,
      "loss": 0.6282,
      "step": 8883
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3662886664739353,
      "learning_rate": 4.275033009385155e-06,
      "loss": 0.6159,
      "step": 8884
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33092865088878565,
      "learning_rate": 4.2714547170441275e-06,
      "loss": 0.6199,
      "step": 8885
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4101723535508872,
      "learning_rate": 4.267877743837205e-06,
      "loss": 0.6723,
      "step": 8886
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4309523284867535,
      "learning_rate": 4.264302090064372e-06,
      "loss": 0.6606,
      "step": 8887
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4156215702085114,
      "learning_rate": 4.260727756025518e-06,
      "loss": 0.7109,
      "step": 8888
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42457298038208596,
      "learning_rate": 4.257154742020424e-06,
      "loss": 0.6488,
      "step": 8889
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5491064271936161,
      "learning_rate": 4.2535830483487435e-06,
      "loss": 0.3683,
      "step": 8890
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39846319967852956,
      "learning_rate": 4.250012675310031e-06,
      "loss": 0.6581,
      "step": 8891
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4102019903185014,
      "learning_rate": 4.2464436232037445e-06,
      "loss": 0.647,
      "step": 8892
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40679264039059165,
      "learning_rate": 4.242875892329201e-06,
      "loss": 0.6749,
      "step": 8893
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42110896604293685,
      "learning_rate": 4.2393094829856275e-06,
      "loss": 0.6698,
      "step": 8894
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4039766127191629,
      "learning_rate": 4.235744395472136e-06,
      "loss": 0.636,
      "step": 8895
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40067105187253177,
      "learning_rate": 4.2321806300877214e-06,
      "loss": 0.6284,
      "step": 8896
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4245886780122794,
      "learning_rate": 4.2286181871312745e-06,
      "loss": 0.6638,
      "step": 8897
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37719384397873046,
      "learning_rate": 4.225057066901572e-06,
      "loss": 0.6277,
      "step": 8898
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4077350784761205,
      "learning_rate": 4.221497269697281e-06,
      "loss": 0.6106,
      "step": 8899
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5414736857294123,
      "learning_rate": 4.217938795816965e-06,
      "loss": 0.3573,
      "step": 8900
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4506076416338279,
      "learning_rate": 4.214381645559058e-06,
      "loss": 0.674,
      "step": 8901
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3974907295558513,
      "learning_rate": 4.210825819221897e-06,
      "loss": 0.6391,
      "step": 8902
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36995657861461384,
      "learning_rate": 4.207271317103712e-06,
      "loss": 0.6417,
      "step": 8903
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38292847031738564,
      "learning_rate": 4.203718139502601e-06,
      "loss": 0.6798,
      "step": 8904
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.32578764159062445,
      "learning_rate": 4.200166286716571e-06,
      "loss": 0.626,
      "step": 8905
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3717863595070777,
      "learning_rate": 4.19661575904352e-06,
      "loss": 0.6478,
      "step": 8906
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36078379980878683,
      "learning_rate": 4.193066556781218e-06,
      "loss": 0.6332,
      "step": 8907
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37668931012647905,
      "learning_rate": 4.189518680227331e-06,
      "loss": 0.6371,
      "step": 8908
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36333841827081254,
      "learning_rate": 4.185972129679426e-06,
      "loss": 0.6457,
      "step": 8909
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3323506294853629,
      "learning_rate": 4.182426905434933e-06,
      "loss": 0.6171,
      "step": 8910
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34648257423317613,
      "learning_rate": 4.178883007791192e-06,
      "loss": 0.6647,
      "step": 8911
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4340153541263022,
      "learning_rate": 4.175340437045428e-06,
      "loss": 0.6766,
      "step": 8912
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4110363547545475,
      "learning_rate": 4.17179919349475e-06,
      "loss": 0.6497,
      "step": 8913
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3752809380836642,
      "learning_rate": 4.168259277436162e-06,
      "loss": 0.6293,
      "step": 8914
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3420728982253326,
      "learning_rate": 4.164720689166544e-06,
      "loss": 0.6123,
      "step": 8915
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3870563509741772,
      "learning_rate": 4.161183428982678e-06,
      "loss": 0.6046,
      "step": 8916
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38365243227103807,
      "learning_rate": 4.157647497181227e-06,
      "loss": 0.6491,
      "step": 8917
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3797890608148196,
      "learning_rate": 4.154112894058755e-06,
      "loss": 0.5996,
      "step": 8918
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35972615608174047,
      "learning_rate": 4.150579619911687e-06,
      "loss": 0.644,
      "step": 8919
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3507693458124115,
      "learning_rate": 4.1470476750363755e-06,
      "loss": 0.6377,
      "step": 8920
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3995137293021421,
      "learning_rate": 4.1435170597290255e-06,
      "loss": 0.6666,
      "step": 8921
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34895329156247407,
      "learning_rate": 4.1399877742857475e-06,
      "loss": 0.5977,
      "step": 8922
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35525917051996747,
      "learning_rate": 4.13645981900255e-06,
      "loss": 0.6359,
      "step": 8923
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40493572416922546,
      "learning_rate": 4.132933194175299e-06,
      "loss": 0.684,
      "step": 8924
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37046102716349844,
      "learning_rate": 4.129407900099782e-06,
      "loss": 0.6274,
      "step": 8925
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4054498548389552,
      "learning_rate": 4.1258839370716575e-06,
      "loss": 0.652,
      "step": 8926
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3695690222466866,
      "learning_rate": 4.1223613053864734e-06,
      "loss": 0.6282,
      "step": 8927
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3875321232606968,
      "learning_rate": 4.1188400053396725e-06,
      "loss": 0.6493,
      "step": 8928
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40555342081138146,
      "learning_rate": 4.115320037226586e-06,
      "loss": 0.6696,
      "step": 8929
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4080838113880915,
      "learning_rate": 4.111801401342421e-06,
      "loss": 0.6385,
      "step": 8930
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4068812053229785,
      "learning_rate": 4.1082840979822824e-06,
      "loss": 0.6364,
      "step": 8931
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37646271133939735,
      "learning_rate": 4.10476812744117e-06,
      "loss": 0.6352,
      "step": 8932
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40277066209832413,
      "learning_rate": 4.101253490013948e-06,
      "loss": 0.6327,
      "step": 8933
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34296858768592653,
      "learning_rate": 4.0977401859954045e-06,
      "loss": 0.6306,
      "step": 8934
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42927578060951915,
      "learning_rate": 4.094228215680183e-06,
      "loss": 0.6556,
      "step": 8935
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3727693926104641,
      "learning_rate": 4.090717579362828e-06,
      "loss": 0.6338,
      "step": 8936
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42110195308222925,
      "learning_rate": 4.0872082773377864e-06,
      "loss": 0.6418,
      "step": 8937
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39423051539273707,
      "learning_rate": 4.083700309899359e-06,
      "loss": 0.6568,
      "step": 8938
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35404120336878525,
      "learning_rate": 4.080193677341762e-06,
      "loss": 0.633,
      "step": 8939
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4258511218247746,
      "learning_rate": 4.076688379959104e-06,
      "loss": 0.6576,
      "step": 8940
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3557475215996708,
      "learning_rate": 4.073184418045357e-06,
      "loss": 0.6431,
      "step": 8941
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43197014596730154,
      "learning_rate": 4.069681791894397e-06,
      "loss": 0.6818,
      "step": 8942
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3894119190546675,
      "learning_rate": 4.066180501799994e-06,
      "loss": 0.6419,
      "step": 8943
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39314235903526507,
      "learning_rate": 4.062680548055782e-06,
      "loss": 0.6608,
      "step": 8944
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41350552582982014,
      "learning_rate": 4.059181930955305e-06,
      "loss": 0.6395,
      "step": 8945
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38786123190265465,
      "learning_rate": 4.055684650791987e-06,
      "loss": 0.6923,
      "step": 8946
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40053388018675506,
      "learning_rate": 4.0521887078591436e-06,
      "loss": 0.6405,
      "step": 8947
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35656898185833885,
      "learning_rate": 4.04869410244997e-06,
      "loss": 0.6221,
      "step": 8948
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3919826502257475,
      "learning_rate": 4.045200834857566e-06,
      "loss": 0.6485,
      "step": 8949
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4033202191198641,
      "learning_rate": 4.0417089053748945e-06,
      "loss": 0.6457,
      "step": 8950
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3525367195022531,
      "learning_rate": 4.0382183142948216e-06,
      "loss": 0.6084,
      "step": 8951
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3261161368672086,
      "learning_rate": 4.03472906191011e-06,
      "loss": 0.64,
      "step": 8952
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35057203092273287,
      "learning_rate": 4.0312411485133784e-06,
      "loss": 0.6201,
      "step": 8953
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38585548835526456,
      "learning_rate": 4.027754574397176e-06,
      "loss": 0.6499,
      "step": 8954
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37838323511110705,
      "learning_rate": 4.024269339853904e-06,
      "loss": 0.6436,
      "step": 8955
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3933989911692002,
      "learning_rate": 4.020785445175868e-06,
      "loss": 0.6595,
      "step": 8956
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39458836786575346,
      "learning_rate": 4.017302890655266e-06,
      "loss": 0.6641,
      "step": 8957
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3687248577603108,
      "learning_rate": 4.0138216765841595e-06,
      "loss": 0.6481,
      "step": 8958
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41237414548264745,
      "learning_rate": 4.010341803254525e-06,
      "loss": 0.6401,
      "step": 8959
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3849542428956629,
      "learning_rate": 4.0068632709582125e-06,
      "loss": 0.6159,
      "step": 8960
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.432196352643197,
      "learning_rate": 4.003386079986964e-06,
      "loss": 0.6626,
      "step": 8961
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37336270262085464,
      "learning_rate": 3.999910230632405e-06,
      "loss": 0.6284,
      "step": 8962
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3839058779024817,
      "learning_rate": 3.996435723186056e-06,
      "loss": 0.6791,
      "step": 8963
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35152935613284064,
      "learning_rate": 3.9929625579393106e-06,
      "loss": 0.6466,
      "step": 8964
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3616100030743354,
      "learning_rate": 3.9894907351834636e-06,
      "loss": 0.6103,
      "step": 8965
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4583951244767807,
      "learning_rate": 3.986020255209697e-06,
      "loss": 0.6908,
      "step": 8966
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34962100539300434,
      "learning_rate": 3.982551118309063e-06,
      "loss": 0.6271,
      "step": 8967
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3396413370762468,
      "learning_rate": 3.979083324772533e-06,
      "loss": 0.6333,
      "step": 8968
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4014051721223772,
      "learning_rate": 3.975616874890931e-06,
      "loss": 0.6628,
      "step": 8969
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4126126353281486,
      "learning_rate": 3.972151768954988e-06,
      "loss": 0.6389,
      "step": 8970
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43105272259813493,
      "learning_rate": 3.968688007255325e-06,
      "loss": 0.6411,
      "step": 8971
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4099546756249124,
      "learning_rate": 3.965225590082433e-06,
      "loss": 0.6783,
      "step": 8972
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39442572928711833,
      "learning_rate": 3.9617645177267075e-06,
      "loss": 0.6396,
      "step": 8973
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37148528394289854,
      "learning_rate": 3.95830479047842e-06,
      "loss": 0.6651,
      "step": 8974
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3880775746250334,
      "learning_rate": 3.954846408627737e-06,
      "loss": 0.6624,
      "step": 8975
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4546098948025512,
      "learning_rate": 3.951389372464707e-06,
      "loss": 0.61,
      "step": 8976
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4028441522910166,
      "learning_rate": 3.947933682279277e-06,
      "loss": 0.6523,
      "step": 8977
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43024393886196244,
      "learning_rate": 3.944479338361253e-06,
      "loss": 0.6614,
      "step": 8978
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41185174355139964,
      "learning_rate": 3.941026341000361e-06,
      "loss": 0.6495,
      "step": 8979
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3946953853625982,
      "learning_rate": 3.937574690486199e-06,
      "loss": 0.668,
      "step": 8980
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.386972897167912,
      "learning_rate": 3.934124387108238e-06,
      "loss": 0.6469,
      "step": 8981
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37116405970028205,
      "learning_rate": 3.930675431155868e-06,
      "loss": 0.678,
      "step": 8982
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41226763862138505,
      "learning_rate": 3.9272278229183495e-06,
      "loss": 0.6363,
      "step": 8983
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4234486799856316,
      "learning_rate": 3.923781562684816e-06,
      "loss": 0.6419,
      "step": 8984
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36410872872930866,
      "learning_rate": 3.9203366507443075e-06,
      "loss": 0.6644,
      "step": 8985
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43173972281651035,
      "learning_rate": 3.9168930873857495e-06,
      "loss": 0.6545,
      "step": 8986
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38854656035420787,
      "learning_rate": 3.913450872897939e-06,
      "loss": 0.6412,
      "step": 8987
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3490025401843877,
      "learning_rate": 3.910010007569578e-06,
      "loss": 0.6446,
      "step": 8988
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4142333456166791,
      "learning_rate": 3.906570491689246e-06,
      "loss": 0.6617,
      "step": 8989
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35904656258031753,
      "learning_rate": 3.903132325545409e-06,
      "loss": 0.6184,
      "step": 8990
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4197579235503576,
      "learning_rate": 3.899695509426431e-06,
      "loss": 0.6699,
      "step": 8991
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4087363205627457,
      "learning_rate": 3.896260043620539e-06,
      "loss": 0.6812,
      "step": 8992
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41095361646650796,
      "learning_rate": 3.8928259284158706e-06,
      "loss": 0.6487,
      "step": 8993
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4101868499769167,
      "learning_rate": 3.8893931641004435e-06,
      "loss": 0.6752,
      "step": 8994
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34857301680700786,
      "learning_rate": 3.885961750962144e-06,
      "loss": 0.623,
      "step": 8995
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38850180969479053,
      "learning_rate": 3.882531689288778e-06,
      "loss": 0.6408,
      "step": 8996
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3599267215444946,
      "learning_rate": 3.8791029793680195e-06,
      "loss": 0.6328,
      "step": 8997
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3716267377554833,
      "learning_rate": 3.8756756214874204e-06,
      "loss": 0.6238,
      "step": 8998
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3876180885804561,
      "learning_rate": 3.8722496159344315e-06,
      "loss": 0.6772,
      "step": 8999
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35466716568513873,
      "learning_rate": 3.868824962996396e-06,
      "loss": 0.6222,
      "step": 9000
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3661221500825474,
      "learning_rate": 3.865401662960524e-06,
      "loss": 0.644,
      "step": 9001
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38237403428726174,
      "learning_rate": 3.86197971611393e-06,
      "loss": 0.6569,
      "step": 9002
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3701254675025103,
      "learning_rate": 3.8585591227436055e-06,
      "loss": 0.611,
      "step": 9003
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44575745682302265,
      "learning_rate": 3.855139883136436e-06,
      "loss": 0.6667,
      "step": 9004
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36878894436107446,
      "learning_rate": 3.851721997579185e-06,
      "loss": 0.6513,
      "step": 9005
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3674215312043573,
      "learning_rate": 3.848305466358513e-06,
      "loss": 0.627,
      "step": 9006
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38839692808192955,
      "learning_rate": 3.84489028976095e-06,
      "loss": 0.6724,
      "step": 9007
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3651971603101297,
      "learning_rate": 3.841476468072926e-06,
      "loss": 0.6495,
      "step": 9008
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39983373874438455,
      "learning_rate": 3.838064001580759e-06,
      "loss": 0.6644,
      "step": 9009
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.35754027295851826,
      "learning_rate": 3.834652890570645e-06,
      "loss": 0.6134,
      "step": 9010
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.35464654699950493,
      "learning_rate": 3.831243135328675e-06,
      "loss": 0.6337,
      "step": 9011
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41648212097388465,
      "learning_rate": 3.827834736140812e-06,
      "loss": 0.6165,
      "step": 9012
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3541657595145435,
      "learning_rate": 3.824427693292918e-06,
      "loss": 0.6218,
      "step": 9013
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3659909757471799,
      "learning_rate": 3.821022007070743e-06,
      "loss": 0.6409,
      "step": 9014
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39156995004144984,
      "learning_rate": 3.817617677759908e-06,
      "loss": 0.6509,
      "step": 9015
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4281336777128291,
      "learning_rate": 3.814214705645931e-06,
      "loss": 0.6579,
      "step": 9016
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43437912845810517,
      "learning_rate": 3.810813091014227e-06,
      "loss": 0.655,
      "step": 9017
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38268039218342137,
      "learning_rate": 3.807412834150075e-06,
      "loss": 0.6612,
      "step": 9018
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3850362072162934,
      "learning_rate": 3.8040139353386507e-06,
      "loss": 0.6915,
      "step": 9019
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3797710425974411,
      "learning_rate": 3.8006163948650244e-06,
      "loss": 0.6655,
      "step": 9020
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38792900863174185,
      "learning_rate": 3.797220213014132e-06,
      "loss": 0.6247,
      "step": 9021
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34092140138089805,
      "learning_rate": 3.793825390070811e-06,
      "loss": 0.6203,
      "step": 9022
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5571590311238152,
      "learning_rate": 3.7904319263197843e-06,
      "loss": 0.3651,
      "step": 9023
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36024200567132414,
      "learning_rate": 3.787039822045655e-06,
      "loss": 0.6445,
      "step": 9024
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3805994107180619,
      "learning_rate": 3.7836490775329206e-06,
      "loss": 0.6568,
      "step": 9025
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3761742372495596,
      "learning_rate": 3.7802596930659486e-06,
      "loss": 0.6266,
      "step": 9026
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38035717491456433,
      "learning_rate": 3.7768716689290053e-06,
      "loss": 0.6357,
      "step": 9027
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3982554612537408,
      "learning_rate": 3.77348500540625e-06,
      "loss": 0.599,
      "step": 9028
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4025326224630824,
      "learning_rate": 3.7700997027816997e-06,
      "loss": 0.6514,
      "step": 9029
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4134145854523954,
      "learning_rate": 3.7667157613392924e-06,
      "loss": 0.6101,
      "step": 9030
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39182783877329497,
      "learning_rate": 3.763333181362836e-06,
      "loss": 0.6266,
      "step": 9031
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3378799097708136,
      "learning_rate": 3.759951963136008e-06,
      "loss": 0.6435,
      "step": 9032
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4105870722877104,
      "learning_rate": 3.7565721069424e-06,
      "loss": 0.6408,
      "step": 9033
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3866310111210337,
      "learning_rate": 3.7531936130654735e-06,
      "loss": 0.6289,
      "step": 9034
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3880924456278459,
      "learning_rate": 3.749816481788575e-06,
      "loss": 0.6019,
      "step": 9035
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3850334624967869,
      "learning_rate": 3.746440713394941e-06,
      "loss": 0.6115,
      "step": 9036
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41888756381794456,
      "learning_rate": 3.7430663081676977e-06,
      "loss": 0.6193,
      "step": 9037
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37198872011524375,
      "learning_rate": 3.7396932663898476e-06,
      "loss": 0.6148,
      "step": 9038
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3873044511657211,
      "learning_rate": 3.736321588344285e-06,
      "loss": 0.6362,
      "step": 9039
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37513391294026893,
      "learning_rate": 3.7329512743137962e-06,
      "loss": 0.6652,
      "step": 9040
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4011759210579428,
      "learning_rate": 3.7295823245810338e-06,
      "loss": 0.6075,
      "step": 9041
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.379363045890615,
      "learning_rate": 3.726214739428551e-06,
      "loss": 0.6569,
      "step": 9042
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37195168113520377,
      "learning_rate": 3.7228485191387818e-06,
      "loss": 0.6213,
      "step": 9043
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.40505384329748334,
      "learning_rate": 3.7194836639940523e-06,
      "loss": 0.6379,
      "step": 9044
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.32777007014096315,
      "learning_rate": 3.716120174276572e-06,
      "loss": 0.6287,
      "step": 9045
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39687653802791717,
      "learning_rate": 3.712758050268419e-06,
      "loss": 0.6237,
      "step": 9046
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37865452099798264,
      "learning_rate": 3.7093972922515797e-06,
      "loss": 0.607,
      "step": 9047
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31738158483700213,
      "learning_rate": 3.7060379005079216e-06,
      "loss": 0.5872,
      "step": 9048
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37836087327395007,
      "learning_rate": 3.7026798753191795e-06,
      "loss": 0.6459,
      "step": 9049
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3704408281943181,
      "learning_rate": 3.69932321696699e-06,
      "loss": 0.6285,
      "step": 9050
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.382942764405021,
      "learning_rate": 3.6959679257328862e-06,
      "loss": 0.6774,
      "step": 9051
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37434415502800217,
      "learning_rate": 3.692614001898258e-06,
      "loss": 0.5987,
      "step": 9052
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41224965858922197,
      "learning_rate": 3.6892614457443986e-06,
      "loss": 0.6395,
      "step": 9053
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3567484962307369,
      "learning_rate": 3.6859102575524874e-06,
      "loss": 0.6322,
      "step": 9054
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4256843626846218,
      "learning_rate": 3.682560437603577e-06,
      "loss": 0.6051,
      "step": 9055
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39128779392947016,
      "learning_rate": 3.6792119861786146e-06,
      "loss": 0.6236,
      "step": 9056
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3779077104413625,
      "learning_rate": 3.675864903558433e-06,
      "loss": 0.6178,
      "step": 9057
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36629153763077454,
      "learning_rate": 3.6725191900237465e-06,
      "loss": 0.6473,
      "step": 9058
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3520226293597004,
      "learning_rate": 3.6691748458551633e-06,
      "loss": 0.6186,
      "step": 9059
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4023901429252936,
      "learning_rate": 3.6658318713331586e-06,
      "loss": 0.6395,
      "step": 9060
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5716295003469091,
      "learning_rate": 3.662490266738108e-06,
      "loss": 0.395,
      "step": 9061
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3799453126007264,
      "learning_rate": 3.6591500323502672e-06,
      "loss": 0.6467,
      "step": 9062
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38583215830654916,
      "learning_rate": 3.655811168449783e-06,
      "loss": 0.6389,
      "step": 9063
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39591198286603446,
      "learning_rate": 3.652473675316668e-06,
      "loss": 0.6447,
      "step": 9064
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3734997483363022,
      "learning_rate": 3.6491375532308526e-06,
      "loss": 0.617,
      "step": 9065
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3838335888568233,
      "learning_rate": 3.645802802472118e-06,
      "loss": 0.646,
      "step": 9066
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39794540877137285,
      "learning_rate": 3.642469423320154e-06,
      "loss": 0.6737,
      "step": 9067
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.40476475202189727,
      "learning_rate": 3.639137416054528e-06,
      "loss": 0.6814,
      "step": 9068
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3513332243879989,
      "learning_rate": 3.635806780954683e-06,
      "loss": 0.5978,
      "step": 9069
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38423903550671024,
      "learning_rate": 3.6324775182999593e-06,
      "loss": 0.6405,
      "step": 9070
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41181652496237275,
      "learning_rate": 3.6291496283695814e-06,
      "loss": 0.6139,
      "step": 9071
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.40221803991166033,
      "learning_rate": 3.6258231114426524e-06,
      "loss": 0.654,
      "step": 9072
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41349612972224503,
      "learning_rate": 3.622497967798164e-06,
      "loss": 0.6271,
      "step": 9073
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39935790726122017,
      "learning_rate": 3.6191741977149965e-06,
      "loss": 0.6497,
      "step": 9074
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4339924201686552,
      "learning_rate": 3.6158518014719015e-06,
      "loss": 0.6674,
      "step": 9075
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39763527130873416,
      "learning_rate": 3.6125307793475297e-06,
      "loss": 0.6284,
      "step": 9076
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.40201050651886705,
      "learning_rate": 3.609211131620414e-06,
      "loss": 0.6204,
      "step": 9077
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.651400832467915,
      "learning_rate": 3.6058928585689556e-06,
      "loss": 0.3644,
      "step": 9078
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4261742072051221,
      "learning_rate": 3.602575960471475e-06,
      "loss": 0.6996,
      "step": 9079
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39072809780999734,
      "learning_rate": 3.59926043760614e-06,
      "loss": 0.633,
      "step": 9080
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36558834140508284,
      "learning_rate": 3.5959462902510243e-06,
      "loss": 0.6305,
      "step": 9081
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4185945302238802,
      "learning_rate": 3.592633518684088e-06,
      "loss": 0.6925,
      "step": 9082
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.42701875349571594,
      "learning_rate": 3.5893221231831586e-06,
      "loss": 0.6535,
      "step": 9083
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4023650314449678,
      "learning_rate": 3.5860121040259623e-06,
      "loss": 0.6608,
      "step": 9084
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3934257090067663,
      "learning_rate": 3.582703461490109e-06,
      "loss": 0.6526,
      "step": 9085
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4798553854156726,
      "learning_rate": 3.5793961958530886e-06,
      "loss": 0.6675,
      "step": 9086
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.35111690333889894,
      "learning_rate": 3.5760903073922792e-06,
      "loss": 0.6287,
      "step": 9087
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.35379847154948585,
      "learning_rate": 3.5727857963849456e-06,
      "loss": 0.6551,
      "step": 9088
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37044477933390235,
      "learning_rate": 3.5694826631082235e-06,
      "loss": 0.6491,
      "step": 9089
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38121681667119167,
      "learning_rate": 3.566180907839145e-06,
      "loss": 0.6128,
      "step": 9090
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.431790977201695,
      "learning_rate": 3.562880530854633e-06,
      "loss": 0.6715,
      "step": 9091
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3989025053347957,
      "learning_rate": 3.559581532431471e-06,
      "loss": 0.6522,
      "step": 9092
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4384407642640649,
      "learning_rate": 3.556283912846361e-06,
      "loss": 0.629,
      "step": 9093
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4315874715768025,
      "learning_rate": 3.552987672375854e-06,
      "loss": 0.6755,
      "step": 9094
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4152242600734739,
      "learning_rate": 3.549692811296408e-06,
      "loss": 0.6526,
      "step": 9095
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37134289283880656,
      "learning_rate": 3.54639932988436e-06,
      "loss": 0.5955,
      "step": 9096
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39597818656106504,
      "learning_rate": 3.5431072284159342e-06,
      "loss": 0.6401,
      "step": 9097
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4162183259970209,
      "learning_rate": 3.539816507167224e-06,
      "loss": 0.6511,
      "step": 9098
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39135408843890007,
      "learning_rate": 3.5365271664142254e-06,
      "loss": 0.6227,
      "step": 9099
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3787876009012638,
      "learning_rate": 3.53323920643281e-06,
      "loss": 0.6434,
      "step": 9100
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34789478145702224,
      "learning_rate": 3.5299526274987363e-06,
      "loss": 0.6305,
      "step": 9101
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3889926546157256,
      "learning_rate": 3.526667429887649e-06,
      "loss": 0.6229,
      "step": 9102
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4322244468889642,
      "learning_rate": 3.5233836138750644e-06,
      "loss": 0.6519,
      "step": 9103
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36732088676636643,
      "learning_rate": 3.520101179736399e-06,
      "loss": 0.6424,
      "step": 9104
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36584851501634214,
      "learning_rate": 3.5168201277469494e-06,
      "loss": 0.6114,
      "step": 9105
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43907935044759083,
      "learning_rate": 3.5135404581818786e-06,
      "loss": 0.6747,
      "step": 9106
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.40978034777282996,
      "learning_rate": 3.5102621713162632e-06,
      "loss": 0.643,
      "step": 9107
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4267476831481659,
      "learning_rate": 3.5069852674250514e-06,
      "loss": 0.6917,
      "step": 9108
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39041629905203,
      "learning_rate": 3.5037097467830616e-06,
      "loss": 0.6245,
      "step": 9109
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3616869175154895,
      "learning_rate": 3.500435609665012e-06,
      "loss": 0.6434,
      "step": 9110
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3750606749286921,
      "learning_rate": 3.497162856345506e-06,
      "loss": 0.5952,
      "step": 9111
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4219101011135237,
      "learning_rate": 3.4938914870990125e-06,
      "loss": 0.6672,
      "step": 9112
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4010018973460541,
      "learning_rate": 3.4906215021999133e-06,
      "loss": 0.6826,
      "step": 9113
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4000575419011763,
      "learning_rate": 3.4873529019224472e-06,
      "loss": 0.673,
      "step": 9114
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42950056647585866,
      "learning_rate": 3.4840856865407503e-06,
      "loss": 0.6743,
      "step": 9115
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38423584776204234,
      "learning_rate": 3.4808198563288476e-06,
      "loss": 0.6543,
      "step": 9116
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3994602215525997,
      "learning_rate": 3.477555411560627e-06,
      "loss": 0.6457,
      "step": 9117
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4160691958665512,
      "learning_rate": 3.474292352509878e-06,
      "loss": 0.6284,
      "step": 9118
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38049094242026243,
      "learning_rate": 3.4710306794502733e-06,
      "loss": 0.6456,
      "step": 9119
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36259444041554445,
      "learning_rate": 3.4677703926553653e-06,
      "loss": 0.6578,
      "step": 9120
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3891611219353967,
      "learning_rate": 3.4645114923985855e-06,
      "loss": 0.6405,
      "step": 9121
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.47291252883940377,
      "learning_rate": 3.4612539789532627e-06,
      "loss": 0.6566,
      "step": 9122
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.361683964479977,
      "learning_rate": 3.457997852592592e-06,
      "loss": 0.6045,
      "step": 9123
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4061526974353539,
      "learning_rate": 3.4547431135896626e-06,
      "loss": 0.6287,
      "step": 9124
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.439854450280269,
      "learning_rate": 3.4514897622174505e-06,
      "loss": 0.6471,
      "step": 9125
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.46821876614128743,
      "learning_rate": 3.448237798748797e-06,
      "loss": 0.6515,
      "step": 9126
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.396430111114302,
      "learning_rate": 3.444987223456455e-06,
      "loss": 0.6497,
      "step": 9127
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34929558124317034,
      "learning_rate": 3.4417380366130472e-06,
      "loss": 0.6252,
      "step": 9128
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4242930099411721,
      "learning_rate": 3.438490238491068e-06,
      "loss": 0.6562,
      "step": 9129
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37722606272376347,
      "learning_rate": 3.4352438293629085e-06,
      "loss": 0.6435,
      "step": 9130
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4188243877748656,
      "learning_rate": 3.4319988095008516e-06,
      "loss": 0.6577,
      "step": 9131
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33120271044971794,
      "learning_rate": 3.428755179177039e-06,
      "loss": 0.623,
      "step": 9132
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.45328202333044376,
      "learning_rate": 3.425512938663518e-06,
      "loss": 0.6558,
      "step": 9133
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4045031798731739,
      "learning_rate": 3.422272088232208e-06,
      "loss": 0.6433,
      "step": 9134
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4093068009574819,
      "learning_rate": 3.419032628154919e-06,
      "loss": 0.601,
      "step": 9135
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3809680557397637,
      "learning_rate": 3.4157945587033426e-06,
      "loss": 0.629,
      "step": 9136
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5513817124876902,
      "learning_rate": 3.4125578801490433e-06,
      "loss": 0.329,
      "step": 9137
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44462160519550264,
      "learning_rate": 3.409322592763482e-06,
      "loss": 0.655,
      "step": 9138
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36358722237639685,
      "learning_rate": 3.406088696818004e-06,
      "loss": 0.6416,
      "step": 9139
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.402078250000331,
      "learning_rate": 3.4028561925838166e-06,
      "loss": 0.6763,
      "step": 9140
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5312609531463152,
      "learning_rate": 3.3996250803320407e-06,
      "loss": 0.3451,
      "step": 9141
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3914618133408959,
      "learning_rate": 3.396395360333664e-06,
      "loss": 0.6141,
      "step": 9142
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3742539365485692,
      "learning_rate": 3.3931670328595544e-06,
      "loss": 0.6143,
      "step": 9143
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38720317945911453,
      "learning_rate": 3.3899400981804688e-06,
      "loss": 0.6481,
      "step": 9144
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4561953237910925,
      "learning_rate": 3.386714556567052e-06,
      "loss": 0.6971,
      "step": 9145
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4278381187246851,
      "learning_rate": 3.3834904082898155e-06,
      "loss": 0.6536,
      "step": 9146
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38979490751120327,
      "learning_rate": 3.380267653619174e-06,
      "loss": 0.6413,
      "step": 9147
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3685989545387734,
      "learning_rate": 3.37704629282541e-06,
      "loss": 0.6442,
      "step": 9148
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3710900547354697,
      "learning_rate": 3.3738263261786996e-06,
      "loss": 0.6077,
      "step": 9149
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37220294305869683,
      "learning_rate": 3.3706077539490933e-06,
      "loss": 0.6058,
      "step": 9150
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3962420505585346,
      "learning_rate": 3.3673905764065397e-06,
      "loss": 0.6389,
      "step": 9151
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38938568386817524,
      "learning_rate": 3.3641747938208447e-06,
      "loss": 0.6278,
      "step": 9152
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4304477688478357,
      "learning_rate": 3.3609604064617196e-06,
      "loss": 0.6529,
      "step": 9153
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4637036366509159,
      "learning_rate": 3.357747414598749e-06,
      "loss": 0.6218,
      "step": 9154
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3388288408060203,
      "learning_rate": 3.3545358185014053e-06,
      "loss": 0.6371,
      "step": 9155
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4082752733400625,
      "learning_rate": 3.351325618439043e-06,
      "loss": 0.6416,
      "step": 9156
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3574859061943763,
      "learning_rate": 3.348116814680891e-06,
      "loss": 0.6421,
      "step": 9157
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37359855979466433,
      "learning_rate": 3.3449094074960706e-06,
      "loss": 0.6446,
      "step": 9158
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40779385143039787,
      "learning_rate": 3.3417033971535904e-06,
      "loss": 0.6281,
      "step": 9159
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42144425780331607,
      "learning_rate": 3.3384987839223217e-06,
      "loss": 0.6237,
      "step": 9160
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3934775564980917,
      "learning_rate": 3.335295568071033e-06,
      "loss": 0.6574,
      "step": 9161
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40315225594852105,
      "learning_rate": 3.3320937498683904e-06,
      "loss": 0.6249,
      "step": 9162
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3811318426694017,
      "learning_rate": 3.3288933295829097e-06,
      "loss": 0.6592,
      "step": 9163
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41532381177566013,
      "learning_rate": 3.3256943074830116e-06,
      "loss": 0.6384,
      "step": 9164
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3514645468085412,
      "learning_rate": 3.3224966838369997e-06,
      "loss": 0.5899,
      "step": 9165
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4161348772868435,
      "learning_rate": 3.319300458913044e-06,
      "loss": 0.6151,
      "step": 9166
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3738741983406517,
      "learning_rate": 3.316105632979214e-06,
      "loss": 0.607,
      "step": 9167
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39799810905857813,
      "learning_rate": 3.3129122063034556e-06,
      "loss": 0.6255,
      "step": 9168
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35944692952062324,
      "learning_rate": 3.309720179153599e-06,
      "loss": 0.6245,
      "step": 9169
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3810541553250401,
      "learning_rate": 3.306529551797357e-06,
      "loss": 0.6382,
      "step": 9170
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3917214042217601,
      "learning_rate": 3.3033403245023176e-06,
      "loss": 0.618,
      "step": 9171
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39911261245859914,
      "learning_rate": 3.30015249753596e-06,
      "loss": 0.6637,
      "step": 9172
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36378359126398413,
      "learning_rate": 3.296966071165648e-06,
      "loss": 0.612,
      "step": 9173
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4222444395477025,
      "learning_rate": 3.293781045658615e-06,
      "loss": 0.6531,
      "step": 9174
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40162727556763944,
      "learning_rate": 3.2905974212819868e-06,
      "loss": 0.6579,
      "step": 9175
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3867328715478748,
      "learning_rate": 3.287415198302781e-06,
      "loss": 0.6936,
      "step": 9176
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4338487781540911,
      "learning_rate": 3.2842343769878738e-06,
      "loss": 0.6498,
      "step": 9177
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.443009856582301,
      "learning_rate": 3.281054957604042e-06,
      "loss": 0.662,
      "step": 9178
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35882871012386375,
      "learning_rate": 3.2778769404179457e-06,
      "loss": 0.6536,
      "step": 9179
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4101722116453609,
      "learning_rate": 3.2747003256961073e-06,
      "loss": 0.6611,
      "step": 9180
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38106589679651043,
      "learning_rate": 3.2715251137049566e-06,
      "loss": 0.6025,
      "step": 9181
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3588514428724689,
      "learning_rate": 3.2683513047107905e-06,
      "loss": 0.6262,
      "step": 9182
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39845225439726717,
      "learning_rate": 3.265178898979795e-06,
      "loss": 0.622,
      "step": 9183
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39948945499520105,
      "learning_rate": 3.262007896778032e-06,
      "loss": 0.6348,
      "step": 9184
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4258218461453155,
      "learning_rate": 3.2588382983714585e-06,
      "loss": 0.6859,
      "step": 9185
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3702487015756002,
      "learning_rate": 3.255670104025894e-06,
      "loss": 0.626,
      "step": 9186
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3615547149902091,
      "learning_rate": 3.252503314007056e-06,
      "loss": 0.6121,
      "step": 9187
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.47818961226781126,
      "learning_rate": 3.249337928580545e-06,
      "loss": 0.664,
      "step": 9188
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36480007553879357,
      "learning_rate": 3.2461739480118215e-06,
      "loss": 0.6289,
      "step": 9189
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36111709286898613,
      "learning_rate": 3.2430113725662647e-06,
      "loss": 0.6266,
      "step": 9190
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4004240442610535,
      "learning_rate": 3.239850202509103e-06,
      "loss": 0.6599,
      "step": 9191
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43743566754787566,
      "learning_rate": 3.2366904381054633e-06,
      "loss": 0.6681,
      "step": 9192
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4166523328975566,
      "learning_rate": 3.2335320796203564e-06,
      "loss": 0.6348,
      "step": 9193
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3580294436608107,
      "learning_rate": 3.230375127318661e-06,
      "loss": 0.6175,
      "step": 9194
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41578095149064365,
      "learning_rate": 3.2272195814651443e-06,
      "loss": 0.6555,
      "step": 9195
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36796107179705906,
      "learning_rate": 3.2240654423244754e-06,
      "loss": 0.5924,
      "step": 9196
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.416988009381854,
      "learning_rate": 3.220912710161175e-06,
      "loss": 0.6595,
      "step": 9197
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41167577574109765,
      "learning_rate": 3.2177613852396594e-06,
      "loss": 0.6519,
      "step": 9198
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39015361299607015,
      "learning_rate": 3.2146114678242337e-06,
      "loss": 0.596,
      "step": 9199
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4106497946751342,
      "learning_rate": 3.211462958179068e-06,
      "loss": 0.6107,
      "step": 9200
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39423886372769457,
      "learning_rate": 3.2083158565682293e-06,
      "loss": 0.6232,
      "step": 9201
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3864912921808689,
      "learning_rate": 3.2051701632556596e-06,
      "loss": 0.6302,
      "step": 9202
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38867466234037856,
      "learning_rate": 3.202025878505184e-06,
      "loss": 0.6388,
      "step": 9203
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39271257197381143,
      "learning_rate": 3.1988830025805174e-06,
      "loss": 0.6642,
      "step": 9204
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3904135950831271,
      "learning_rate": 3.195741535745238e-06,
      "loss": 0.6312,
      "step": 9205
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39720247949644083,
      "learning_rate": 3.1926014782628223e-06,
      "loss": 0.6384,
      "step": 9206
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4114812794232406,
      "learning_rate": 3.1894628303966212e-06,
      "loss": 0.6581,
      "step": 9207
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40345630572058655,
      "learning_rate": 3.186325592409876e-06,
      "loss": 0.6485,
      "step": 9208
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3863359799579856,
      "learning_rate": 3.1831897645656884e-06,
      "loss": 0.6158,
      "step": 9209
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38286083035933893,
      "learning_rate": 3.1800553471270733e-06,
      "loss": 0.6354,
      "step": 9210
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37533761246260766,
      "learning_rate": 3.1769223403569005e-06,
      "loss": 0.6287,
      "step": 9211
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43138921334220653,
      "learning_rate": 3.1737907445179326e-06,
      "loss": 0.6362,
      "step": 9212
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4013454899548731,
      "learning_rate": 3.170660559872818e-06,
      "loss": 0.6601,
      "step": 9213
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4042576015283434,
      "learning_rate": 3.167531786684075e-06,
      "loss": 0.6446,
      "step": 9214
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3447157343447536,
      "learning_rate": 3.1644044252141117e-06,
      "loss": 0.6199,
      "step": 9215
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42155438996233624,
      "learning_rate": 3.1612784757252157e-06,
      "loss": 0.6644,
      "step": 9216
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39449244864747596,
      "learning_rate": 3.15815393847956e-06,
      "loss": 0.64,
      "step": 9217
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3788382815039189,
      "learning_rate": 3.1550308137391927e-06,
      "loss": 0.591,
      "step": 9218
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4262753534420329,
      "learning_rate": 3.151909101766051e-06,
      "loss": 0.6139,
      "step": 9219
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4102393535420365,
      "learning_rate": 3.1487888028219426e-06,
      "loss": 0.6386,
      "step": 9220
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34783576405511285,
      "learning_rate": 3.1456699171685657e-06,
      "loss": 0.6222,
      "step": 9221
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.371817579220387,
      "learning_rate": 3.1425524450675017e-06,
      "loss": 0.6335,
      "step": 9222
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3656811542431954,
      "learning_rate": 3.139436386780197e-06,
      "loss": 0.6316,
      "step": 9223
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41842086184948557,
      "learning_rate": 3.136321742568009e-06,
      "loss": 0.6326,
      "step": 9224
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3884417563193357,
      "learning_rate": 3.1332085126921473e-06,
      "loss": 0.6272,
      "step": 9225
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3420892905454308,
      "learning_rate": 3.130096697413716e-06,
      "loss": 0.5851,
      "step": 9226
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.33933325361161915,
      "learning_rate": 3.1269862969937057e-06,
      "loss": 0.5841,
      "step": 9227
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38570087544902953,
      "learning_rate": 3.123877311692973e-06,
      "loss": 0.5908,
      "step": 9228
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36371044732671337,
      "learning_rate": 3.1207697417722692e-06,
      "loss": 0.6016,
      "step": 9229
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3904102976468618,
      "learning_rate": 3.1176635874922214e-06,
      "loss": 0.6423,
      "step": 9230
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38315027953149644,
      "learning_rate": 3.1145588491133405e-06,
      "loss": 0.6194,
      "step": 9231
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3450036152551259,
      "learning_rate": 3.111455526896017e-06,
      "loss": 0.6128,
      "step": 9232
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39031080100916493,
      "learning_rate": 3.1083536211005263e-06,
      "loss": 0.6673,
      "step": 9233
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34622639181043985,
      "learning_rate": 3.1052531319870137e-06,
      "loss": 0.6617,
      "step": 9234
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35478052161167345,
      "learning_rate": 3.1021540598155166e-06,
      "loss": 0.5946,
      "step": 9235
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4333202921663244,
      "learning_rate": 3.099056404845955e-06,
      "loss": 0.6677,
      "step": 9236
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42332520411412616,
      "learning_rate": 3.0959601673381124e-06,
      "loss": 0.6387,
      "step": 9237
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39667146666677605,
      "learning_rate": 3.0928653475516855e-06,
      "loss": 0.6385,
      "step": 9238
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3470445851831574,
      "learning_rate": 3.0897719457462185e-06,
      "loss": 0.6157,
      "step": 9239
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3195240464054686,
      "learning_rate": 3.086679962181156e-06,
      "loss": 0.5903,
      "step": 9240
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3953850239035844,
      "learning_rate": 3.083589397115818e-06,
      "loss": 0.6371,
      "step": 9241
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37997039518127834,
      "learning_rate": 3.0805002508094127e-06,
      "loss": 0.6602,
      "step": 9242
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3856575951559732,
      "learning_rate": 3.0774125235210127e-06,
      "loss": 0.6436,
      "step": 9243
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41374746823780767,
      "learning_rate": 3.074326215509589e-06,
      "loss": 0.6385,
      "step": 9244
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4073366529246739,
      "learning_rate": 3.0712413270339826e-06,
      "loss": 0.6566,
      "step": 9245
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39681983405185983,
      "learning_rate": 3.0681578583529226e-06,
      "loss": 0.6351,
      "step": 9246
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4242415612469035,
      "learning_rate": 3.0650758097250178e-06,
      "loss": 0.6611,
      "step": 9247
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3456562444324981,
      "learning_rate": 3.06199518140875e-06,
      "loss": 0.5795,
      "step": 9248
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4096045518380708,
      "learning_rate": 3.0589159736624885e-06,
      "loss": 0.6719,
      "step": 9249
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.40057981883602223,
      "learning_rate": 3.0558381867444884e-06,
      "loss": 0.6567,
      "step": 9250
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.45028981710045424,
      "learning_rate": 3.0527618209128705e-06,
      "loss": 0.6916,
      "step": 9251
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3697870165198183,
      "learning_rate": 3.0496868764256527e-06,
      "loss": 0.6295,
      "step": 9252
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39133317961168534,
      "learning_rate": 3.0466133535407325e-06,
      "loss": 0.6463,
      "step": 9253
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3915930422330396,
      "learning_rate": 3.0435412525158713e-06,
      "loss": 0.6236,
      "step": 9254
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34013814857578245,
      "learning_rate": 3.0404705736087247e-06,
      "loss": 0.6566,
      "step": 9255
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3781863556943021,
      "learning_rate": 3.0374013170768356e-06,
      "loss": 0.6426,
      "step": 9256
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.33483743286941686,
      "learning_rate": 3.0343334831776096e-06,
      "loss": 0.6272,
      "step": 9257
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38114316869386655,
      "learning_rate": 3.0312670721683424e-06,
      "loss": 0.6589,
      "step": 9258
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41770387856853086,
      "learning_rate": 3.028202084306213e-06,
      "loss": 0.6139,
      "step": 9259
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3719579867356701,
      "learning_rate": 3.0251385198482786e-06,
      "loss": 0.6391,
      "step": 9260
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4009868905032946,
      "learning_rate": 3.0220763790514816e-06,
      "loss": 0.6482,
      "step": 9261
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36591652019472076,
      "learning_rate": 3.0190156621726284e-06,
      "loss": 0.6414,
      "step": 9262
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39362389684117205,
      "learning_rate": 3.0159563694684245e-06,
      "loss": 0.6456,
      "step": 9263
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3908944606499783,
      "learning_rate": 3.012898501195447e-06,
      "loss": 0.6701,
      "step": 9264
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4029992934769904,
      "learning_rate": 3.0098420576101593e-06,
      "loss": 0.6458,
      "step": 9265
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34032239118802726,
      "learning_rate": 3.006787038968897e-06,
      "loss": 0.6199,
      "step": 9266
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3829248641843338,
      "learning_rate": 3.00373344552789e-06,
      "loss": 0.6596,
      "step": 9267
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38516962583568315,
      "learning_rate": 3.000681277543227e-06,
      "loss": 0.6469,
      "step": 9268
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36671698547330245,
      "learning_rate": 2.9976305352708945e-06,
      "loss": 0.6051,
      "step": 9269
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3976708331271486,
      "learning_rate": 2.994581218966761e-06,
      "loss": 0.6496,
      "step": 9270
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4247436843746974,
      "learning_rate": 2.991533328886558e-06,
      "loss": 0.6668,
      "step": 9271
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3868834564090882,
      "learning_rate": 2.9884868652859087e-06,
      "loss": 0.6276,
      "step": 9272
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3969257330104578,
      "learning_rate": 2.9854418284203324e-06,
      "loss": 0.6374,
      "step": 9273
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3964034874427025,
      "learning_rate": 2.982398218545197e-06,
      "loss": 0.6392,
      "step": 9274
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38553056021799714,
      "learning_rate": 2.979356035915768e-06,
      "loss": 0.6255,
      "step": 9275
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4443106491344702,
      "learning_rate": 2.976315280787201e-06,
      "loss": 0.6332,
      "step": 9276
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39807586920678323,
      "learning_rate": 2.973275953414505e-06,
      "loss": 0.6528,
      "step": 9277
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3659298200282909,
      "learning_rate": 2.970238054052592e-06,
      "loss": 0.6294,
      "step": 9278
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4113128228205552,
      "learning_rate": 2.967201582956245e-06,
      "loss": 0.6731,
      "step": 9279
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39521625503855773,
      "learning_rate": 2.9641665403801332e-06,
      "loss": 0.647,
      "step": 9280
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3924414504855645,
      "learning_rate": 2.961132926578805e-06,
      "loss": 0.6437,
      "step": 9281
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3415474715592874,
      "learning_rate": 2.9581007418066732e-06,
      "loss": 0.6463,
      "step": 9282
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37880622334853353,
      "learning_rate": 2.9550699863180533e-06,
      "loss": 0.6523,
      "step": 9283
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36700869861558877,
      "learning_rate": 2.9520406603671303e-06,
      "loss": 0.6407,
      "step": 9284
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35177018264950344,
      "learning_rate": 2.949012764207961e-06,
      "loss": 0.6263,
      "step": 9285
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42095093700645875,
      "learning_rate": 2.945986298094503e-06,
      "loss": 0.6566,
      "step": 9286
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4323611211286386,
      "learning_rate": 2.942961262280584e-06,
      "loss": 0.6721,
      "step": 9287
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3555859044836329,
      "learning_rate": 2.9399376570198978e-06,
      "loss": 0.6325,
      "step": 9288
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43106293237468174,
      "learning_rate": 2.9369154825660385e-06,
      "loss": 0.6661,
      "step": 9289
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4106252202281621,
      "learning_rate": 2.9338947391724738e-06,
      "loss": 0.6523,
      "step": 9290
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3538005460771011,
      "learning_rate": 2.930875427092543e-06,
      "loss": 0.6249,
      "step": 9291
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3397719371546869,
      "learning_rate": 2.9278575465794755e-06,
      "loss": 0.599,
      "step": 9292
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3931098840364766,
      "learning_rate": 2.9248410978863775e-06,
      "loss": 0.6308,
      "step": 9293
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4109793579917686,
      "learning_rate": 2.9218260812662345e-06,
      "loss": 0.6361,
      "step": 9294
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34669986367819144,
      "learning_rate": 2.918812496971919e-06,
      "loss": 0.6101,
      "step": 9295
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.40358607583103795,
      "learning_rate": 2.9158003452561656e-06,
      "loss": 0.6705,
      "step": 9296
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37072766845092375,
      "learning_rate": 2.9127896263716037e-06,
      "loss": 0.6254,
      "step": 9297
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3879058645708424,
      "learning_rate": 2.9097803405707405e-06,
      "loss": 0.6213,
      "step": 9298
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37738466938304277,
      "learning_rate": 2.90677248810596e-06,
      "loss": 0.6488,
      "step": 9299
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4132659928851204,
      "learning_rate": 2.903766069229528e-06,
      "loss": 0.6497,
      "step": 9300
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.367392897296839,
      "learning_rate": 2.900761084193593e-06,
      "loss": 0.6132,
      "step": 9301
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3833533642906427,
      "learning_rate": 2.8977575332501716e-06,
      "loss": 0.639,
      "step": 9302
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3806002042090486,
      "learning_rate": 2.8947554166511717e-06,
      "loss": 0.6349,
      "step": 9303
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38618665715747325,
      "learning_rate": 2.89175473464838e-06,
      "loss": 0.6524,
      "step": 9304
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34086108474268256,
      "learning_rate": 2.888755487493453e-06,
      "loss": 0.6115,
      "step": 9305
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3797034849234981,
      "learning_rate": 2.8857576754379345e-06,
      "loss": 0.6425,
      "step": 9306
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43123333093539723,
      "learning_rate": 2.882761298733259e-06,
      "loss": 0.6228,
      "step": 9307
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3523115715544204,
      "learning_rate": 2.879766357630718e-06,
      "loss": 0.5982,
      "step": 9308
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4016862331157527,
      "learning_rate": 2.876772852381495e-06,
      "loss": 0.6459,
      "step": 9309
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4095057447045219,
      "learning_rate": 2.8737807832366595e-06,
      "loss": 0.6567,
      "step": 9310
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36965818129364664,
      "learning_rate": 2.8707901504471404e-06,
      "loss": 0.61,
      "step": 9311
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4417055954304493,
      "learning_rate": 2.867800954263764e-06,
      "loss": 0.6305,
      "step": 9312
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38675925721611626,
      "learning_rate": 2.8648131949372325e-06,
      "loss": 0.6684,
      "step": 9313
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38569772152508475,
      "learning_rate": 2.861826872718123e-06,
      "loss": 0.612,
      "step": 9314
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4260691691992199,
      "learning_rate": 2.8588419878569017e-06,
      "loss": 0.6627,
      "step": 9315
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36686620641685247,
      "learning_rate": 2.8558585406038976e-06,
      "loss": 0.6382,
      "step": 9316
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.402583298164639,
      "learning_rate": 2.8528765312093323e-06,
      "loss": 0.6349,
      "step": 9317
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42784307908203356,
      "learning_rate": 2.8498959599233077e-06,
      "loss": 0.6612,
      "step": 9318
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38755068450630825,
      "learning_rate": 2.8469168269957915e-06,
      "loss": 0.6468,
      "step": 9319
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.521452535973859,
      "learning_rate": 2.843939132676643e-06,
      "loss": 0.3306,
      "step": 9320
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41568390085864065,
      "learning_rate": 2.840962877215607e-06,
      "loss": 0.7012,
      "step": 9321
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3685516078624111,
      "learning_rate": 2.837988060862289e-06,
      "loss": 0.5943,
      "step": 9322
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36015287743448693,
      "learning_rate": 2.835014683866184e-06,
      "loss": 0.6092,
      "step": 9323
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3804778181243941,
      "learning_rate": 2.8320427464766733e-06,
      "loss": 0.6176,
      "step": 9324
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39727031399682344,
      "learning_rate": 2.829072248942999e-06,
      "loss": 0.6051,
      "step": 9325
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.32865451585535693,
      "learning_rate": 2.826103191514298e-06,
      "loss": 0.6235,
      "step": 9326
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37589016539475895,
      "learning_rate": 2.8231355744395815e-06,
      "loss": 0.6453,
      "step": 9327
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4159939159427395,
      "learning_rate": 2.82016939796774e-06,
      "loss": 0.6169,
      "step": 9328
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3793883903995833,
      "learning_rate": 2.8172046623475436e-06,
      "loss": 0.6299,
      "step": 9329
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38208071787689624,
      "learning_rate": 2.814241367827644e-06,
      "loss": 0.6419,
      "step": 9330
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3617518425823954,
      "learning_rate": 2.811279514656562e-06,
      "loss": 0.6225,
      "step": 9331
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4021333920831698,
      "learning_rate": 2.8083191030827083e-06,
      "loss": 0.6421,
      "step": 9332
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3715533302616507,
      "learning_rate": 2.8053601333543736e-06,
      "loss": 0.614,
      "step": 9333
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6942385658119348,
      "learning_rate": 2.8024026057197096e-06,
      "loss": 0.3518,
      "step": 9334
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4012500447708531,
      "learning_rate": 2.7994465204267784e-06,
      "loss": 0.6719,
      "step": 9335
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3543292917404302,
      "learning_rate": 2.7964918777234905e-06,
      "loss": 0.6321,
      "step": 9336
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4068708749133484,
      "learning_rate": 2.7935386778576522e-06,
      "loss": 0.6897,
      "step": 9337
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3734544775936103,
      "learning_rate": 2.79058692107695e-06,
      "loss": 0.5885,
      "step": 9338
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3915062437678885,
      "learning_rate": 2.787636607628934e-06,
      "loss": 0.6802,
      "step": 9339
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.404168718700581,
      "learning_rate": 2.784687737761047e-06,
      "loss": 0.6393,
      "step": 9340
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.45791579128650517,
      "learning_rate": 2.781740311720611e-06,
      "loss": 0.6245,
      "step": 9341
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.46271204546045497,
      "learning_rate": 2.778794329754819e-06,
      "loss": 0.639,
      "step": 9342
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39639660814380584,
      "learning_rate": 2.7758497921107474e-06,
      "loss": 0.6374,
      "step": 9343
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4112345299643687,
      "learning_rate": 2.772906699035358e-06,
      "loss": 0.6506,
      "step": 9344
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43024563194024473,
      "learning_rate": 2.7699650507754763e-06,
      "loss": 0.7038,
      "step": 9345
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3821819674981597,
      "learning_rate": 2.767024847577815e-06,
      "loss": 0.6319,
      "step": 9346
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6482215962244717,
      "learning_rate": 2.7640860896889734e-06,
      "loss": 0.3625,
      "step": 9347
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3970088498012323,
      "learning_rate": 2.761148777355405e-06,
      "loss": 0.6625,
      "step": 9348
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3989065354156734,
      "learning_rate": 2.758212910823479e-06,
      "loss": 0.6245,
      "step": 9349
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42491446953319506,
      "learning_rate": 2.755278490339408e-06,
      "loss": 0.6531,
      "step": 9350
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36391786486325933,
      "learning_rate": 2.7523455161493063e-06,
      "loss": 0.6356,
      "step": 9351
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39826663766531195,
      "learning_rate": 2.749413988499159e-06,
      "loss": 0.6544,
      "step": 9352
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4433670648004283,
      "learning_rate": 2.7464839076348224e-06,
      "loss": 0.6677,
      "step": 9353
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4099518896964439,
      "learning_rate": 2.7435552738020434e-06,
      "loss": 0.5964,
      "step": 9354
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37818598252586405,
      "learning_rate": 2.7406280872464442e-06,
      "loss": 0.6277,
      "step": 9355
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36097638083470895,
      "learning_rate": 2.7377023482135223e-06,
      "loss": 0.6363,
      "step": 9356
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3664015469483397,
      "learning_rate": 2.734778056948657e-06,
      "loss": 0.6079,
      "step": 9357
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4048813838015409,
      "learning_rate": 2.73185521369711e-06,
      "loss": 0.6734,
      "step": 9358
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4214114175819338,
      "learning_rate": 2.728933818704007e-06,
      "loss": 0.628,
      "step": 9359
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3957486633190488,
      "learning_rate": 2.726013872214366e-06,
      "loss": 0.6437,
      "step": 9360
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3641453215797681,
      "learning_rate": 2.7230953744730838e-06,
      "loss": 0.6264,
      "step": 9361
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5397231349519341,
      "learning_rate": 2.7201783257249203e-06,
      "loss": 0.3512,
      "step": 9362
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40579263347937805,
      "learning_rate": 2.7172627262145334e-06,
      "loss": 0.6387,
      "step": 9363
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3518219817695069,
      "learning_rate": 2.714348576186456e-06,
      "loss": 0.6438,
      "step": 9364
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3846685698554148,
      "learning_rate": 2.7114358758850823e-06,
      "loss": 0.6332,
      "step": 9365
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3863750238082032,
      "learning_rate": 2.7085246255547048e-06,
      "loss": 0.6641,
      "step": 9366
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43199560306299095,
      "learning_rate": 2.7056148254394863e-06,
      "loss": 0.5994,
      "step": 9367
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40170975170836526,
      "learning_rate": 2.7027064757834566e-06,
      "loss": 0.6109,
      "step": 9368
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.431877694797561,
      "learning_rate": 2.6997995768305553e-06,
      "loss": 0.6465,
      "step": 9369
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3780853432319936,
      "learning_rate": 2.6968941288245652e-06,
      "loss": 0.6419,
      "step": 9370
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4154999398639617,
      "learning_rate": 2.6939901320091676e-06,
      "loss": 0.6332,
      "step": 9371
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37582889209371423,
      "learning_rate": 2.691087586627923e-06,
      "loss": 0.632,
      "step": 9372
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39452467711966893,
      "learning_rate": 2.688186492924252e-06,
      "loss": 0.6249,
      "step": 9373
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3913518574945614,
      "learning_rate": 2.6852868511414733e-06,
      "loss": 0.6714,
      "step": 9374
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35213935074970454,
      "learning_rate": 2.6823886615227767e-06,
      "loss": 0.6305,
      "step": 9375
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4006176785152176,
      "learning_rate": 2.679491924311226e-06,
      "loss": 0.6172,
      "step": 9376
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36318733271422465,
      "learning_rate": 2.6765966397497712e-06,
      "loss": 0.6326,
      "step": 9377
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33596211919328367,
      "learning_rate": 2.67370280808124e-06,
      "loss": 0.6342,
      "step": 9378
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4549101334202761,
      "learning_rate": 2.670810429548325e-06,
      "loss": 0.651,
      "step": 9379
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.387124380386869,
      "learning_rate": 2.66791950439361e-06,
      "loss": 0.5917,
      "step": 9380
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3916383743469831,
      "learning_rate": 2.6650300328595593e-06,
      "loss": 0.6382,
      "step": 9381
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40516640864545383,
      "learning_rate": 2.6621420151884957e-06,
      "loss": 0.6483,
      "step": 9382
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42329036000392367,
      "learning_rate": 2.659255451622651e-06,
      "loss": 0.6345,
      "step": 9383
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.430615569930347,
      "learning_rate": 2.6563703424041045e-06,
      "loss": 0.6799,
      "step": 9384
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4221696617271051,
      "learning_rate": 2.653486687774831e-06,
      "loss": 0.6298,
      "step": 9385
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4348905882362803,
      "learning_rate": 2.6506044879766823e-06,
      "loss": 0.6656,
      "step": 9386
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41286020530664874,
      "learning_rate": 2.647723743251385e-06,
      "loss": 0.6577,
      "step": 9387
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3873904629422948,
      "learning_rate": 2.644844453840536e-06,
      "loss": 0.6578,
      "step": 9388
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4394246038102725,
      "learning_rate": 2.641966619985623e-06,
      "loss": 0.6764,
      "step": 9389
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3964533355952351,
      "learning_rate": 2.6390902419280084e-06,
      "loss": 0.6575,
      "step": 9390
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38458364100963843,
      "learning_rate": 2.6362153199089257e-06,
      "loss": 0.6397,
      "step": 9391
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3932835744813359,
      "learning_rate": 2.633341854169502e-06,
      "loss": 0.6444,
      "step": 9392
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3903411716764969,
      "learning_rate": 2.630469844950716e-06,
      "loss": 0.666,
      "step": 9393
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42083907645097185,
      "learning_rate": 2.627599292493448e-06,
      "loss": 0.6257,
      "step": 9394
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39281610979235876,
      "learning_rate": 2.624730197038452e-06,
      "loss": 0.6624,
      "step": 9395
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40437895999102175,
      "learning_rate": 2.62186255882634e-06,
      "loss": 0.6641,
      "step": 9396
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37804173316907913,
      "learning_rate": 2.6189963780976337e-06,
      "loss": 0.5898,
      "step": 9397
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4206008084508404,
      "learning_rate": 2.616131655092713e-06,
      "loss": 0.6127,
      "step": 9398
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3820497921412183,
      "learning_rate": 2.613268390051833e-06,
      "loss": 0.6412,
      "step": 9399
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39225667004743087,
      "learning_rate": 2.610406583215135e-06,
      "loss": 0.6473,
      "step": 9400
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4595504004238127,
      "learning_rate": 2.6075462348226378e-06,
      "loss": 0.699,
      "step": 9401
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3919977273179385,
      "learning_rate": 2.6046873451142317e-06,
      "loss": 0.6201,
      "step": 9402
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3580009031857855,
      "learning_rate": 2.6018299143296878e-06,
      "loss": 0.6375,
      "step": 9403
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3649777876906394,
      "learning_rate": 2.598973942708658e-06,
      "loss": 0.6619,
      "step": 9404
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38832693118085015,
      "learning_rate": 2.596119430490669e-06,
      "loss": 0.5985,
      "step": 9405
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34034202752099474,
      "learning_rate": 2.593266377915129e-06,
      "loss": 0.6318,
      "step": 9406
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32172070209468706,
      "learning_rate": 2.5904147852213115e-06,
      "loss": 0.6129,
      "step": 9407
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3792102981060469,
      "learning_rate": 2.58756465264838e-06,
      "loss": 0.6463,
      "step": 9408
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4088393785624794,
      "learning_rate": 2.584715980435377e-06,
      "loss": 0.6487,
      "step": 9409
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38173778675789094,
      "learning_rate": 2.581868768821203e-06,
      "loss": 0.6401,
      "step": 9410
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4104814502285524,
      "learning_rate": 2.579023018044664e-06,
      "loss": 0.6319,
      "step": 9411
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37619994788468836,
      "learning_rate": 2.5761787283444293e-06,
      "loss": 0.6442,
      "step": 9412
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39658601145748157,
      "learning_rate": 2.57333589995904e-06,
      "loss": 0.6408,
      "step": 9413
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3440634151246426,
      "learning_rate": 2.5704945331269217e-06,
      "loss": 0.6526,
      "step": 9414
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4298592006275715,
      "learning_rate": 2.56765462808638e-06,
      "loss": 0.6495,
      "step": 9415
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3812689340643122,
      "learning_rate": 2.5648161850755894e-06,
      "loss": 0.6524,
      "step": 9416
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39643861985100687,
      "learning_rate": 2.5619792043326075e-06,
      "loss": 0.6518,
      "step": 9417
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3752620373244215,
      "learning_rate": 2.5591436860953713e-06,
      "loss": 0.6464,
      "step": 9418
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3757139529624021,
      "learning_rate": 2.55630963060169e-06,
      "loss": 0.6315,
      "step": 9419
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.395775286209588,
      "learning_rate": 2.5534770380892516e-06,
      "loss": 0.6576,
      "step": 9420
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.383689412331667,
      "learning_rate": 2.55064590879563e-06,
      "loss": 0.6501,
      "step": 9421
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44368915526269986,
      "learning_rate": 2.5478162429582588e-06,
      "loss": 0.6514,
      "step": 9422
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35904180793125956,
      "learning_rate": 2.5449880408144624e-06,
      "loss": 0.6477,
      "step": 9423
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3489245760845706,
      "learning_rate": 2.542161302601438e-06,
      "loss": 0.6015,
      "step": 9424
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4523170678087906,
      "learning_rate": 2.539336028556263e-06,
      "loss": 0.6697,
      "step": 9425
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3820626979525061,
      "learning_rate": 2.5365122189158898e-06,
      "loss": 0.6667,
      "step": 9426
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4177259430470233,
      "learning_rate": 2.5336898739171444e-06,
      "loss": 0.6578,
      "step": 9427
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4171828601094885,
      "learning_rate": 2.5308689937967334e-06,
      "loss": 0.6352,
      "step": 9428
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34594328105820793,
      "learning_rate": 2.528049578791247e-06,
      "loss": 0.6253,
      "step": 9429
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3821087772445881,
      "learning_rate": 2.525231629137139e-06,
      "loss": 0.5974,
      "step": 9430
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3776171169827115,
      "learning_rate": 2.5224151450707446e-06,
      "loss": 0.6414,
      "step": 9431
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4332245328689044,
      "learning_rate": 2.519600126828292e-06,
      "loss": 0.649,
      "step": 9432
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3894272416781094,
      "learning_rate": 2.5167865746458643e-06,
      "loss": 0.6539,
      "step": 9433
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41200340601088453,
      "learning_rate": 2.5139744887594296e-06,
      "loss": 0.6102,
      "step": 9434
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3419957525212094,
      "learning_rate": 2.5111638694048423e-06,
      "loss": 0.6079,
      "step": 9435
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37058419594070713,
      "learning_rate": 2.508354716817816e-06,
      "loss": 0.659,
      "step": 9436
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3467030275279212,
      "learning_rate": 2.5055470312339548e-06,
      "loss": 0.6596,
      "step": 9437
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40278512682198436,
      "learning_rate": 2.5027408128887354e-06,
      "loss": 0.6409,
      "step": 9438
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38057842608375936,
      "learning_rate": 2.499936062017514e-06,
      "loss": 0.6223,
      "step": 9439
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41956928186858033,
      "learning_rate": 2.4971327788555222e-06,
      "loss": 0.6732,
      "step": 9440
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36516241652543474,
      "learning_rate": 2.494330963637863e-06,
      "loss": 0.6226,
      "step": 9441
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3835381149713839,
      "learning_rate": 2.491530616599527e-06,
      "loss": 0.6569,
      "step": 9442
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36683669947108344,
      "learning_rate": 2.4887317379753697e-06,
      "loss": 0.6609,
      "step": 9443
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4099993418130601,
      "learning_rate": 2.485934328000139e-06,
      "loss": 0.6505,
      "step": 9444
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43215221895174,
      "learning_rate": 2.483138386908437e-06,
      "loss": 0.6726,
      "step": 9445
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5617200017513275,
      "learning_rate": 2.4803439149347708e-06,
      "loss": 0.6359,
      "step": 9446
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3664144235862583,
      "learning_rate": 2.477550912313498e-06,
      "loss": 0.6114,
      "step": 9447
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41236716815300295,
      "learning_rate": 2.4747593792788684e-06,
      "loss": 0.6109,
      "step": 9448
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4022790159544066,
      "learning_rate": 2.4719693160650084e-06,
      "loss": 0.6182,
      "step": 9449
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4093576714936662,
      "learning_rate": 2.469180722905911e-06,
      "loss": 0.6636,
      "step": 9450
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3310380704940746,
      "learning_rate": 2.466393600035448e-06,
      "loss": 0.6157,
      "step": 9451
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38799188533171686,
      "learning_rate": 2.4636079476873877e-06,
      "loss": 0.6691,
      "step": 9452
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38235503015147043,
      "learning_rate": 2.460823766095346e-06,
      "loss": 0.6273,
      "step": 9453
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4460709905631803,
      "learning_rate": 2.4580410554928325e-06,
      "loss": 0.6328,
      "step": 9454
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3732040374354682,
      "learning_rate": 2.4552598161132355e-06,
      "loss": 0.6337,
      "step": 9455
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3779744897460809,
      "learning_rate": 2.4524800481898047e-06,
      "loss": 0.6552,
      "step": 9456
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3496195223892286,
      "learning_rate": 2.44970175195568e-06,
      "loss": 0.6121,
      "step": 9457
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5970728323806249,
      "learning_rate": 2.4469249276438743e-06,
      "loss": 0.3733,
      "step": 9458
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41075181723422455,
      "learning_rate": 2.4441495754872756e-06,
      "loss": 0.6839,
      "step": 9459
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.418416539672194,
      "learning_rate": 2.441375695718655e-06,
      "loss": 0.6472,
      "step": 9460
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4018969211711221,
      "learning_rate": 2.4386032885706447e-06,
      "loss": 0.6194,
      "step": 9461
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.411849183168733,
      "learning_rate": 2.435832354275769e-06,
      "loss": 0.632,
      "step": 9462
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40569360805143206,
      "learning_rate": 2.4330628930664248e-06,
      "loss": 0.6352,
      "step": 9463
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41704873093243683,
      "learning_rate": 2.430294905174875e-06,
      "loss": 0.6444,
      "step": 9464
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41567537472716526,
      "learning_rate": 2.427528390833269e-06,
      "loss": 0.6898,
      "step": 9465
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38100428312576196,
      "learning_rate": 2.4247633502736423e-06,
      "loss": 0.6365,
      "step": 9466
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41520101895512673,
      "learning_rate": 2.4219997837278863e-06,
      "loss": 0.6578,
      "step": 9467
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41017780634174217,
      "learning_rate": 2.419237691427778e-06,
      "loss": 0.6089,
      "step": 9468
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4197835781057121,
      "learning_rate": 2.416477073604977e-06,
      "loss": 0.6523,
      "step": 9469
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37959166106474224,
      "learning_rate": 2.413717930491002e-06,
      "loss": 0.6253,
      "step": 9470
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3939398888414501,
      "learning_rate": 2.410960262317268e-06,
      "loss": 0.6117,
      "step": 9471
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35657856041339275,
      "learning_rate": 2.4082040693150542e-06,
      "loss": 0.6038,
      "step": 9472
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36736279425657303,
      "learning_rate": 2.4054493517155187e-06,
      "loss": 0.6493,
      "step": 9473
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4324522766033738,
      "learning_rate": 2.4026961097497026e-06,
      "loss": 0.7075,
      "step": 9474
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3982941257850371,
      "learning_rate": 2.399944343648506e-06,
      "loss": 0.6302,
      "step": 9475
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3772053632441035,
      "learning_rate": 2.397194053642722e-06,
      "loss": 0.6204,
      "step": 9476
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4049555226315867,
      "learning_rate": 2.394445239963015e-06,
      "loss": 0.6457,
      "step": 9477
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3971687212286916,
      "learning_rate": 2.391697902839927e-06,
      "loss": 0.6617,
      "step": 9478
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4220038424104673,
      "learning_rate": 2.3889520425038624e-06,
      "loss": 0.6346,
      "step": 9479
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.39812253564125843,
      "learning_rate": 2.3862076591851293e-06,
      "loss": 0.6332,
      "step": 9480
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3776005348651291,
      "learning_rate": 2.383464753113882e-06,
      "loss": 0.5797,
      "step": 9481
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4123175582041569,
      "learning_rate": 2.3807233245201733e-06,
      "loss": 0.6647,
      "step": 9482
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.556292491230713,
      "learning_rate": 2.377983373633923e-06,
      "loss": 0.3831,
      "step": 9483
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3884876243341743,
      "learning_rate": 2.375244900684923e-06,
      "loss": 0.6378,
      "step": 9484
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3879660839864328,
      "learning_rate": 2.3725079059028456e-06,
      "loss": 0.637,
      "step": 9485
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38823032982423483,
      "learning_rate": 2.369772389517242e-06,
      "loss": 0.6286,
      "step": 9486
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3619487151081129,
      "learning_rate": 2.367038351757538e-06,
      "loss": 0.6377,
      "step": 9487
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43382921098710875,
      "learning_rate": 2.36430579285303e-06,
      "loss": 0.6677,
      "step": 9488
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43102690190321935,
      "learning_rate": 2.3615747130329013e-06,
      "loss": 0.6662,
      "step": 9489
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.39395748754831855,
      "learning_rate": 2.358845112526196e-06,
      "loss": 0.6773,
      "step": 9490
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3972140789313807,
      "learning_rate": 2.3561169915618455e-06,
      "loss": 0.6157,
      "step": 9491
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36678903899890436,
      "learning_rate": 2.3533903503686585e-06,
      "loss": 0.6313,
      "step": 9492
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.39316490494709144,
      "learning_rate": 2.350665189175303e-06,
      "loss": 0.5863,
      "step": 9493
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3364400146499346,
      "learning_rate": 2.347941508210352e-06,
      "loss": 0.6032,
      "step": 9494
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3650832675333902,
      "learning_rate": 2.3452193077022245e-06,
      "loss": 0.6268,
      "step": 9495
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3889389530843826,
      "learning_rate": 2.3424985878792317e-06,
      "loss": 0.6402,
      "step": 9496
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3895432628468296,
      "learning_rate": 2.339779348969562e-06,
      "loss": 0.6366,
      "step": 9497
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3792919024099338,
      "learning_rate": 2.337061591201264e-06,
      "loss": 0.6142,
      "step": 9498
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3884777665857025,
      "learning_rate": 2.33434531480228e-06,
      "loss": 0.6261,
      "step": 9499
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38819891164562786,
      "learning_rate": 2.331630520000421e-06,
      "loss": 0.6386,
      "step": 9500
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38179963578050075,
      "learning_rate": 2.328917207023371e-06,
      "loss": 0.6145,
      "step": 9501
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40157852434912966,
      "learning_rate": 2.3262053760986936e-06,
      "loss": 0.6151,
      "step": 9502
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40702609746337154,
      "learning_rate": 2.3234950274538302e-06,
      "loss": 0.6456,
      "step": 9503
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3748944773392223,
      "learning_rate": 2.3207861613160885e-06,
      "loss": 0.6384,
      "step": 9504
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34081169814235995,
      "learning_rate": 2.31807877791266e-06,
      "loss": 0.6251,
      "step": 9505
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3673198984556771,
      "learning_rate": 2.315372877470614e-06,
      "loss": 0.651,
      "step": 9506
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36208022978837423,
      "learning_rate": 2.312668460216878e-06,
      "loss": 0.646,
      "step": 9507
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3822861020595651,
      "learning_rate": 2.309965526378284e-06,
      "loss": 0.6463,
      "step": 9508
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4122728269393771,
      "learning_rate": 2.307264076181519e-06,
      "loss": 0.6731,
      "step": 9509
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4232068768821799,
      "learning_rate": 2.3045641098531448e-06,
      "loss": 0.6201,
      "step": 9510
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3432970972195774,
      "learning_rate": 2.3018656276196083e-06,
      "loss": 0.6227,
      "step": 9511
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3708890821245883,
      "learning_rate": 2.2991686297072314e-06,
      "loss": 0.6166,
      "step": 9512
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4554871109508235,
      "learning_rate": 2.2964731163422016e-06,
      "loss": 0.6877,
      "step": 9513
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4132714514922408,
      "learning_rate": 2.2937790877505894e-06,
      "loss": 0.682,
      "step": 9514
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34335779788256837,
      "learning_rate": 2.291086544158343e-06,
      "loss": 0.6332,
      "step": 9515
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3504824080905409,
      "learning_rate": 2.2883954857912796e-06,
      "loss": 0.6577,
      "step": 9516
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3134685694914176,
      "learning_rate": 2.2857059128751e-06,
      "loss": 0.5855,
      "step": 9517
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.45362428562912094,
      "learning_rate": 2.2830178256353698e-06,
      "loss": 0.6204,
      "step": 9518
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40110342141087807,
      "learning_rate": 2.2803312242975383e-06,
      "loss": 0.6576,
      "step": 9519
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41131334338101017,
      "learning_rate": 2.2776461090869304e-06,
      "loss": 0.6137,
      "step": 9520
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41242411869723083,
      "learning_rate": 2.2749624802287307e-06,
      "loss": 0.6614,
      "step": 9521
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3859231056767469,
      "learning_rate": 2.272280337948027e-06,
      "loss": 0.6066,
      "step": 9522
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.42185391002606176,
      "learning_rate": 2.269599682469765e-06,
      "loss": 0.6584,
      "step": 9523
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4078582516243538,
      "learning_rate": 2.26692051401876e-06,
      "loss": 0.669,
      "step": 9524
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.45691311291349274,
      "learning_rate": 2.2642428328197164e-06,
      "loss": 0.6261,
      "step": 9525
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41716831065142773,
      "learning_rate": 2.2615666390972126e-06,
      "loss": 0.6466,
      "step": 9526
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3884071933610098,
      "learning_rate": 2.2588919330756866e-06,
      "loss": 0.6491,
      "step": 9527
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37467907723930716,
      "learning_rate": 2.256218714979468e-06,
      "loss": 0.6425,
      "step": 9528
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4226507668735887,
      "learning_rate": 2.2535469850327575e-06,
      "loss": 0.6601,
      "step": 9529
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3943262837534145,
      "learning_rate": 2.2508767434596267e-06,
      "loss": 0.662,
      "step": 9530
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35660503108591196,
      "learning_rate": 2.2482079904840348e-06,
      "loss": 0.6083,
      "step": 9531
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3254008744575738,
      "learning_rate": 2.2455407263297957e-06,
      "loss": 0.6293,
      "step": 9532
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3515297632403925,
      "learning_rate": 2.2428749512206127e-06,
      "loss": 0.6019,
      "step": 9533
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.403774798413687,
      "learning_rate": 2.2402106653800626e-06,
      "loss": 0.6109,
      "step": 9534
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38962210800787456,
      "learning_rate": 2.2375478690315957e-06,
      "loss": 0.6018,
      "step": 9535
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35324760058588034,
      "learning_rate": 2.234886562398537e-06,
      "loss": 0.6154,
      "step": 9536
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3575140397492035,
      "learning_rate": 2.2322267457040935e-06,
      "loss": 0.6276,
      "step": 9537
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38529349484799286,
      "learning_rate": 2.2295684191713286e-06,
      "loss": 0.6244,
      "step": 9538
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3623685083689534,
      "learning_rate": 2.226911583023199e-06,
      "loss": 0.6636,
      "step": 9539
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36604675453008056,
      "learning_rate": 2.224256237482536e-06,
      "loss": 0.5859,
      "step": 9540
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38939833353711173,
      "learning_rate": 2.2216023827720256e-06,
      "loss": 0.6399,
      "step": 9541
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43755181947284216,
      "learning_rate": 2.2189500191142566e-06,
      "loss": 0.6619,
      "step": 9542
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3972623510023357,
      "learning_rate": 2.216299146731682e-06,
      "loss": 0.6616,
      "step": 9543
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3730074434306991,
      "learning_rate": 2.2136497658466127e-06,
      "loss": 0.6387,
      "step": 9544
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3850341940314014,
      "learning_rate": 2.2110018766812602e-06,
      "loss": 0.6494,
      "step": 9545
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4030433585930942,
      "learning_rate": 2.208355479457698e-06,
      "loss": 0.6463,
      "step": 9546
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3548238413480553,
      "learning_rate": 2.2057105743978725e-06,
      "loss": 0.6175,
      "step": 9547
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40606906791091824,
      "learning_rate": 2.203067161723611e-06,
      "loss": 0.6449,
      "step": 9548
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3817459268529194,
      "learning_rate": 2.200425241656614e-06,
      "loss": 0.6496,
      "step": 9549
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4920859606225971,
      "learning_rate": 2.1977848144184535e-06,
      "loss": 0.6609,
      "step": 9550
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3975486664262601,
      "learning_rate": 2.195145880230587e-06,
      "loss": 0.6247,
      "step": 9551
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4113118541059737,
      "learning_rate": 2.192508439314327e-06,
      "loss": 0.658,
      "step": 9552
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3866382113797613,
      "learning_rate": 2.1898724918908786e-06,
      "loss": 0.6453,
      "step": 9553
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3437535615563323,
      "learning_rate": 2.1872380381813207e-06,
      "loss": 0.6151,
      "step": 9554
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38105778618390584,
      "learning_rate": 2.1846050784065875e-06,
      "loss": 0.6414,
      "step": 9555
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40625751832854395,
      "learning_rate": 2.1819736127875137e-06,
      "loss": 0.6468,
      "step": 9556
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3813883212425264,
      "learning_rate": 2.179343641544798e-06,
      "loss": 0.6746,
      "step": 9557
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3980182220833444,
      "learning_rate": 2.176715164899006e-06,
      "loss": 0.6493,
      "step": 9558
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4185562628403926,
      "learning_rate": 2.1740881830705884e-06,
      "loss": 0.6557,
      "step": 9559
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38892837641147493,
      "learning_rate": 2.171462696279871e-06,
      "loss": 0.6183,
      "step": 9560
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4219571276477524,
      "learning_rate": 2.1688387047470404e-06,
      "loss": 0.6825,
      "step": 9561
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3371056147140001,
      "learning_rate": 2.1662162086921733e-06,
      "loss": 0.6396,
      "step": 9562
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.42062347098908026,
      "learning_rate": 2.163595208335216e-06,
      "loss": 0.654,
      "step": 9563
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4124975795826338,
      "learning_rate": 2.1609757038959887e-06,
      "loss": 0.6681,
      "step": 9564
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41864407740493326,
      "learning_rate": 2.158357695594182e-06,
      "loss": 0.602,
      "step": 9565
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4363271458683408,
      "learning_rate": 2.1557411836493735e-06,
      "loss": 0.6341,
      "step": 9566
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4146264321948832,
      "learning_rate": 2.1531261682809988e-06,
      "loss": 0.6656,
      "step": 9567
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36501682675301556,
      "learning_rate": 2.1505126497083783e-06,
      "loss": 0.6767,
      "step": 9568
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40278547058906544,
      "learning_rate": 2.1479006281507032e-06,
      "loss": 0.6635,
      "step": 9569
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38199242673118333,
      "learning_rate": 2.1452901038270445e-06,
      "loss": 0.6246,
      "step": 9570
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38560281327495305,
      "learning_rate": 2.1426810769563457e-06,
      "loss": 0.6468,
      "step": 9571
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4089882988375371,
      "learning_rate": 2.140073547757413e-06,
      "loss": 0.6389,
      "step": 9572
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3916309702126256,
      "learning_rate": 2.1374675164489454e-06,
      "loss": 0.6657,
      "step": 9573
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3533337918394356,
      "learning_rate": 2.134862983249506e-06,
      "loss": 0.5939,
      "step": 9574
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41360010225796523,
      "learning_rate": 2.132259948377531e-06,
      "loss": 0.6179,
      "step": 9575
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41502347681379126,
      "learning_rate": 2.12965841205133e-06,
      "loss": 0.6142,
      "step": 9576
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3858816047976051,
      "learning_rate": 2.127058374489104e-06,
      "loss": 0.6233,
      "step": 9577
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38759238048563455,
      "learning_rate": 2.124459835908903e-06,
      "loss": 0.644,
      "step": 9578
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4221357811287132,
      "learning_rate": 2.1218627965286664e-06,
      "loss": 0.699,
      "step": 9579
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.558772728819646,
      "learning_rate": 2.119267256566211e-06,
      "loss": 0.333,
      "step": 9580
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36191482449940215,
      "learning_rate": 2.1166732162392134e-06,
      "loss": 0.5801,
      "step": 9581
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39667471027427337,
      "learning_rate": 2.1140806757652334e-06,
      "loss": 0.6504,
      "step": 9582
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3580172497631336,
      "learning_rate": 2.1114896353617074e-06,
      "loss": 0.6367,
      "step": 9583
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4102855844457624,
      "learning_rate": 2.1089000952459427e-06,
      "loss": 0.6612,
      "step": 9584
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4265661917718217,
      "learning_rate": 2.1063120556351225e-06,
      "loss": 0.6198,
      "step": 9585
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3933710946861989,
      "learning_rate": 2.1037255167462954e-06,
      "loss": 0.6279,
      "step": 9586
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36508963662558963,
      "learning_rate": 2.1011404787963975e-06,
      "loss": 0.604,
      "step": 9587
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38934464136397234,
      "learning_rate": 2.0985569420022366e-06,
      "loss": 0.6277,
      "step": 9588
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4100114182464772,
      "learning_rate": 2.0959749065804803e-06,
      "loss": 0.6574,
      "step": 9589
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.403966701457644,
      "learning_rate": 2.093394372747681e-06,
      "loss": 0.6776,
      "step": 9590
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41357468982134654,
      "learning_rate": 2.090815340720278e-06,
      "loss": 0.6429,
      "step": 9591
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3645523301201165,
      "learning_rate": 2.088237810714562e-06,
      "loss": 0.6266,
      "step": 9592
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4283871472831142,
      "learning_rate": 2.085661782946706e-06,
      "loss": 0.6544,
      "step": 9593
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4130801258908318,
      "learning_rate": 2.0830872576327652e-06,
      "loss": 0.6832,
      "step": 9594
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3973054785833224,
      "learning_rate": 2.080514234988655e-06,
      "loss": 0.6027,
      "step": 9595
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35857534194894225,
      "learning_rate": 2.0779427152301766e-06,
      "loss": 0.6091,
      "step": 9596
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.428236393676321,
      "learning_rate": 2.075372698572995e-06,
      "loss": 0.6843,
      "step": 9597
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3902359651239232,
      "learning_rate": 2.0728041852326597e-06,
      "loss": 0.6606,
      "step": 9598
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40577305445651046,
      "learning_rate": 2.0702371754245875e-06,
      "loss": 0.6306,
      "step": 9599
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38331653526543924,
      "learning_rate": 2.067671669364071e-06,
      "loss": 0.6063,
      "step": 9600
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4111117959195694,
      "learning_rate": 2.0651076672662727e-06,
      "loss": 0.6247,
      "step": 9601
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3936997633513213,
      "learning_rate": 2.062545169346235e-06,
      "loss": 0.6182,
      "step": 9602
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4124661925226277,
      "learning_rate": 2.0599841758188742e-06,
      "loss": 0.6249,
      "step": 9603
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.380432813331573,
      "learning_rate": 2.0574246868989658e-06,
      "loss": 0.6131,
      "step": 9604
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3630660521275188,
      "learning_rate": 2.0548667028011883e-06,
      "loss": 0.6121,
      "step": 9605
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4287816455202489,
      "learning_rate": 2.052310223740064e-06,
      "loss": 0.6334,
      "step": 9606
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4066773639750855,
      "learning_rate": 2.0497552499300054e-06,
      "loss": 0.6211,
      "step": 9607
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4219385592351442,
      "learning_rate": 2.047201781585302e-06,
      "loss": 0.6404,
      "step": 9608
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3592469309248957,
      "learning_rate": 2.044649818920097e-06,
      "loss": 0.6348,
      "step": 9609
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3383608049406182,
      "learning_rate": 2.0420993621484285e-06,
      "loss": 0.6365,
      "step": 9610
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3888062622667818,
      "learning_rate": 2.039550411484199e-06,
      "loss": 0.6514,
      "step": 9611
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.579499358765538,
      "learning_rate": 2.037002967141184e-06,
      "loss": 0.3326,
      "step": 9612
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36146334608626496,
      "learning_rate": 2.034457029333039e-06,
      "loss": 0.6409,
      "step": 9613
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3962155817252462,
      "learning_rate": 2.031912598273289e-06,
      "loss": 0.6546,
      "step": 9614
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3758473799216981,
      "learning_rate": 2.0293696741753256e-06,
      "loss": 0.6213,
      "step": 9615
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.43007745981472684,
      "learning_rate": 2.0268282572524223e-06,
      "loss": 0.6779,
      "step": 9616
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3724373283677475,
      "learning_rate": 2.0242883477177287e-06,
      "loss": 0.6258,
      "step": 9617
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6397343715454656,
      "learning_rate": 2.021749945784264e-06,
      "loss": 0.3423,
      "step": 9618
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3288521031806091,
      "learning_rate": 2.019213051664921e-06,
      "loss": 0.6258,
      "step": 9619
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37900009308773697,
      "learning_rate": 2.0166776655724595e-06,
      "loss": 0.6571,
      "step": 9620
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3723302019039987,
      "learning_rate": 2.014143787719525e-06,
      "loss": 0.6023,
      "step": 9621
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3468991245706609,
      "learning_rate": 2.011611418318631e-06,
      "loss": 0.572,
      "step": 9622
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4261868845206999,
      "learning_rate": 2.0090805575821637e-06,
      "loss": 0.6806,
      "step": 9623
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4363793562052315,
      "learning_rate": 2.006551205722376e-06,
      "loss": 0.663,
      "step": 9624
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34977798874969424,
      "learning_rate": 2.004023362951417e-06,
      "loss": 0.5934,
      "step": 9625
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4041042632684715,
      "learning_rate": 2.0014970294812807e-06,
      "loss": 0.6288,
      "step": 9626
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35671294593010333,
      "learning_rate": 1.99897220552385e-06,
      "loss": 0.5958,
      "step": 9627
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3718206627386042,
      "learning_rate": 1.996448891290885e-06,
      "loss": 0.6713,
      "step": 9628
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40978664420638033,
      "learning_rate": 1.9939270869940074e-06,
      "loss": 0.6665,
      "step": 9629
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38583986147751503,
      "learning_rate": 1.991406792844717e-06,
      "loss": 0.6709,
      "step": 9630
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3700621696110798,
      "learning_rate": 1.9888880090543907e-06,
      "loss": 0.6422,
      "step": 9631
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4029675620834309,
      "learning_rate": 1.9863707358342753e-06,
      "loss": 0.6342,
      "step": 9632
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4155849853992864,
      "learning_rate": 1.9838549733954913e-06,
      "loss": 0.6321,
      "step": 9633
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.432015630224607,
      "learning_rate": 1.9813407219490346e-06,
      "loss": 0.6756,
      "step": 9634
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41227515808240717,
      "learning_rate": 1.978827981705769e-06,
      "loss": 0.6519,
      "step": 9635
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3547581020533248,
      "learning_rate": 1.976316752876435e-06,
      "loss": 0.6221,
      "step": 9636
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.384236313302828,
      "learning_rate": 1.9738070356716534e-06,
      "loss": 0.6128,
      "step": 9637
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41054655155314335,
      "learning_rate": 1.9712988303018977e-06,
      "loss": 0.5839,
      "step": 9638
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4012452456956974,
      "learning_rate": 1.968792136977542e-06,
      "loss": 0.6269,
      "step": 9639
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41531665385443944,
      "learning_rate": 1.9662869559088116e-06,
      "loss": 0.6269,
      "step": 9640
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39896016143828317,
      "learning_rate": 1.963783287305814e-06,
      "loss": 0.6123,
      "step": 9641
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39998594181095615,
      "learning_rate": 1.961281131378534e-06,
      "loss": 0.6509,
      "step": 9642
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.433923015924447,
      "learning_rate": 1.9587804883368156e-06,
      "loss": 0.6673,
      "step": 9643
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41228325352023776,
      "learning_rate": 1.9562813583903904e-06,
      "loss": 0.6203,
      "step": 9644
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38248774074717823,
      "learning_rate": 1.953783741748856e-06,
      "loss": 0.6153,
      "step": 9645
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37585649327596743,
      "learning_rate": 1.9512876386216838e-06,
      "loss": 0.6238,
      "step": 9646
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4219907281082755,
      "learning_rate": 1.9487930492182206e-06,
      "loss": 0.6893,
      "step": 9647
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3736021690407515,
      "learning_rate": 1.9462999737476897e-06,
      "loss": 0.5851,
      "step": 9648
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3885940788461833,
      "learning_rate": 1.9438084124191703e-06,
      "loss": 0.6166,
      "step": 9649
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3885797909236319,
      "learning_rate": 1.9413183654416336e-06,
      "loss": 0.6379,
      "step": 9650
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38594329028195135,
      "learning_rate": 1.9388298330239207e-06,
      "loss": 0.5935,
      "step": 9651
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.346620427248502,
      "learning_rate": 1.93634281537473e-06,
      "loss": 0.5944,
      "step": 9652
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4065599125202095,
      "learning_rate": 1.9338573127026583e-06,
      "loss": 0.6027,
      "step": 9653
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44343500949915476,
      "learning_rate": 1.9313733252161526e-06,
      "loss": 0.6254,
      "step": 9654
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37867347954164415,
      "learning_rate": 1.928890853123546e-06,
      "loss": 0.6379,
      "step": 9655
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4167685688510639,
      "learning_rate": 1.926409896633037e-06,
      "loss": 0.6385,
      "step": 9656
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3825195530258408,
      "learning_rate": 1.923930455952707e-06,
      "loss": 0.6329,
      "step": 9657
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40826942972079017,
      "learning_rate": 1.921452531290495e-06,
      "loss": 0.6414,
      "step": 9658
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44040321031671176,
      "learning_rate": 1.9189761228542282e-06,
      "loss": 0.6701,
      "step": 9659
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.383227593385256,
      "learning_rate": 1.9165012308515953e-06,
      "loss": 0.6225,
      "step": 9660
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37397037179960796,
      "learning_rate": 1.914027855490166e-06,
      "loss": 0.6618,
      "step": 9661
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4063696872889069,
      "learning_rate": 1.9115559969773812e-06,
      "loss": 0.6709,
      "step": 9662
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3719683434837992,
      "learning_rate": 1.909085655520546e-06,
      "loss": 0.6664,
      "step": 9663
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4524112469564943,
      "learning_rate": 1.9066168313268485e-06,
      "loss": 0.6503,
      "step": 9664
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38884436220832563,
      "learning_rate": 1.9041495246033492e-06,
      "loss": 0.6294,
      "step": 9665
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3997484972834933,
      "learning_rate": 1.9016837355569696e-06,
      "loss": 0.6261,
      "step": 9666
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4134970803563094,
      "learning_rate": 1.8992194643945194e-06,
      "loss": 0.632,
      "step": 9667
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3994546071728687,
      "learning_rate": 1.8967567113226782e-06,
      "loss": 0.6424,
      "step": 9668
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4011328136830771,
      "learning_rate": 1.8942954765479826e-06,
      "loss": 0.6305,
      "step": 9669
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3910753104711504,
      "learning_rate": 1.8918357602768611e-06,
      "loss": 0.6454,
      "step": 9670
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3989037340010975,
      "learning_rate": 1.8893775627156062e-06,
      "loss": 0.6187,
      "step": 9671
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4166992047755368,
      "learning_rate": 1.8869208840703823e-06,
      "loss": 0.6438,
      "step": 9672
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41026966166688056,
      "learning_rate": 1.8844657245472287e-06,
      "loss": 0.6468,
      "step": 9673
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3648637267293363,
      "learning_rate": 1.882012084352054e-06,
      "loss": 0.5957,
      "step": 9674
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42052434655504095,
      "learning_rate": 1.879559963690647e-06,
      "loss": 0.6608,
      "step": 9675
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3943996999030349,
      "learning_rate": 1.8771093627686653e-06,
      "loss": 0.6453,
      "step": 9676
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43985447101976705,
      "learning_rate": 1.8746602817916315e-06,
      "loss": 0.6355,
      "step": 9677
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39259946838540344,
      "learning_rate": 1.8722127209649476e-06,
      "loss": 0.6342,
      "step": 9678
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.365544201317594,
      "learning_rate": 1.869766680493892e-06,
      "loss": 0.6562,
      "step": 9679
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37365657772217287,
      "learning_rate": 1.867322160583609e-06,
      "loss": 0.6424,
      "step": 9680
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4264455018255255,
      "learning_rate": 1.8648791614391172e-06,
      "loss": 0.662,
      "step": 9681
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4139554551155716,
      "learning_rate": 1.8624376832653125e-06,
      "loss": 0.6411,
      "step": 9682
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.363399538929682,
      "learning_rate": 1.859997726266951e-06,
      "loss": 0.5858,
      "step": 9683
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3969998628889357,
      "learning_rate": 1.8575592906486717e-06,
      "loss": 0.6564,
      "step": 9684
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3704529718430886,
      "learning_rate": 1.8551223766149884e-06,
      "loss": 0.6515,
      "step": 9685
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3795945699776216,
      "learning_rate": 1.8526869843702754e-06,
      "loss": 0.6696,
      "step": 9686
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3684940551652808,
      "learning_rate": 1.8502531141187829e-06,
      "loss": 0.5951,
      "step": 9687
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4454125366750561,
      "learning_rate": 1.847820766064652e-06,
      "loss": 0.6892,
      "step": 9688
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44963851900225543,
      "learning_rate": 1.8453899404118658e-06,
      "loss": 0.6609,
      "step": 9689
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4212208012189099,
      "learning_rate": 1.8429606373642994e-06,
      "loss": 0.6425,
      "step": 9690
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43746813391384026,
      "learning_rate": 1.8405328571257008e-06,
      "loss": 0.618,
      "step": 9691
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4146501128620556,
      "learning_rate": 1.8381065998996783e-06,
      "loss": 0.6178,
      "step": 9692
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43434261831727133,
      "learning_rate": 1.8356818658897203e-06,
      "loss": 0.6767,
      "step": 9693
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38981935592338407,
      "learning_rate": 1.833258655299186e-06,
      "loss": 0.6421,
      "step": 9694
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42511955832378845,
      "learning_rate": 1.8308369683313089e-06,
      "loss": 0.6514,
      "step": 9695
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3677742760008382,
      "learning_rate": 1.828416805189197e-06,
      "loss": 0.5998,
      "step": 9696
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38628314429249183,
      "learning_rate": 1.8259981660758176e-06,
      "loss": 0.6672,
      "step": 9697
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36965649570685377,
      "learning_rate": 1.8235810511940232e-06,
      "loss": 0.6294,
      "step": 9698
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3913324188731941,
      "learning_rate": 1.821165460746539e-06,
      "loss": 0.6467,
      "step": 9699
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3927969246099352,
      "learning_rate": 1.8187513949359493e-06,
      "loss": 0.6566,
      "step": 9700
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40285638295821546,
      "learning_rate": 1.8163388539647164e-06,
      "loss": 0.6225,
      "step": 9701
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4190868894388734,
      "learning_rate": 1.8139278380351943e-06,
      "loss": 0.6537,
      "step": 9702
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.433948598404522,
      "learning_rate": 1.8115183473495767e-06,
      "loss": 0.6926,
      "step": 9703
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4214645932630802,
      "learning_rate": 1.8091103821099486e-06,
      "loss": 0.6429,
      "step": 9704
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38435589626951816,
      "learning_rate": 1.8067039425182663e-06,
      "loss": 0.6516,
      "step": 9705
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3958071633479981,
      "learning_rate": 1.8042990287763507e-06,
      "loss": 0.6588,
      "step": 9706
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41214917664989925,
      "learning_rate": 1.8018956410859002e-06,
      "loss": 0.6373,
      "step": 9707
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37021015801237633,
      "learning_rate": 1.7994937796484847e-06,
      "loss": 0.6432,
      "step": 9708
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41804747417923016,
      "learning_rate": 1.7970934446655452e-06,
      "loss": 0.6395,
      "step": 9709
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40248039430818966,
      "learning_rate": 1.7946946363383987e-06,
      "loss": 0.6559,
      "step": 9710
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3497025856851546,
      "learning_rate": 1.792297354868222e-06,
      "loss": 0.6192,
      "step": 9711
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43803491265471484,
      "learning_rate": 1.7899016004560765e-06,
      "loss": 0.6558,
      "step": 9712
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4027088851156484,
      "learning_rate": 1.7875073733028924e-06,
      "loss": 0.6199,
      "step": 9713
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3577337571316196,
      "learning_rate": 1.7851146736094716e-06,
      "loss": 0.6221,
      "step": 9714
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41606560136696014,
      "learning_rate": 1.782723501576482e-06,
      "loss": 0.6514,
      "step": 9715
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3559268185316292,
      "learning_rate": 1.780333857404477e-06,
      "loss": 0.6126,
      "step": 9716
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40993092321339325,
      "learning_rate": 1.7779457412938628e-06,
      "loss": 0.6374,
      "step": 9717
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36320620346194427,
      "learning_rate": 1.7755591534449345e-06,
      "loss": 0.6528,
      "step": 9718
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40149756856077945,
      "learning_rate": 1.7731740940578524e-06,
      "loss": 0.6226,
      "step": 9719
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37011136207887946,
      "learning_rate": 1.7707905633326428e-06,
      "loss": 0.6424,
      "step": 9720
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.404792558013358,
      "learning_rate": 1.7684085614692104e-06,
      "loss": 0.637,
      "step": 9721
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42590334324147777,
      "learning_rate": 1.7660280886673397e-06,
      "loss": 0.6466,
      "step": 9722
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3513177979379632,
      "learning_rate": 1.7636491451266713e-06,
      "loss": 0.6471,
      "step": 9723
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3985712878847535,
      "learning_rate": 1.7612717310467232e-06,
      "loss": 0.6826,
      "step": 9724
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5711005686037219,
      "learning_rate": 1.7588958466268912e-06,
      "loss": 0.3653,
      "step": 9725
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3900885463692811,
      "learning_rate": 1.7565214920664341e-06,
      "loss": 0.6608,
      "step": 9726
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4107226434038905,
      "learning_rate": 1.7541486675644836e-06,
      "loss": 0.6168,
      "step": 9727
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38314896697948586,
      "learning_rate": 1.7517773733200517e-06,
      "loss": 0.6225,
      "step": 9728
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41764360427744895,
      "learning_rate": 1.74940760953201e-06,
      "loss": 0.6485,
      "step": 9729
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4037636351256051,
      "learning_rate": 1.7470393763991178e-06,
      "loss": 0.6414,
      "step": 9730
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.31398548168910395,
      "learning_rate": 1.744672674119985e-06,
      "loss": 0.6082,
      "step": 9731
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3773158270557904,
      "learning_rate": 1.7423075028931057e-06,
      "loss": 0.6176,
      "step": 9732
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41930689675128313,
      "learning_rate": 1.739943862916853e-06,
      "loss": 0.6439,
      "step": 9733
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38597679873781876,
      "learning_rate": 1.7375817543894503e-06,
      "loss": 0.6272,
      "step": 9734
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41064525826416726,
      "learning_rate": 1.7352211775090077e-06,
      "loss": 0.6541,
      "step": 9735
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4035223573617612,
      "learning_rate": 1.7328621324735139e-06,
      "loss": 0.6353,
      "step": 9736
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4332487110985776,
      "learning_rate": 1.7305046194808084e-06,
      "loss": 0.6315,
      "step": 9737
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3670482276127797,
      "learning_rate": 1.728148638728615e-06,
      "loss": 0.6523,
      "step": 9738
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3410936187198369,
      "learning_rate": 1.7257941904145336e-06,
      "loss": 0.6151,
      "step": 9739
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.384159263551356,
      "learning_rate": 1.7234412747360196e-06,
      "loss": 0.6141,
      "step": 9740
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39046652216743555,
      "learning_rate": 1.721089891890415e-06,
      "loss": 0.6149,
      "step": 9741
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3909209224312975,
      "learning_rate": 1.7187400420749244e-06,
      "loss": 0.6254,
      "step": 9742
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4094696894032068,
      "learning_rate": 1.716391725486628e-06,
      "loss": 0.6153,
      "step": 9743
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.32524218576251845,
      "learning_rate": 1.714044942322477e-06,
      "loss": 0.5866,
      "step": 9744
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3938995831258278,
      "learning_rate": 1.7116996927792961e-06,
      "loss": 0.6228,
      "step": 9745
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3810578385361069,
      "learning_rate": 1.7093559770537726e-06,
      "loss": 0.6274,
      "step": 9746
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3714218939236322,
      "learning_rate": 1.7070137953424737e-06,
      "loss": 0.658,
      "step": 9747
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3838604433234249,
      "learning_rate": 1.7046731478418377e-06,
      "loss": 0.6613,
      "step": 9748
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.395984652529613,
      "learning_rate": 1.7023340347481632e-06,
      "loss": 0.6363,
      "step": 9749
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40314351357359585,
      "learning_rate": 1.6999964562576442e-06,
      "loss": 0.6254,
      "step": 9750
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4319149678488623,
      "learning_rate": 1.6976604125663155e-06,
      "loss": 0.6394,
      "step": 9751
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39531747445608345,
      "learning_rate": 1.6953259038701065e-06,
      "loss": 0.6329,
      "step": 9752
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3902477872098509,
      "learning_rate": 1.6929929303648096e-06,
      "loss": 0.6943,
      "step": 9753
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.398768062488237,
      "learning_rate": 1.6906614922460817e-06,
      "loss": 0.629,
      "step": 9754
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3672973805613785,
      "learning_rate": 1.6883315897094643e-06,
      "loss": 0.6214,
      "step": 9755
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3485488395932651,
      "learning_rate": 1.6860032229503609e-06,
      "loss": 0.6327,
      "step": 9756
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39717620396275066,
      "learning_rate": 1.6836763921640487e-06,
      "loss": 0.6102,
      "step": 9757
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3320159634997314,
      "learning_rate": 1.6813510975456782e-06,
      "loss": 0.6098,
      "step": 9758
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4266551968630787,
      "learning_rate": 1.679027339290271e-06,
      "loss": 0.6378,
      "step": 9759
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3936240102218581,
      "learning_rate": 1.6767051175927117e-06,
      "loss": 0.6406,
      "step": 9760
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.47945119068700287,
      "learning_rate": 1.6743844326477643e-06,
      "loss": 0.6445,
      "step": 9761
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4071480642839718,
      "learning_rate": 1.6720652846500661e-06,
      "loss": 0.6183,
      "step": 9762
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36793903131020356,
      "learning_rate": 1.6697476737941132e-06,
      "loss": 0.6358,
      "step": 9763
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37601281661026265,
      "learning_rate": 1.6674316002742896e-06,
      "loss": 0.6479,
      "step": 9764
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43478511049040075,
      "learning_rate": 1.6651170642848358e-06,
      "loss": 0.6375,
      "step": 9765
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4001588555236629,
      "learning_rate": 1.6628040660198696e-06,
      "loss": 0.6288,
      "step": 9766
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37180404515630217,
      "learning_rate": 1.660492605673385e-06,
      "loss": 0.6486,
      "step": 9767
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.391401879569794,
      "learning_rate": 1.6581826834392355e-06,
      "loss": 0.6423,
      "step": 9768
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4172422342793989,
      "learning_rate": 1.655874299511151e-06,
      "loss": 0.6286,
      "step": 9769
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4231704895445976,
      "learning_rate": 1.653567454082734e-06,
      "loss": 0.6437,
      "step": 9770
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37331399931681825,
      "learning_rate": 1.6512621473474588e-06,
      "loss": 0.5916,
      "step": 9771
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3818166257434745,
      "learning_rate": 1.6489583794986685e-06,
      "loss": 0.6257,
      "step": 9772
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41919603280019635,
      "learning_rate": 1.6466561507295774e-06,
      "loss": 0.6529,
      "step": 9773
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3830906583666384,
      "learning_rate": 1.644355461233269e-06,
      "loss": 0.6163,
      "step": 9774
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38461010854497085,
      "learning_rate": 1.6420563112026977e-06,
      "loss": 0.6211,
      "step": 9775
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3830120710996957,
      "learning_rate": 1.6397587008306982e-06,
      "loss": 0.6294,
      "step": 9776
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34416129692705577,
      "learning_rate": 1.6374626303099562e-06,
      "loss": 0.6237,
      "step": 9777
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3828072853620772,
      "learning_rate": 1.6351680998330489e-06,
      "loss": 0.6296,
      "step": 9778
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37783773404240695,
      "learning_rate": 1.632875109592418e-06,
      "loss": 0.6268,
      "step": 9779
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.579766196109303,
      "learning_rate": 1.630583659780367e-06,
      "loss": 0.3339,
      "step": 9780
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34047321768288147,
      "learning_rate": 1.6282937505890783e-06,
      "loss": 0.6235,
      "step": 9781
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38444768128513657,
      "learning_rate": 1.6260053822106093e-06,
      "loss": 0.6189,
      "step": 9782
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38349285944974365,
      "learning_rate": 1.6237185548368772e-06,
      "loss": 0.6172,
      "step": 9783
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43904572033035866,
      "learning_rate": 1.6214332686596756e-06,
      "loss": 0.6661,
      "step": 9784
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3867737704517823,
      "learning_rate": 1.6191495238706688e-06,
      "loss": 0.6417,
      "step": 9785
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3616740716026924,
      "learning_rate": 1.616867320661395e-06,
      "loss": 0.6409,
      "step": 9786
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.392442405305156,
      "learning_rate": 1.6145866592232607e-06,
      "loss": 0.6468,
      "step": 9787
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41741821885823543,
      "learning_rate": 1.612307539747535e-06,
      "loss": 0.6671,
      "step": 9788
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4226581081123815,
      "learning_rate": 1.61002996242537e-06,
      "loss": 0.6442,
      "step": 9789
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4571084996740291,
      "learning_rate": 1.6077539274477816e-06,
      "loss": 0.6285,
      "step": 9790
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38340370627411047,
      "learning_rate": 1.6054794350056613e-06,
      "loss": 0.6629,
      "step": 9791
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39421518953630674,
      "learning_rate": 1.6032064852897633e-06,
      "loss": 0.6627,
      "step": 9792
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33351614701554905,
      "learning_rate": 1.600935078490724e-06,
      "loss": 0.598,
      "step": 9793
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4343468326382649,
      "learning_rate": 1.5986652147990378e-06,
      "loss": 0.6521,
      "step": 9794
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35309846334141914,
      "learning_rate": 1.5963968944050745e-06,
      "loss": 0.64,
      "step": 9795
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4385191225111722,
      "learning_rate": 1.594130117499082e-06,
      "loss": 0.682,
      "step": 9796
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38491173184550365,
      "learning_rate": 1.5918648842711592e-06,
      "loss": 0.6116,
      "step": 9797
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44458001429506644,
      "learning_rate": 1.5896011949113078e-06,
      "loss": 0.657,
      "step": 9798
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43832762078091386,
      "learning_rate": 1.5873390496093643e-06,
      "loss": 0.672,
      "step": 9799
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3823127913347938,
      "learning_rate": 1.5850784485550573e-06,
      "loss": 0.629,
      "step": 9800
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41342231878201224,
      "learning_rate": 1.5828193919379819e-06,
      "loss": 0.6348,
      "step": 9801
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4129348271911272,
      "learning_rate": 1.5805618799476042e-06,
      "loss": 0.6261,
      "step": 9802
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3515795113086674,
      "learning_rate": 1.5783059127732547e-06,
      "loss": 0.5863,
      "step": 9803
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43664798357005724,
      "learning_rate": 1.5760514906041401e-06,
      "loss": 0.6425,
      "step": 9804
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3577429051980092,
      "learning_rate": 1.573798613629336e-06,
      "loss": 0.6252,
      "step": 9805
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33524313125825944,
      "learning_rate": 1.571547282037791e-06,
      "loss": 0.6237,
      "step": 9806
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38112950628324943,
      "learning_rate": 1.5692974960183204e-06,
      "loss": 0.628,
      "step": 9807
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35642744090271766,
      "learning_rate": 1.5670492557596073e-06,
      "loss": 0.6302,
      "step": 9808
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4304339346570845,
      "learning_rate": 1.5648025614502117e-06,
      "loss": 0.6466,
      "step": 9809
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3957605527460831,
      "learning_rate": 1.5625574132785649e-06,
      "loss": 0.6282,
      "step": 9810
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4257109199201669,
      "learning_rate": 1.560313811432954e-06,
      "loss": 0.6693,
      "step": 9811
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38664172997054613,
      "learning_rate": 1.558071756101558e-06,
      "loss": 0.6412,
      "step": 9812
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42578876390183235,
      "learning_rate": 1.555831247472417e-06,
      "loss": 0.6508,
      "step": 9813
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3887094558400414,
      "learning_rate": 1.5535922857334273e-06,
      "loss": 0.6354,
      "step": 9814
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3852464825902837,
      "learning_rate": 1.551354871072377e-06,
      "loss": 0.6112,
      "step": 9815
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3861628365831335,
      "learning_rate": 1.5491190036769154e-06,
      "loss": 0.6212,
      "step": 9816
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35617127200323784,
      "learning_rate": 1.5468846837345575e-06,
      "loss": 0.6115,
      "step": 9817
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3874459181104983,
      "learning_rate": 1.5446519114326974e-06,
      "loss": 0.6437,
      "step": 9818
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4427089826620956,
      "learning_rate": 1.54242068695859e-06,
      "loss": 0.6627,
      "step": 9819
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43330031598564067,
      "learning_rate": 1.54019101049937e-06,
      "loss": 0.6686,
      "step": 9820
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3348587343171674,
      "learning_rate": 1.537962882242039e-06,
      "loss": 0.5991,
      "step": 9821
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4145386112435814,
      "learning_rate": 1.5357363023734606e-06,
      "loss": 0.6302,
      "step": 9822
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41642314111856177,
      "learning_rate": 1.5335112710803812e-06,
      "loss": 0.6848,
      "step": 9823
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4366054135773981,
      "learning_rate": 1.5312877885494114e-06,
      "loss": 0.6888,
      "step": 9824
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4204223931769678,
      "learning_rate": 1.5290658549670246e-06,
      "loss": 0.6657,
      "step": 9825
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37519879210200996,
      "learning_rate": 1.52684547051958e-06,
      "loss": 0.6386,
      "step": 9826
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36869319778609067,
      "learning_rate": 1.524626635393298e-06,
      "loss": 0.6218,
      "step": 9827
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39672744968537793,
      "learning_rate": 1.5224093497742654e-06,
      "loss": 0.6439,
      "step": 9828
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.420002675055485,
      "learning_rate": 1.5201936138484463e-06,
      "loss": 0.6713,
      "step": 9829
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.417170873724027,
      "learning_rate": 1.5179794278016724e-06,
      "loss": 0.6543,
      "step": 9830
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4050698169974712,
      "learning_rate": 1.5157667918196417e-06,
      "loss": 0.6229,
      "step": 9831
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39863797279366786,
      "learning_rate": 1.5135557060879214e-06,
      "loss": 0.6363,
      "step": 9832
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.40379204492469556,
      "learning_rate": 1.5113461707919652e-06,
      "loss": 0.6424,
      "step": 9833
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5513407019479447,
      "learning_rate": 1.5091381861170762e-06,
      "loss": 0.3494,
      "step": 9834
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3939704957833224,
      "learning_rate": 1.5069317522484328e-06,
      "loss": 0.6824,
      "step": 9835
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3747661248586859,
      "learning_rate": 1.5047268693710959e-06,
      "loss": 0.649,
      "step": 9836
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36597415503600195,
      "learning_rate": 1.5025235376699732e-06,
      "loss": 0.6131,
      "step": 9837
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3558932319581414,
      "learning_rate": 1.5003217573298657e-06,
      "loss": 0.6309,
      "step": 9838
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4555318113113442,
      "learning_rate": 1.4981215285354278e-06,
      "loss": 0.6481,
      "step": 9839
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3884468281348556,
      "learning_rate": 1.4959228514711944e-06,
      "loss": 0.6494,
      "step": 9840
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.40325542999903863,
      "learning_rate": 1.4937257263215643e-06,
      "loss": 0.6494,
      "step": 9841
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4073631250131714,
      "learning_rate": 1.4915301532708081e-06,
      "loss": 0.6494,
      "step": 9842
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3574327331896184,
      "learning_rate": 1.489336132503063e-06,
      "loss": 0.6362,
      "step": 9843
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.40365267543294353,
      "learning_rate": 1.4871436642023463e-06,
      "loss": 0.5682,
      "step": 9844
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36047505795537516,
      "learning_rate": 1.4849527485525284e-06,
      "loss": 0.6659,
      "step": 9845
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42306408941680396,
      "learning_rate": 1.482763385737358e-06,
      "loss": 0.6704,
      "step": 9846
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3497548364546083,
      "learning_rate": 1.4805755759404684e-06,
      "loss": 0.6262,
      "step": 9847
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41490624551486854,
      "learning_rate": 1.4783893193453347e-06,
      "loss": 0.641,
      "step": 9848
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4246297951356156,
      "learning_rate": 1.4762046161353215e-06,
      "loss": 0.5976,
      "step": 9849
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4050420751087175,
      "learning_rate": 1.4740214664936581e-06,
      "loss": 0.6363,
      "step": 9850
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36035876049557436,
      "learning_rate": 1.4718398706034399e-06,
      "loss": 0.6035,
      "step": 9851
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42088085987721036,
      "learning_rate": 1.4696598286476315e-06,
      "loss": 0.6192,
      "step": 9852
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39792992955627016,
      "learning_rate": 1.467481340809076e-06,
      "loss": 0.6164,
      "step": 9853
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36758061484381566,
      "learning_rate": 1.4653044072704802e-06,
      "loss": 0.6306,
      "step": 9854
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4079844563279123,
      "learning_rate": 1.4631290282144205e-06,
      "loss": 0.6233,
      "step": 9855
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3800720497573155,
      "learning_rate": 1.4609552038233422e-06,
      "loss": 0.6448,
      "step": 9856
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33691408619510055,
      "learning_rate": 1.458782934279559e-06,
      "loss": 0.6093,
      "step": 9857
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39764400716784276,
      "learning_rate": 1.4566122197652588e-06,
      "loss": 0.6308,
      "step": 9858
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4152934517807072,
      "learning_rate": 1.4544430604625005e-06,
      "loss": 0.6352,
      "step": 9859
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37418040985368883,
      "learning_rate": 1.4522754565532004e-06,
      "loss": 0.6691,
      "step": 9860
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3905774994853481,
      "learning_rate": 1.4501094082191625e-06,
      "loss": 0.6165,
      "step": 9861
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38338778591279776,
      "learning_rate": 1.4479449156420433e-06,
      "loss": 0.6265,
      "step": 9862
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41812968899513114,
      "learning_rate": 1.44578197900338e-06,
      "loss": 0.6056,
      "step": 9863
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3483224632193022,
      "learning_rate": 1.4436205984845763e-06,
      "loss": 0.6307,
      "step": 9864
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4101598032818147,
      "learning_rate": 1.4414607742668985e-06,
      "loss": 0.6575,
      "step": 9865
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3997165818213873,
      "learning_rate": 1.4393025065314947e-06,
      "loss": 0.6463,
      "step": 9866
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3932741746459078,
      "learning_rate": 1.4371457954593716e-06,
      "loss": 0.6404,
      "step": 9867
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36344987013712793,
      "learning_rate": 1.434990641231413e-06,
      "loss": 0.6241,
      "step": 9868
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38288470011969017,
      "learning_rate": 1.43283704402837e-06,
      "loss": 0.6221,
      "step": 9869
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39990126856124103,
      "learning_rate": 1.4306850040308629e-06,
      "loss": 0.6435,
      "step": 9870
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3914470683341018,
      "learning_rate": 1.4285345214193757e-06,
      "loss": 0.5965,
      "step": 9871
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3559445682152262,
      "learning_rate": 1.426385596374269e-06,
      "loss": 0.6057,
      "step": 9872
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3959412479886679,
      "learning_rate": 1.4242382290757717e-06,
      "loss": 0.6676,
      "step": 9873
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39353540842073437,
      "learning_rate": 1.42209241970398e-06,
      "loss": 0.6202,
      "step": 9874
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4007921083621584,
      "learning_rate": 1.4199481684388628e-06,
      "loss": 0.6179,
      "step": 9875
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.414206139118521,
      "learning_rate": 1.41780547546025e-06,
      "loss": 0.6683,
      "step": 9876
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5423859941221649,
      "learning_rate": 1.4156643409478532e-06,
      "loss": 0.3335,
      "step": 9877
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43941096002761615,
      "learning_rate": 1.4135247650812445e-06,
      "loss": 0.6606,
      "step": 9878
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3669587550634759,
      "learning_rate": 1.4113867480398647e-06,
      "loss": 0.6438,
      "step": 9879
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.40259399186483746,
      "learning_rate": 1.409250290003026e-06,
      "loss": 0.6467,
      "step": 9880
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3993944700585375,
      "learning_rate": 1.4071153911499203e-06,
      "loss": 0.6299,
      "step": 9881
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42240027161771154,
      "learning_rate": 1.4049820516595891e-06,
      "loss": 0.6458,
      "step": 9882
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37299961168879653,
      "learning_rate": 1.402850271710956e-06,
      "loss": 0.6565,
      "step": 9883
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3429363861528861,
      "learning_rate": 1.4007200514828134e-06,
      "loss": 0.6206,
      "step": 9884
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4105490029943435,
      "learning_rate": 1.398591391153814e-06,
      "loss": 0.6343,
      "step": 9885
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42810834192817665,
      "learning_rate": 1.3964642909024907e-06,
      "loss": 0.6636,
      "step": 9886
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.373993244733209,
      "learning_rate": 1.3943387509072404e-06,
      "loss": 0.6547,
      "step": 9887
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3999379036499265,
      "learning_rate": 1.3922147713463296e-06,
      "loss": 0.634,
      "step": 9888
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4118427078310878,
      "learning_rate": 1.3900923523978914e-06,
      "loss": 0.6432,
      "step": 9889
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4046596186948214,
      "learning_rate": 1.387971494239937e-06,
      "loss": 0.6533,
      "step": 9890
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3516287816649335,
      "learning_rate": 1.3858521970503325e-06,
      "loss": 0.6026,
      "step": 9891
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4111726679737722,
      "learning_rate": 1.383734461006825e-06,
      "loss": 0.6621,
      "step": 9892
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42594023510421647,
      "learning_rate": 1.3816182862870276e-06,
      "loss": 0.6616,
      "step": 9893
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4123194939110865,
      "learning_rate": 1.3795036730684119e-06,
      "loss": 0.6634,
      "step": 9894
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34711542375618226,
      "learning_rate": 1.3773906215283428e-06,
      "loss": 0.6201,
      "step": 9895
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4121872641452675,
      "learning_rate": 1.3752791318440296e-06,
      "loss": 0.6201,
      "step": 9896
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3843526462444412,
      "learning_rate": 1.3731692041925614e-06,
      "loss": 0.6252,
      "step": 9897
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3790489414079751,
      "learning_rate": 1.3710608387508995e-06,
      "loss": 0.6321,
      "step": 9898
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4000225111883845,
      "learning_rate": 1.3689540356958642e-06,
      "loss": 0.6224,
      "step": 9899
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.43255722561920407,
      "learning_rate": 1.3668487952041543e-06,
      "loss": 0.6527,
      "step": 9900
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39173734379410485,
      "learning_rate": 1.3647451174523307e-06,
      "loss": 0.648,
      "step": 9901
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3758247356790737,
      "learning_rate": 1.3626430026168302e-06,
      "loss": 0.6273,
      "step": 9902
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39269050474216116,
      "learning_rate": 1.3605424508739518e-06,
      "loss": 0.6163,
      "step": 9903
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3879549870538864,
      "learning_rate": 1.358443462399872e-06,
      "loss": 0.6325,
      "step": 9904
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.372428047483107,
      "learning_rate": 1.3563460373706217e-06,
      "loss": 0.6069,
      "step": 9905
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37130363400469635,
      "learning_rate": 1.3542501759621128e-06,
      "loss": 0.6177,
      "step": 9906
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3798665125447249,
      "learning_rate": 1.3521558783501255e-06,
      "loss": 0.6435,
      "step": 9907
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40001192444314876,
      "learning_rate": 1.3500631447102985e-06,
      "loss": 0.6456,
      "step": 9908
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33047469367302007,
      "learning_rate": 1.3479719752181586e-06,
      "loss": 0.6018,
      "step": 9909
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3846595627265563,
      "learning_rate": 1.3458823700490808e-06,
      "loss": 0.6084,
      "step": 9910
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34292035728652764,
      "learning_rate": 1.3437943293783207e-06,
      "loss": 0.6019,
      "step": 9911
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38106275910025916,
      "learning_rate": 1.341707853381e-06,
      "loss": 0.6053,
      "step": 9912
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4131294691653658,
      "learning_rate": 1.3396229422321061e-06,
      "loss": 0.6298,
      "step": 9913
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39045965851523545,
      "learning_rate": 1.3375395961065008e-06,
      "loss": 0.6249,
      "step": 9914
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34270818795566627,
      "learning_rate": 1.3354578151789087e-06,
      "loss": 0.621,
      "step": 9915
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41844373697649334,
      "learning_rate": 1.3333775996239306e-06,
      "loss": 0.6573,
      "step": 9916
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3738821633499172,
      "learning_rate": 1.3312989496160288e-06,
      "loss": 0.6307,
      "step": 9917
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44510840633169396,
      "learning_rate": 1.329221865329542e-06,
      "loss": 0.6844,
      "step": 9918
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35270266352125307,
      "learning_rate": 1.3271463469386658e-06,
      "loss": 0.6365,
      "step": 9919
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35148368506961214,
      "learning_rate": 1.3250723946174727e-06,
      "loss": 0.607,
      "step": 9920
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36098347037502193,
      "learning_rate": 1.3230000085399097e-06,
      "loss": 0.6222,
      "step": 9921
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4195248485723595,
      "learning_rate": 1.3209291888797715e-06,
      "loss": 0.6784,
      "step": 9922
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3516570424678083,
      "learning_rate": 1.318859935810748e-06,
      "loss": 0.6257,
      "step": 9923
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40305989593765407,
      "learning_rate": 1.3167922495063823e-06,
      "loss": 0.6299,
      "step": 9924
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3684485271333077,
      "learning_rate": 1.3147261301400849e-06,
      "loss": 0.614,
      "step": 9925
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3681402426152319,
      "learning_rate": 1.3126615778851414e-06,
      "loss": 0.6124,
      "step": 9926
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.45500351002360623,
      "learning_rate": 1.3105985929147047e-06,
      "loss": 0.6232,
      "step": 9927
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4372214756475813,
      "learning_rate": 1.3085371754017917e-06,
      "loss": 0.6408,
      "step": 9928
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4283264660726262,
      "learning_rate": 1.3064773255192908e-06,
      "loss": 0.665,
      "step": 9929
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3493611352254033,
      "learning_rate": 1.3044190434399596e-06,
      "loss": 0.6469,
      "step": 9930
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38942672020620644,
      "learning_rate": 1.3023623293364263e-06,
      "loss": 0.6565,
      "step": 9931
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4071941180340788,
      "learning_rate": 1.3003071833811864e-06,
      "loss": 0.6232,
      "step": 9932
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3795292350795822,
      "learning_rate": 1.2982536057465955e-06,
      "loss": 0.6382,
      "step": 9933
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37916605422605726,
      "learning_rate": 1.2962015966048891e-06,
      "loss": 0.6205,
      "step": 9934
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4020763328175942,
      "learning_rate": 1.2941511561281695e-06,
      "loss": 0.6616,
      "step": 9935
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39682478924113435,
      "learning_rate": 1.2921022844883968e-06,
      "loss": 0.6268,
      "step": 9936
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41643131478042356,
      "learning_rate": 1.2900549818574138e-06,
      "loss": 0.6565,
      "step": 9937
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4008581600407027,
      "learning_rate": 1.288009248406925e-06,
      "loss": 0.6356,
      "step": 9938
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4037892299736944,
      "learning_rate": 1.2859650843085026e-06,
      "loss": 0.6252,
      "step": 9939
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41354039940746606,
      "learning_rate": 1.2839224897335844e-06,
      "loss": 0.6418,
      "step": 9940
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4105335945221328,
      "learning_rate": 1.2818814648534895e-06,
      "loss": 0.6331,
      "step": 9941
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36824415835965635,
      "learning_rate": 1.279842009839385e-06,
      "loss": 0.6136,
      "step": 9942
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34303627552828464,
      "learning_rate": 1.2778041248623253e-06,
      "loss": 0.623,
      "step": 9943
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40088178218056814,
      "learning_rate": 1.2757678100932203e-06,
      "loss": 0.6587,
      "step": 9944
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36212949438127345,
      "learning_rate": 1.273733065702858e-06,
      "loss": 0.623,
      "step": 9945
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4221780614493408,
      "learning_rate": 1.271699891861886e-06,
      "loss": 0.6912,
      "step": 9946
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4147841644500801,
      "learning_rate": 1.2696682887408307e-06,
      "loss": 0.6571,
      "step": 9947
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39915356344402414,
      "learning_rate": 1.267638256510071e-06,
      "loss": 0.6563,
      "step": 9948
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4204437089032612,
      "learning_rate": 1.2656097953398706e-06,
      "loss": 0.6753,
      "step": 9949
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3558772366555428,
      "learning_rate": 1.263582905400349e-06,
      "loss": 0.6138,
      "step": 9950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3992347452970932,
      "learning_rate": 1.2615575868615015e-06,
      "loss": 0.6707,
      "step": 9951
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4282848214508232,
      "learning_rate": 1.2595338398931922e-06,
      "loss": 0.6334,
      "step": 9952
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3643457327142641,
      "learning_rate": 1.257511664665143e-06,
      "loss": 0.6035,
      "step": 9953
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38159841777245895,
      "learning_rate": 1.2554910613469563e-06,
      "loss": 0.634,
      "step": 9954
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4107846590053465,
      "learning_rate": 1.2534720301080983e-06,
      "loss": 0.6297,
      "step": 9955
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3843410191931551,
      "learning_rate": 1.251454571117896e-06,
      "loss": 0.6203,
      "step": 9956
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.403247036738938,
      "learning_rate": 1.2494386845455608e-06,
      "loss": 0.6378,
      "step": 9957
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3765317819084231,
      "learning_rate": 1.2474243705601596e-06,
      "loss": 0.6449,
      "step": 9958
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3490681003583699,
      "learning_rate": 1.2454116293306263e-06,
      "loss": 0.6064,
      "step": 9959
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40149654463553,
      "learning_rate": 1.24340046102577e-06,
      "loss": 0.6279,
      "step": 9960
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3816489159205274,
      "learning_rate": 1.241390865814267e-06,
      "loss": 0.651,
      "step": 9961
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41953958262595764,
      "learning_rate": 1.239382843864656e-06,
      "loss": 0.599,
      "step": 9962
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3867126487667234,
      "learning_rate": 1.2373763953453466e-06,
      "loss": 0.6713,
      "step": 9963
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4294696608364429,
      "learning_rate": 1.235371520424622e-06,
      "loss": 0.6718,
      "step": 9964
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37010331862316637,
      "learning_rate": 1.2333682192706253e-06,
      "loss": 0.6151,
      "step": 9965
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3775869940822821,
      "learning_rate": 1.2313664920513736e-06,
      "loss": 0.6185,
      "step": 9966
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3932648267840015,
      "learning_rate": 1.2293663389347454e-06,
      "loss": 0.6134,
      "step": 9967
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4030120378968237,
      "learning_rate": 1.2273677600884915e-06,
      "loss": 0.6397,
      "step": 9968
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40042380771932357,
      "learning_rate": 1.2253707556802374e-06,
      "loss": 0.6168,
      "step": 9969
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3722868013539368,
      "learning_rate": 1.223375325877456e-06,
      "loss": 0.6066,
      "step": 9970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3837593761242737,
      "learning_rate": 1.2213814708475136e-06,
      "loss": 0.6434,
      "step": 9971
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35892739594916523,
      "learning_rate": 1.2193891907576317e-06,
      "loss": 0.6012,
      "step": 9972
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39059918282593403,
      "learning_rate": 1.2173984857748944e-06,
      "loss": 0.6278,
      "step": 9973
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4284605330345051,
      "learning_rate": 1.2154093560662616e-06,
      "loss": 0.6628,
      "step": 9974
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39875197494052617,
      "learning_rate": 1.2134218017985622e-06,
      "loss": 0.6266,
      "step": 9975
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36779030971219057,
      "learning_rate": 1.2114358231384871e-06,
      "loss": 0.6474,
      "step": 9976
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40161079299319424,
      "learning_rate": 1.2094514202525964e-06,
      "loss": 0.6415,
      "step": 9977
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3954551299309309,
      "learning_rate": 1.2074685933073239e-06,
      "loss": 0.6403,
      "step": 9978
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37444252185879656,
      "learning_rate": 1.2054873424689627e-06,
      "loss": 0.6566,
      "step": 9979
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3685651841915447,
      "learning_rate": 1.2035076679036806e-06,
      "loss": 0.6593,
      "step": 9980
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3624707248619205,
      "learning_rate": 1.2015295697775132e-06,
      "loss": 0.6271,
      "step": 9981
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.43480986228853113,
      "learning_rate": 1.1995530482563567e-06,
      "loss": 0.5748,
      "step": 9982
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41632917721796786,
      "learning_rate": 1.1975781035059786e-06,
      "loss": 0.675,
      "step": 9983
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37397838244125664,
      "learning_rate": 1.1956047356920197e-06,
      "loss": 0.6141,
      "step": 9984
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40357562277639264,
      "learning_rate": 1.1936329449799788e-06,
      "loss": 0.6089,
      "step": 9985
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36356820396664197,
      "learning_rate": 1.1916627315352348e-06,
      "loss": 0.6483,
      "step": 9986
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3941268096011666,
      "learning_rate": 1.1896940955230196e-06,
      "loss": 0.6498,
      "step": 9987
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.419767230394131,
      "learning_rate": 1.1877270371084437e-06,
      "loss": 0.6253,
      "step": 9988
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41202687537729427,
      "learning_rate": 1.1857615564564862e-06,
      "loss": 0.6328,
      "step": 9989
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4137889436052378,
      "learning_rate": 1.1837976537319795e-06,
      "loss": 0.6544,
      "step": 9990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36498569646614837,
      "learning_rate": 1.1818353290996388e-06,
      "loss": 0.6189,
      "step": 9991
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.45124831971357154,
      "learning_rate": 1.1798745827240476e-06,
      "loss": 0.6918,
      "step": 9992
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3831951579554835,
      "learning_rate": 1.1779154147696414e-06,
      "loss": 0.6615,
      "step": 9993
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41661045188207135,
      "learning_rate": 1.1759578254007398e-06,
      "loss": 0.6463,
      "step": 9994
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3471696019809176,
      "learning_rate": 1.1740018147815246e-06,
      "loss": 0.6232,
      "step": 9995
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4183133138570486,
      "learning_rate": 1.1720473830760381e-06,
      "loss": 0.6475,
      "step": 9996
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39429940813824865,
      "learning_rate": 1.170094530448198e-06,
      "loss": 0.6464,
      "step": 9997
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35672086773528455,
      "learning_rate": 1.1681432570617913e-06,
      "loss": 0.6315,
      "step": 9998
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.376971954721255,
      "learning_rate": 1.1661935630804643e-06,
      "loss": 0.6224,
      "step": 9999
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3987227510615833,
      "learning_rate": 1.1642454486677425e-06,
      "loss": 0.6507,
      "step": 10000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4307789793339291,
      "learning_rate": 1.1622989139870033e-06,
      "loss": 0.6518,
      "step": 10001
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3918694909234812,
      "learning_rate": 1.1603539592015033e-06,
      "loss": 0.6375,
      "step": 10002
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41568268271160025,
      "learning_rate": 1.1584105844743654e-06,
      "loss": 0.6546,
      "step": 10003
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41276695988565293,
      "learning_rate": 1.1564687899685788e-06,
      "loss": 0.6566,
      "step": 10004
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38358278521985156,
      "learning_rate": 1.1545285758469916e-06,
      "loss": 0.6553,
      "step": 10005
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4182612265846655,
      "learning_rate": 1.1525899422723396e-06,
      "loss": 0.6545,
      "step": 10006
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4067043749641215,
      "learning_rate": 1.1506528894072045e-06,
      "loss": 0.6275,
      "step": 10007
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4405271908449597,
      "learning_rate": 1.148717417414047e-06,
      "loss": 0.6415,
      "step": 10008
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4016849363740405,
      "learning_rate": 1.1467835264551952e-06,
      "loss": 0.6655,
      "step": 10009
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3422130944581917,
      "learning_rate": 1.144851216692837e-06,
      "loss": 0.6288,
      "step": 10010
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3160457817421023,
      "learning_rate": 1.1429204882890389e-06,
      "loss": 0.5847,
      "step": 10011
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3643194580419885,
      "learning_rate": 1.140991341405724e-06,
      "loss": 0.6277,
      "step": 10012
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3812032817185053,
      "learning_rate": 1.1390637762046875e-06,
      "loss": 0.6138,
      "step": 10013
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3363230702692709,
      "learning_rate": 1.1371377928475958e-06,
      "loss": 0.6393,
      "step": 10014
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3531764713253002,
      "learning_rate": 1.1352133914959794e-06,
      "loss": 0.6122,
      "step": 10015
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43485381379358173,
      "learning_rate": 1.133290572311232e-06,
      "loss": 0.6409,
      "step": 10016
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3802745415469206,
      "learning_rate": 1.1313693354546173e-06,
      "loss": 0.6207,
      "step": 10017
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.563686855030161,
      "learning_rate": 1.1294496810872713e-06,
      "loss": 0.3565,
      "step": 10018
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37231052232619505,
      "learning_rate": 1.127531609370185e-06,
      "loss": 0.6212,
      "step": 10019
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37781377917998876,
      "learning_rate": 1.1256151204642362e-06,
      "loss": 0.615,
      "step": 10020
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.421536638390259,
      "learning_rate": 1.1237002145301523e-06,
      "loss": 0.6185,
      "step": 10021
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3417014371275203,
      "learning_rate": 1.1217868917285313e-06,
      "loss": 0.6033,
      "step": 10022
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.401498094523064,
      "learning_rate": 1.1198751522198492e-06,
      "loss": 0.6543,
      "step": 10023
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3348162690995183,
      "learning_rate": 1.1179649961644335e-06,
      "loss": 0.6421,
      "step": 10024
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4065523227054106,
      "learning_rate": 1.116056423722489e-06,
      "loss": 0.6706,
      "step": 10025
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3994915918722655,
      "learning_rate": 1.1141494350540883e-06,
      "loss": 0.6716,
      "step": 10026
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3939163477594735,
      "learning_rate": 1.1122440303191652e-06,
      "loss": 0.6139,
      "step": 10027
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.409218545380764,
      "learning_rate": 1.1103402096775229e-06,
      "loss": 0.6174,
      "step": 10028
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41292767869095354,
      "learning_rate": 1.1084379732888406e-06,
      "loss": 0.6053,
      "step": 10029
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5393343311373826,
      "learning_rate": 1.106537321312644e-06,
      "loss": 0.3395,
      "step": 10030
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3921157052145726,
      "learning_rate": 1.1046382539083456e-06,
      "loss": 0.6356,
      "step": 10031
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38081647078027225,
      "learning_rate": 1.1027407712352224e-06,
      "loss": 0.6072,
      "step": 10032
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40519140830438694,
      "learning_rate": 1.1008448734524002e-06,
      "loss": 0.6511,
      "step": 10033
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39221028542700087,
      "learning_rate": 1.098950560718901e-06,
      "loss": 0.6465,
      "step": 10034
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38967336742283654,
      "learning_rate": 1.0970578331935867e-06,
      "loss": 0.6171,
      "step": 10035
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3128372241708718,
      "learning_rate": 1.0951666910352055e-06,
      "loss": 0.6138,
      "step": 10036
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44094158603201905,
      "learning_rate": 1.0932771344023619e-06,
      "loss": 0.6483,
      "step": 10037
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43580166697484296,
      "learning_rate": 1.0913891634535334e-06,
      "loss": 0.7004,
      "step": 10038
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43122223839633517,
      "learning_rate": 1.089502778347058e-06,
      "loss": 0.6655,
      "step": 10039
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3808119781998887,
      "learning_rate": 1.0876179792411468e-06,
      "loss": 0.6146,
      "step": 10040
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3844324529290577,
      "learning_rate": 1.0857347662938734e-06,
      "loss": 0.6114,
      "step": 10041
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41025440791419027,
      "learning_rate": 1.0838531396631845e-06,
      "loss": 0.6362,
      "step": 10042
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.373622382448375,
      "learning_rate": 1.08197309950689e-06,
      "loss": 0.6597,
      "step": 10043
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3735771774519317,
      "learning_rate": 1.0800946459826611e-06,
      "loss": 0.6248,
      "step": 10044
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37651193019195484,
      "learning_rate": 1.0782177792480453e-06,
      "loss": 0.6843,
      "step": 10045
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39399748275760205,
      "learning_rate": 1.076342499460452e-06,
      "loss": 0.6568,
      "step": 10046
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.35118827580074247,
      "learning_rate": 1.0744688067771603e-06,
      "loss": 0.6527,
      "step": 10047
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4058814017584438,
      "learning_rate": 1.0725967013553129e-06,
      "loss": 0.5861,
      "step": 10048
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43821654736322874,
      "learning_rate": 1.0707261833519244e-06,
      "loss": 0.6669,
      "step": 10049
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37664084408385323,
      "learning_rate": 1.0688572529238694e-06,
      "loss": 0.6126,
      "step": 10050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40292598404868396,
      "learning_rate": 1.0669899102278914e-06,
      "loss": 0.6162,
      "step": 10051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5294916697198224,
      "learning_rate": 1.0651241554206072e-06,
      "loss": 0.346,
      "step": 10052
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4003820939878441,
      "learning_rate": 1.0632599886584872e-06,
      "loss": 0.6227,
      "step": 10053
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3542881034076042,
      "learning_rate": 1.0613974100978885e-06,
      "loss": 0.6149,
      "step": 10054
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3815499087262135,
      "learning_rate": 1.0595364198950132e-06,
      "loss": 0.6594,
      "step": 10055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42407322641345124,
      "learning_rate": 1.0576770182059447e-06,
      "loss": 0.633,
      "step": 10056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38858151143557484,
      "learning_rate": 1.05581920518663e-06,
      "loss": 0.6439,
      "step": 10057
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.383494718083952,
      "learning_rate": 1.0539629809928776e-06,
      "loss": 0.6205,
      "step": 10058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3588164367899994,
      "learning_rate": 1.0521083457803693e-06,
      "loss": 0.6123,
      "step": 10059
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3645026862146538,
      "learning_rate": 1.05025529970465e-06,
      "loss": 0.6114,
      "step": 10060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3791815679321733,
      "learning_rate": 1.048403842921133e-06,
      "loss": 0.6401,
      "step": 10061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39874261105748254,
      "learning_rate": 1.0465539755851006e-06,
      "loss": 0.6855,
      "step": 10062
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42024465989052007,
      "learning_rate": 1.0447056978516956e-06,
      "loss": 0.6254,
      "step": 10063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.408679789298673,
      "learning_rate": 1.0428590098759317e-06,
      "loss": 0.6281,
      "step": 10064
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37260602232920975,
      "learning_rate": 1.041013911812685e-06,
      "loss": 0.6138,
      "step": 10065
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38451075939689977,
      "learning_rate": 1.0391704038167115e-06,
      "loss": 0.6282,
      "step": 10066
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44391793042738387,
      "learning_rate": 1.037328486042608e-06,
      "loss": 0.6413,
      "step": 10067
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39997072099087533,
      "learning_rate": 1.0354881586448683e-06,
      "loss": 0.6816,
      "step": 10068
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4181362524528998,
      "learning_rate": 1.0336494217778358e-06,
      "loss": 0.6096,
      "step": 10069
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38995885463324526,
      "learning_rate": 1.0318122755957182e-06,
      "loss": 0.6346,
      "step": 10070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3834593888167329,
      "learning_rate": 1.0299767202525968e-06,
      "loss": 0.6325,
      "step": 10071
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3930828141439505,
      "learning_rate": 1.0281427559024214e-06,
      "loss": 0.6421,
      "step": 10072
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4089338654595246,
      "learning_rate": 1.0263103826989961e-06,
      "loss": 0.629,
      "step": 10073
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3735573488543558,
      "learning_rate": 1.0244796007960066e-06,
      "loss": 0.6162,
      "step": 10074
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.393860201740045,
      "learning_rate": 1.0226504103469947e-06,
      "loss": 0.6303,
      "step": 10075
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41803422928455924,
      "learning_rate": 1.0208228115053753e-06,
      "loss": 0.6647,
      "step": 10076
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4546821351557964,
      "learning_rate": 1.018996804424428e-06,
      "loss": 0.6144,
      "step": 10077
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4135357614150613,
      "learning_rate": 1.0171723892572927e-06,
      "loss": 0.6576,
      "step": 10078
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40294257011845114,
      "learning_rate": 1.0153495661569845e-06,
      "loss": 0.6035,
      "step": 10079
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40958611196906103,
      "learning_rate": 1.0135283352763836e-06,
      "loss": 0.6568,
      "step": 10080
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.46881530797619353,
      "learning_rate": 1.0117086967682233e-06,
      "loss": 0.6327,
      "step": 10081
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3413135935370418,
      "learning_rate": 1.009890650785128e-06,
      "loss": 0.6234,
      "step": 10082
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39534924358636836,
      "learning_rate": 1.0080741974795717e-06,
      "loss": 0.6252,
      "step": 10083
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41577780608362674,
      "learning_rate": 1.006259337003892e-06,
      "loss": 0.6454,
      "step": 10084
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4006917132482172,
      "learning_rate": 1.004446069510303e-06,
      "loss": 0.606,
      "step": 10085
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37463953830921337,
      "learning_rate": 1.0026343951508855e-06,
      "loss": 0.6122,
      "step": 10086
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3601683342879473,
      "learning_rate": 1.0008243140775752e-06,
      "loss": 0.6211,
      "step": 10087
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34456474527377995,
      "learning_rate": 9.990158264421846e-07,
      "loss": 0.6197,
      "step": 10088
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38607263429400324,
      "learning_rate": 9.972089323963896e-07,
      "loss": 0.6378,
      "step": 10089
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4248576304031767,
      "learning_rate": 9.954036320917293e-07,
      "loss": 0.6461,
      "step": 10090
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40172331450458426,
      "learning_rate": 9.935999256796202e-07,
      "loss": 0.6464,
      "step": 10091
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.445120905362593,
      "learning_rate": 9.917978133113282e-07,
      "loss": 0.659,
      "step": 10092
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42780986686884936,
      "learning_rate": 9.899972951379966e-07,
      "loss": 0.6499,
      "step": 10093
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4501218718838311,
      "learning_rate": 9.881983713106335e-07,
      "loss": 0.6545,
      "step": 10094
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3967555021553057,
      "learning_rate": 9.86401041980114e-07,
      "loss": 0.6414,
      "step": 10095
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5672426096979224,
      "learning_rate": 9.84605307297175e-07,
      "loss": 0.3716,
      "step": 10096
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3976699617911146,
      "learning_rate": 9.828111674124274e-07,
      "loss": 0.6238,
      "step": 10097
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3410332048984601,
      "learning_rate": 9.810186224763375e-07,
      "loss": 0.631,
      "step": 10098
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3732278601682825,
      "learning_rate": 9.79227672639247e-07,
      "loss": 0.66,
      "step": 10099
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3899182168460446,
      "learning_rate": 9.774383180513624e-07,
      "loss": 0.6689,
      "step": 10100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39090371766576903,
      "learning_rate": 9.756505588627507e-07,
      "loss": 0.6324,
      "step": 10101
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37412417712911045,
      "learning_rate": 9.738643952233474e-07,
      "loss": 0.64,
      "step": 10102
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4172420985736481,
      "learning_rate": 9.720798272829657e-07,
      "loss": 0.6379,
      "step": 10103
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38419218215826095,
      "learning_rate": 9.702968551912662e-07,
      "loss": 0.6067,
      "step": 10104
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.416137637977892,
      "learning_rate": 9.685154790977868e-07,
      "loss": 0.617,
      "step": 10105
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38838765971394007,
      "learning_rate": 9.667356991519327e-07,
      "loss": 0.6099,
      "step": 10106
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4006881881063914,
      "learning_rate": 9.649575155029667e-07,
      "loss": 0.6315,
      "step": 10107
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3944358032899454,
      "learning_rate": 9.631809283000248e-07,
      "loss": 0.5919,
      "step": 10108
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4170690076772611,
      "learning_rate": 9.614059376921081e-07,
      "loss": 0.618,
      "step": 10109
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37013420219877696,
      "learning_rate": 9.59632543828084e-07,
      "loss": 0.622,
      "step": 10110
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4150944863411257,
      "learning_rate": 9.578607468566848e-07,
      "loss": 0.6171,
      "step": 10111
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37648469746805746,
      "learning_rate": 9.560905469265048e-07,
      "loss": 0.6513,
      "step": 10112
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4302092427843192,
      "learning_rate": 9.54321944186012e-07,
      "loss": 0.6047,
      "step": 10113
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4463224028190381,
      "learning_rate": 9.525549387835386e-07,
      "loss": 0.6731,
      "step": 10114
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.35836880727710835,
      "learning_rate": 9.507895308672777e-07,
      "loss": 0.6161,
      "step": 10115
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3921348996409979,
      "learning_rate": 9.490257205852882e-07,
      "loss": 0.6055,
      "step": 10116
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4029461366745006,
      "learning_rate": 9.472635080855119e-07,
      "loss": 0.6546,
      "step": 10117
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3367388855770499,
      "learning_rate": 9.455028935157306e-07,
      "loss": 0.6135,
      "step": 10118
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43673777159537625,
      "learning_rate": 9.437438770236107e-07,
      "loss": 0.669,
      "step": 10119
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42777324285846435,
      "learning_rate": 9.419864587566785e-07,
      "loss": 0.5756,
      "step": 10120
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3789910470564084,
      "learning_rate": 9.40230638862325e-07,
      "loss": 0.6386,
      "step": 10121
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3945683392287799,
      "learning_rate": 9.384764174878081e-07,
      "loss": 0.6143,
      "step": 10122
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42076725477176186,
      "learning_rate": 9.367237947802543e-07,
      "loss": 0.6823,
      "step": 10123
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36412585338304154,
      "learning_rate": 9.34972770886653e-07,
      "loss": 0.6578,
      "step": 10124
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4107948076039189,
      "learning_rate": 9.33223345953862e-07,
      "loss": 0.6031,
      "step": 10125
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38485433230021454,
      "learning_rate": 9.314755201286041e-07,
      "loss": 0.659,
      "step": 10126
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40275990295897723,
      "learning_rate": 9.29729293557462e-07,
      "loss": 0.6281,
      "step": 10127
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3700773501465141,
      "learning_rate": 9.279846663868941e-07,
      "loss": 0.6074,
      "step": 10128
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42714560612304336,
      "learning_rate": 9.262416387632167e-07,
      "loss": 0.6013,
      "step": 10129
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40682552346134787,
      "learning_rate": 9.245002108326195e-07,
      "loss": 0.6206,
      "step": 10130
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4075205043049482,
      "learning_rate": 9.227603827411546e-07,
      "loss": 0.6245,
      "step": 10131
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42035986189920976,
      "learning_rate": 9.210221546347365e-07,
      "loss": 0.6218,
      "step": 10132
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4041277678773955,
      "learning_rate": 9.19285526659146e-07,
      "loss": 0.61,
      "step": 10133
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3687249672462128,
      "learning_rate": 9.175504989600382e-07,
      "loss": 0.6097,
      "step": 10134
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3920336324176904,
      "learning_rate": 9.158170716829207e-07,
      "loss": 0.6306,
      "step": 10135
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39567281602260423,
      "learning_rate": 9.140852449731752e-07,
      "loss": 0.6384,
      "step": 10136
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4260369064622306,
      "learning_rate": 9.123550189760566e-07,
      "loss": 0.6371,
      "step": 10137
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4211537530369172,
      "learning_rate": 9.106263938366666e-07,
      "loss": 0.6214,
      "step": 10138
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41824832090120057,
      "learning_rate": 9.088993696999892e-07,
      "loss": 0.6452,
      "step": 10139
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3899694923889808,
      "learning_rate": 9.071739467108665e-07,
      "loss": 0.6021,
      "step": 10140
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3462467901070334,
      "learning_rate": 9.054501250140024e-07,
      "loss": 0.6269,
      "step": 10141
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37582741343441106,
      "learning_rate": 9.037279047539793e-07,
      "loss": 0.6109,
      "step": 10142
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.425417772940289,
      "learning_rate": 9.020072860752305e-07,
      "loss": 0.6732,
      "step": 10143
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3940210498030194,
      "learning_rate": 9.002882691220693e-07,
      "loss": 0.6347,
      "step": 10144
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40526544857091235,
      "learning_rate": 8.98570854038665e-07,
      "loss": 0.6517,
      "step": 10145
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5254798232599001,
      "learning_rate": 8.968550409690536e-07,
      "loss": 0.3582,
      "step": 10146
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41430873265280754,
      "learning_rate": 8.951408300571373e-07,
      "loss": 0.658,
      "step": 10147
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37953883029768465,
      "learning_rate": 8.934282214466905e-07,
      "loss": 0.5796,
      "step": 10148
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4181852086894024,
      "learning_rate": 8.917172152813402e-07,
      "loss": 0.6697,
      "step": 10149
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3893307732237017,
      "learning_rate": 8.900078117045897e-07,
      "loss": 0.6181,
      "step": 10150
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3787218664930866,
      "learning_rate": 8.883000108598083e-07,
      "loss": 0.6165,
      "step": 10151
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40931127908137577,
      "learning_rate": 8.865938128902218e-07,
      "loss": 0.6109,
      "step": 10152
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4423972357035262,
      "learning_rate": 8.84889217938929e-07,
      "loss": 0.6765,
      "step": 10153
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36011125103787106,
      "learning_rate": 8.831862261488955e-07,
      "loss": 0.6311,
      "step": 10154
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4107705461095146,
      "learning_rate": 8.814848376629426e-07,
      "loss": 0.5799,
      "step": 10155
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36409154566840796,
      "learning_rate": 8.797850526237673e-07,
      "loss": 0.5948,
      "step": 10156
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.372656473341017,
      "learning_rate": 8.780868711739287e-07,
      "loss": 0.6343,
      "step": 10157
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3897893146640491,
      "learning_rate": 8.763902934558488e-07,
      "loss": 0.618,
      "step": 10158
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43264915864532494,
      "learning_rate": 8.746953196118224e-07,
      "loss": 0.618,
      "step": 10159
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3935760423625791,
      "learning_rate": 8.730019497840025e-07,
      "loss": 0.6336,
      "step": 10160
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3921009157571301,
      "learning_rate": 8.71310184114409e-07,
      "loss": 0.6675,
      "step": 10161
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4854238429176563,
      "learning_rate": 8.696200227449259e-07,
      "loss": 0.6931,
      "step": 10162
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35382599845910173,
      "learning_rate": 8.679314658173133e-07,
      "loss": 0.614,
      "step": 10163
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38160740824787476,
      "learning_rate": 8.662445134731757e-07,
      "loss": 0.6018,
      "step": 10164
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37999282803058193,
      "learning_rate": 8.645591658540086e-07,
      "loss": 0.6271,
      "step": 10165
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3798855236697858,
      "learning_rate": 8.628754231011527e-07,
      "loss": 0.6351,
      "step": 10166
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37777645371751734,
      "learning_rate": 8.611932853558236e-07,
      "loss": 0.6249,
      "step": 10167
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41082756024986433,
      "learning_rate": 8.595127527591018e-07,
      "loss": 0.6529,
      "step": 10168
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41002304073863943,
      "learning_rate": 8.578338254519258e-07,
      "loss": 0.6163,
      "step": 10169
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37827149639696983,
      "learning_rate": 8.561565035751096e-07,
      "loss": 0.6236,
      "step": 10170
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41369133162647287,
      "learning_rate": 8.544807872693295e-07,
      "loss": 0.6347,
      "step": 10171
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35932979950710014,
      "learning_rate": 8.52806676675122e-07,
      "loss": 0.6305,
      "step": 10172
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5674889550679433,
      "learning_rate": 8.511341719328947e-07,
      "loss": 0.3823,
      "step": 10173
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39017951026642,
      "learning_rate": 8.4946327318292e-07,
      "loss": 0.6344,
      "step": 10174
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3886768493058735,
      "learning_rate": 8.477939805653324e-07,
      "loss": 0.6549,
      "step": 10175
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3856873321024434,
      "learning_rate": 8.46126294220131e-07,
      "loss": 0.6013,
      "step": 10176
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3991361667679436,
      "learning_rate": 8.444602142871905e-07,
      "loss": 0.6435,
      "step": 10177
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3761888646614751,
      "learning_rate": 8.427957409062304e-07,
      "loss": 0.6798,
      "step": 10178
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38091007525949827,
      "learning_rate": 8.411328742168612e-07,
      "loss": 0.6185,
      "step": 10179
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34667660568896913,
      "learning_rate": 8.394716143585379e-07,
      "loss": 0.6369,
      "step": 10180
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39194543313628755,
      "learning_rate": 8.378119614705915e-07,
      "loss": 0.6344,
      "step": 10181
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3859342718658171,
      "learning_rate": 8.361539156922127e-07,
      "loss": 0.6591,
      "step": 10182
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40448131521677233,
      "learning_rate": 8.344974771624614e-07,
      "loss": 0.6377,
      "step": 10183
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4418818461142567,
      "learning_rate": 8.328426460202599e-07,
      "loss": 0.6445,
      "step": 10184
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39549945517060653,
      "learning_rate": 8.311894224043971e-07,
      "loss": 0.5995,
      "step": 10185
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4007182851325226,
      "learning_rate": 8.295378064535286e-07,
      "loss": 0.6659,
      "step": 10186
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35862896069962874,
      "learning_rate": 8.278877983061706e-07,
      "loss": 0.6177,
      "step": 10187
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.47086499739613796,
      "learning_rate": 8.262393981007122e-07,
      "loss": 0.6591,
      "step": 10188
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3764446559063791,
      "learning_rate": 8.245926059753983e-07,
      "loss": 0.6352,
      "step": 10189
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3905342591581925,
      "learning_rate": 8.229474220683431e-07,
      "loss": 0.6325,
      "step": 10190
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4381821665606366,
      "learning_rate": 8.213038465175294e-07,
      "loss": 0.6748,
      "step": 10191
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39459154500003224,
      "learning_rate": 8.196618794607958e-07,
      "loss": 0.5978,
      "step": 10192
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3507864562294139,
      "learning_rate": 8.180215210358589e-07,
      "loss": 0.6607,
      "step": 10193
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3469759900935207,
      "learning_rate": 8.163827713802952e-07,
      "loss": 0.6502,
      "step": 10194
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39176622561906876,
      "learning_rate": 8.147456306315349e-07,
      "loss": 0.6448,
      "step": 10195
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4061855065669079,
      "learning_rate": 8.131100989268903e-07,
      "loss": 0.6046,
      "step": 10196
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3885355654460683,
      "learning_rate": 8.114761764035317e-07,
      "loss": 0.6228,
      "step": 10197
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4023451162192957,
      "learning_rate": 8.098438631984917e-07,
      "loss": 0.64,
      "step": 10198
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4039916188320664,
      "learning_rate": 8.082131594486698e-07,
      "loss": 0.658,
      "step": 10199
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3506718539548303,
      "learning_rate": 8.065840652908319e-07,
      "loss": 0.5979,
      "step": 10200
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.390320617387084,
      "learning_rate": 8.049565808616089e-07,
      "loss": 0.6291,
      "step": 10201
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38403110807891844,
      "learning_rate": 8.033307062974982e-07,
      "loss": 0.6101,
      "step": 10202
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3838450309982688,
      "learning_rate": 8.017064417348552e-07,
      "loss": 0.6161,
      "step": 10203
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39515646128955556,
      "learning_rate": 8.000837873099065e-07,
      "loss": 0.6371,
      "step": 10204
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5983007023611217,
      "learning_rate": 7.984627431587455e-07,
      "loss": 0.3599,
      "step": 10205
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3970677129265147,
      "learning_rate": 7.968433094173211e-07,
      "loss": 0.6171,
      "step": 10206
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37821043572889523,
      "learning_rate": 7.952254862214581e-07,
      "loss": 0.6064,
      "step": 10207
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3417205125274471,
      "learning_rate": 7.936092737068435e-07,
      "loss": 0.6135,
      "step": 10208
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4468117449669395,
      "learning_rate": 7.919946720090199e-07,
      "loss": 0.6644,
      "step": 10209
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42326960123752555,
      "learning_rate": 7.90381681263408e-07,
      "loss": 0.6185,
      "step": 10210
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39330796539627644,
      "learning_rate": 7.887703016052862e-07,
      "loss": 0.6592,
      "step": 10211
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41831443653550066,
      "learning_rate": 7.871605331697907e-07,
      "loss": 0.6226,
      "step": 10212
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.422383491466956,
      "learning_rate": 7.85552376091947e-07,
      "loss": 0.6352,
      "step": 10213
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38882434541199234,
      "learning_rate": 7.839458305066161e-07,
      "loss": 0.6153,
      "step": 10214
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39691109919085055,
      "learning_rate": 7.823408965485413e-07,
      "loss": 0.6593,
      "step": 10215
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38380106764202065,
      "learning_rate": 7.80737574352326e-07,
      "loss": 0.6509,
      "step": 10216
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38290504370495093,
      "learning_rate": 7.791358640524405e-07,
      "loss": 0.6518,
      "step": 10217
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3989872766420037,
      "learning_rate": 7.775357657832172e-07,
      "loss": 0.6334,
      "step": 10218
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41271097159355985,
      "learning_rate": 7.759372796788534e-07,
      "loss": 0.6974,
      "step": 10219
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4289134420888172,
      "learning_rate": 7.743404058734127e-07,
      "loss": 0.5968,
      "step": 10220
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4015341678032029,
      "learning_rate": 7.727451445008238e-07,
      "loss": 0.6807,
      "step": 10221
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3759193437987576,
      "learning_rate": 7.711514956948818e-07,
      "loss": 0.6294,
      "step": 10222
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39541450590779037,
      "learning_rate": 7.695594595892398e-07,
      "loss": 0.5731,
      "step": 10223
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5601409945790459,
      "learning_rate": 7.679690363174197e-07,
      "loss": 0.3701,
      "step": 10224
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39069670401031376,
      "learning_rate": 7.663802260128128e-07,
      "loss": 0.6386,
      "step": 10225
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3365045929442413,
      "learning_rate": 7.647930288086636e-07,
      "loss": 0.6293,
      "step": 10226
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4258231980778842,
      "learning_rate": 7.632074448380966e-07,
      "loss": 0.6521,
      "step": 10227
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39028173690104634,
      "learning_rate": 7.6162347423409e-07,
      "loss": 0.622,
      "step": 10228
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3681052623097597,
      "learning_rate": 7.600411171294841e-07,
      "loss": 0.6131,
      "step": 10229
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.38483814518330833,
      "learning_rate": 7.584603736569974e-07,
      "loss": 0.62,
      "step": 10230
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4124365452307054,
      "learning_rate": 7.568812439492013e-07,
      "loss": 0.6241,
      "step": 10231
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36671627853611954,
      "learning_rate": 7.553037281385344e-07,
      "loss": 0.6297,
      "step": 10232
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4091475378099595,
      "learning_rate": 7.537278263573022e-07,
      "loss": 0.6245,
      "step": 10233
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40889290102921266,
      "learning_rate": 7.521535387376722e-07,
      "loss": 0.633,
      "step": 10234
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4178393894312159,
      "learning_rate": 7.505808654116809e-07,
      "loss": 0.6876,
      "step": 10235
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3764427946151326,
      "learning_rate": 7.49009806511225e-07,
      "loss": 0.6515,
      "step": 10236
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3899984658185646,
      "learning_rate": 7.474403621680682e-07,
      "loss": 0.6586,
      "step": 10237
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3790271683310663,
      "learning_rate": 7.458725325138338e-07,
      "loss": 0.6306,
      "step": 10238
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3716272396105943,
      "learning_rate": 7.443063176800192e-07,
      "loss": 0.6608,
      "step": 10239
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5738368000585614,
      "learning_rate": 7.42741717797979e-07,
      "loss": 0.3752,
      "step": 10240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38825430739646427,
      "learning_rate": 7.411787329989329e-07,
      "loss": 0.645,
      "step": 10241
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3713615212733995,
      "learning_rate": 7.396173634139692e-07,
      "loss": 0.6404,
      "step": 10242
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43272795621370014,
      "learning_rate": 7.380576091740366e-07,
      "loss": 0.6322,
      "step": 10243
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3726401251482928,
      "learning_rate": 7.364994704099482e-07,
      "loss": 0.6299,
      "step": 10244
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39391130010257186,
      "learning_rate": 7.349429472523862e-07,
      "loss": 0.6302,
      "step": 10245
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.376942518224923,
      "learning_rate": 7.333880398318926e-07,
      "loss": 0.6263,
      "step": 10246
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37463713039372576,
      "learning_rate": 7.318347482788723e-07,
      "loss": 0.608,
      "step": 10247
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3684827671836201,
      "learning_rate": 7.302830727236077e-07,
      "loss": 0.6372,
      "step": 10248
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35236233213922374,
      "learning_rate": 7.287330132962255e-07,
      "loss": 0.6179,
      "step": 10249
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40560031658157497,
      "learning_rate": 7.271845701267333e-07,
      "loss": 0.6176,
      "step": 10250
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3747590067710463,
      "learning_rate": 7.256377433449979e-07,
      "loss": 0.6336,
      "step": 10251
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37318283468616187,
      "learning_rate": 7.240925330807447e-07,
      "loss": 0.6409,
      "step": 10252
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4223939716853092,
      "learning_rate": 7.225489394635698e-07,
      "loss": 0.6561,
      "step": 10253
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4064534534631369,
      "learning_rate": 7.210069626229365e-07,
      "loss": 0.6281,
      "step": 10254
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39270214690876737,
      "learning_rate": 7.194666026881658e-07,
      "loss": 0.6489,
      "step": 10255
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3736620901835954,
      "learning_rate": 7.179278597884476e-07,
      "loss": 0.6156,
      "step": 10256
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36543083837327756,
      "learning_rate": 7.163907340528297e-07,
      "loss": 0.6385,
      "step": 10257
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36388536847015696,
      "learning_rate": 7.148552256102337e-07,
      "loss": 0.6368,
      "step": 10258
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4050095395229897,
      "learning_rate": 7.133213345894407e-07,
      "loss": 0.6293,
      "step": 10259
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40358186047891226,
      "learning_rate": 7.117890611190948e-07,
      "loss": 0.6278,
      "step": 10260
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37690092953240356,
      "learning_rate": 7.102584053277017e-07,
      "loss": 0.6076,
      "step": 10261
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3922223439438687,
      "learning_rate": 7.087293673436457e-07,
      "loss": 0.6398,
      "step": 10262
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4177429841625243,
      "learning_rate": 7.072019472951575e-07,
      "loss": 0.6271,
      "step": 10263
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4197173219700345,
      "learning_rate": 7.056761453103412e-07,
      "loss": 0.6637,
      "step": 10264
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38099830016868474,
      "learning_rate": 7.041519615171677e-07,
      "loss": 0.6135,
      "step": 10265
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3951841861571363,
      "learning_rate": 7.026293960434615e-07,
      "loss": 0.6284,
      "step": 10266
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.45651471816672023,
      "learning_rate": 7.011084490169251e-07,
      "loss": 0.6968,
      "step": 10267
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40420231237112153,
      "learning_rate": 6.995891205651162e-07,
      "loss": 0.6107,
      "step": 10268
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40364915423520065,
      "learning_rate": 6.980714108154573e-07,
      "loss": 0.6564,
      "step": 10269
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3906188238147811,
      "learning_rate": 6.965553198952402e-07,
      "loss": 0.6547,
      "step": 10270
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.424036920677474,
      "learning_rate": 6.950408479316162e-07,
      "loss": 0.6192,
      "step": 10271
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4123689364407162,
      "learning_rate": 6.935279950515994e-07,
      "loss": 0.6471,
      "step": 10272
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36718327160452197,
      "learning_rate": 6.920167613820728e-07,
      "loss": 0.6329,
      "step": 10273
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4214536201594885,
      "learning_rate": 6.905071470497859e-07,
      "loss": 0.6477,
      "step": 10274
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36285268645693536,
      "learning_rate": 6.8899915218134e-07,
      "loss": 0.6449,
      "step": 10275
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41999873553551975,
      "learning_rate": 6.874927769032181e-07,
      "loss": 0.643,
      "step": 10276
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4293214239111961,
      "learning_rate": 6.859880213417502e-07,
      "loss": 0.6435,
      "step": 10277
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42006796191061074,
      "learning_rate": 6.844848856231423e-07,
      "loss": 0.6409,
      "step": 10278
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3897329063158213,
      "learning_rate": 6.829833698734623e-07,
      "loss": 0.6518,
      "step": 10279
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3830556746718003,
      "learning_rate": 6.814834742186361e-07,
      "loss": 0.5866,
      "step": 10280
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42851201394105326,
      "learning_rate": 6.799851987844585e-07,
      "loss": 0.6202,
      "step": 10281
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5675099080512512,
      "learning_rate": 6.784885436965915e-07,
      "loss": 0.3563,
      "step": 10282
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3372741155203317,
      "learning_rate": 6.769935090805546e-07,
      "loss": 0.5716,
      "step": 10283
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3928231718874965,
      "learning_rate": 6.755000950617363e-07,
      "loss": 0.6216,
      "step": 10284
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39546570384060453,
      "learning_rate": 6.740083017653875e-07,
      "loss": 0.5539,
      "step": 10285
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43849315401102323,
      "learning_rate": 6.725181293166217e-07,
      "loss": 0.6195,
      "step": 10286
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36895423217462925,
      "learning_rate": 6.710295778404185e-07,
      "loss": 0.6188,
      "step": 10287
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4191893255987076,
      "learning_rate": 6.695426474616251e-07,
      "loss": 0.66,
      "step": 10288
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4378159694277018,
      "learning_rate": 6.680573383049371e-07,
      "loss": 0.6666,
      "step": 10289
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3989994684382331,
      "learning_rate": 6.665736504949394e-07,
      "loss": 0.6376,
      "step": 10290
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3700042216699241,
      "learning_rate": 6.65091584156059e-07,
      "loss": 0.6258,
      "step": 10291
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38525017432872033,
      "learning_rate": 6.636111394125966e-07,
      "loss": 0.6302,
      "step": 10292
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36984558976617243,
      "learning_rate": 6.621323163887172e-07,
      "loss": 0.6127,
      "step": 10293
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40065875697182235,
      "learning_rate": 6.606551152084439e-07,
      "loss": 0.6179,
      "step": 10294
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.45354351143864646,
      "learning_rate": 6.591795359956665e-07,
      "loss": 0.7109,
      "step": 10295
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38488621430618863,
      "learning_rate": 6.577055788741505e-07,
      "loss": 0.5906,
      "step": 10296
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41671380832087596,
      "learning_rate": 6.562332439675034e-07,
      "loss": 0.641,
      "step": 10297
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39687915441707117,
      "learning_rate": 6.547625313992134e-07,
      "loss": 0.6393,
      "step": 10298
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3906619208780261,
      "learning_rate": 6.532934412926306e-07,
      "loss": 0.6417,
      "step": 10299
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3618201565138143,
      "learning_rate": 6.518259737709587e-07,
      "loss": 0.5987,
      "step": 10300
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.385805357569824,
      "learning_rate": 6.50360128957277e-07,
      "loss": 0.626,
      "step": 10301
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4042243386673011,
      "learning_rate": 6.488959069745204e-07,
      "loss": 0.6691,
      "step": 10302
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4234018113911961,
      "learning_rate": 6.474333079454953e-07,
      "loss": 0.6524,
      "step": 10303
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38793805793681774,
      "learning_rate": 6.459723319928657e-07,
      "loss": 0.6652,
      "step": 10304
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41045905018310525,
      "learning_rate": 6.445129792391668e-07,
      "loss": 0.6366,
      "step": 10305
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4043092299652242,
      "learning_rate": 6.430552498067854e-07,
      "loss": 0.6362,
      "step": 10306
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34579621413517336,
      "learning_rate": 6.415991438179836e-07,
      "loss": 0.5935,
      "step": 10307
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3515465131810933,
      "learning_rate": 6.401446613948836e-07,
      "loss": 0.605,
      "step": 10308
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4276744193024326,
      "learning_rate": 6.386918026594657e-07,
      "loss": 0.6652,
      "step": 10309
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39060791075710066,
      "learning_rate": 6.372405677335902e-07,
      "loss": 0.6527,
      "step": 10310
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3966400177686848,
      "learning_rate": 6.357909567389597e-07,
      "loss": 0.642,
      "step": 10311
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3472303764711765,
      "learning_rate": 6.34342969797157e-07,
      "loss": 0.6399,
      "step": 10312
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41722031649778013,
      "learning_rate": 6.328966070296227e-07,
      "loss": 0.6359,
      "step": 10313
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4007590948863999,
      "learning_rate": 6.314518685576598e-07,
      "loss": 0.6593,
      "step": 10314
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4153849243893747,
      "learning_rate": 6.300087545024358e-07,
      "loss": 0.6537,
      "step": 10315
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3490432160499428,
      "learning_rate": 6.28567264984985e-07,
      "loss": 0.6176,
      "step": 10316
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39966963747447937,
      "learning_rate": 6.27127400126204e-07,
      "loss": 0.6277,
      "step": 10317
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3441451063862424,
      "learning_rate": 6.256891600468496e-07,
      "loss": 0.6069,
      "step": 10318
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3776175167826139,
      "learning_rate": 6.242525448675496e-07,
      "loss": 0.6611,
      "step": 10319
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37095345724102047,
      "learning_rate": 6.228175547087878e-07,
      "loss": 0.5915,
      "step": 10320
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.46167683152719086,
      "learning_rate": 6.213841896909145e-07,
      "loss": 0.7046,
      "step": 10321
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40293100250249214,
      "learning_rate": 6.199524499341492e-07,
      "loss": 0.6377,
      "step": 10322
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4200628891951265,
      "learning_rate": 6.185223355585623e-07,
      "loss": 0.6204,
      "step": 10323
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4130943860630399,
      "learning_rate": 6.170938466841026e-07,
      "loss": 0.655,
      "step": 10324
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3954777467641533,
      "learning_rate": 6.156669834305739e-07,
      "loss": 0.6203,
      "step": 10325
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4133463154596414,
      "learning_rate": 6.142417459176431e-07,
      "loss": 0.6361,
      "step": 10326
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40810616174491515,
      "learning_rate": 6.128181342648476e-07,
      "loss": 0.6647,
      "step": 10327
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.44707939252361795,
      "learning_rate": 6.113961485915809e-07,
      "loss": 0.6819,
      "step": 10328
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3616417522511146,
      "learning_rate": 6.09975789017101e-07,
      "loss": 0.5977,
      "step": 10329
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.49743004173748157,
      "learning_rate": 6.085570556605369e-07,
      "loss": 0.3595,
      "step": 10330
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4373457339456352,
      "learning_rate": 6.071399486408713e-07,
      "loss": 0.6755,
      "step": 10331
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4843572562880243,
      "learning_rate": 6.057244680769603e-07,
      "loss": 0.6907,
      "step": 10332
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3794680488590656,
      "learning_rate": 6.043106140875155e-07,
      "loss": 0.6146,
      "step": 10333
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34775844276510226,
      "learning_rate": 6.028983867911132e-07,
      "loss": 0.5973,
      "step": 10334
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40870809663268315,
      "learning_rate": 6.014877863061985e-07,
      "loss": 0.6178,
      "step": 10335
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3760929098007324,
      "learning_rate": 6.000788127510792e-07,
      "loss": 0.6395,
      "step": 10336
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40820798780473233,
      "learning_rate": 5.986714662439142e-07,
      "loss": 0.6316,
      "step": 10337
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.361645936524467,
      "learning_rate": 5.972657469027443e-07,
      "loss": 0.6353,
      "step": 10338
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4102177831689443,
      "learning_rate": 5.958616548454688e-07,
      "loss": 0.6461,
      "step": 10339
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4019677094210011,
      "learning_rate": 5.944591901898377e-07,
      "loss": 0.6305,
      "step": 10340
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4016747434304648,
      "learning_rate": 5.930583530534772e-07,
      "loss": 0.6311,
      "step": 10341
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3766546906624927,
      "learning_rate": 5.916591435538798e-07,
      "loss": 0.642,
      "step": 10342
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35524113856219813,
      "learning_rate": 5.902615618083874e-07,
      "loss": 0.6343,
      "step": 10343
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.45400454303901067,
      "learning_rate": 5.888656079342169e-07,
      "loss": 0.6856,
      "step": 10344
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4114066116286767,
      "learning_rate": 5.87471282048444e-07,
      "loss": 0.574,
      "step": 10345
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4155086523963529,
      "learning_rate": 5.860785842680106e-07,
      "loss": 0.6803,
      "step": 10346
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37244199297137204,
      "learning_rate": 5.846875147097231e-07,
      "loss": 0.647,
      "step": 10347
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41271013014123126,
      "learning_rate": 5.832980734902438e-07,
      "loss": 0.6721,
      "step": 10348
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.357628582941606,
      "learning_rate": 5.819102607261062e-07,
      "loss": 0.6376,
      "step": 10349
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3966471069775732,
      "learning_rate": 5.805240765337039e-07,
      "loss": 0.6455,
      "step": 10350
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42314616839888525,
      "learning_rate": 5.791395210292927e-07,
      "loss": 0.6242,
      "step": 10351
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3968661634376716,
      "learning_rate": 5.777565943289975e-07,
      "loss": 0.6419,
      "step": 10352
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3750400850166692,
      "learning_rate": 5.763752965488012e-07,
      "loss": 0.6689,
      "step": 10353
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4660917083864201,
      "learning_rate": 5.749956278045488e-07,
      "loss": 0.6092,
      "step": 10354
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37704683564635383,
      "learning_rate": 5.736175882119544e-07,
      "loss": 0.6538,
      "step": 10355
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3897397638114895,
      "learning_rate": 5.722411778865943e-07,
      "loss": 0.6395,
      "step": 10356
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39430809405511585,
      "learning_rate": 5.70866396943901e-07,
      "loss": 0.6658,
      "step": 10357
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4084855870646372,
      "learning_rate": 5.694932454991798e-07,
      "loss": 0.6448,
      "step": 10358
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3897610011511937,
      "learning_rate": 5.681217236675918e-07,
      "loss": 0.6423,
      "step": 10359
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3875902447415527,
      "learning_rate": 5.667518315641673e-07,
      "loss": 0.6464,
      "step": 10360
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.38945939704946286,
      "learning_rate": 5.65383569303799e-07,
      "loss": 0.6445,
      "step": 10361
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36037607424553936,
      "learning_rate": 5.640169370012416e-07,
      "loss": 0.5943,
      "step": 10362
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4042644431268441,
      "learning_rate": 5.626519347711079e-07,
      "loss": 0.6181,
      "step": 10363
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4134422903909015,
      "learning_rate": 5.612885627278819e-07,
      "loss": 0.6791,
      "step": 10364
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5268547028118833,
      "learning_rate": 5.599268209859077e-07,
      "loss": 0.3321,
      "step": 10365
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40670120350731875,
      "learning_rate": 5.585667096593961e-07,
      "loss": 0.6382,
      "step": 10366
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3926445599889307,
      "learning_rate": 5.572082288624136e-07,
      "loss": 0.6198,
      "step": 10367
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3685203230913156,
      "learning_rate": 5.558513787088959e-07,
      "loss": 0.6416,
      "step": 10368
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42763185824980887,
      "learning_rate": 5.544961593126408e-07,
      "loss": 0.5943,
      "step": 10369
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.38702787228725904,
      "learning_rate": 5.531425707873106e-07,
      "loss": 0.6522,
      "step": 10370
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39135980928927266,
      "learning_rate": 5.517906132464257e-07,
      "loss": 0.5941,
      "step": 10371
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3772566612476143,
      "learning_rate": 5.504402868033709e-07,
      "loss": 0.6087,
      "step": 10372
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37314961878012104,
      "learning_rate": 5.490915915714046e-07,
      "loss": 0.6302,
      "step": 10373
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5839960816348488,
      "learning_rate": 5.477445276636339e-07,
      "loss": 0.3835,
      "step": 10374
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3901145179332551,
      "learning_rate": 5.463990951930376e-07,
      "loss": 0.6323,
      "step": 10375
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4004618865136483,
      "learning_rate": 5.450552942724563e-07,
      "loss": 0.6325,
      "step": 10376
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4208629337787624,
      "learning_rate": 5.43713125014591e-07,
      "loss": 0.6245,
      "step": 10377
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43828827956556476,
      "learning_rate": 5.423725875320074e-07,
      "loss": 0.6346,
      "step": 10378
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40589101032624075,
      "learning_rate": 5.410336819371375e-07,
      "loss": 0.6495,
      "step": 10379
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37804352367337435,
      "learning_rate": 5.396964083422718e-07,
      "loss": 0.6585,
      "step": 10380
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3583003390211956,
      "learning_rate": 5.383607668595669e-07,
      "loss": 0.6129,
      "step": 10381
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3976917748594757,
      "learning_rate": 5.370267576010402e-07,
      "loss": 0.6524,
      "step": 10382
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3989070038863459,
      "learning_rate": 5.356943806785731e-07,
      "loss": 0.6421,
      "step": 10383
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3906231088222767,
      "learning_rate": 5.34363636203914e-07,
      "loss": 0.6458,
      "step": 10384
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3876382254259484,
      "learning_rate": 5.330345242886648e-07,
      "loss": 0.6298,
      "step": 10385
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.423452905142734,
      "learning_rate": 5.317070450443007e-07,
      "loss": 0.6085,
      "step": 10386
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.38479556850870966,
      "learning_rate": 5.303811985821572e-07,
      "loss": 0.6609,
      "step": 10387
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4179884869186322,
      "learning_rate": 5.290569850134275e-07,
      "loss": 0.637,
      "step": 10388
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40815353626093254,
      "learning_rate": 5.277344044491739e-07,
      "loss": 0.6418,
      "step": 10389
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39185023871197217,
      "learning_rate": 5.26413457000321e-07,
      "loss": 0.587,
      "step": 10390
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43025995203054257,
      "learning_rate": 5.250941427776513e-07,
      "loss": 0.6429,
      "step": 10391
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40109653459114203,
      "learning_rate": 5.237764618918162e-07,
      "loss": 0.6684,
      "step": 10392
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.405576920520548,
      "learning_rate": 5.224604144533274e-07,
      "loss": 0.633,
      "step": 10393
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3949372441903522,
      "learning_rate": 5.21146000572561e-07,
      "loss": 0.6245,
      "step": 10394
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3976759836229344,
      "learning_rate": 5.198332203597555e-07,
      "loss": 0.6062,
      "step": 10395
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3484830680465546,
      "learning_rate": 5.185220739250141e-07,
      "loss": 0.6197,
      "step": 10396
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.38756873233895467,
      "learning_rate": 5.172125613782952e-07,
      "loss": 0.6132,
      "step": 10397
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.418776946216264,
      "learning_rate": 5.15904682829429e-07,
      "loss": 0.6496,
      "step": 10398
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39287970935957234,
      "learning_rate": 5.145984383881075e-07,
      "loss": 0.6155,
      "step": 10399
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.426746945220549,
      "learning_rate": 5.132938281638833e-07,
      "loss": 0.6481,
      "step": 10400
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4135123319501924,
      "learning_rate": 5.119908522661732e-07,
      "loss": 0.6161,
      "step": 10401
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37276159272534665,
      "learning_rate": 5.10689510804252e-07,
      "loss": 0.6264,
      "step": 10402
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4077930336730328,
      "learning_rate": 5.093898038872658e-07,
      "loss": 0.6565,
      "step": 10403
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3823648501638938,
      "learning_rate": 5.080917316242184e-07,
      "loss": 0.638,
      "step": 10404
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39228794810291584,
      "learning_rate": 5.067952941239762e-07,
      "loss": 0.6405,
      "step": 10405
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40047020924052384,
      "learning_rate": 5.0550049149527e-07,
      "loss": 0.6449,
      "step": 10406
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41940787176909927,
      "learning_rate": 5.042073238466949e-07,
      "loss": 0.6493,
      "step": 10407
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43776501380544314,
      "learning_rate": 5.029157912867067e-07,
      "loss": 0.6538,
      "step": 10408
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4250325017476537,
      "learning_rate": 5.016258939236252e-07,
      "loss": 0.6318,
      "step": 10409
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3604511913193395,
      "learning_rate": 5.003376318656328e-07,
      "loss": 0.6056,
      "step": 10410
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41399805750366553,
      "learning_rate": 4.99051005220772e-07,
      "loss": 0.6089,
      "step": 10411
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36953258066771444,
      "learning_rate": 4.97766014096952e-07,
      "loss": 0.608,
      "step": 10412
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40759046951124606,
      "learning_rate": 4.964826586019444e-07,
      "loss": 0.6386,
      "step": 10413
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39824664879965876,
      "learning_rate": 4.952009388433809e-07,
      "loss": 0.6261,
      "step": 10414
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3976102819947407,
      "learning_rate": 4.939208549287622e-07,
      "loss": 0.6677,
      "step": 10415
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3911619488299982,
      "learning_rate": 4.9264240696544e-07,
      "loss": 0.6452,
      "step": 10416
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40706284194890524,
      "learning_rate": 4.913655950606422e-07,
      "loss": 0.6528,
      "step": 10417
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37828899180885356,
      "learning_rate": 4.900904193214495e-07,
      "loss": 0.6384,
      "step": 10418
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39456084475287617,
      "learning_rate": 4.888168798548143e-07,
      "loss": 0.6432,
      "step": 10419
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3929012878268707,
      "learning_rate": 4.875449767675399e-07,
      "loss": 0.5978,
      "step": 10420
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39006459944595645,
      "learning_rate": 4.862747101663079e-07,
      "loss": 0.6387,
      "step": 10421
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41811140287630666,
      "learning_rate": 4.850060801576461e-07,
      "loss": 0.6335,
      "step": 10422
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39973726403251897,
      "learning_rate": 4.837390868479564e-07,
      "loss": 0.6136,
      "step": 10423
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3851079002299025,
      "learning_rate": 4.824737303435023e-07,
      "loss": 0.6125,
      "step": 10424
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3999830542052127,
      "learning_rate": 4.812100107504037e-07,
      "loss": 0.6328,
      "step": 10425
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34523448342908636,
      "learning_rate": 4.799479281746488e-07,
      "loss": 0.6072,
      "step": 10426
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3577005775514754,
      "learning_rate": 4.786874827220844e-07,
      "loss": 0.6464,
      "step": 10427
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37008324688436006,
      "learning_rate": 4.77428674498428e-07,
      "loss": 0.5975,
      "step": 10428
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4357750943371115,
      "learning_rate": 4.761715036092507e-07,
      "loss": 0.6731,
      "step": 10429
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5639950233992789,
      "learning_rate": 4.7491597015999257e-07,
      "loss": 0.3422,
      "step": 10430
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36637565522607224,
      "learning_rate": 4.7366207425594945e-07,
      "loss": 0.6417,
      "step": 10431
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40184795176482174,
      "learning_rate": 4.724098160022861e-07,
      "loss": 0.6325,
      "step": 10432
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3692285077273809,
      "learning_rate": 4.7115919550402953e-07,
      "loss": 0.6368,
      "step": 10433
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43970582572772904,
      "learning_rate": 4.6991021286606263e-07,
      "loss": 0.6161,
      "step": 10434
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3643419428345923,
      "learning_rate": 4.686628681931438e-07,
      "loss": 0.6172,
      "step": 10435
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5587988156846689,
      "learning_rate": 4.674171615898826e-07,
      "loss": 0.355,
      "step": 10436
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4027918797904255,
      "learning_rate": 4.661730931607511e-07,
      "loss": 0.6249,
      "step": 10437
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3626862448652229,
      "learning_rate": 4.649306630100947e-07,
      "loss": 0.5875,
      "step": 10438
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37838110014280746,
      "learning_rate": 4.636898712421101e-07,
      "loss": 0.6211,
      "step": 10439
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40764243796576166,
      "learning_rate": 4.624507179608628e-07,
      "loss": 0.6138,
      "step": 10440
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4241281618214134,
      "learning_rate": 4.612132032702765e-07,
      "loss": 0.6618,
      "step": 10441
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4205420854770735,
      "learning_rate": 4.599773272741437e-07,
      "loss": 0.6175,
      "step": 10442
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36441661516469565,
      "learning_rate": 4.587430900761125e-07,
      "loss": 0.665,
      "step": 10443
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3739275977444322,
      "learning_rate": 4.575104917797002e-07,
      "loss": 0.6426,
      "step": 10444
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.377280088953712,
      "learning_rate": 4.5627953248828183e-07,
      "loss": 0.5951,
      "step": 10445
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4830701852992341,
      "learning_rate": 4.550502123050948e-07,
      "loss": 0.6716,
      "step": 10446
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4154331996186397,
      "learning_rate": 4.538225313332434e-07,
      "loss": 0.6507,
      "step": 10447
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4181770995592794,
      "learning_rate": 4.525964896756874e-07,
      "loss": 0.6573,
      "step": 10448
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.45597138452779734,
      "learning_rate": 4.5137208743526005e-07,
      "loss": 0.6556,
      "step": 10449
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.33816385948918715,
      "learning_rate": 4.50149324714646e-07,
      "loss": 0.6016,
      "step": 10450
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39529837053342215,
      "learning_rate": 4.489282016163965e-07,
      "loss": 0.6319,
      "step": 10451
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36600767182589544,
      "learning_rate": 4.477087182429274e-07,
      "loss": 0.6344,
      "step": 10452
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3908380813073999,
      "learning_rate": 4.4649087469651685e-07,
      "loss": 0.6428,
      "step": 10453
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34304536282965215,
      "learning_rate": 4.45274671079301e-07,
      "loss": 0.585,
      "step": 10454
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5488730117459035,
      "learning_rate": 4.4406010749328265e-07,
      "loss": 0.3606,
      "step": 10455
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36308039494143696,
      "learning_rate": 4.428471840403248e-07,
      "loss": 0.6014,
      "step": 10456
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3775639931291771,
      "learning_rate": 4.416359008221527e-07,
      "loss": 0.6104,
      "step": 10457
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3888105814922887,
      "learning_rate": 4.4042625794036063e-07,
      "loss": 0.6672,
      "step": 10458
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3547405125811424,
      "learning_rate": 4.392182554963942e-07,
      "loss": 0.605,
      "step": 10459
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37105029008358426,
      "learning_rate": 4.3801189359157006e-07,
      "loss": 0.641,
      "step": 10460
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3827205255940511,
      "learning_rate": 4.3680717232706506e-07,
      "loss": 0.6055,
      "step": 10461
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3726109167016907,
      "learning_rate": 4.3560409180391174e-07,
      "loss": 0.639,
      "step": 10462
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4136661795335209,
      "learning_rate": 4.3440265212301826e-07,
      "loss": 0.6537,
      "step": 10463
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3606495292856476,
      "learning_rate": 4.332028533851462e-07,
      "loss": 0.6046,
      "step": 10464
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3774226319025783,
      "learning_rate": 4.320046956909174e-07,
      "loss": 0.6369,
      "step": 10465
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40276247142479665,
      "learning_rate": 4.308081791408247e-07,
      "loss": 0.6476,
      "step": 10466
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4492097760770383,
      "learning_rate": 4.296133038352168e-07,
      "loss": 0.6459,
      "step": 10467
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41255929739758035,
      "learning_rate": 4.284200698743024e-07,
      "loss": 0.6442,
      "step": 10468
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5367868364081959,
      "learning_rate": 4.2722847735816585e-07,
      "loss": 0.3352,
      "step": 10469
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.43420611889151217,
      "learning_rate": 4.2603852638673393e-07,
      "loss": 0.6574,
      "step": 10470
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37212844359943437,
      "learning_rate": 4.2485021705981343e-07,
      "loss": 0.6355,
      "step": 10471
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4167608161819887,
      "learning_rate": 4.236635494770669e-07,
      "loss": 0.6551,
      "step": 10472
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41425645679376555,
      "learning_rate": 4.2247852373801245e-07,
      "loss": 0.6189,
      "step": 10473
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40430584232218547,
      "learning_rate": 4.2129513994204176e-07,
      "loss": 0.6351,
      "step": 10474
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36520664899261684,
      "learning_rate": 4.20113398188402e-07,
      "loss": 0.6361,
      "step": 10475
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.435627329840717,
      "learning_rate": 4.1893329857620735e-07,
      "loss": 0.6554,
      "step": 10476
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37226445325118174,
      "learning_rate": 4.1775484120442524e-07,
      "loss": 0.6321,
      "step": 10477
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4248005701529895,
      "learning_rate": 4.16578026171901e-07,
      "loss": 0.6371,
      "step": 10478
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3599238806041827,
      "learning_rate": 4.1540285357732246e-07,
      "loss": 0.6228,
      "step": 10479
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36994829667261675,
      "learning_rate": 4.142293235192552e-07,
      "loss": 0.626,
      "step": 10480
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38858331246064953,
      "learning_rate": 4.130574360961226e-07,
      "loss": 0.6629,
      "step": 10481
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38441548196281017,
      "learning_rate": 4.118871914062039e-07,
      "loss": 0.6617,
      "step": 10482
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3601777468332952,
      "learning_rate": 4.107185895476518e-07,
      "loss": 0.6254,
      "step": 10483
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4270496809550844,
      "learning_rate": 4.0955163061847434e-07,
      "loss": 0.624,
      "step": 10484
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3554205895358206,
      "learning_rate": 4.083863147165423e-07,
      "loss": 0.6238,
      "step": 10485
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40096625417357884,
      "learning_rate": 4.072226419395886e-07,
      "loss": 0.645,
      "step": 10486
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37245162079679905,
      "learning_rate": 4.0606061238521287e-07,
      "loss": 0.6412,
      "step": 10487
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4078096009056408,
      "learning_rate": 4.049002261508683e-07,
      "loss": 0.6148,
      "step": 10488
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4310627400865359,
      "learning_rate": 4.0374148333387484e-07,
      "loss": 0.6201,
      "step": 10489
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4055859145538487,
      "learning_rate": 4.025843840314192e-07,
      "loss": 0.6445,
      "step": 10490
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39864292811908925,
      "learning_rate": 4.014289283405415e-07,
      "loss": 0.6455,
      "step": 10491
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.435000514290472,
      "learning_rate": 4.0027511635815306e-07,
      "loss": 0.6762,
      "step": 10492
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3805523374050386,
      "learning_rate": 3.991229481810188e-07,
      "loss": 0.6175,
      "step": 10493
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36806582258364307,
      "learning_rate": 3.979724239057703e-07,
      "loss": 0.618,
      "step": 10494
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40003363880513243,
      "learning_rate": 3.968235436289036e-07,
      "loss": 0.6165,
      "step": 10495
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4061281404465298,
      "learning_rate": 3.9567630744676843e-07,
      "loss": 0.6485,
      "step": 10496
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39586644172222135,
      "learning_rate": 3.9453071545558776e-07,
      "loss": 0.6031,
      "step": 10497
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3836581236354688,
      "learning_rate": 3.93386767751438e-07,
      "loss": 0.6165,
      "step": 10498
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4154530662773746,
      "learning_rate": 3.922444644302625e-07,
      "loss": 0.6487,
      "step": 10499
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3752726976884155,
      "learning_rate": 3.9110380558786023e-07,
      "loss": 0.6378,
      "step": 10500
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4035716435464034,
      "learning_rate": 3.899647913199056e-07,
      "loss": 0.6469,
      "step": 10501
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38465503215762137,
      "learning_rate": 3.888274217219157e-07,
      "loss": 0.6597,
      "step": 10502
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41815950361727355,
      "learning_rate": 3.876916968892874e-07,
      "loss": 0.6678,
      "step": 10503
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3896671082338443,
      "learning_rate": 3.865576169172713e-07,
      "loss": 0.6436,
      "step": 10504
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36472851583622523,
      "learning_rate": 3.8542518190098023e-07,
      "loss": 0.646,
      "step": 10505
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36268006447421686,
      "learning_rate": 3.842943919353914e-07,
      "loss": 0.5966,
      "step": 10506
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3500737887128337,
      "learning_rate": 3.831652471153424e-07,
      "loss": 0.5913,
      "step": 10507
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3999852579637542,
      "learning_rate": 3.82037747535533e-07,
      "loss": 0.6201,
      "step": 10508
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3883848928264151,
      "learning_rate": 3.809118932905231e-07,
      "loss": 0.6347,
      "step": 10509
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40643318310080756,
      "learning_rate": 3.7978768447473947e-07,
      "loss": 0.6322,
      "step": 10510
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4068443524673454,
      "learning_rate": 3.7866512118246876e-07,
      "loss": 0.6468,
      "step": 10511
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3889858023216157,
      "learning_rate": 3.77544203507858e-07,
      "loss": 0.6416,
      "step": 10512
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33974243469589066,
      "learning_rate": 3.7642493154491644e-07,
      "loss": 0.6093,
      "step": 10513
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4174320547312435,
      "learning_rate": 3.753073053875178e-07,
      "loss": 0.6693,
      "step": 10514
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37958802625955756,
      "learning_rate": 3.7419132512939384e-07,
      "loss": 0.5867,
      "step": 10515
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3803512751484847,
      "learning_rate": 3.730769908641407e-07,
      "loss": 0.6225,
      "step": 10516
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42200389460773713,
      "learning_rate": 3.7196430268521486e-07,
      "loss": 0.6016,
      "step": 10517
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45449101185964375,
      "learning_rate": 3.7085326068594383e-07,
      "loss": 0.6169,
      "step": 10518
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39870388663089856,
      "learning_rate": 3.6974386495949977e-07,
      "loss": 0.6519,
      "step": 10519
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3708192428768857,
      "learning_rate": 3.686361155989304e-07,
      "loss": 0.6425,
      "step": 10520
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41439305575942914,
      "learning_rate": 3.6753001269714597e-07,
      "loss": 0.612,
      "step": 10521
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40146042374563684,
      "learning_rate": 3.6642555634690547e-07,
      "loss": 0.6327,
      "step": 10522
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39327871121139535,
      "learning_rate": 3.6532274664084157e-07,
      "loss": 0.6502,
      "step": 10523
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4036490689354809,
      "learning_rate": 3.6422158367144687e-07,
      "loss": 0.6391,
      "step": 10524
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42530869484253536,
      "learning_rate": 3.6312206753107427e-07,
      "loss": 0.6428,
      "step": 10525
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3626955528293656,
      "learning_rate": 3.620241983119388e-07,
      "loss": 0.6285,
      "step": 10526
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38338463694554303,
      "learning_rate": 3.6092797610611794e-07,
      "loss": 0.6281,
      "step": 10527
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3641600509097555,
      "learning_rate": 3.598334010055471e-07,
      "loss": 0.6489,
      "step": 10528
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.381771359106889,
      "learning_rate": 3.5874047310203276e-07,
      "loss": 0.6486,
      "step": 10529
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3608087290383516,
      "learning_rate": 3.5764919248723274e-07,
      "loss": 0.6309,
      "step": 10530
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39233097561686,
      "learning_rate": 3.565595592526716e-07,
      "loss": 0.6618,
      "step": 10531
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3515448657507715,
      "learning_rate": 3.554715734897385e-07,
      "loss": 0.6051,
      "step": 10532
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42911844823449924,
      "learning_rate": 3.543852352896804e-07,
      "loss": 0.629,
      "step": 10533
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37363430851708657,
      "learning_rate": 3.5330054474360666e-07,
      "loss": 0.6492,
      "step": 10534
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38707358680223775,
      "learning_rate": 3.5221750194248896e-07,
      "loss": 0.6013,
      "step": 10535
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4137573482128373,
      "learning_rate": 3.5113610697716126e-07,
      "loss": 0.6901,
      "step": 10536
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38974749878882886,
      "learning_rate": 3.5005635993831554e-07,
      "loss": 0.6896,
      "step": 10537
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4343140819000777,
      "learning_rate": 3.489782609165149e-07,
      "loss": 0.6392,
      "step": 10538
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36778206605074604,
      "learning_rate": 3.479018100021736e-07,
      "loss": 0.5952,
      "step": 10539
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3859237086704909,
      "learning_rate": 3.4682700728557286e-07,
      "loss": 0.6014,
      "step": 10540
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38177761828571266,
      "learning_rate": 3.4575385285685824e-07,
      "loss": 0.6456,
      "step": 10541
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3423826741653555,
      "learning_rate": 3.446823468060312e-07,
      "loss": 0.6277,
      "step": 10542
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3963459035318057,
      "learning_rate": 3.436124892229575e-07,
      "loss": 0.6628,
      "step": 10543
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3871021993057459,
      "learning_rate": 3.4254428019736774e-07,
      "loss": 0.6288,
      "step": 10544
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40310549590324835,
      "learning_rate": 3.414777198188435e-07,
      "loss": 0.6152,
      "step": 10545
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.35218622135501626,
      "learning_rate": 3.4041280817684654e-07,
      "loss": 0.6338,
      "step": 10546
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3881517436352289,
      "learning_rate": 3.393495453606832e-07,
      "loss": 0.6842,
      "step": 10547
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41191466864756654,
      "learning_rate": 3.382879314595289e-07,
      "loss": 0.6578,
      "step": 10548
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3716304552988499,
      "learning_rate": 3.372279665624212e-07,
      "loss": 0.6265,
      "step": 10549
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3943914577994359,
      "learning_rate": 3.3616965075825794e-07,
      "loss": 0.6247,
      "step": 10550
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3715901187079162,
      "learning_rate": 3.3511298413579475e-07,
      "loss": 0.6115,
      "step": 10551
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3936401277895846,
      "learning_rate": 3.340579667836585e-07,
      "loss": 0.6452,
      "step": 10552
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36949807690551034,
      "learning_rate": 3.3300459879033185e-07,
      "loss": 0.6679,
      "step": 10553
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3293169590413432,
      "learning_rate": 3.3195288024415515e-07,
      "loss": 0.5896,
      "step": 10554
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33712123600842064,
      "learning_rate": 3.309028112333401e-07,
      "loss": 0.639,
      "step": 10555
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3788663382797778,
      "learning_rate": 3.298543918459496e-07,
      "loss": 0.6266,
      "step": 10556
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3366481501072782,
      "learning_rate": 3.2880762216991544e-07,
      "loss": 0.6511,
      "step": 10557
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3730388027705992,
      "learning_rate": 3.2776250229302977e-07,
      "loss": 0.6428,
      "step": 10558
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3859403888672382,
      "learning_rate": 3.2671903230294677e-07,
      "loss": 0.6186,
      "step": 10559
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4058297309249051,
      "learning_rate": 3.256772122871787e-07,
      "loss": 0.613,
      "step": 10560
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3756258006736735,
      "learning_rate": 3.2463704233310025e-07,
      "loss": 0.6584,
      "step": 10561
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5316021706322871,
      "learning_rate": 3.2359852252795256e-07,
      "loss": 0.3503,
      "step": 10562
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4223055814433001,
      "learning_rate": 3.225616529588349e-07,
      "loss": 0.6258,
      "step": 10563
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36575173449131676,
      "learning_rate": 3.215264337127066e-07,
      "loss": 0.625,
      "step": 10564
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44681787840473824,
      "learning_rate": 3.204928648763872e-07,
      "loss": 0.6028,
      "step": 10565
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37088445647162616,
      "learning_rate": 3.1946094653656945e-07,
      "loss": 0.6042,
      "step": 10566
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3889177565265099,
      "learning_rate": 3.184306787797908e-07,
      "loss": 0.6564,
      "step": 10567
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38955716392949863,
      "learning_rate": 3.1740206169246223e-07,
      "loss": 0.6128,
      "step": 10568
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.40897745948312525,
      "learning_rate": 3.1637509536085464e-07,
      "loss": 0.6464,
      "step": 10569
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3940966733650188,
      "learning_rate": 3.1534977987109474e-07,
      "loss": 0.6494,
      "step": 10570
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3919964708528753,
      "learning_rate": 3.1432611530917367e-07,
      "loss": 0.6411,
      "step": 10571
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3643504331988885,
      "learning_rate": 3.133041017609495e-07,
      "loss": 0.6233,
      "step": 10572
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3761960549924599,
      "learning_rate": 3.1228373931213365e-07,
      "loss": 0.6366,
      "step": 10573
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4108180058941569,
      "learning_rate": 3.112650280483021e-07,
      "loss": 0.6364,
      "step": 10574
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39406361155202524,
      "learning_rate": 3.1024796805489753e-07,
      "loss": 0.6453,
      "step": 10575
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34900770853198265,
      "learning_rate": 3.092325594172163e-07,
      "loss": 0.5993,
      "step": 10576
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4145646641687293,
      "learning_rate": 3.0821880222041913e-07,
      "loss": 0.6128,
      "step": 10577
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3665629061897476,
      "learning_rate": 3.072066965495313e-07,
      "loss": 0.6279,
      "step": 10578
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.36146733245767637,
      "learning_rate": 3.061962424894316e-07,
      "loss": 0.6336,
      "step": 10579
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41450634731836605,
      "learning_rate": 3.0518744012487e-07,
      "loss": 0.6075,
      "step": 10580
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42175080098168805,
      "learning_rate": 3.041802895404544e-07,
      "loss": 0.6491,
      "step": 10581
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4482915202557734,
      "learning_rate": 3.031747908206484e-07,
      "loss": 0.6788,
      "step": 10582
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3925664270895786,
      "learning_rate": 3.0217094404978886e-07,
      "loss": 0.661,
      "step": 10583
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4071803167417945,
      "learning_rate": 3.0116874931205965e-07,
      "loss": 0.6572,
      "step": 10584
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3447908771447225,
      "learning_rate": 3.0016820669152016e-07,
      "loss": 0.624,
      "step": 10585
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39901950338005926,
      "learning_rate": 2.991693162720788e-07,
      "loss": 0.63,
      "step": 10586
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42496514594088586,
      "learning_rate": 2.9817207813751527e-07,
      "loss": 0.6518,
      "step": 10587
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4331611462097737,
      "learning_rate": 2.9717649237146704e-07,
      "loss": 0.6707,
      "step": 10588
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5180516531452282,
      "learning_rate": 2.9618255905743187e-07,
      "loss": 0.3291,
      "step": 10589
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.36698398014378736,
      "learning_rate": 2.951902782787674e-07,
      "loss": 0.6181,
      "step": 10590
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.36688912357839365,
      "learning_rate": 2.9419965011869613e-07,
      "loss": 0.642,
      "step": 10591
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42967254247903847,
      "learning_rate": 2.932106746603047e-07,
      "loss": 0.6454,
      "step": 10592
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3783305383192957,
      "learning_rate": 2.922233519865314e-07,
      "loss": 0.5979,
      "step": 10593
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4027333983578536,
      "learning_rate": 2.9123768218018764e-07,
      "loss": 0.6245,
      "step": 10594
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38711460341462006,
      "learning_rate": 2.9025366532393404e-07,
      "loss": 0.6455,
      "step": 10595
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5370647512146648,
      "learning_rate": 2.8927130150030457e-07,
      "loss": 0.2874,
      "step": 10596
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3880873233829323,
      "learning_rate": 2.8829059079168445e-07,
      "loss": 0.6433,
      "step": 10597
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4298124604082479,
      "learning_rate": 2.8731153328033e-07,
      "loss": 0.612,
      "step": 10598
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42549361842544503,
      "learning_rate": 2.863341290483468e-07,
      "loss": 0.6582,
      "step": 10599
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.416495169037481,
      "learning_rate": 2.853583781777136e-07,
      "loss": 0.6443,
      "step": 10600
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37697997105053704,
      "learning_rate": 2.843842807502628e-07,
      "loss": 0.633,
      "step": 10601
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3930533689539878,
      "learning_rate": 2.834118368476935e-07,
      "loss": 0.6579,
      "step": 10602
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44228685346869623,
      "learning_rate": 2.8244104655156257e-07,
      "loss": 0.6487,
      "step": 10603
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42610830655780096,
      "learning_rate": 2.814719099432872e-07,
      "loss": 0.6424,
      "step": 10604
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3842575467024574,
      "learning_rate": 2.8050442710414685e-07,
      "loss": 0.6065,
      "step": 10605
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3763341219635447,
      "learning_rate": 2.7953859811528984e-07,
      "loss": 0.6156,
      "step": 10606
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3196470751721111,
      "learning_rate": 2.785744230577092e-07,
      "loss": 0.6233,
      "step": 10607
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3913383201933172,
      "learning_rate": 2.776119020122758e-07,
      "loss": 0.6166,
      "step": 10608
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43339620073968876,
      "learning_rate": 2.766510350597162e-07,
      "loss": 0.6452,
      "step": 10609
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3411907137144919,
      "learning_rate": 2.756918222806104e-07,
      "loss": 0.6162,
      "step": 10610
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41533255018824194,
      "learning_rate": 2.7473426375541403e-07,
      "loss": 0.6625,
      "step": 10611
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34093025589976395,
      "learning_rate": 2.7377835956443166e-07,
      "loss": 0.6251,
      "step": 10612
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3843365617595448,
      "learning_rate": 2.7282410978783483e-07,
      "loss": 0.633,
      "step": 10613
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3545495336598006,
      "learning_rate": 2.7187151450565495e-07,
      "loss": 0.6182,
      "step": 10614
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37534387259250424,
      "learning_rate": 2.7092057379778604e-07,
      "loss": 0.6294,
      "step": 10615
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39016012273816747,
      "learning_rate": 2.6997128774398195e-07,
      "loss": 0.6112,
      "step": 10616
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4075477886290518,
      "learning_rate": 2.690236564238591e-07,
      "loss": 0.6291,
      "step": 10617
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4292264972087194,
      "learning_rate": 2.680776799168916e-07,
      "loss": 0.6082,
      "step": 10618
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3703146836839447,
      "learning_rate": 2.671333583024205e-07,
      "loss": 0.6076,
      "step": 10619
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38226098937294933,
      "learning_rate": 2.661906916596424e-07,
      "loss": 0.6188,
      "step": 10620
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4207450683149772,
      "learning_rate": 2.652496800676185e-07,
      "loss": 0.6226,
      "step": 10621
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37389855632473507,
      "learning_rate": 2.643103236052702e-07,
      "loss": 0.6514,
      "step": 10622
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41864673554319437,
      "learning_rate": 2.633726223513811e-07,
      "loss": 0.6353,
      "step": 10623
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3999657796875753,
      "learning_rate": 2.6243657638459487e-07,
      "loss": 0.6455,
      "step": 10624
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.40751508819512783,
      "learning_rate": 2.6150218578341547e-07,
      "loss": 0.6263,
      "step": 10625
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3949188730859744,
      "learning_rate": 2.605694506262091e-07,
      "loss": 0.625,
      "step": 10626
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4149444097498621,
      "learning_rate": 2.5963837099120423e-07,
      "loss": 0.6078,
      "step": 10627
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.40174438419193487,
      "learning_rate": 2.587089469564874e-07,
      "loss": 0.6626,
      "step": 10628
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39568507525259006,
      "learning_rate": 2.5778117860001173e-07,
      "loss": 0.6209,
      "step": 10629
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37472075035940944,
      "learning_rate": 2.568550659995861e-07,
      "loss": 0.6503,
      "step": 10630
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6588769336945041,
      "learning_rate": 2.5593060923288173e-07,
      "loss": 0.3618,
      "step": 10631
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3477061078673319,
      "learning_rate": 2.550078083774321e-07,
      "loss": 0.6136,
      "step": 10632
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4085518166788789,
      "learning_rate": 2.5408666351063094e-07,
      "loss": 0.6055,
      "step": 10633
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4196345564645069,
      "learning_rate": 2.5316717470973637e-07,
      "loss": 0.6565,
      "step": 10634
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3723632396411527,
      "learning_rate": 2.522493420518601e-07,
      "loss": 0.6583,
      "step": 10635
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4346052802464658,
      "learning_rate": 2.51333165613985e-07,
      "loss": 0.6396,
      "step": 10636
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3720166788494559,
      "learning_rate": 2.5041864547294513e-07,
      "loss": 0.6241,
      "step": 10637
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3795601901243021,
      "learning_rate": 2.495057817054436e-07,
      "loss": 0.6246,
      "step": 10638
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34957835830119577,
      "learning_rate": 2.485945743880369e-07,
      "loss": 0.6092,
      "step": 10639
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3404287568147881,
      "learning_rate": 2.4768502359715284e-07,
      "loss": 0.6569,
      "step": 10640
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3946368122199539,
      "learning_rate": 2.4677712940906816e-07,
      "loss": 0.6288,
      "step": 10641
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3851771366296556,
      "learning_rate": 2.4587089189993086e-07,
      "loss": 0.6266,
      "step": 10642
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3880420814693246,
      "learning_rate": 2.4496631114574675e-07,
      "loss": 0.6204,
      "step": 10643
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3742203736252213,
      "learning_rate": 2.440633872223774e-07,
      "loss": 0.6444,
      "step": 10644
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3840152865529274,
      "learning_rate": 2.4316212020555343e-07,
      "loss": 0.6341,
      "step": 10645
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37216214417903226,
      "learning_rate": 2.4226251017086534e-07,
      "loss": 0.6293,
      "step": 10646
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39346685970224987,
      "learning_rate": 2.413645571937551e-07,
      "loss": 0.6456,
      "step": 10647
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.35487048794138015,
      "learning_rate": 2.404682613495379e-07,
      "loss": 0.6269,
      "step": 10648
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39534341974112724,
      "learning_rate": 2.395736227133849e-07,
      "loss": 0.6484,
      "step": 10649
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37511163262837416,
      "learning_rate": 2.386806413603271e-07,
      "loss": 0.628,
      "step": 10650
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3919452433344514,
      "learning_rate": 2.377893173652601e-07,
      "loss": 0.6432,
      "step": 10651
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3863104130651691,
      "learning_rate": 2.368996508029331e-07,
      "loss": 0.6326,
      "step": 10652
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3989776298578768,
      "learning_rate": 2.360116417479663e-07,
      "loss": 0.6242,
      "step": 10653
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37391107616801317,
      "learning_rate": 2.3512529027483355e-07,
      "loss": 0.6288,
      "step": 10654
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3570594531202396,
      "learning_rate": 2.342405964578731e-07,
      "loss": 0.605,
      "step": 10655
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4023236958039928,
      "learning_rate": 2.3335756037128343e-07,
      "loss": 0.6141,
      "step": 10656
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.36025381359301195,
      "learning_rate": 2.3247618208912525e-07,
      "loss": 0.6108,
      "step": 10657
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4221176801818702,
      "learning_rate": 2.315964616853128e-07,
      "loss": 0.6641,
      "step": 10658
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4039796046868611,
      "learning_rate": 2.307183992336315e-07,
      "loss": 0.612,
      "step": 10659
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3542953689052541,
      "learning_rate": 2.2984199480772462e-07,
      "loss": 0.6418,
      "step": 10660
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38688411967590225,
      "learning_rate": 2.2896724848109342e-07,
      "loss": 0.646,
      "step": 10661
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4013898364718689,
      "learning_rate": 2.280941603270992e-07,
      "loss": 0.6467,
      "step": 10662
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43306919930506915,
      "learning_rate": 2.2722273041897002e-07,
      "loss": 0.6398,
      "step": 10663
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4098445489271313,
      "learning_rate": 2.263529588297919e-07,
      "loss": 0.6699,
      "step": 10664
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4146678104913882,
      "learning_rate": 2.2548484563251095e-07,
      "loss": 0.6499,
      "step": 10665
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42502530480420997,
      "learning_rate": 2.2461839089993555e-07,
      "loss": 0.6462,
      "step": 10666
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38276512502950555,
      "learning_rate": 2.237535947047298e-07,
      "loss": 0.6439,
      "step": 10667
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3985944497720354,
      "learning_rate": 2.2289045711942902e-07,
      "loss": 0.6294,
      "step": 10668
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.35412759028140184,
      "learning_rate": 2.2202897821641979e-07,
      "loss": 0.6151,
      "step": 10669
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5744190698675482,
      "learning_rate": 2.2116915806795314e-07,
      "loss": 0.3799,
      "step": 10670
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.40780555800955254,
      "learning_rate": 2.2031099674614476e-07,
      "loss": 0.6286,
      "step": 10671
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39122029033871164,
      "learning_rate": 2.1945449432296373e-07,
      "loss": 0.6534,
      "step": 10672
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38151946830235617,
      "learning_rate": 2.1859965087024592e-07,
      "loss": 0.6501,
      "step": 10673
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42053966618056965,
      "learning_rate": 2.1774646645968734e-07,
      "loss": 0.6703,
      "step": 10674
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4010871943732927,
      "learning_rate": 2.168949411628396e-07,
      "loss": 0.6212,
      "step": 10675
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41915592480735386,
      "learning_rate": 2.1604507505112115e-07,
      "loss": 0.6417,
      "step": 10676
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3967714614713454,
      "learning_rate": 2.1519686819581053e-07,
      "loss": 0.6128,
      "step": 10677
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4217808597543272,
      "learning_rate": 2.1435032066804195e-07,
      "loss": 0.6293,
      "step": 10678
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37433441146864355,
      "learning_rate": 2.135054325388186e-07,
      "loss": 0.6301,
      "step": 10679
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3607326994675934,
      "learning_rate": 2.126622038790016e-07,
      "loss": 0.6072,
      "step": 10680
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.369480107994288,
      "learning_rate": 2.1182063475930548e-07,
      "loss": 0.6302,
      "step": 10681
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38922253210044117,
      "learning_rate": 2.1098072525031376e-07,
      "loss": 0.6451,
      "step": 10682
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3558924832233839,
      "learning_rate": 2.1014247542247013e-07,
      "loss": 0.6276,
      "step": 10683
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4143344221981721,
      "learning_rate": 2.0930588534607833e-07,
      "loss": 0.6603,
      "step": 10684
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3666767359905734,
      "learning_rate": 2.0847095509129777e-07,
      "loss": 0.6321,
      "step": 10685
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41182106935199964,
      "learning_rate": 2.0763768472815914e-07,
      "loss": 0.6385,
      "step": 10686
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3681367798925317,
      "learning_rate": 2.068060743265443e-07,
      "loss": 0.6327,
      "step": 10687
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4137315991428941,
      "learning_rate": 2.059761239561997e-07,
      "loss": 0.6223,
      "step": 10688
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3537897600040781,
      "learning_rate": 2.051478336867341e-07,
      "loss": 0.6023,
      "step": 10689
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41351592501980294,
      "learning_rate": 2.0432120358760964e-07,
      "loss": 0.6596,
      "step": 10690
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3389169722514696,
      "learning_rate": 2.0349623372816207e-07,
      "loss": 0.6394,
      "step": 10691
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4501511521487626,
      "learning_rate": 2.02672924177576e-07,
      "loss": 0.682,
      "step": 10692
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3879733412446707,
      "learning_rate": 2.0185127500490066e-07,
      "loss": 0.6469,
      "step": 10693
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39233686759877534,
      "learning_rate": 2.01031286279052e-07,
      "loss": 0.6685,
      "step": 10694
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4116483353624465,
      "learning_rate": 2.0021295806879726e-07,
      "loss": 0.6683,
      "step": 10695
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3938624202051021,
      "learning_rate": 1.9939629044277042e-07,
      "loss": 0.6211,
      "step": 10696
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4257069618093459,
      "learning_rate": 1.9858128346946115e-07,
      "loss": 0.637,
      "step": 10697
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3971480282810612,
      "learning_rate": 1.977679372172281e-07,
      "loss": 0.6391,
      "step": 10698
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3767025101393302,
      "learning_rate": 1.9695625175428114e-07,
      "loss": 0.5725,
      "step": 10699
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37199095349559835,
      "learning_rate": 1.9614622714869914e-07,
      "loss": 0.6164,
      "step": 10700
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3801437713790676,
      "learning_rate": 1.9533786346841444e-07,
      "loss": 0.6432,
      "step": 10701
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3404306488484956,
      "learning_rate": 1.9453116078122615e-07,
      "loss": 0.6328,
      "step": 10702
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43315880311474003,
      "learning_rate": 1.9372611915479123e-07,
      "loss": 0.6554,
      "step": 10703
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43318720431287044,
      "learning_rate": 1.929227386566246e-07,
      "loss": 0.673,
      "step": 10704
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3723334554015788,
      "learning_rate": 1.9212101935410786e-07,
      "loss": 0.6576,
      "step": 10705
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38274734982843867,
      "learning_rate": 1.9132096131448064e-07,
      "loss": 0.6383,
      "step": 10706
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.398755856491053,
      "learning_rate": 1.905225646048403e-07,
      "loss": 0.632,
      "step": 10707
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34821170867612883,
      "learning_rate": 1.897258292921489e-07,
      "loss": 0.6289,
      "step": 10708
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33685645026607347,
      "learning_rate": 1.8893075544322846e-07,
      "loss": 0.6114,
      "step": 10709
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3806764588597368,
      "learning_rate": 1.8813734312475685e-07,
      "loss": 0.6652,
      "step": 10710
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40913292290537034,
      "learning_rate": 1.87345592403283e-07,
      "loss": 0.6553,
      "step": 10711
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42285754623663663,
      "learning_rate": 1.8655550334520488e-07,
      "loss": 0.6251,
      "step": 10712
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3827782698974999,
      "learning_rate": 1.8576707601678733e-07,
      "loss": 0.6424,
      "step": 10713
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3966390745536905,
      "learning_rate": 1.8498031048415743e-07,
      "loss": 0.6203,
      "step": 10714
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4027736070705185,
      "learning_rate": 1.8419520681329795e-07,
      "loss": 0.6273,
      "step": 10715
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41104131775439656,
      "learning_rate": 1.83411765070054e-07,
      "loss": 0.677,
      "step": 10716
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36928106682654943,
      "learning_rate": 1.8262998532013522e-07,
      "loss": 0.6193,
      "step": 10717
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3628319310593638,
      "learning_rate": 1.8184986762910473e-07,
      "loss": 0.6236,
      "step": 10718
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3657052204504256,
      "learning_rate": 1.810714120623902e-07,
      "loss": 0.6317,
      "step": 10719
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3876700874015148,
      "learning_rate": 1.8029461868528386e-07,
      "loss": 0.6481,
      "step": 10720
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4096530421859981,
      "learning_rate": 1.7951948756292915e-07,
      "loss": 0.662,
      "step": 10721
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4053582628358076,
      "learning_rate": 1.787460187603407e-07,
      "loss": 0.6267,
      "step": 10722
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3689620515269039,
      "learning_rate": 1.7797421234238443e-07,
      "loss": 0.6127,
      "step": 10723
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41524912541736003,
      "learning_rate": 1.7720406837379078e-07,
      "loss": 0.6496,
      "step": 10724
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36735647129367993,
      "learning_rate": 1.7643558691915252e-07,
      "loss": 0.6334,
      "step": 10725
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37675204250852606,
      "learning_rate": 1.7566876804292254e-07,
      "loss": 0.5876,
      "step": 10726
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.400408356629067,
      "learning_rate": 1.749036118094094e-07,
      "loss": 0.6592,
      "step": 10727
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3952885166627704,
      "learning_rate": 1.7414011828278844e-07,
      "loss": 0.6405,
      "step": 10728
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37844963586406477,
      "learning_rate": 1.7337828752709284e-07,
      "loss": 0.6411,
      "step": 10729
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4102364547080063,
      "learning_rate": 1.7261811960621378e-07,
      "loss": 0.6379,
      "step": 10730
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4490247693714986,
      "learning_rate": 1.718596145839091e-07,
      "loss": 0.6899,
      "step": 10731
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3875041949383487,
      "learning_rate": 1.7110277252379238e-07,
      "loss": 0.6404,
      "step": 10732
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36240473562262443,
      "learning_rate": 1.7034759348933728e-07,
      "loss": 0.5945,
      "step": 10733
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39093748257600547,
      "learning_rate": 1.6959407754388424e-07,
      "loss": 0.6186,
      "step": 10734
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35765096729396617,
      "learning_rate": 1.6884222475062494e-07,
      "loss": 0.6333,
      "step": 10735
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4000659385553214,
      "learning_rate": 1.6809203517261784e-07,
      "loss": 0.6481,
      "step": 10736
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40880872267829726,
      "learning_rate": 1.673435088727815e-07,
      "loss": 0.6801,
      "step": 10737
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36798734463846305,
      "learning_rate": 1.6659664591389236e-07,
      "loss": 0.6043,
      "step": 10738
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3973747616018779,
      "learning_rate": 1.658514463585914e-07,
      "loss": 0.6313,
      "step": 10739
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.396737156480543,
      "learning_rate": 1.6510791026937534e-07,
      "loss": 0.6444,
      "step": 10740
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37398969323011616,
      "learning_rate": 1.6436603770860538e-07,
      "loss": 0.6332,
      "step": 10741
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36829361380974945,
      "learning_rate": 1.636258287384984e-07,
      "loss": 0.6248,
      "step": 10742
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37872693019550246,
      "learning_rate": 1.6288728342113813e-07,
      "loss": 0.6269,
      "step": 10743
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43372389642009135,
      "learning_rate": 1.6215040181846385e-07,
      "loss": 0.6492,
      "step": 10744
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3587933906589637,
      "learning_rate": 1.6141518399227728e-07,
      "loss": 0.6279,
      "step": 10745
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40617455070937347,
      "learning_rate": 1.6068163000424018e-07,
      "loss": 0.6164,
      "step": 10746
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5403640032966444,
      "learning_rate": 1.599497399158767e-07,
      "loss": 0.3517,
      "step": 10747
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3838140354670904,
      "learning_rate": 1.592195137885666e-07,
      "loss": 0.6375,
      "step": 10748
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40727101482576455,
      "learning_rate": 1.584909516835542e-07,
      "loss": 0.6415,
      "step": 10749
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3688260774218516,
      "learning_rate": 1.577640536619418e-07,
      "loss": 0.6019,
      "step": 10750
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39868380613814486,
      "learning_rate": 1.5703881978469838e-07,
      "loss": 0.6176,
      "step": 10751
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3951753210598931,
      "learning_rate": 1.5631525011264192e-07,
      "loss": 0.6291,
      "step": 10752
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39921276209175294,
      "learning_rate": 1.5559334470645948e-07,
      "loss": 0.623,
      "step": 10753
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39430322836051956,
      "learning_rate": 1.5487310362670037e-07,
      "loss": 0.6664,
      "step": 10754
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34490810252988174,
      "learning_rate": 1.541545269337652e-07,
      "loss": 0.6167,
      "step": 10755
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39248867076326466,
      "learning_rate": 1.534376146879235e-07,
      "loss": 0.665,
      "step": 10756
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41899318638996413,
      "learning_rate": 1.5272236694930054e-07,
      "loss": 0.6516,
      "step": 10757
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43277496512190733,
      "learning_rate": 1.5200878377788163e-07,
      "loss": 0.6521,
      "step": 10758
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3912013805565523,
      "learning_rate": 1.512968652335167e-07,
      "loss": 0.6336,
      "step": 10759
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43094360564419165,
      "learning_rate": 1.5058661137591136e-07,
      "loss": 0.6687,
      "step": 10760
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39319674267890914,
      "learning_rate": 1.4987802226463566e-07,
      "loss": 0.6818,
      "step": 10761
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3694786609199786,
      "learning_rate": 1.4917109795911765e-07,
      "loss": 0.6062,
      "step": 10762
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38120599330540955,
      "learning_rate": 1.484658385186455e-07,
      "loss": 0.6282,
      "step": 10763
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43033868646282436,
      "learning_rate": 1.477622440023696e-07,
      "loss": 0.6512,
      "step": 10764
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3792217708610424,
      "learning_rate": 1.4706031446929836e-07,
      "loss": 0.634,
      "step": 10765
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3758251049405728,
      "learning_rate": 1.4636004997830245e-07,
      "loss": 0.6086,
      "step": 10766
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3509256572940214,
      "learning_rate": 1.4566145058811042e-07,
      "loss": 0.6312,
      "step": 10767
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41477768922337105,
      "learning_rate": 1.4496451635731767e-07,
      "loss": 0.6217,
      "step": 10768
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.353139774477445,
      "learning_rate": 1.4426924734437075e-07,
      "loss": 0.594,
      "step": 10769
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3500761071071037,
      "learning_rate": 1.435756436075808e-07,
      "loss": 0.621,
      "step": 10770
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4082664423141416,
      "learning_rate": 1.4288370520512352e-07,
      "loss": 0.6208,
      "step": 10771
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4162341027756451,
      "learning_rate": 1.4219343219502803e-07,
      "loss": 0.6385,
      "step": 10772
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39969863720840787,
      "learning_rate": 1.415048246351858e-07,
      "loss": 0.629,
      "step": 10773
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40835597180304334,
      "learning_rate": 1.4081788258335281e-07,
      "loss": 0.6482,
      "step": 10774
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3250180110313795,
      "learning_rate": 1.4013260609713864e-07,
      "loss": 0.632,
      "step": 10775
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39846363862298,
      "learning_rate": 1.3944899523401723e-07,
      "loss": 0.565,
      "step": 10776
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42407391204581124,
      "learning_rate": 1.387670500513272e-07,
      "loss": 0.6751,
      "step": 10777
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3815164571833953,
      "learning_rate": 1.3808677060625385e-07,
      "loss": 0.6678,
      "step": 10778
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37724332745581873,
      "learning_rate": 1.374081569558583e-07,
      "loss": 0.6294,
      "step": 10779
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3759562612955956,
      "learning_rate": 1.3673120915705275e-07,
      "loss": 0.5933,
      "step": 10780
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4197549610491877,
      "learning_rate": 1.3605592726661176e-07,
      "loss": 0.6266,
      "step": 10781
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3942323491115711,
      "learning_rate": 1.3538231134117008e-07,
      "loss": 0.6244,
      "step": 10782
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35909086064849827,
      "learning_rate": 1.347103614372247e-07,
      "loss": 0.6009,
      "step": 10783
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34841293731048895,
      "learning_rate": 1.3404007761112836e-07,
      "loss": 0.6231,
      "step": 10784
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36465387643572145,
      "learning_rate": 1.333714599191005e-07,
      "loss": 0.6033,
      "step": 10785
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4022025054140796,
      "learning_rate": 1.3270450841721406e-07,
      "loss": 0.6116,
      "step": 10786
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43237584110612304,
      "learning_rate": 1.320392231614065e-07,
      "loss": 0.6256,
      "step": 10787
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43264461321006387,
      "learning_rate": 1.3137560420747542e-07,
      "loss": 0.6207,
      "step": 10788
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.359991339157085,
      "learning_rate": 1.3071365161107852e-07,
      "loss": 0.6456,
      "step": 10789
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3450255260160794,
      "learning_rate": 1.3005336542772917e-07,
      "loss": 0.6215,
      "step": 10790
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40486948037152964,
      "learning_rate": 1.2939474571280975e-07,
      "loss": 0.6036,
      "step": 10791
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4126028411506529,
      "learning_rate": 1.2873779252155382e-07,
      "loss": 0.687,
      "step": 10792
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4037968550170916,
      "learning_rate": 1.280825059090618e-07,
      "loss": 0.6489,
      "step": 10793
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4492301964626022,
      "learning_rate": 1.2742888593028967e-07,
      "loss": 0.6324,
      "step": 10794
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3676924132343738,
      "learning_rate": 1.2677693264005808e-07,
      "loss": 0.6371,
      "step": 10795
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38373396914266567,
      "learning_rate": 1.2612664609304547e-07,
      "loss": 0.6266,
      "step": 10796
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4119756066405625,
      "learning_rate": 1.2547802634378603e-07,
      "loss": 0.6318,
      "step": 10797
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3587793123787392,
      "learning_rate": 1.2483107344668512e-07,
      "loss": 0.6474,
      "step": 10798
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37723452254148665,
      "learning_rate": 1.2418578745599708e-07,
      "loss": 0.6483,
      "step": 10799
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37568356907416356,
      "learning_rate": 1.2354216842584532e-07,
      "loss": 0.606,
      "step": 10800
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3919746776712307,
      "learning_rate": 1.2290021641020444e-07,
      "loss": 0.6596,
      "step": 10801
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3675918449558352,
      "learning_rate": 1.2225993146291804e-07,
      "loss": 0.6193,
      "step": 10802
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40135046047169337,
      "learning_rate": 1.2162131363768536e-07,
      "loss": 0.6278,
      "step": 10803
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3846263445554743,
      "learning_rate": 1.2098436298806582e-07,
      "loss": 0.6392,
      "step": 10804
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4017326733449647,
      "learning_rate": 1.203490795674811e-07,
      "loss": 0.6584,
      "step": 10805
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.419627146450574,
      "learning_rate": 1.1971546342920858e-07,
      "loss": 0.6406,
      "step": 10806
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3648306700895493,
      "learning_rate": 1.1908351462639245e-07,
      "loss": 0.6012,
      "step": 10807
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44365321059983215,
      "learning_rate": 1.1845323321203029e-07,
      "loss": 0.6584,
      "step": 10808
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43158725782065344,
      "learning_rate": 1.1782461923898647e-07,
      "loss": 0.6511,
      "step": 10809
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3883002816605299,
      "learning_rate": 1.1719767275997885e-07,
      "loss": 0.6478,
      "step": 10810
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3678516107798183,
      "learning_rate": 1.16572393827592e-07,
      "loss": 0.6207,
      "step": 10811
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.33740081100036917,
      "learning_rate": 1.15948782494264e-07,
      "loss": 0.607,
      "step": 10812
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4001941391364907,
      "learning_rate": 1.1532683881229744e-07,
      "loss": 0.6606,
      "step": 10813
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3480491031456663,
      "learning_rate": 1.1470656283385285e-07,
      "loss": 0.6104,
      "step": 10814
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3740105856215511,
      "learning_rate": 1.1408795461095523e-07,
      "loss": 0.634,
      "step": 10815
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38659915539112183,
      "learning_rate": 1.1347101419548534e-07,
      "loss": 0.6326,
      "step": 10816
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4206902165937535,
      "learning_rate": 1.1285574163918401e-07,
      "loss": 0.6446,
      "step": 10817
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4162338925073436,
      "learning_rate": 1.1224213699365439e-07,
      "loss": 0.6517,
      "step": 10818
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3670965070933756,
      "learning_rate": 1.1163020031035977e-07,
      "loss": 0.5811,
      "step": 10819
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3842077742043028,
      "learning_rate": 1.1101993164061908e-07,
      "loss": 0.5913,
      "step": 10820
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40657961722353697,
      "learning_rate": 1.1041133103561585e-07,
      "loss": 0.6273,
      "step": 10821
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3672053912718973,
      "learning_rate": 1.0980439854639591e-07,
      "loss": 0.5951,
      "step": 10822
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38835673987851166,
      "learning_rate": 1.0919913422385853e-07,
      "loss": 0.639,
      "step": 10823
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3516528845840457,
      "learning_rate": 1.0859553811876978e-07,
      "loss": 0.6183,
      "step": 10824
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4020993423887209,
      "learning_rate": 1.0799361028174915e-07,
      "loss": 0.6401,
      "step": 10825
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3847741767001923,
      "learning_rate": 1.0739335076328072e-07,
      "loss": 0.6326,
      "step": 10826
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40592884675810864,
      "learning_rate": 1.0679475961370867e-07,
      "loss": 0.6045,
      "step": 10827
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38466822662407346,
      "learning_rate": 1.0619783688323282e-07,
      "loss": 0.6645,
      "step": 10828
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38778761445716037,
      "learning_rate": 1.0560258262191981e-07,
      "loss": 0.5991,
      "step": 10829
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3773943022225486,
      "learning_rate": 1.0500899687969412e-07,
      "loss": 0.6587,
      "step": 10830
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37570364214889745,
      "learning_rate": 1.0441707970633597e-07,
      "loss": 0.6295,
      "step": 10831
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39761302859849884,
      "learning_rate": 1.0382683115148784e-07,
      "loss": 0.6278,
      "step": 10832
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42056086380908964,
      "learning_rate": 1.0323825126465458e-07,
      "loss": 0.6209,
      "step": 10833
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36757726284588776,
      "learning_rate": 1.0265134009520339e-07,
      "loss": 0.6457,
      "step": 10834
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40441357853867516,
      "learning_rate": 1.0206609769235043e-07,
      "loss": 0.6235,
      "step": 10835
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38045923296201145,
      "learning_rate": 1.0148252410518534e-07,
      "loss": 0.6298,
      "step": 10836
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3948703441474455,
      "learning_rate": 1.0090061938264894e-07,
      "loss": 0.594,
      "step": 10837
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36030232358521813,
      "learning_rate": 1.0032038357354668e-07,
      "loss": 0.5528,
      "step": 10838
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37588166944949886,
      "learning_rate": 9.974181672654182e-08,
      "loss": 0.6216,
      "step": 10839
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3701457058319863,
      "learning_rate": 9.916491889015556e-08,
      "loss": 0.6242,
      "step": 10840
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3760533412644414,
      "learning_rate": 9.858969011277363e-08,
      "loss": 0.6153,
      "step": 10841
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4333712103820674,
      "learning_rate": 9.801613044263969e-08,
      "loss": 0.6833,
      "step": 10842
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4317329542654611,
      "learning_rate": 9.744423992785746e-08,
      "loss": 0.6525,
      "step": 10843
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40956532533940904,
      "learning_rate": 9.687401861639078e-08,
      "loss": 0.63,
      "step": 10844
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4314666369929353,
      "learning_rate": 9.630546655606365e-08,
      "loss": 0.6331,
      "step": 10845
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43612742374789404,
      "learning_rate": 9.573858379455791e-08,
      "loss": 0.6009,
      "step": 10846
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3946127283073377,
      "learning_rate": 9.517337037941998e-08,
      "loss": 0.6572,
      "step": 10847
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40968932492188775,
      "learning_rate": 9.460982635805193e-08,
      "loss": 0.6087,
      "step": 10848
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3783030705653531,
      "learning_rate": 9.40479517777182e-08,
      "loss": 0.6025,
      "step": 10849
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4112312671089974,
      "learning_rate": 9.348774668554329e-08,
      "loss": 0.6421,
      "step": 10850
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4148964602031256,
      "learning_rate": 9.292921112850961e-08,
      "loss": 0.659,
      "step": 10851
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3807985483869993,
      "learning_rate": 9.237234515346194e-08,
      "loss": 0.6454,
      "step": 10852
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3860106371051406,
      "learning_rate": 9.18171488071029e-08,
      "loss": 0.6609,
      "step": 10853
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3926484990804133,
      "learning_rate": 9.126362213599749e-08,
      "loss": 0.6418,
      "step": 10854
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38109800285246803,
      "learning_rate": 9.071176518656632e-08,
      "loss": 0.621,
      "step": 10855
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39414896771632374,
      "learning_rate": 9.016157800509684e-08,
      "loss": 0.6356,
      "step": 10856
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3932122278493995,
      "learning_rate": 8.96130606377299e-08,
      "loss": 0.6419,
      "step": 10857
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4240731946323455,
      "learning_rate": 8.90662131304687e-08,
      "loss": 0.5939,
      "step": 10858
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3956842414888851,
      "learning_rate": 8.852103552918101e-08,
      "loss": 0.6509,
      "step": 10859
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36625683845856205,
      "learning_rate": 8.79775278795858e-08,
      "loss": 0.6251,
      "step": 10860
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48346043242910614,
      "learning_rate": 8.74356902272666e-08,
      "loss": 0.6353,
      "step": 10861
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4009983902610041,
      "learning_rate": 8.689552261766932e-08,
      "loss": 0.6562,
      "step": 10862
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41086734564333655,
      "learning_rate": 8.635702509609322e-08,
      "loss": 0.6111,
      "step": 10863
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3890253759564216,
      "learning_rate": 8.582019770770667e-08,
      "loss": 0.6854,
      "step": 10864
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4350270169653059,
      "learning_rate": 8.528504049752917e-08,
      "loss": 0.6337,
      "step": 10865
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39359133989377104,
      "learning_rate": 8.475155351044262e-08,
      "loss": 0.6363,
      "step": 10866
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3754487619465424,
      "learning_rate": 8.421973679119344e-08,
      "loss": 0.6174,
      "step": 10867
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40565590277473723,
      "learning_rate": 8.368959038438374e-08,
      "loss": 0.6497,
      "step": 10868
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3940501830393511,
      "learning_rate": 8.316111433447349e-08,
      "loss": 0.6235,
      "step": 10869
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4386889673568042,
      "learning_rate": 8.263430868578726e-08,
      "loss": 0.6534,
      "step": 10870
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3615140350429996,
      "learning_rate": 8.210917348250746e-08,
      "loss": 0.6287,
      "step": 10871
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5719537022602796,
      "learning_rate": 8.158570876867667e-08,
      "loss": 0.3625,
      "step": 10872
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3485363421815663,
      "learning_rate": 8.106391458819752e-08,
      "loss": 0.6288,
      "step": 10873
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3707772709824906,
      "learning_rate": 8.054379098483278e-08,
      "loss": 0.6411,
      "step": 10874
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3953891185167603,
      "learning_rate": 8.002533800220091e-08,
      "loss": 0.6607,
      "step": 10875
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4152636414666574,
      "learning_rate": 7.950855568378934e-08,
      "loss": 0.6552,
      "step": 10876
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3114966213473059,
      "learning_rate": 7.899344407293674e-08,
      "loss": 0.62,
      "step": 10877
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4079160633585654,
      "learning_rate": 7.84800032128441e-08,
      "loss": 0.6295,
      "step": 10878
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34042011847052794,
      "learning_rate": 7.796823314657697e-08,
      "loss": 0.6117,
      "step": 10879
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4233689201928021,
      "learning_rate": 7.745813391705214e-08,
      "loss": 0.6285,
      "step": 10880
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3555066234688289,
      "learning_rate": 7.694970556705317e-08,
      "loss": 0.6329,
      "step": 10881
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37682205911060584,
      "learning_rate": 7.644294813922371e-08,
      "loss": 0.6576,
      "step": 10882
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34669589861432953,
      "learning_rate": 7.593786167606087e-08,
      "loss": 0.6094,
      "step": 10883
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4289096225765925,
      "learning_rate": 7.543444621992857e-08,
      "loss": 0.655,
      "step": 10884
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40341116082844863,
      "learning_rate": 7.493270181304412e-08,
      "loss": 0.6502,
      "step": 10885
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40647678169669077,
      "learning_rate": 7.443262849749389e-08,
      "loss": 0.6177,
      "step": 10886
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5542011337841144,
      "learning_rate": 7.393422631521319e-08,
      "loss": 0.3522,
      "step": 10887
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4079337686004138,
      "learning_rate": 7.343749530800415e-08,
      "loss": 0.643,
      "step": 10888
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3474638189056729,
      "learning_rate": 7.294243551752678e-08,
      "loss": 0.5917,
      "step": 10889
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35913954410090554,
      "learning_rate": 7.244904698530342e-08,
      "loss": 0.5921,
      "step": 10890
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3648442024279044,
      "learning_rate": 7.195732975270986e-08,
      "loss": 0.5688,
      "step": 10891
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3954127455212761,
      "learning_rate": 7.146728386098867e-08,
      "loss": 0.6689,
      "step": 10892
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36991896466013807,
      "learning_rate": 7.09789093512403e-08,
      "loss": 0.6434,
      "step": 10893
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39863483329266863,
      "learning_rate": 7.04922062644231e-08,
      "loss": 0.666,
      "step": 10894
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35349389308651213,
      "learning_rate": 7.000717464135332e-08,
      "loss": 0.6367,
      "step": 10895
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34189638843604847,
      "learning_rate": 6.952381452271395e-08,
      "loss": 0.6262,
      "step": 10896
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3955091902797246,
      "learning_rate": 6.904212594904148e-08,
      "loss": 0.6215,
      "step": 10897
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.377091562529272,
      "learning_rate": 6.85621089607369e-08,
      "loss": 0.59,
      "step": 10898
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3287942372239691,
      "learning_rate": 6.80837635980569e-08,
      "loss": 0.5784,
      "step": 10899
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44227024518697305,
      "learning_rate": 6.76070899011183e-08,
      "loss": 0.6035,
      "step": 10900
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42076719838936133,
      "learning_rate": 6.713208790990244e-08,
      "loss": 0.6688,
      "step": 10901
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35927014857036005,
      "learning_rate": 6.665875766424412e-08,
      "loss": 0.6154,
      "step": 10902
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37209183561517595,
      "learning_rate": 6.61870992038427e-08,
      "loss": 0.6116,
      "step": 10903
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37217331784641267,
      "learning_rate": 6.571711256825541e-08,
      "loss": 0.6101,
      "step": 10904
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39931045868418463,
      "learning_rate": 6.524879779689963e-08,
      "loss": 0.5859,
      "step": 10905
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44273467798319777,
      "learning_rate": 6.478215492905282e-08,
      "loss": 0.6542,
      "step": 10906
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40534709136604147,
      "learning_rate": 6.431718400385035e-08,
      "loss": 0.6585,
      "step": 10907
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36097405343859684,
      "learning_rate": 6.385388506028989e-08,
      "loss": 0.6101,
      "step": 10908
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3502504117694039,
      "learning_rate": 6.339225813722705e-08,
      "loss": 0.6173,
      "step": 10909
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37601702997483744,
      "learning_rate": 6.293230327337974e-08,
      "loss": 0.6447,
      "step": 10910
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40293529992344945,
      "learning_rate": 6.247402050731932e-08,
      "loss": 0.6523,
      "step": 10911
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3697884609469546,
      "learning_rate": 6.201740987748617e-08,
      "loss": 0.6269,
      "step": 10912
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4063952632639306,
      "learning_rate": 6.156247142217409e-08,
      "loss": 0.5931,
      "step": 10913
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40356193596214185,
      "learning_rate": 6.1109205179537e-08,
      "loss": 0.6272,
      "step": 10914
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3864172378171941,
      "learning_rate": 6.065761118759117e-08,
      "loss": 0.6243,
      "step": 10915
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.411464502233849,
      "learning_rate": 6.020768948421074e-08,
      "loss": 0.6365,
      "step": 10916
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4053131275206078,
      "learning_rate": 5.975944010712997e-08,
      "loss": 0.6222,
      "step": 10917
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40384931979031746,
      "learning_rate": 5.931286309394324e-08,
      "loss": 0.6447,
      "step": 10918
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5390936595904283,
      "learning_rate": 5.886795848210502e-08,
      "loss": 0.3268,
      "step": 10919
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38766449331721703,
      "learning_rate": 5.8424726308927707e-08,
      "loss": 0.5947,
      "step": 10920
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3598778271144858,
      "learning_rate": 5.798316661158598e-08,
      "loss": 0.5855,
      "step": 10921
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4021394765833826,
      "learning_rate": 5.754327942711247e-08,
      "loss": 0.639,
      "step": 10922
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39631575896900767,
      "learning_rate": 5.7105064792397636e-08,
      "loss": 0.6492,
      "step": 10923
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40256446244311545,
      "learning_rate": 5.666852274419876e-08,
      "loss": 0.6567,
      "step": 10924
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39846902611597756,
      "learning_rate": 5.623365331912434e-08,
      "loss": 0.6626,
      "step": 10925
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39468550010299774,
      "learning_rate": 5.5800456553647406e-08,
      "loss": 0.6765,
      "step": 10926
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4322737817804487,
      "learning_rate": 5.536893248409891e-08,
      "loss": 0.6165,
      "step": 10927
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44472451154177783,
      "learning_rate": 5.493908114667212e-08,
      "loss": 0.6854,
      "step": 10928
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4029798333398549,
      "learning_rate": 5.451090257741598e-08,
      "loss": 0.6119,
      "step": 10929
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4050130089306766,
      "learning_rate": 5.408439681224398e-08,
      "loss": 0.6389,
      "step": 10930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3896639949680407,
      "learning_rate": 5.365956388692528e-08,
      "loss": 0.6228,
      "step": 10931
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3558741230897277,
      "learning_rate": 5.323640383708695e-08,
      "loss": 0.6478,
      "step": 10932
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.448541263364041,
      "learning_rate": 5.281491669822503e-08,
      "loss": 0.6787,
      "step": 10933
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4462969459217344,
      "learning_rate": 5.239510250568458e-08,
      "loss": 0.6339,
      "step": 10934
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42789567282736635,
      "learning_rate": 5.197696129467522e-08,
      "loss": 0.6566,
      "step": 10935
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4377817449346232,
      "learning_rate": 5.1560493100268896e-08,
      "loss": 0.6193,
      "step": 10936
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37944015193669695,
      "learning_rate": 5.1145697957391e-08,
      "loss": 0.6735,
      "step": 10937
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3528178690734947,
      "learning_rate": 5.073257590083147e-08,
      "loss": 0.6225,
      "step": 10938
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40684241999167825,
      "learning_rate": 5.032112696523817e-08,
      "loss": 0.6632,
      "step": 10939
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38211899584132175,
      "learning_rate": 4.9911351185121246e-08,
      "loss": 0.6435,
      "step": 10940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46942792655746934,
      "learning_rate": 4.9503248594842125e-08,
      "loss": 0.6512,
      "step": 10941
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42249668409414404,
      "learning_rate": 4.909681922863341e-08,
      "loss": 0.6581,
      "step": 10942
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42211935865246303,
      "learning_rate": 4.8692063120578945e-08,
      "loss": 0.6581,
      "step": 10943
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3867868624673164,
      "learning_rate": 4.8288980304627144e-08,
      "loss": 0.6102,
      "step": 10944
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38703852006735934,
      "learning_rate": 4.788757081458428e-08,
      "loss": 0.644,
      "step": 10945
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4164175634691759,
      "learning_rate": 4.7487834684112335e-08,
      "loss": 0.6521,
      "step": 10946
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39082386257795887,
      "learning_rate": 4.7089771946740024e-08,
      "loss": 0.6265,
      "step": 10947
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3697698402696975,
      "learning_rate": 4.669338263584955e-08,
      "loss": 0.6289,
      "step": 10948
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3818831367401993,
      "learning_rate": 4.629866678468986e-08,
      "loss": 0.6399,
      "step": 10949
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3893775719762393,
      "learning_rate": 4.590562442636337e-08,
      "loss": 0.6441,
      "step": 10950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.395588900443373,
      "learning_rate": 4.55142555938326e-08,
      "loss": 0.6283,
      "step": 10951
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3338413075041475,
      "learning_rate": 4.5124560319922406e-08,
      "loss": 0.5548,
      "step": 10952
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5253354603798125,
      "learning_rate": 4.473653863731553e-08,
      "loss": 0.33,
      "step": 10953
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40424332275462427,
      "learning_rate": 4.435019057855705e-08,
      "loss": 0.6534,
      "step": 10954
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5735327111082981,
      "learning_rate": 4.39655161760455e-08,
      "loss": 0.3574,
      "step": 10955
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4067244807644443,
      "learning_rate": 4.3582515462048394e-08,
      "loss": 0.6503,
      "step": 10956
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35418940605986715,
      "learning_rate": 4.320118846868227e-08,
      "loss": 0.6355,
      "step": 10957
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36538713503374876,
      "learning_rate": 4.2821535227930424e-08,
      "loss": 0.6361,
      "step": 10958
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3779319899257022,
      "learning_rate": 4.244355577163406e-08,
      "loss": 0.6179,
      "step": 10959
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.368077424507063,
      "learning_rate": 4.206725013149449e-08,
      "loss": 0.6119,
      "step": 10960
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44802892550347373,
      "learning_rate": 4.1692618339073124e-08,
      "loss": 0.6709,
      "step": 10961
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.416729831688334,
      "learning_rate": 4.131966042578706e-08,
      "loss": 0.6395,
      "step": 10962
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42191130281869993,
      "learning_rate": 4.0948376422915713e-08,
      "loss": 0.6256,
      "step": 10963
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40687815437175645,
      "learning_rate": 4.057876636160307e-08,
      "loss": 0.5877,
      "step": 10964
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4151645026801042,
      "learning_rate": 4.021083027284212e-08,
      "loss": 0.6434,
      "step": 10965
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4088295608182904,
      "learning_rate": 3.984456818749261e-08,
      "loss": 0.5977,
      "step": 10966
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3747465676542944,
      "learning_rate": 3.9479980136274406e-08,
      "loss": 0.6268,
      "step": 10967
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41427735667497156,
      "learning_rate": 3.911706614976529e-08,
      "loss": 0.6019,
      "step": 10968
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4215847961238235,
      "learning_rate": 3.875582625839869e-08,
      "loss": 0.64,
      "step": 10969
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40573246846621347,
      "learning_rate": 3.8396260492477024e-08,
      "loss": 0.6565,
      "step": 10970
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3772159550835728,
      "learning_rate": 3.803836888215173e-08,
      "loss": 0.6102,
      "step": 10971
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3608617874466704,
      "learning_rate": 3.768215145743881e-08,
      "loss": 0.635,
      "step": 10972
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41504340270304807,
      "learning_rate": 3.7327608248216576e-08,
      "loss": 0.6314,
      "step": 10973
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37261373065718484,
      "learning_rate": 3.697473928422124e-08,
      "loss": 0.6323,
      "step": 10974
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33212125867206155,
      "learning_rate": 3.662354459504247e-08,
      "loss": 0.68,
      "step": 10975
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3857638421657313,
      "learning_rate": 3.627402421013893e-08,
      "loss": 0.637,
      "step": 10976
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38169821518667674,
      "learning_rate": 3.592617815882271e-08,
      "loss": 0.6021,
      "step": 10977
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3882770308268865,
      "learning_rate": 3.558000647026827e-08,
      "loss": 0.6171,
      "step": 10978
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43694202108836505,
      "learning_rate": 3.523550917350571e-08,
      "loss": 0.6378,
      "step": 10979
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4011630160460598,
      "learning_rate": 3.48926862974297e-08,
      "loss": 0.6017,
      "step": 10980
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.535812119960977,
      "learning_rate": 3.455153787079501e-08,
      "loss": 0.3376,
      "step": 10981
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36609561458382844,
      "learning_rate": 3.421206392220766e-08,
      "loss": 0.6431,
      "step": 10982
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4048175557232285,
      "learning_rate": 3.3874264480144856e-08,
      "loss": 0.6407,
      "step": 10983
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42898687227309085,
      "learning_rate": 3.3538139572932834e-08,
      "loss": 0.6701,
      "step": 10984
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4149510456702275,
      "learning_rate": 3.32036892287646e-08,
      "loss": 0.691,
      "step": 10985
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44122992404113015,
      "learning_rate": 3.287091347568661e-08,
      "loss": 0.665,
      "step": 10986
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4566408923413349,
      "learning_rate": 3.2539812341614296e-08,
      "loss": 0.6583,
      "step": 10987
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35563991578268894,
      "learning_rate": 3.221038585430991e-08,
      "loss": 0.6195,
      "step": 10988
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3959302494275378,
      "learning_rate": 3.18826340414069e-08,
      "loss": 0.641,
      "step": 10989
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38625393898623095,
      "learning_rate": 3.155655693039439e-08,
      "loss": 0.6333,
      "step": 10990
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34948060084402116,
      "learning_rate": 3.123215454861495e-08,
      "loss": 0.5985,
      "step": 10991
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3968171923514228,
      "learning_rate": 3.090942692328014e-08,
      "loss": 0.6199,
      "step": 10992
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40700567052542586,
      "learning_rate": 3.058837408145277e-08,
      "loss": 0.6791,
      "step": 10993
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3887853854785083,
      "learning_rate": 3.02689960500624e-08,
      "loss": 0.663,
      "step": 10994
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3399545592401217,
      "learning_rate": 2.995129285589649e-08,
      "loss": 0.6314,
      "step": 10995
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40240420347603467,
      "learning_rate": 2.9635264525595953e-08,
      "loss": 0.6491,
      "step": 10996
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3455829329293731,
      "learning_rate": 2.932091108566626e-08,
      "loss": 0.6033,
      "step": 10997
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4092764065341778,
      "learning_rate": 2.9008232562475202e-08,
      "loss": 0.5982,
      "step": 10998
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42167889409405407,
      "learning_rate": 2.869722898224625e-08,
      "loss": 0.6468,
      "step": 10999
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3950864227332274,
      "learning_rate": 2.8387900371058542e-08,
      "loss": 0.6217,
      "step": 11000
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5751841431881245,
      "learning_rate": 2.8080246754857988e-08,
      "loss": 0.3805,
      "step": 11001
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3733376292221912,
      "learning_rate": 2.777426815944839e-08,
      "loss": 0.6336,
      "step": 11002
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3990125670731284,
      "learning_rate": 2.7469964610491452e-08,
      "loss": 0.6276,
      "step": 11003
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5805434696321929,
      "learning_rate": 2.716733613350675e-08,
      "loss": 0.3668,
      "step": 11004
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3794005102138005,
      "learning_rate": 2.6866382753876207e-08,
      "loss": 0.6566,
      "step": 11005
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3903328108605114,
      "learning_rate": 2.6567104496841855e-08,
      "loss": 0.6252,
      "step": 11006
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3571724997141076,
      "learning_rate": 2.6269501387503617e-08,
      "loss": 0.6492,
      "step": 11007
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40378931636281756,
      "learning_rate": 2.597357345081708e-08,
      "loss": 0.6356,
      "step": 11008
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4143248423681194,
      "learning_rate": 2.5679320711606835e-08,
      "loss": 0.6195,
      "step": 11009
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39358400191296267,
      "learning_rate": 2.5386743194546483e-08,
      "loss": 0.6538,
      "step": 11010
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41327746936303084,
      "learning_rate": 2.5095840924178604e-08,
      "loss": 0.6133,
      "step": 11011
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3954013876545662,
      "learning_rate": 2.4806613924899248e-08,
      "loss": 0.6445,
      "step": 11012
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3660687473999031,
      "learning_rate": 2.451906222096456e-08,
      "loss": 0.6169,
      "step": 11013
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3546269300715711,
      "learning_rate": 2.4233185836493034e-08,
      "loss": 0.6118,
      "step": 11014
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37890437070680155,
      "learning_rate": 2.3948984795458818e-08,
      "loss": 0.6333,
      "step": 11015
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4571746670385362,
      "learning_rate": 2.3666459121696184e-08,
      "loss": 0.6377,
      "step": 11016
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3870720910581661,
      "learning_rate": 2.338560883890395e-08,
      "loss": 0.6542,
      "step": 11017
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38867965939323956,
      "learning_rate": 2.310643397063661e-08,
      "loss": 0.6098,
      "step": 11018
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40945510790062023,
      "learning_rate": 2.2828934540304325e-08,
      "loss": 0.641,
      "step": 11019
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39908086037989043,
      "learning_rate": 2.2553110571184034e-08,
      "loss": 0.6253,
      "step": 11020
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39356816584503956,
      "learning_rate": 2.227896208640834e-08,
      "loss": 0.6178,
      "step": 11021
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3423391784087861,
      "learning_rate": 2.2006489108967744e-08,
      "loss": 0.6247,
      "step": 11022
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4417070534353903,
      "learning_rate": 2.1735691661717297e-08,
      "loss": 0.6756,
      "step": 11023
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35360651805244836,
      "learning_rate": 2.1466569767365496e-08,
      "loss": 0.6078,
      "step": 11024
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38413842290741324,
      "learning_rate": 2.1199123448485403e-08,
      "loss": 0.6631,
      "step": 11025
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3994942174081421,
      "learning_rate": 2.0933352727505742e-08,
      "loss": 0.6411,
      "step": 11026
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5460024222075132,
      "learning_rate": 2.0669257626717564e-08,
      "loss": 0.3741,
      "step": 11027
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39458991510789454,
      "learning_rate": 2.0406838168269828e-08,
      "loss": 0.6139,
      "step": 11028
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38810972948589983,
      "learning_rate": 2.0146094374171587e-08,
      "loss": 0.6301,
      "step": 11029
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3957412429084831,
      "learning_rate": 1.98870262662898e-08,
      "loss": 0.6107,
      "step": 11030
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3338307221717749,
      "learning_rate": 1.962963386635375e-08,
      "loss": 0.5982,
      "step": 11031
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40052483372257697,
      "learning_rate": 1.937391719595061e-08,
      "loss": 0.6535,
      "step": 11032
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40373027800562167,
      "learning_rate": 1.9119876276527672e-08,
      "loss": 0.6202,
      "step": 11033
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4078595065642107,
      "learning_rate": 1.8867511129387893e-08,
      "loss": 0.6299,
      "step": 11034
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42551374847376794,
      "learning_rate": 1.861682177569879e-08,
      "loss": 0.655,
      "step": 11035
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3681321967088795,
      "learning_rate": 1.8367808236485763e-08,
      "loss": 0.5801,
      "step": 11036
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4030693558133395,
      "learning_rate": 1.8120470532634325e-08,
      "loss": 0.6262,
      "step": 11037
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.36033780272635724,
      "learning_rate": 1.787480868488567e-08,
      "loss": 0.6352,
      "step": 11038
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39218841503606083,
      "learning_rate": 1.763082271384553e-08,
      "loss": 0.6368,
      "step": 11039
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38125863979063573,
      "learning_rate": 1.7388512639975317e-08,
      "loss": 0.6145,
      "step": 11040
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.435022892853258,
      "learning_rate": 1.7147878483596557e-08,
      "loss": 0.6347,
      "step": 11041
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3530271584534477,
      "learning_rate": 1.6908920264893104e-08,
      "loss": 0.6396,
      "step": 11042
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3784809877141828,
      "learning_rate": 1.6671638003904477e-08,
      "loss": 0.6673,
      "step": 11043
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3867969893829758,
      "learning_rate": 1.6436031720532543e-08,
      "loss": 0.6059,
      "step": 11044
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4204696892729868,
      "learning_rate": 1.620210143453482e-08,
      "loss": 0.6665,
      "step": 11045
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4057947966636675,
      "learning_rate": 1.5969847165535623e-08,
      "loss": 0.6405,
      "step": 11046
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3491162725249784,
      "learning_rate": 1.5739268933008256e-08,
      "loss": 0.6182,
      "step": 11047
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3819817910739564,
      "learning_rate": 1.5510366756295025e-08,
      "loss": 0.6666,
      "step": 11048
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4007543902246534,
      "learning_rate": 1.5283140654591687e-08,
      "loss": 0.6186,
      "step": 11049
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4068003772431228,
      "learning_rate": 1.5057590646954113e-08,
      "loss": 0.64,
      "step": 11050
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4042690399611848,
      "learning_rate": 1.4833716752302718e-08,
      "loss": 0.6298,
      "step": 11051
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3952075740547098,
      "learning_rate": 1.4611518989411378e-08,
      "loss": 0.6252,
      "step": 11052
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3753156533577882,
      "learning_rate": 1.4390997376916293e-08,
      "loss": 0.6573,
      "step": 11053
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37146086803586104,
      "learning_rate": 1.4172151933309342e-08,
      "loss": 0.5969,
      "step": 11054
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38979379015011184,
      "learning_rate": 1.3954982676949169e-08,
      "loss": 0.6608,
      "step": 11055
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37902654484986237,
      "learning_rate": 1.3739489626047875e-08,
      "loss": 0.5914,
      "step": 11056
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.45540194664278305,
      "learning_rate": 1.3525672798677669e-08,
      "loss": 0.6743,
      "step": 11057
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3653857305615471,
      "learning_rate": 1.3313532212770874e-08,
      "loss": 0.6326,
      "step": 11058
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3761555863157903,
      "learning_rate": 1.3103067886122144e-08,
      "loss": 0.6513,
      "step": 11059
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3743418571548514,
      "learning_rate": 1.2894279836379586e-08,
      "loss": 0.6583,
      "step": 11060
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4015319620661872,
      "learning_rate": 1.2687168081055855e-08,
      "loss": 0.6164,
      "step": 11061
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3988930028955285,
      "learning_rate": 1.2481732637521505e-08,
      "loss": 0.6152,
      "step": 11062
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.36074343916651974,
      "learning_rate": 1.227797352300497e-08,
      "loss": 0.6311,
      "step": 11063
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37980946862788956,
      "learning_rate": 1.207589075459481e-08,
      "loss": 0.6702,
      "step": 11064
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3610244951216,
      "learning_rate": 1.1875484349241906e-08,
      "loss": 0.6128,
      "step": 11065
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37954592609249227,
      "learning_rate": 1.1676754323750594e-08,
      "loss": 0.622,
      "step": 11066
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3964595546173524,
      "learning_rate": 1.1479700694791985e-08,
      "loss": 0.6521,
      "step": 11067
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43585404898796065,
      "learning_rate": 1.1284323478890635e-08,
      "loss": 0.6436,
      "step": 11068
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41662461700121284,
      "learning_rate": 1.1090622692431218e-08,
      "loss": 0.6261,
      "step": 11069
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41533077372352556,
      "learning_rate": 1.089859835166296e-08,
      "loss": 0.6323,
      "step": 11070
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3808329767229404,
      "learning_rate": 1.070825047268631e-08,
      "loss": 0.6241,
      "step": 11071
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41188673115046187,
      "learning_rate": 1.051957907147072e-08,
      "loss": 0.6059,
      "step": 11072
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43082344099320025,
      "learning_rate": 1.0332584163834647e-08,
      "loss": 0.6719,
      "step": 11073
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4423072688602353,
      "learning_rate": 1.014726576546332e-08,
      "loss": 0.6344,
      "step": 11074
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37504621641472946,
      "learning_rate": 9.963623891899865e-09,
      "loss": 0.6206,
      "step": 11075
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37433109000791026,
      "learning_rate": 9.781658558545293e-09,
      "loss": 0.626,
      "step": 11076
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37178956605232677,
      "learning_rate": 9.60136978066073e-09,
      "loss": 0.577,
      "step": 11077
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.427231541950481,
      "learning_rate": 9.422757573367413e-09,
      "loss": 0.655,
      "step": 11078
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4026130632903426,
      "learning_rate": 9.245821951644474e-09,
      "loss": 0.6496,
      "step": 11079
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41333847787949873,
      "learning_rate": 9.07056293033115e-09,
      "loss": 0.6581,
      "step": 11080
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4200642111302362,
      "learning_rate": 8.896980524129018e-09,
      "loss": 0.631,
      "step": 11081
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4072178601543913,
      "learning_rate": 8.725074747590878e-09,
      "loss": 0.6345,
      "step": 11082
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4164618385973962,
      "learning_rate": 8.55484561514075e-09,
      "loss": 0.6637,
      "step": 11083
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40230506177085196,
      "learning_rate": 8.386293141049439e-09,
      "loss": 0.6271,
      "step": 11084
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41781205744970157,
      "learning_rate": 8.219417339456748e-09,
      "loss": 0.6629,
      "step": 11085
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3662363003000296,
      "learning_rate": 8.054218224358145e-09,
      "loss": 0.6242,
      "step": 11086
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43836488961014175,
      "learning_rate": 7.890695809606996e-09,
      "loss": 0.6567,
      "step": 11087
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4082568681840304,
      "learning_rate": 7.728850108918995e-09,
      "loss": 0.6399,
      "step": 11088
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.46501445079753123,
      "learning_rate": 7.568681135869948e-09,
      "loss": 0.6877,
      "step": 11089
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3794333679691941,
      "learning_rate": 7.410188903889115e-09,
      "loss": 0.663,
      "step": 11090
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3810857380960041,
      "learning_rate": 7.2533734262703046e-09,
      "loss": 0.6221,
      "step": 11091
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39579621276970783,
      "learning_rate": 7.09823471616744e-09,
      "loss": 0.6323,
      "step": 11092
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.49985237601032306,
      "learning_rate": 6.944772786587894e-09,
      "loss": 0.6607,
      "step": 11093
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.359222615999686,
      "learning_rate": 6.7929876504080335e-09,
      "loss": 0.6309,
      "step": 11094
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4022953606057045,
      "learning_rate": 6.642879320353235e-09,
      "loss": 0.6371,
      "step": 11095
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39524282247250325,
      "learning_rate": 6.494447809013426e-09,
      "loss": 0.6406,
      "step": 11096
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3815133401804654,
      "learning_rate": 6.34769312883865e-09,
      "loss": 0.6331,
      "step": 11097
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42464305366633653,
      "learning_rate": 6.202615292136838e-09,
      "loss": 0.6342,
      "step": 11098
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43175642973083767,
      "learning_rate": 6.059214311073813e-09,
      "loss": 0.6616,
      "step": 11099
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35937863986143975,
      "learning_rate": 5.917490197679954e-09,
      "loss": 0.5771,
      "step": 11100
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.36525343516518644,
      "learning_rate": 5.7774429638390865e-09,
      "loss": 0.6351,
      "step": 11101
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38545473330451907,
      "learning_rate": 5.6390726212973705e-09,
      "loss": 0.6324,
      "step": 11102
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40568067055300233,
      "learning_rate": 5.502379181658857e-09,
      "loss": 0.6238,
      "step": 11103
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3737429445064496,
      "learning_rate": 5.367362656387709e-09,
      "loss": 0.6337,
      "step": 11104
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35820095157082926,
      "learning_rate": 5.234023056810422e-09,
      "loss": 0.6472,
      "step": 11105
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40859156566730115,
      "learning_rate": 5.102360394106942e-09,
      "loss": 0.5864,
      "step": 11106
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41025430604434454,
      "learning_rate": 4.972374679321768e-09,
      "loss": 0.6673,
      "step": 11107
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41498904954410804,
      "learning_rate": 4.844065923355068e-09,
      "loss": 0.6511,
      "step": 11108
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41113257925343455,
      "learning_rate": 4.717434136967125e-09,
      "loss": 0.6551,
      "step": 11109
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4199983797221217,
      "learning_rate": 4.592479330782773e-09,
      "loss": 0.6616,
      "step": 11110
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4208059131170257,
      "learning_rate": 4.469201515275856e-09,
      "loss": 0.6213,
      "step": 11111
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4064184465813797,
      "learning_rate": 4.347600700789212e-09,
      "loss": 0.6387,
      "step": 11112
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4279704183051969,
      "learning_rate": 4.227676897521349e-09,
      "loss": 0.6235,
      "step": 11113
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40978997731082395,
      "learning_rate": 4.109430115530888e-09,
      "loss": 0.6604,
      "step": 11114
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4274873073129044,
      "learning_rate": 3.9928603647321205e-09,
      "loss": 0.6437,
      "step": 11115
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4167073762479992,
      "learning_rate": 3.877967654901671e-09,
      "loss": 0.6135,
      "step": 11116
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3828730124812717,
      "learning_rate": 3.764751995678495e-09,
      "loss": 0.6368,
      "step": 11117
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40969271761440523,
      "learning_rate": 3.653213396555e-09,
      "loss": 0.6374,
      "step": 11118
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3407504574841296,
      "learning_rate": 3.543351866888145e-09,
      "loss": 0.5768,
      "step": 11119
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4086444166000833,
      "learning_rate": 3.4351674158905613e-09,
      "loss": 0.6406,
      "step": 11120
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38091578388173847,
      "learning_rate": 3.3286600526372114e-09,
      "loss": 0.6426,
      "step": 11121
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4572050598618931,
      "learning_rate": 3.223829786056509e-09,
      "loss": 0.6209,
      "step": 11122
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4090776758837956,
      "learning_rate": 3.120676624945862e-09,
      "loss": 0.5631,
      "step": 11123
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5715277622005133,
      "learning_rate": 3.0192005779516863e-09,
      "loss": 0.3421,
      "step": 11124
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3823769685727061,
      "learning_rate": 2.9194016535871727e-09,
      "loss": 0.6223,
      "step": 11125
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3765036289778797,
      "learning_rate": 2.8212798602234027e-09,
      "loss": 0.6238,
      "step": 11126
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.372946946523004,
      "learning_rate": 2.7248352060871286e-09,
      "loss": 0.6319,
      "step": 11127
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38023618333798787,
      "learning_rate": 2.630067699269656e-09,
      "loss": 0.6398,
      "step": 11128
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.37452717358339027,
      "learning_rate": 2.5369773477179618e-09,
      "loss": 0.62,
      "step": 11129
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.37369660056344933,
      "learning_rate": 2.445564159236913e-09,
      "loss": 0.6305,
      "step": 11130
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42197622775800475,
      "learning_rate": 2.355828141498151e-09,
      "loss": 0.6136,
      "step": 11131
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42003811684931713,
      "learning_rate": 2.267769302022327e-09,
      "loss": 0.6597,
      "step": 11132
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42766538309281127,
      "learning_rate": 2.181387648199085e-09,
      "loss": 0.672,
      "step": 11133
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4587138350812924,
      "learning_rate": 2.0966831872715197e-09,
      "loss": 0.6906,
      "step": 11134
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3993818500947023,
      "learning_rate": 2.0136559263450596e-09,
      "loss": 0.6368,
      "step": 11135
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3636570199083667,
      "learning_rate": 1.9323058723785813e-09,
      "loss": 0.6256,
      "step": 11136
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4330386690065566,
      "learning_rate": 1.8526330321999575e-09,
      "loss": 0.6826,
      "step": 11137
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3724481426270223,
      "learning_rate": 1.7746374124905097e-09,
      "loss": 0.6306,
      "step": 11138
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4016068429842866,
      "learning_rate": 1.6983190197894517e-09,
      "loss": 0.6488,
      "step": 11139
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4034354575226016,
      "learning_rate": 1.6236778604983295e-09,
      "loss": 0.6718,
      "step": 11140
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40929500086844495,
      "learning_rate": 1.5507139408765803e-09,
      "loss": 0.5918,
      "step": 11141
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3228844720983819,
      "learning_rate": 1.4794272670459742e-09,
      "loss": 0.623,
      "step": 11142
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4015156474664243,
      "learning_rate": 1.4098178449839517e-09,
      "loss": 0.641,
      "step": 11143
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38084236517326286,
      "learning_rate": 1.3418856805258451e-09,
      "loss": 0.6333,
      "step": 11144
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40512677396971986,
      "learning_rate": 1.27563077937376e-09,
      "loss": 0.6192,
      "step": 11145
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.37249899777337675,
      "learning_rate": 1.2110531470810317e-09,
      "loss": 0.6715,
      "step": 11146
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4136571604829972,
      "learning_rate": 1.1481527890655486e-09,
      "loss": 0.6453,
      "step": 11147
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4201465493397515,
      "learning_rate": 1.0869297106030908e-09,
      "loss": 0.5943,
      "step": 11148
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.420320906753387,
      "learning_rate": 1.0273839168251087e-09,
      "loss": 0.6327,
      "step": 11149
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3312022004681953,
      "learning_rate": 9.695154127298267e-10,
      "loss": 0.6234,
      "step": 11150
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3499693097172766,
      "learning_rate": 9.133242031666988e-10,
      "loss": 0.6308,
      "step": 11151
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4056489336801469,
      "learning_rate": 8.588102928519526e-10,
      "loss": 0.6094,
      "step": 11152
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3851698522814828,
      "learning_rate": 8.059736863530454e-10,
      "loss": 0.6497,
      "step": 11153
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5343320919823563,
      "learning_rate": 7.548143881064285e-10,
      "loss": 0.3496,
      "step": 11154
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.44709137167823626,
      "learning_rate": 7.053324024020036e-10,
      "loss": 0.5934,
      "step": 11155
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3916232135732933,
      "learning_rate": 6.575277333853436e-10,
      "loss": 0.6386,
      "step": 11156
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3891411474762733,
      "learning_rate": 6.114003850710148e-10,
      "loss": 0.6369,
      "step": 11157
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.36842314847468427,
      "learning_rate": 5.669503613225935e-10,
      "loss": 0.6163,
      "step": 11158
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.368801159782034,
      "learning_rate": 5.241776658726494e-10,
      "loss": 0.6255,
      "step": 11159
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42252595891036565,
      "learning_rate": 4.830823023072028e-10,
      "loss": 0.6553,
      "step": 11160
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3873071596754028,
      "learning_rate": 4.4366427407238623e-10,
      "loss": 0.6517,
      "step": 11161
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40149133495636086,
      "learning_rate": 4.059235844744436e-10,
      "loss": 0.6283,
      "step": 11162
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.34760195773761676,
      "learning_rate": 3.6986023667751056e-10,
      "loss": 0.619,
      "step": 11163
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.39461269315846265,
      "learning_rate": 3.3547423370583475e-10,
      "loss": 0.6565,
      "step": 11164
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.39427666968285713,
      "learning_rate": 3.02765578445996e-10,
      "loss": 0.6236,
      "step": 11165
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3814852256804435,
      "learning_rate": 2.717342736380246e-10,
      "loss": 0.6018,
      "step": 11166
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4595202083429211,
      "learning_rate": 2.423803218887244e-10,
      "loss": 0.6718,
      "step": 11167
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3488324724774409,
      "learning_rate": 2.1470372565390863e-10,
      "loss": 0.6216,
      "step": 11168
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3853917312425673,
      "learning_rate": 1.8870448726060474e-10,
      "loss": 0.6335,
      "step": 11169
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.37119342446719866,
      "learning_rate": 1.6438260888707036e-10,
      "loss": 0.6096,
      "step": 11170
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3858140268116964,
      "learning_rate": 1.4173809257389538e-10,
      "loss": 0.6784,
      "step": 11171
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3612671893855952,
      "learning_rate": 1.207709402173407e-10,
      "loss": 0.6255,
      "step": 11172
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3821357541619113,
      "learning_rate": 1.0148115357822008e-10,
      "loss": 0.6679,
      "step": 11173
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.372900818355495,
      "learning_rate": 8.386873427523867e-11,
      "loss": 0.6069,
      "step": 11174
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4006830186354071,
      "learning_rate": 6.79336837849931e-11,
      "loss": 0.6332,
      "step": 11175
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4232531565573043,
      "learning_rate": 5.367600344197144e-11,
      "loss": 0.6538,
      "step": 11176
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3357769144834524,
      "learning_rate": 4.10956944429941e-11,
      "loss": 0.5601,
      "step": 11177
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.41331738608296514,
      "learning_rate": 3.0192757844993427e-11,
      "loss": 0.6399,
      "step": 11178
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.39262224402316004,
      "learning_rate": 2.0967194560572723e-11,
      "loss": 0.6591,
      "step": 11179
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5315850486423755,
      "learning_rate": 1.3419005362447223e-11,
      "loss": 0.3445,
      "step": 11180
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3655677859134506,
      "learning_rate": 7.548190885664496e-12,
      "loss": 0.6383,
      "step": 11181
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.39104169414591144,
      "learning_rate": 3.3547516209431195e-12,
      "loss": 0.6193,
      "step": 11182
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.48953134690504196,
      "learning_rate": 8.386879235544598e-13,
      "loss": 0.572,
      "step": 11183
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5353899287007698,
      "learning_rate": 0.0,
      "loss": 0.4046,
      "step": 11184
    },
    {
      "epoch": 1.0,
      "step": 11184,
      "total_flos": 4556966994247680.0,
      "train_loss": 0.021413467747181954,
      "train_runtime": 5327.3926,
      "train_samples_per_second": 268.709,
      "train_steps_per_second": 2.099
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 11184,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 300,
  "total_flos": 4556966994247680.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}