{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 7812,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 40.219005259896065,
      "learning_rate": 4.2553191489361707e-08,
      "loss": 2.1051,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.042119196771521,
      "learning_rate": 8.510638297872341e-08,
      "loss": 0.9134,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.134603905358697,
      "learning_rate": 1.276595744680851e-07,
      "loss": 0.9045,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.826987457331256,
      "learning_rate": 1.7021276595744683e-07,
      "loss": 2.1364,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.556636918557775,
      "learning_rate": 2.1276595744680852e-07,
      "loss": 0.9214,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 41.87903691395654,
      "learning_rate": 2.553191489361702e-07,
      "loss": 2.0463,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 63.396493094400235,
      "learning_rate": 2.9787234042553196e-07,
      "loss": 2.2249,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 117.04574761448605,
      "learning_rate": 3.4042553191489365e-07,
      "loss": 2.0021,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 37.49740907652403,
      "learning_rate": 3.8297872340425535e-07,
      "loss": 1.9915,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 35.174959012088436,
      "learning_rate": 4.2553191489361704e-07,
      "loss": 1.9467,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 82.43840586499174,
      "learning_rate": 4.6808510638297873e-07,
      "loss": 2.1326,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 102.76353410004569,
      "learning_rate": 5.106382978723404e-07,
      "loss": 1.9238,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 40.01726565259603,
      "learning_rate": 5.531914893617021e-07,
      "loss": 2.1424,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.395014989345192,
      "learning_rate": 5.957446808510639e-07,
      "loss": 0.9007,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.231344560202864,
      "learning_rate": 6.382978723404255e-07,
      "loss": 1.9317,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 176.14386625909324,
      "learning_rate": 6.808510638297873e-07,
      "loss": 1.9534,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 38.121481891794275,
      "learning_rate": 7.234042553191489e-07,
      "loss": 2.0709,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.71215130059222,
      "learning_rate": 7.659574468085107e-07,
      "loss": 1.9181,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.33911160064958,
      "learning_rate": 8.085106382978725e-07,
      "loss": 1.9274,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.589149499006048,
      "learning_rate": 8.510638297872341e-07,
      "loss": 1.9364,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 37.03427650251182,
      "learning_rate": 8.936170212765959e-07,
      "loss": 1.8537,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.84023273248525,
      "learning_rate": 9.361702127659575e-07,
      "loss": 1.8317,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.109703384717417,
      "learning_rate": 9.787234042553193e-07,
      "loss": 0.9411,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 46.30526464190145,
      "learning_rate": 1.0212765957446809e-06,
      "loss": 1.8557,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.200284886463773,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 1.6845,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 41.059511459759534,
      "learning_rate": 1.1063829787234042e-06,
      "loss": 1.8341,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.69890016153909,
      "learning_rate": 1.148936170212766e-06,
      "loss": 1.7871,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.002158712406995,
      "learning_rate": 1.1914893617021278e-06,
      "loss": 1.7116,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.44583379323745,
      "learning_rate": 1.2340425531914894e-06,
      "loss": 1.6493,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 57.19582739220289,
      "learning_rate": 1.276595744680851e-06,
      "loss": 1.4772,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.977198145062886,
      "learning_rate": 1.3191489361702128e-06,
      "loss": 1.4573,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.165408879500408,
      "learning_rate": 1.3617021276595746e-06,
      "loss": 0.7933,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.332163561525032,
      "learning_rate": 1.4042553191489364e-06,
      "loss": 1.4835,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.68967339221501,
      "learning_rate": 1.4468085106382978e-06,
      "loss": 1.4816,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.834571254240316,
      "learning_rate": 1.4893617021276596e-06,
      "loss": 1.3661,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.330872545748381,
      "learning_rate": 1.5319148936170214e-06,
      "loss": 1.4457,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 39.89681076928846,
      "learning_rate": 1.5744680851063832e-06,
      "loss": 1.3509,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.780213947615602,
      "learning_rate": 1.617021276595745e-06,
      "loss": 1.3671,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.312041534299382,
      "learning_rate": 1.6595744680851064e-06,
      "loss": 1.298,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.850446265377265,
      "learning_rate": 1.7021276595744682e-06,
      "loss": 1.3895,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.035541901603018,
      "learning_rate": 1.74468085106383e-06,
      "loss": 1.3402,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.355550159483872,
      "learning_rate": 1.7872340425531918e-06,
      "loss": 1.2768,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.788031608459184,
      "learning_rate": 1.8297872340425531e-06,
      "loss": 1.2394,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.157173610364187,
      "learning_rate": 1.872340425531915e-06,
      "loss": 1.2446,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.634015693468678,
      "learning_rate": 1.9148936170212767e-06,
      "loss": 1.2838,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.554607715509729,
      "learning_rate": 1.9574468085106385e-06,
      "loss": 1.1509,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 25.17705038999618,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.1574,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.102800562044436,
      "learning_rate": 2.0425531914893617e-06,
      "loss": 1.1376,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.477667941524698,
      "learning_rate": 2.0851063829787235e-06,
      "loss": 1.1773,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.505251325108991,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 1.2198,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.627599008009353,
      "learning_rate": 2.170212765957447e-06,
      "loss": 1.181,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.335531267513149,
      "learning_rate": 2.2127659574468085e-06,
      "loss": 1.1153,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.32319240488554,
      "learning_rate": 2.2553191489361703e-06,
      "loss": 1.1455,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.546047819147103,
      "learning_rate": 2.297872340425532e-06,
      "loss": 1.0698,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.652393414227529,
      "learning_rate": 2.340425531914894e-06,
      "loss": 1.0993,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.429523414343901,
      "learning_rate": 2.3829787234042557e-06,
      "loss": 1.1867,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.319199118210774,
      "learning_rate": 2.425531914893617e-06,
      "loss": 1.1354,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.935604538324647,
      "learning_rate": 2.468085106382979e-06,
      "loss": 1.1818,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.788842389723689,
      "learning_rate": 2.5106382978723402e-06,
      "loss": 1.1919,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3501517514061878,
      "learning_rate": 2.553191489361702e-06,
      "loss": 0.7367,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.61491620369947,
      "learning_rate": 2.595744680851064e-06,
      "loss": 1.1574,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.515249664502457,
      "learning_rate": 2.6382978723404256e-06,
      "loss": 1.2034,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 29.20034519313824,
      "learning_rate": 2.6808510638297874e-06,
      "loss": 1.1171,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.584524635424275,
      "learning_rate": 2.7234042553191492e-06,
      "loss": 1.1149,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.179393358315973,
      "learning_rate": 2.765957446808511e-06,
      "loss": 1.1667,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 63.44703097169444,
      "learning_rate": 2.808510638297873e-06,
      "loss": 1.0747,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.99989891649076,
      "learning_rate": 2.8510638297872346e-06,
      "loss": 1.173,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.679620970564953,
      "learning_rate": 2.8936170212765956e-06,
      "loss": 1.0458,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2803078442579547,
      "learning_rate": 2.9361702127659574e-06,
      "loss": 0.666,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.349834559122638,
      "learning_rate": 2.978723404255319e-06,
      "loss": 0.7618,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9222458929565946,
      "learning_rate": 3.021276595744681e-06,
      "loss": 1.1778,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.354515796781913,
      "learning_rate": 3.0638297872340428e-06,
      "loss": 0.7507,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.31208752468611,
      "learning_rate": 3.1063829787234046e-06,
      "loss": 1.1785,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.62203667203515,
      "learning_rate": 3.1489361702127664e-06,
      "loss": 1.0814,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.5350111469672605,
      "learning_rate": 3.191489361702128e-06,
      "loss": 1.1384,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4730596455174765,
      "learning_rate": 3.23404255319149e-06,
      "loss": 1.1249,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.444812511570502,
      "learning_rate": 3.276595744680851e-06,
      "loss": 1.1665,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8877974061095015,
      "learning_rate": 3.3191489361702127e-06,
      "loss": 0.9745,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.123814771637216,
      "learning_rate": 3.3617021276595745e-06,
      "loss": 1.1714,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9963637364976194,
      "learning_rate": 3.4042553191489363e-06,
      "loss": 1.0802,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.894446972431274,
      "learning_rate": 3.446808510638298e-06,
      "loss": 1.1014,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 35.068318897167536,
      "learning_rate": 3.48936170212766e-06,
      "loss": 1.1182,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0003306596035286,
      "learning_rate": 3.5319148936170217e-06,
      "loss": 0.5707,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1442989143708917,
      "learning_rate": 3.5744680851063835e-06,
      "loss": 0.6997,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.159050311031946,
      "learning_rate": 3.6170212765957453e-06,
      "loss": 0.9961,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.090181090254616,
      "learning_rate": 3.6595744680851063e-06,
      "loss": 1.0188,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.585972873867637,
      "learning_rate": 3.702127659574468e-06,
      "loss": 1.1555,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.677277923433559,
      "learning_rate": 3.74468085106383e-06,
      "loss": 1.1058,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.562702212141085,
      "learning_rate": 3.7872340425531917e-06,
      "loss": 1.1027,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.443845104988814,
      "learning_rate": 3.8297872340425535e-06,
      "loss": 1.1617,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0709375456926975,
      "learning_rate": 3.872340425531915e-06,
      "loss": 1.1019,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1642717806840257,
      "learning_rate": 3.914893617021277e-06,
      "loss": 0.9524,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.74993465669042,
      "learning_rate": 3.957446808510639e-06,
      "loss": 1.1271,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.881635420199256,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.0719,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.022866403077271,
      "learning_rate": 4.042553191489362e-06,
      "loss": 0.9695,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.480443527857745,
      "learning_rate": 4.085106382978723e-06,
      "loss": 1.0668,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.42731272646531,
      "learning_rate": 4.127659574468085e-06,
      "loss": 1.1802,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.745984389077082,
      "learning_rate": 4.170212765957447e-06,
      "loss": 1.0091,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0103439846255107,
      "learning_rate": 4.212765957446809e-06,
      "loss": 0.9875,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.544074654750107,
      "learning_rate": 4.255319148936171e-06,
      "loss": 1.1486,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.617425625850544,
      "learning_rate": 4.297872340425532e-06,
      "loss": 1.048,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 69.3519836714979,
      "learning_rate": 4.340425531914894e-06,
      "loss": 1.0459,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.75641826059657,
      "learning_rate": 4.382978723404256e-06,
      "loss": 0.9826,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.852489703982116,
      "learning_rate": 4.425531914893617e-06,
      "loss": 1.0941,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9760323720563455,
      "learning_rate": 4.468085106382979e-06,
      "loss": 0.6422,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7518188376247417,
      "learning_rate": 4.5106382978723406e-06,
      "loss": 1.0405,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.85380841800239,
      "learning_rate": 4.553191489361702e-06,
      "loss": 0.98,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.094933959493089,
      "learning_rate": 4.595744680851064e-06,
      "loss": 1.0384,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.858361658158021,
      "learning_rate": 4.638297872340426e-06,
      "loss": 1.0361,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9365495674384369,
      "learning_rate": 4.680851063829788e-06,
      "loss": 0.6744,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2359411151331647,
      "learning_rate": 4.7234042553191496e-06,
      "loss": 1.0034,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.997846318651466,
      "learning_rate": 4.765957446808511e-06,
      "loss": 1.0351,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.009734974859843,
      "learning_rate": 4.808510638297872e-06,
      "loss": 1.0239,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.596163691654454,
      "learning_rate": 4.851063829787234e-06,
      "loss": 0.9805,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.68379456883951,
      "learning_rate": 4.893617021276596e-06,
      "loss": 1.1193,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0385066458256405,
      "learning_rate": 4.936170212765958e-06,
      "loss": 0.6553,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.521196610538548,
      "learning_rate": 4.9787234042553195e-06,
      "loss": 1.163,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.168524860873551,
      "learning_rate": 5.0212765957446805e-06,
      "loss": 1.0735,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.179945199752059,
      "learning_rate": 5.063829787234042e-06,
      "loss": 1.0312,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.445504406002343,
      "learning_rate": 5.106382978723404e-06,
      "loss": 1.0781,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3578735046353594,
      "learning_rate": 5.148936170212766e-06,
      "loss": 0.9894,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.17451331374913,
      "learning_rate": 5.191489361702128e-06,
      "loss": 0.9459,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.549988614197763,
      "learning_rate": 5.2340425531914895e-06,
      "loss": 1.0535,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.982484454181189,
      "learning_rate": 5.276595744680851e-06,
      "loss": 0.9913,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1101181984468003,
      "learning_rate": 5.319148936170213e-06,
      "loss": 0.9898,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.996495826005205,
      "learning_rate": 5.361702127659575e-06,
      "loss": 0.7128,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.517049744959034,
      "learning_rate": 5.404255319148937e-06,
      "loss": 1.0493,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.949778057028481,
      "learning_rate": 5.4468085106382985e-06,
      "loss": 1.0924,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.980944156750867,
      "learning_rate": 5.48936170212766e-06,
      "loss": 1.0304,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.535566744305586,
      "learning_rate": 5.531914893617022e-06,
      "loss": 1.05,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3741980515652905,
      "learning_rate": 5.574468085106384e-06,
      "loss": 1.0239,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.40455169480816,
      "learning_rate": 5.617021276595746e-06,
      "loss": 1.0019,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.3451816007740485,
      "learning_rate": 5.6595744680851075e-06,
      "loss": 0.9512,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0629519831394676,
      "learning_rate": 5.702127659574469e-06,
      "loss": 0.6178,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0256367546760496,
      "learning_rate": 5.744680851063831e-06,
      "loss": 0.633,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.058023326089286,
      "learning_rate": 5.787234042553191e-06,
      "loss": 0.987,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7433857245254,
      "learning_rate": 5.829787234042553e-06,
      "loss": 1.1229,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5343214857829364,
      "learning_rate": 5.872340425531915e-06,
      "loss": 1.0074,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.055182341680217,
      "learning_rate": 5.9148936170212766e-06,
      "loss": 1.0646,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.044411980465289,
      "learning_rate": 5.957446808510638e-06,
      "loss": 0.9743,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 31.870632197516862,
      "learning_rate": 6e-06,
      "loss": 0.8613,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.100006061360409,
      "learning_rate": 6.042553191489362e-06,
      "loss": 1.0708,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.22034085254299,
      "learning_rate": 6.085106382978724e-06,
      "loss": 1.0168,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9423085316200759,
      "learning_rate": 6.1276595744680855e-06,
      "loss": 0.6463,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.251699870130561,
      "learning_rate": 6.170212765957447e-06,
      "loss": 1.0311,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5794883489948006,
      "learning_rate": 6.212765957446809e-06,
      "loss": 0.9961,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5307615616725108,
      "learning_rate": 6.255319148936171e-06,
      "loss": 0.98,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7944579940980327,
      "learning_rate": 6.297872340425533e-06,
      "loss": 1.036,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9994284162905136,
      "learning_rate": 6.3404255319148945e-06,
      "loss": 0.8887,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.505677898461023,
      "learning_rate": 6.382978723404256e-06,
      "loss": 1.0009,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.374313784643987,
      "learning_rate": 6.425531914893618e-06,
      "loss": 1.0499,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.23436891089678,
      "learning_rate": 6.46808510638298e-06,
      "loss": 1.0107,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.798619648874734,
      "learning_rate": 6.510638297872342e-06,
      "loss": 0.9981,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.24368696389334,
      "learning_rate": 6.553191489361702e-06,
      "loss": 1.025,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1894802309053745,
      "learning_rate": 6.595744680851064e-06,
      "loss": 0.9546,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.004667705384093,
      "learning_rate": 6.6382978723404254e-06,
      "loss": 1.0133,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.009892857626557,
      "learning_rate": 6.680851063829787e-06,
      "loss": 1.0944,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.606957096073383,
      "learning_rate": 6.723404255319149e-06,
      "loss": 1.0118,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0383618708179787,
      "learning_rate": 6.765957446808511e-06,
      "loss": 0.7196,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2768447086058696,
      "learning_rate": 6.808510638297873e-06,
      "loss": 1.0503,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.00019500260969,
      "learning_rate": 6.8510638297872344e-06,
      "loss": 0.9569,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.033945233788027,
      "learning_rate": 6.893617021276596e-06,
      "loss": 0.9037,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.286685226452114,
      "learning_rate": 6.936170212765958e-06,
      "loss": 0.9582,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7502901866936664,
      "learning_rate": 6.97872340425532e-06,
      "loss": 1.0666,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.576602381125577,
      "learning_rate": 7.021276595744682e-06,
      "loss": 1.0132,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.271177828500013,
      "learning_rate": 7.0638297872340434e-06,
      "loss": 1.0086,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.747227284945915,
      "learning_rate": 7.106382978723405e-06,
      "loss": 1.0066,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9744970522866997,
      "learning_rate": 7.148936170212767e-06,
      "loss": 0.6393,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6808627088127532,
      "learning_rate": 7.191489361702129e-06,
      "loss": 0.9211,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.1544469245419355,
      "learning_rate": 7.234042553191491e-06,
      "loss": 0.9306,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.411335952527084,
      "learning_rate": 7.2765957446808524e-06,
      "loss": 0.946,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3945755477915744,
      "learning_rate": 7.3191489361702125e-06,
      "loss": 1.1173,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.17009217245091,
      "learning_rate": 7.361702127659574e-06,
      "loss": 0.9392,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1798082292499035,
      "learning_rate": 7.404255319148936e-06,
      "loss": 1.0745,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.725363726700604,
      "learning_rate": 7.446808510638298e-06,
      "loss": 0.948,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6435083553802845,
      "learning_rate": 7.48936170212766e-06,
      "loss": 1.0598,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0079667006332362,
      "learning_rate": 7.5319148936170215e-06,
      "loss": 0.7003,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.120075556395121,
      "learning_rate": 7.574468085106383e-06,
      "loss": 0.7636,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9780965175610922,
      "learning_rate": 7.617021276595745e-06,
      "loss": 0.5901,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2482178136316855,
      "learning_rate": 7.659574468085107e-06,
      "loss": 0.9528,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9719025978347897,
      "learning_rate": 7.702127659574469e-06,
      "loss": 0.9905,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0289529547408893,
      "learning_rate": 7.74468085106383e-06,
      "loss": 1.0139,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2378593158923867,
      "learning_rate": 7.787234042553192e-06,
      "loss": 0.9946,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.001664995933612,
      "learning_rate": 7.829787234042554e-06,
      "loss": 1.0299,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.270144277018234,
      "learning_rate": 7.872340425531916e-06,
      "loss": 0.9524,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.136091052989623,
      "learning_rate": 7.914893617021278e-06,
      "loss": 1.0377,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3933844917686846,
      "learning_rate": 7.95744680851064e-06,
      "loss": 0.9961,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.672756311310243,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.9524,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6700034854647274,
      "learning_rate": 8.042553191489363e-06,
      "loss": 1.0031,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8756961385741286,
      "learning_rate": 8.085106382978723e-06,
      "loss": 0.9738,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.845851176541437,
      "learning_rate": 8.127659574468085e-06,
      "loss": 0.9585,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.338500367149041,
      "learning_rate": 8.170212765957447e-06,
      "loss": 1.0001,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.04985925925997,
      "learning_rate": 8.212765957446809e-06,
      "loss": 1.0425,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0802144265871707,
      "learning_rate": 8.25531914893617e-06,
      "loss": 1.0105,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.133505713677714,
      "learning_rate": 8.297872340425532e-06,
      "loss": 0.9886,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2156836709950305,
      "learning_rate": 8.340425531914894e-06,
      "loss": 1.0013,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6925630728249523,
      "learning_rate": 8.382978723404256e-06,
      "loss": 1.0396,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.810463608643269,
      "learning_rate": 8.425531914893618e-06,
      "loss": 0.8613,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.591555791205093,
      "learning_rate": 8.46808510638298e-06,
      "loss": 1.0263,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6046278127580005,
      "learning_rate": 8.510638297872341e-06,
      "loss": 0.9784,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.27628949350642,
      "learning_rate": 8.553191489361703e-06,
      "loss": 0.681,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.967020937206017,
      "learning_rate": 8.595744680851065e-06,
      "loss": 0.9522,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.615718026870681,
      "learning_rate": 8.638297872340427e-06,
      "loss": 1.0193,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6433024430031415,
      "learning_rate": 8.680851063829788e-06,
      "loss": 0.9874,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8766476153002125,
      "learning_rate": 8.72340425531915e-06,
      "loss": 1.0381,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7394832421913868,
      "learning_rate": 8.765957446808512e-06,
      "loss": 1.0318,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.118009275444568,
      "learning_rate": 8.808510638297874e-06,
      "loss": 0.8958,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2037314296424824,
      "learning_rate": 8.851063829787234e-06,
      "loss": 0.9717,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.977726656607224,
      "learning_rate": 8.893617021276596e-06,
      "loss": 1.0999,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.537333171065344,
      "learning_rate": 8.936170212765958e-06,
      "loss": 0.8918,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6691731096075826,
      "learning_rate": 8.97872340425532e-06,
      "loss": 0.97,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.4503357511977484,
      "learning_rate": 9.021276595744681e-06,
      "loss": 0.9384,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8056970081527646,
      "learning_rate": 9.063829787234043e-06,
      "loss": 0.911,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8030458146760469,
      "learning_rate": 9.106382978723405e-06,
      "loss": 0.9683,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7444896300755472,
      "learning_rate": 9.148936170212767e-06,
      "loss": 1.0356,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7400649631843366,
      "learning_rate": 9.191489361702128e-06,
      "loss": 0.9219,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.439171040140551,
      "learning_rate": 9.23404255319149e-06,
      "loss": 1.0301,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6864320394045564,
      "learning_rate": 9.276595744680852e-06,
      "loss": 0.9824,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0584781815242632,
      "learning_rate": 9.319148936170214e-06,
      "loss": 0.6708,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4354082924197147,
      "learning_rate": 9.361702127659576e-06,
      "loss": 0.8665,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6438169528508546,
      "learning_rate": 9.404255319148937e-06,
      "loss": 0.9644,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5319951987452514,
      "learning_rate": 9.446808510638299e-06,
      "loss": 0.9236,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.037836437597852,
      "learning_rate": 9.489361702127661e-06,
      "loss": 1.0359,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.492244030594403,
      "learning_rate": 9.531914893617023e-06,
      "loss": 0.9346,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8102312196022399,
      "learning_rate": 9.574468085106385e-06,
      "loss": 1.0984,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5057339251481912,
      "learning_rate": 9.617021276595745e-06,
      "loss": 0.9177,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.423211267074831,
      "learning_rate": 9.659574468085106e-06,
      "loss": 0.9254,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7720403138855463,
      "learning_rate": 9.702127659574468e-06,
      "loss": 0.8473,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6689373957520652,
      "learning_rate": 9.74468085106383e-06,
      "loss": 0.9857,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8882723591159885,
      "learning_rate": 9.787234042553192e-06,
      "loss": 0.9209,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5897708617160935,
      "learning_rate": 9.829787234042554e-06,
      "loss": 0.9891,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6549863667343943,
      "learning_rate": 9.872340425531915e-06,
      "loss": 0.9705,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7068275934006503,
      "learning_rate": 9.914893617021277e-06,
      "loss": 0.9897,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9085002485080302,
      "learning_rate": 9.957446808510639e-06,
      "loss": 0.6823,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4164362536699788,
      "learning_rate": 1e-05,
      "loss": 0.9828,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5541904188666114,
      "learning_rate": 9.999999570221018e-06,
      "loss": 0.9041,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3515613839828873,
      "learning_rate": 9.999998280884144e-06,
      "loss": 0.9163,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.704069961650007,
      "learning_rate": 9.999996131989602e-06,
      "loss": 0.8681,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.132629297025691,
      "learning_rate": 9.99999312353776e-06,
      "loss": 1.0553,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.043521367837014,
      "learning_rate": 9.999989255529133e-06,
      "loss": 0.7416,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.201841279079125,
      "learning_rate": 9.99998452796439e-06,
      "loss": 0.9774,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5316041351819534,
      "learning_rate": 9.99997894084434e-06,
      "loss": 0.9818,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8931744856465698,
      "learning_rate": 9.999972494169947e-06,
      "loss": 0.9942,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3804910442799208,
      "learning_rate": 9.999965187942317e-06,
      "loss": 0.8751,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.313159819863083,
      "learning_rate": 9.999957022162707e-06,
      "loss": 0.949,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0952275325613177,
      "learning_rate": 9.99994799683252e-06,
      "loss": 1.0052,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.958161346723847,
      "learning_rate": 9.999938111953306e-06,
      "loss": 0.947,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.052415349576133,
      "learning_rate": 9.99992736752677e-06,
      "loss": 1.029,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.065056426986825,
      "learning_rate": 9.999915763554754e-06,
      "loss": 0.695,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.87412053147013,
      "learning_rate": 9.999903300039253e-06,
      "loss": 0.8865,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9920206297522022,
      "learning_rate": 9.999889976982413e-06,
      "loss": 0.6505,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.051541343733892,
      "learning_rate": 9.99987579438652e-06,
      "loss": 0.8916,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4929339290311692,
      "learning_rate": 9.999860752254016e-06,
      "loss": 0.9625,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.986823528008541,
      "learning_rate": 9.999844850587486e-06,
      "loss": 0.6717,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1567504145273224,
      "learning_rate": 9.99982808938966e-06,
      "loss": 0.7248,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0328977852296255,
      "learning_rate": 9.999810468663424e-06,
      "loss": 0.8524,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9738241979394312,
      "learning_rate": 9.999791988411807e-06,
      "loss": 0.6498,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2869085250987626,
      "learning_rate": 9.999772648637984e-06,
      "loss": 0.9335,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6303664172778671,
      "learning_rate": 9.999752449345279e-06,
      "loss": 0.8468,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5119251481170377,
      "learning_rate": 9.999731390537168e-06,
      "loss": 0.9787,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.030028491335317,
      "learning_rate": 9.999709472217268e-06,
      "loss": 0.9604,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.7868090662352274,
      "learning_rate": 9.999686694389348e-06,
      "loss": 0.9461,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.63527309517783,
      "learning_rate": 9.999663057057324e-06,
      "loss": 0.8875,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6694816617557489,
      "learning_rate": 9.999638560225259e-06,
      "loss": 0.9943,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6415696254653818,
      "learning_rate": 9.999613203897365e-06,
      "loss": 0.9175,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0964636933057648,
      "learning_rate": 9.999586988078e-06,
      "loss": 0.7382,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.760958775561538,
      "learning_rate": 9.999559912771673e-06,
      "loss": 0.9899,
      "step": 267
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3757975288553363,
      "learning_rate": 9.999531977983038e-06,
      "loss": 0.9021,
      "step": 268
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8662005825160166,
      "learning_rate": 9.999503183716894e-06,
      "loss": 1.0035,
      "step": 269
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4786628279682588,
      "learning_rate": 9.999473529978194e-06,
      "loss": 0.9834,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1243192740743235,
      "learning_rate": 9.999443016772037e-06,
      "loss": 0.9927,
      "step": 271
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5831858211741447,
      "learning_rate": 9.999411644103665e-06,
      "loss": 0.879,
      "step": 272
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5387407968742304,
      "learning_rate": 9.999379411978474e-06,
      "loss": 0.9895,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.224925574510128,
      "learning_rate": 9.999346320402003e-06,
      "loss": 0.9479,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5992000310733485,
      "learning_rate": 9.999312369379944e-06,
      "loss": 1.0159,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2262451413801125,
      "learning_rate": 9.99927755891813e-06,
      "loss": 0.8844,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4463194002369564,
      "learning_rate": 9.99924188902255e-06,
      "loss": 0.9169,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9437786184038349,
      "learning_rate": 9.99920535969933e-06,
      "loss": 0.6731,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9119797774770435,
      "learning_rate": 9.999167970954756e-06,
      "loss": 0.966,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.152043252993726,
      "learning_rate": 9.999129722795248e-06,
      "loss": 0.974,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6216703112547766,
      "learning_rate": 9.999090615227389e-06,
      "loss": 0.9722,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6111937553845246,
      "learning_rate": 9.999050648257898e-06,
      "loss": 0.8569,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.948070756768925,
      "learning_rate": 9.999009821893648e-06,
      "loss": 0.6736,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.004610684430267,
      "learning_rate": 9.998968136141655e-06,
      "loss": 0.9681,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4359165579177584,
      "learning_rate": 9.998925591009086e-06,
      "loss": 0.9067,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6862638012969187,
      "learning_rate": 9.998882186503256e-06,
      "loss": 0.913,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8374431619415217,
      "learning_rate": 9.998837922631625e-06,
      "loss": 0.9132,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1502626582867435,
      "learning_rate": 9.998792799401804e-06,
      "loss": 0.978,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4582875854913802,
      "learning_rate": 9.998746816821551e-06,
      "loss": 0.8685,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.58511278059945,
      "learning_rate": 9.99869997489877e-06,
      "loss": 1.034,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5692175638215613,
      "learning_rate": 9.99865227364151e-06,
      "loss": 0.8561,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8424183130530334,
      "learning_rate": 9.998603713057977e-06,
      "loss": 0.9527,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9914528527408994,
      "learning_rate": 9.998554293156518e-06,
      "loss": 1.0063,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1095625673755842,
      "learning_rate": 9.998504013945627e-06,
      "loss": 0.9772,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8895400233495175,
      "learning_rate": 9.998452875433948e-06,
      "loss": 0.7451,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.951518320205812,
      "learning_rate": 9.998400877630272e-06,
      "loss": 0.9788,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9645676783584697,
      "learning_rate": 9.99834802054354e-06,
      "loss": 0.6965,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9602507422258505,
      "learning_rate": 9.998294304182837e-06,
      "loss": 0.6491,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8496322999619172,
      "learning_rate": 9.998239728557399e-06,
      "loss": 0.985,
      "step": 299
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5547745967230584,
      "learning_rate": 9.998184293676606e-06,
      "loss": 0.8706,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4967031933710768,
      "learning_rate": 9.998127999549988e-06,
      "loss": 0.8072,
      "step": 301
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7996901633011686,
      "learning_rate": 9.998070846187225e-06,
      "loss": 0.9181,
      "step": 302
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3511108268485423,
      "learning_rate": 9.99801283359814e-06,
      "loss": 0.8984,
      "step": 303
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0727665040386425,
      "learning_rate": 9.997953961792708e-06,
      "loss": 0.8772,
      "step": 304
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6629671857503756,
      "learning_rate": 9.997894230781048e-06,
      "loss": 0.8741,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9679466111603052,
      "learning_rate": 9.99783364057343e-06,
      "loss": 1.0255,
      "step": 306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9030964344911276,
      "learning_rate": 9.997772191180269e-06,
      "loss": 0.955,
      "step": 307
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.42128618878279,
      "learning_rate": 9.997709882612128e-06,
      "loss": 0.9767,
      "step": 308
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7461098717406411,
      "learning_rate": 9.99764671487972e-06,
      "loss": 0.8239,
      "step": 309
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8732092661791147,
      "learning_rate": 9.997582687993905e-06,
      "loss": 0.9254,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5065747189857175,
      "learning_rate": 9.99751780196569e-06,
      "loss": 0.8651,
      "step": 311
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4036759506957006,
      "learning_rate": 9.997452056806226e-06,
      "loss": 0.852,
      "step": 312
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.526301645279762,
      "learning_rate": 9.99738545252682e-06,
      "loss": 0.8929,
      "step": 313
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.984762718465852,
      "learning_rate": 9.99731798913892e-06,
      "loss": 1.0139,
      "step": 314
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.511332541732339,
      "learning_rate": 9.997249666654122e-06,
      "loss": 0.861,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.529362678793336,
      "learning_rate": 9.997180485084175e-06,
      "loss": 0.8982,
      "step": 316
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7859977653388943,
      "learning_rate": 9.99711044444097e-06,
      "loss": 0.8576,
      "step": 317
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0822251049708576,
      "learning_rate": 9.997039544736547e-06,
      "loss": 1.0156,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5985966582975448,
      "learning_rate": 9.996967785983097e-06,
      "loss": 0.9511,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8674448336394829,
      "learning_rate": 9.996895168192954e-06,
      "loss": 0.9426,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.932498214634118,
      "learning_rate": 9.996821691378603e-06,
      "loss": 0.6894,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.440688685475695,
      "learning_rate": 9.996747355552675e-06,
      "loss": 0.8597,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8084533927865747,
      "learning_rate": 9.99667216072795e-06,
      "loss": 0.9845,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8909762763079551,
      "learning_rate": 9.996596106917353e-06,
      "loss": 0.675,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5838279171746938,
      "learning_rate": 9.99651919413396e-06,
      "loss": 0.9027,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.882874488582247,
      "learning_rate": 9.996441422390994e-06,
      "loss": 0.8993,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9266090633355012,
      "learning_rate": 9.996362791701822e-06,
      "loss": 0.7192,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.717696100330419,
      "learning_rate": 9.996283302079965e-06,
      "loss": 0.9376,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3814622868584188,
      "learning_rate": 9.996202953539085e-06,
      "loss": 0.9053,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.919095212405375,
      "learning_rate": 9.996121746092996e-06,
      "loss": 0.9407,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4618283066885165,
      "learning_rate": 9.99603967975566e-06,
      "loss": 0.8762,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5222328889401135,
      "learning_rate": 9.995956754541185e-06,
      "loss": 0.9638,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9427435293841528,
      "learning_rate": 9.995872970463824e-06,
      "loss": 0.9474,
      "step": 333
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.828210031293635,
      "learning_rate": 9.995788327537983e-06,
      "loss": 0.939,
      "step": 334
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.126772533614103,
      "learning_rate": 9.995702825778213e-06,
      "loss": 0.9029,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4999870517870286,
      "learning_rate": 9.995616465199209e-06,
      "loss": 0.9408,
      "step": 336
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3043750915999897,
      "learning_rate": 9.995529245815824e-06,
      "loss": 1.0156,
      "step": 337
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3949353559276796,
      "learning_rate": 9.995441167643048e-06,
      "loss": 0.8547,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9257140071905166,
      "learning_rate": 9.995352230696021e-06,
      "loss": 0.6856,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.090685345887354,
      "learning_rate": 9.995262434990036e-06,
      "loss": 0.8848,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.495395330691129,
      "learning_rate": 9.995171780540528e-06,
      "loss": 0.9394,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.948739883257739,
      "learning_rate": 9.995080267363082e-06,
      "loss": 0.6532,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6305220303198917,
      "learning_rate": 9.994987895473431e-06,
      "loss": 0.9151,
      "step": 343
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.514652017512981,
      "learning_rate": 9.994894664887453e-06,
      "loss": 0.8881,
      "step": 344
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2754505368349365,
      "learning_rate": 9.994800575621176e-06,
      "loss": 0.894,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.221216993650877,
      "learning_rate": 9.994705627690777e-06,
      "loss": 0.9386,
      "step": 346
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5621713555310752,
      "learning_rate": 9.994609821112576e-06,
      "loss": 0.9241,
      "step": 347
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.907443875813683,
      "learning_rate": 9.994513155903042e-06,
      "loss": 1.0385,
      "step": 348
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.762849835088825,
      "learning_rate": 9.994415632078797e-06,
      "loss": 1.0464,
      "step": 349
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5280260352988155,
      "learning_rate": 9.994317249656607e-06,
      "loss": 0.8053,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.082578976147256,
      "learning_rate": 9.994218008653381e-06,
      "loss": 0.6495,
      "step": 351
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.104445659798836,
      "learning_rate": 9.994117909086179e-06,
      "loss": 0.8608,
      "step": 352
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1309724466099444,
      "learning_rate": 9.994016950972214e-06,
      "loss": 0.8856,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6719560406789984,
      "learning_rate": 9.99391513432884e-06,
      "loss": 0.9495,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9193408229426347,
      "learning_rate": 9.993812459173557e-06,
      "loss": 0.6647,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.763840076214024,
      "learning_rate": 9.993708925524022e-06,
      "loss": 0.9322,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1812205911456022,
      "learning_rate": 9.993604533398029e-06,
      "loss": 0.7088,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3840061065270055,
      "learning_rate": 9.993499282813528e-06,
      "loss": 0.8931,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5510173029068322,
      "learning_rate": 9.993393173788608e-06,
      "loss": 0.9016,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7024200774698408,
      "learning_rate": 9.993286206341515e-06,
      "loss": 0.927,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.058456932993282,
      "learning_rate": 9.993178380490636e-06,
      "loss": 0.9976,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7583715805639815,
      "learning_rate": 9.993069696254506e-06,
      "loss": 0.9352,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0472056537367338,
      "learning_rate": 9.992960153651812e-06,
      "loss": 0.7337,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.509270532131139,
      "learning_rate": 9.992849752701384e-06,
      "loss": 0.9435,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8846515512755493,
      "learning_rate": 9.9927384934222e-06,
      "loss": 0.905,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4391658577605944,
      "learning_rate": 9.99262637583339e-06,
      "loss": 0.9131,
      "step": 366
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8491507250407078,
      "learning_rate": 9.992513399954225e-06,
      "loss": 0.9353,
      "step": 367
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9858013833396437,
      "learning_rate": 9.99239956580413e-06,
      "loss": 0.8728,
      "step": 368
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4308386119916146,
      "learning_rate": 9.99228487340267e-06,
      "loss": 0.8995,
      "step": 369
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6760213070906824,
      "learning_rate": 9.992169322769568e-06,
      "loss": 0.9147,
      "step": 370
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.400638659504613,
      "learning_rate": 9.992052913924683e-06,
      "loss": 0.9318,
      "step": 371
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8734837217465818,
      "learning_rate": 9.991935646888031e-06,
      "loss": 0.617,
      "step": 372
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.065190303800916,
      "learning_rate": 9.991817521679769e-06,
      "loss": 0.9453,
      "step": 373
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7144926133828227,
      "learning_rate": 9.991698538320205e-06,
      "loss": 0.8564,
      "step": 374
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9934491843730944,
      "learning_rate": 9.991578696829793e-06,
      "loss": 0.687,
      "step": 375
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9821236300831087,
      "learning_rate": 9.991457997229136e-06,
      "loss": 0.926,
      "step": 376
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9149559898562633,
      "learning_rate": 9.991336439538983e-06,
      "loss": 0.6879,
      "step": 377
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6227999238025788,
      "learning_rate": 9.991214023780232e-06,
      "loss": 1.0005,
      "step": 378
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8300153068334097,
      "learning_rate": 9.991090749973926e-06,
      "loss": 0.9392,
      "step": 379
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.629800095901057,
      "learning_rate": 9.99096661814126e-06,
      "loss": 0.9707,
      "step": 380
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9141689026315685,
      "learning_rate": 9.990841628303571e-06,
      "loss": 0.9047,
      "step": 381
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.593364893119427,
      "learning_rate": 9.990715780482348e-06,
      "loss": 0.8939,
      "step": 382
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5994633789132295,
      "learning_rate": 9.990589074699225e-06,
      "loss": 0.9305,
      "step": 383
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1059481092786103,
      "learning_rate": 9.990461510975983e-06,
      "loss": 0.9392,
      "step": 384
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3963699419504276,
      "learning_rate": 9.990333089334556e-06,
      "loss": 0.9001,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1389092584359317,
      "learning_rate": 9.990203809797014e-06,
      "loss": 0.893,
      "step": 386
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4948865968630745,
      "learning_rate": 9.990073672385588e-06,
      "loss": 0.8624,
      "step": 387
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4864365673085629,
      "learning_rate": 9.989942677122648e-06,
      "loss": 0.7973,
      "step": 388
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.814456532198197,
      "learning_rate": 9.989810824030712e-06,
      "loss": 0.8654,
      "step": 389
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.412809344346012,
      "learning_rate": 9.989678113132451e-06,
      "loss": 0.904,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.490489487995922,
      "learning_rate": 9.989544544450675e-06,
      "loss": 0.8476,
      "step": 391
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.514171888284984,
      "learning_rate": 9.989410118008348e-06,
      "loss": 0.929,
      "step": 392
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4459244732111103,
      "learning_rate": 9.98927483382858e-06,
      "loss": 0.96,
      "step": 393
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6184141485868626,
      "learning_rate": 9.989138691934628e-06,
      "loss": 0.8413,
      "step": 394
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.685547052111598,
      "learning_rate": 9.989001692349894e-06,
      "loss": 0.827,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.456763196863134,
      "learning_rate": 9.988863835097934e-06,
      "loss": 0.9583,
      "step": 396
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9596069203717714,
      "learning_rate": 9.988725120202442e-06,
      "loss": 0.8471,
      "step": 397
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.487559714632089,
      "learning_rate": 9.98858554768727e-06,
      "loss": 0.8184,
      "step": 398
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3312009784622199,
      "learning_rate": 9.988445117576408e-06,
      "loss": 0.906,
      "step": 399
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4369034874644226,
      "learning_rate": 9.988303829894002e-06,
      "loss": 0.8726,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8691491163771085,
      "learning_rate": 9.988161684664336e-06,
      "loss": 0.6169,
      "step": 401
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5359853860591064,
      "learning_rate": 9.988018681911849e-06,
      "loss": 0.9178,
      "step": 402
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4696292946289993,
      "learning_rate": 9.987874821661124e-06,
      "loss": 0.8451,
      "step": 403
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9879155422044806,
      "learning_rate": 9.987730103936895e-06,
      "loss": 0.7218,
      "step": 404
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6039838195753158,
      "learning_rate": 9.987584528764036e-06,
      "loss": 1.0481,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8790423190270883,
      "learning_rate": 9.987438096167577e-06,
      "loss": 0.7409,
      "step": 406
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8978077837393668,
      "learning_rate": 9.987290806172689e-06,
      "loss": 0.9677,
      "step": 407
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9021339135726669,
      "learning_rate": 9.987142658804692e-06,
      "loss": 0.6501,
      "step": 408
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2433706356743017,
      "learning_rate": 9.986993654089059e-06,
      "loss": 0.8382,
      "step": 409
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.621100280082855,
      "learning_rate": 9.986843792051402e-06,
      "loss": 0.8899,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9814302126615053,
      "learning_rate": 9.986693072717483e-06,
      "loss": 0.9404,
      "step": 411
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4450395362853512,
      "learning_rate": 9.986541496113215e-06,
      "loss": 0.9471,
      "step": 412
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4875361862837677,
      "learning_rate": 9.986389062264656e-06,
      "loss": 0.908,
      "step": 413
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9527375911313267,
      "learning_rate": 9.98623577119801e-06,
      "loss": 0.871,
      "step": 414
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.152727981738004,
      "learning_rate": 9.98608162293963e-06,
      "loss": 0.873,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9356104280381223,
      "learning_rate": 9.985926617516012e-06,
      "loss": 0.8943,
      "step": 416
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7684168841858503,
      "learning_rate": 9.98577075495381e-06,
      "loss": 0.8637,
      "step": 417
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5661270628542665,
      "learning_rate": 9.985614035279813e-06,
      "loss": 0.8554,
      "step": 418
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.803991467422983,
      "learning_rate": 9.985456458520968e-06,
      "loss": 0.9609,
      "step": 419
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9911428323665277,
      "learning_rate": 9.98529802470436e-06,
      "loss": 0.8738,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8109060937744286,
      "learning_rate": 9.985138733857225e-06,
      "loss": 0.8869,
      "step": 421
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.011379507172746,
      "learning_rate": 9.984978586006951e-06,
      "loss": 0.8349,
      "step": 422
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4498771725802553,
      "learning_rate": 9.984817581181068e-06,
      "loss": 0.8483,
      "step": 423
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9262788955635746,
      "learning_rate": 9.984655719407252e-06,
      "loss": 0.6788,
      "step": 424
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.933155479534556,
      "learning_rate": 9.98449300071333e-06,
      "loss": 0.8918,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7614640454397439,
      "learning_rate": 9.98432942512728e-06,
      "loss": 0.9577,
      "step": 426
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0224053676545932,
      "learning_rate": 9.984164992677215e-06,
      "loss": 0.7098,
      "step": 427
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.562995880202577,
      "learning_rate": 9.983999703391408e-06,
      "loss": 0.9911,
      "step": 428
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8710324427313426,
      "learning_rate": 9.98383355729827e-06,
      "loss": 0.7253,
      "step": 429
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.636745437465233,
      "learning_rate": 9.983666554426367e-06,
      "loss": 0.9871,
      "step": 430
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8810777828303167,
      "learning_rate": 9.98349869480441e-06,
      "loss": 0.7086,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6564175740004912,
      "learning_rate": 9.983329978461252e-06,
      "loss": 0.9061,
      "step": 432
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7270790747482259,
      "learning_rate": 9.9831604054259e-06,
      "loss": 0.9357,
      "step": 433
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.820074729228359,
      "learning_rate": 9.982989975727502e-06,
      "loss": 0.8784,
      "step": 434
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4835840481085412,
      "learning_rate": 9.982818689395362e-06,
      "loss": 0.902,
      "step": 435
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4279075465995763,
      "learning_rate": 9.982646546458922e-06,
      "loss": 0.8477,
      "step": 436
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.365975468091801,
      "learning_rate": 9.982473546947777e-06,
      "loss": 0.9689,
      "step": 437
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4788501961015508,
      "learning_rate": 9.982299690891668e-06,
      "loss": 0.8134,
      "step": 438
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7083566299256943,
      "learning_rate": 9.982124978320482e-06,
      "loss": 0.9848,
      "step": 439
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9348094945584453,
      "learning_rate": 9.981949409264256e-06,
      "loss": 0.7096,
      "step": 440
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.409531876466612,
      "learning_rate": 9.981772983753169e-06,
      "loss": 0.8861,
      "step": 441
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3692353001445567,
      "learning_rate": 9.981595701817552e-06,
      "loss": 0.9476,
      "step": 442
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.487256274132707,
      "learning_rate": 9.981417563487884e-06,
      "loss": 0.988,
      "step": 443
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5360360751232038,
      "learning_rate": 9.981238568794787e-06,
      "loss": 0.823,
      "step": 444
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.687387333431371,
      "learning_rate": 9.981058717769031e-06,
      "loss": 0.916,
      "step": 445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9211205369780907,
      "learning_rate": 9.980878010441539e-06,
      "loss": 0.6798,
      "step": 446
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.795895359952637,
      "learning_rate": 9.980696446843371e-06,
      "loss": 0.8927,
      "step": 447
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4921872789696033,
      "learning_rate": 9.980514027005743e-06,
      "loss": 0.9281,
      "step": 448
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5336316316024645,
      "learning_rate": 9.980330750960014e-06,
      "loss": 0.9148,
      "step": 449
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9249366529829744,
      "learning_rate": 9.980146618737694e-06,
      "loss": 0.6494,
      "step": 450
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8585869243481463,
      "learning_rate": 9.979961630370435e-06,
      "loss": 0.9263,
      "step": 451
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4370641896799758,
      "learning_rate": 9.979775785890037e-06,
      "loss": 0.8712,
      "step": 452
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5342854844225169,
      "learning_rate": 9.979589085328454e-06,
      "loss": 0.9332,
      "step": 453
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4092828064842802,
      "learning_rate": 9.979401528717777e-06,
      "loss": 0.8365,
      "step": 454
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.477301260598403,
      "learning_rate": 9.979213116090251e-06,
      "loss": 0.888,
      "step": 455
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7198583252036432,
      "learning_rate": 9.979023847478268e-06,
      "loss": 0.8805,
      "step": 456
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7146105023830878,
      "learning_rate": 9.978833722914362e-06,
      "loss": 0.8344,
      "step": 457
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.705970785918567,
      "learning_rate": 9.97864274243122e-06,
      "loss": 0.8922,
      "step": 458
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7949910571284176,
      "learning_rate": 9.978450906061673e-06,
      "loss": 0.9032,
      "step": 459
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6480715671245807,
      "learning_rate": 9.978258213838701e-06,
      "loss": 0.7622,
      "step": 460
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.802821185808631,
      "learning_rate": 9.978064665795429e-06,
      "loss": 0.9216,
      "step": 461
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8562930813547619,
      "learning_rate": 9.977870261965128e-06,
      "loss": 0.6969,
      "step": 462
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.936659107293068,
      "learning_rate": 9.977675002381222e-06,
      "loss": 0.8848,
      "step": 463
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.028915440465218,
      "learning_rate": 9.977478887077277e-06,
      "loss": 0.9198,
      "step": 464
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.865370770639181,
      "learning_rate": 9.977281916087008e-06,
      "loss": 0.9548,
      "step": 465
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5950900047713257,
      "learning_rate": 9.977084089444276e-06,
      "loss": 0.8363,
      "step": 466
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9709539119489183,
      "learning_rate": 9.976885407183088e-06,
      "loss": 0.6541,
      "step": 467
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7142808721895724,
      "learning_rate": 9.976685869337603e-06,
      "loss": 0.9763,
      "step": 468
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4847629971048646,
      "learning_rate": 9.976485475942123e-06,
      "loss": 0.9144,
      "step": 469
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6419470370722928,
      "learning_rate": 9.976284227031097e-06,
      "loss": 0.8917,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.505954264119548,
      "learning_rate": 9.976082122639122e-06,
      "loss": 0.9443,
      "step": 471
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1217623223161053,
      "learning_rate": 9.975879162800942e-06,
      "loss": 0.8835,
      "step": 472
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5295084383032893,
      "learning_rate": 9.975675347551449e-06,
      "loss": 1.0242,
      "step": 473
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3933781990209466,
      "learning_rate": 9.975470676925681e-06,
      "loss": 0.9938,
      "step": 474
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6880142776953593,
      "learning_rate": 9.975265150958822e-06,
      "loss": 0.8796,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.435832088275508,
      "learning_rate": 9.975058769686205e-06,
      "loss": 0.9678,
      "step": 476
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6323399490184938,
      "learning_rate": 9.97485153314331e-06,
      "loss": 0.8431,
      "step": 477
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5679741705503303,
      "learning_rate": 9.974643441365765e-06,
      "loss": 0.8995,
      "step": 478
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2447338464523625,
      "learning_rate": 9.97443449438934e-06,
      "loss": 0.8682,
      "step": 479
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7936392804876804,
      "learning_rate": 9.974224692249956e-06,
      "loss": 0.9039,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4967772355335454,
      "learning_rate": 9.974014034983684e-06,
      "loss": 0.8062,
      "step": 481
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5068501755939459,
      "learning_rate": 9.973802522626731e-06,
      "loss": 0.9119,
      "step": 482
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.373391482749295,
      "learning_rate": 9.973590155215467e-06,
      "loss": 0.7437,
      "step": 483
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0272584880150444,
      "learning_rate": 9.973376932786396e-06,
      "loss": 0.6888,
      "step": 484
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5029242204492068,
      "learning_rate": 9.973162855376173e-06,
      "loss": 0.993,
      "step": 485
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3619288227656734,
      "learning_rate": 9.972947923021602e-06,
      "loss": 0.8576,
      "step": 486
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5227547343734409,
      "learning_rate": 9.972732135759631e-06,
      "loss": 0.9134,
      "step": 487
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8727145009462977,
      "learning_rate": 9.972515493627357e-06,
      "loss": 0.628,
      "step": 488
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2757380177906426,
      "learning_rate": 9.972297996662023e-06,
      "loss": 0.9324,
      "step": 489
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4735299081450226,
      "learning_rate": 9.972079644901021e-06,
      "loss": 0.94,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3942611461756789,
      "learning_rate": 9.971860438381886e-06,
      "loss": 1.0661,
      "step": 491
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4182424298470129,
      "learning_rate": 9.971640377142301e-06,
      "loss": 1.0033,
      "step": 492
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3385728617354542,
      "learning_rate": 9.971419461220102e-06,
      "loss": 0.8839,
      "step": 493
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4372944352412866,
      "learning_rate": 9.971197690653263e-06,
      "loss": 0.9028,
      "step": 494
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6221129373214658,
      "learning_rate": 9.970975065479909e-06,
      "loss": 0.8816,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4347452196122952,
      "learning_rate": 9.970751585738315e-06,
      "loss": 0.8053,
      "step": 496
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.041808440296312,
      "learning_rate": 9.970527251466895e-06,
      "loss": 0.8947,
      "step": 497
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5314386610959372,
      "learning_rate": 9.970302062704218e-06,
      "loss": 0.9728,
      "step": 498
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8409481956645593,
      "learning_rate": 9.970076019488997e-06,
      "loss": 0.8371,
      "step": 499
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4784841034237566,
      "learning_rate": 9.969849121860089e-06,
      "loss": 0.8753,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5948985136201943,
      "learning_rate": 9.969621369856502e-06,
      "loss": 0.904,
      "step": 501
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5223772162374083,
      "learning_rate": 9.969392763517387e-06,
      "loss": 0.947,
      "step": 502
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.335460953568833,
      "learning_rate": 9.969163302882048e-06,
      "loss": 0.9236,
      "step": 503
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5754868075428776,
      "learning_rate": 9.968932987989927e-06,
      "loss": 0.9054,
      "step": 504
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.506387790328525,
      "learning_rate": 9.968701818880624e-06,
      "loss": 0.8856,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.460541635458987,
      "learning_rate": 9.968469795593874e-06,
      "loss": 0.9022,
      "step": 506
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4841729199921412,
      "learning_rate": 9.968236918169567e-06,
      "loss": 0.9366,
      "step": 507
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.591949878277734,
      "learning_rate": 9.968003186647737e-06,
      "loss": 0.9258,
      "step": 508
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7795039355614046,
      "learning_rate": 9.967768601068566e-06,
      "loss": 0.9001,
      "step": 509
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.934844422151346,
      "learning_rate": 9.96753316147238e-06,
      "loss": 0.942,
      "step": 510
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.819546883452969,
      "learning_rate": 9.967296867899656e-06,
      "loss": 0.9319,
      "step": 511
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9934826317171261,
      "learning_rate": 9.967059720391014e-06,
      "loss": 0.8658,
      "step": 512
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.722010609964779,
      "learning_rate": 9.966821718987222e-06,
      "loss": 0.9224,
      "step": 513
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6125091775214078,
      "learning_rate": 9.966582863729198e-06,
      "loss": 0.8812,
      "step": 514
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7654704769839578,
      "learning_rate": 9.966343154658002e-06,
      "loss": 1.0003,
      "step": 515
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3979206829438044,
      "learning_rate": 9.966102591814842e-06,
      "loss": 0.8986,
      "step": 516
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3996636912870373,
      "learning_rate": 9.965861175241075e-06,
      "loss": 0.9975,
      "step": 517
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4648660725746043,
      "learning_rate": 9.965618904978203e-06,
      "loss": 0.8219,
      "step": 518
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.165242193571012,
      "learning_rate": 9.965375781067874e-06,
      "loss": 0.8215,
      "step": 519
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.297662595833426,
      "learning_rate": 9.965131803551885e-06,
      "loss": 0.8884,
      "step": 520
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2743377976413415,
      "learning_rate": 9.96488697247218e-06,
      "loss": 0.8973,
      "step": 521
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9134070042394735,
      "learning_rate": 9.964641287870844e-06,
      "loss": 0.6756,
      "step": 522
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9575128400561603,
      "learning_rate": 9.964394749790117e-06,
      "loss": 0.8926,
      "step": 523
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8846352235372759,
      "learning_rate": 9.964147358272379e-06,
      "loss": 0.9524,
      "step": 524
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4614102713026105,
      "learning_rate": 9.963899113360161e-06,
      "loss": 0.8658,
      "step": 525
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8145936535449907,
      "learning_rate": 9.96365001509614e-06,
      "loss": 0.8395,
      "step": 526
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1141485623893053,
      "learning_rate": 9.963400063523136e-06,
      "loss": 0.6005,
      "step": 527
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5367130970364726,
      "learning_rate": 9.963149258684122e-06,
      "loss": 1.0657,
      "step": 528
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3723324387437956,
      "learning_rate": 9.962897600622212e-06,
      "loss": 0.8392,
      "step": 529
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4636146813325057,
      "learning_rate": 9.962645089380672e-06,
      "loss": 0.8487,
      "step": 530
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5941486846171915,
      "learning_rate": 9.962391725002906e-06,
      "loss": 0.8511,
      "step": 531
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.840547963763196,
      "learning_rate": 9.962137507532474e-06,
      "loss": 0.6763,
      "step": 532
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.695330445381718,
      "learning_rate": 9.961882437013079e-06,
      "loss": 0.8727,
      "step": 533
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9218614086948458,
      "learning_rate": 9.961626513488572e-06,
      "loss": 0.9003,
      "step": 534
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6267892680520522,
      "learning_rate": 9.961369737002943e-06,
      "loss": 0.841,
      "step": 535
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4553423455011938,
      "learning_rate": 9.961112107600343e-06,
      "loss": 0.8722,
      "step": 536
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4420177205240063,
      "learning_rate": 9.960853625325056e-06,
      "loss": 0.9038,
      "step": 537
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3676349896506892,
      "learning_rate": 9.960594290221519e-06,
      "loss": 0.8371,
      "step": 538
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.579128131467461,
      "learning_rate": 9.960334102334316e-06,
      "loss": 0.9048,
      "step": 539
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3948150398594077,
      "learning_rate": 9.960073061708176e-06,
      "loss": 0.8636,
      "step": 540
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7311204049223636,
      "learning_rate": 9.959811168387974e-06,
      "loss": 0.7809,
      "step": 541
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.419072363951772,
      "learning_rate": 9.959548422418734e-06,
      "loss": 0.8585,
      "step": 542
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7749990267690658,
      "learning_rate": 9.959284823845623e-06,
      "loss": 0.8429,
      "step": 543
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.609676860618843,
      "learning_rate": 9.959020372713959e-06,
      "loss": 0.8711,
      "step": 544
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7245180847482537,
      "learning_rate": 9.958755069069202e-06,
      "loss": 0.9616,
      "step": 545
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4251638953549841,
      "learning_rate": 9.958488912956961e-06,
      "loss": 0.8452,
      "step": 546
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3307101929895873,
      "learning_rate": 9.958221904422993e-06,
      "loss": 0.8875,
      "step": 547
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4990689190840136,
      "learning_rate": 9.9579540435132e-06,
      "loss": 0.973,
      "step": 548
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4502241300031433,
      "learning_rate": 9.957685330273628e-06,
      "loss": 0.9679,
      "step": 549
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3446017432882464,
      "learning_rate": 9.957415764750474e-06,
      "loss": 0.9079,
      "step": 550
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3966166700684162,
      "learning_rate": 9.957145346990079e-06,
      "loss": 0.8258,
      "step": 551
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4232823280135196,
      "learning_rate": 9.95687407703893e-06,
      "loss": 0.9031,
      "step": 552
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4608556039590668,
      "learning_rate": 9.956601954943664e-06,
      "loss": 0.9064,
      "step": 553
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4554814997500907,
      "learning_rate": 9.956328980751057e-06,
      "loss": 0.915,
      "step": 554
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.195826090560842,
      "learning_rate": 9.956055154508042e-06,
      "loss": 0.8531,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4067631996753491,
      "learning_rate": 9.955780476261689e-06,
      "loss": 0.8317,
      "step": 556
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9319899069421587,
      "learning_rate": 9.955504946059221e-06,
      "loss": 0.6744,
      "step": 557
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0217009388791336,
      "learning_rate": 9.955228563948003e-06,
      "loss": 0.6627,
      "step": 558
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4563195150603219,
      "learning_rate": 9.954951329975549e-06,
      "loss": 0.9704,
      "step": 559
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8017317725771016,
      "learning_rate": 9.954673244189518e-06,
      "loss": 0.9216,
      "step": 560
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9538741807306853,
      "learning_rate": 9.954394306637719e-06,
      "loss": 0.7132,
      "step": 561
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.963481937963558,
      "learning_rate": 9.9541145173681e-06,
      "loss": 0.677,
      "step": 562
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.374873126201534,
      "learning_rate": 9.953833876428763e-06,
      "loss": 0.8191,
      "step": 563
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9320744846271126,
      "learning_rate": 9.953552383867953e-06,
      "loss": 0.6806,
      "step": 564
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.272060092530244,
      "learning_rate": 9.953270039734063e-06,
      "loss": 0.78,
      "step": 565
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3426114572850154,
      "learning_rate": 9.952986844075628e-06,
      "loss": 0.8825,
      "step": 566
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.784021962078458,
      "learning_rate": 9.952702796941334e-06,
      "loss": 0.7692,
      "step": 567
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5194848617936492,
      "learning_rate": 9.952417898380015e-06,
      "loss": 0.9078,
      "step": 568
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5692771305920268,
      "learning_rate": 9.952132148440644e-06,
      "loss": 0.8733,
      "step": 569
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3754470423096086,
      "learning_rate": 9.951845547172347e-06,
      "loss": 0.911,
      "step": 570
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9415379845005487,
      "learning_rate": 9.951558094624395e-06,
      "loss": 0.9051,
      "step": 571
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4447883307428895,
      "learning_rate": 9.951269790846204e-06,
      "loss": 0.8613,
      "step": 572
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3832335433590628,
      "learning_rate": 9.950980635887332e-06,
      "loss": 0.8363,
      "step": 573
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3666249055406785,
      "learning_rate": 9.950690629797494e-06,
      "loss": 0.9375,
      "step": 574
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8463141558189617,
      "learning_rate": 9.950399772626545e-06,
      "loss": 0.7212,
      "step": 575
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4275991481487256,
      "learning_rate": 9.950108064424482e-06,
      "loss": 0.9369,
      "step": 576
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.834376188822143,
      "learning_rate": 9.949815505241458e-06,
      "loss": 0.8991,
      "step": 577
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3695013211312115,
      "learning_rate": 9.949522095127765e-06,
      "loss": 0.8504,
      "step": 578
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4610018051830167,
      "learning_rate": 9.949227834133845e-06,
      "loss": 0.8777,
      "step": 579
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4402937986263793,
      "learning_rate": 9.948932722310283e-06,
      "loss": 0.9163,
      "step": 580
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.484026509650082,
      "learning_rate": 9.948636759707815e-06,
      "loss": 0.8704,
      "step": 581
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8635280829037856,
      "learning_rate": 9.948339946377316e-06,
      "loss": 0.5993,
      "step": 582
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7232940086335815,
      "learning_rate": 9.948042282369816e-06,
      "loss": 0.8334,
      "step": 583
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6978059719350624,
      "learning_rate": 9.947743767736485e-06,
      "loss": 0.941,
      "step": 584
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9801372512768626,
      "learning_rate": 9.947444402528638e-06,
      "loss": 0.7078,
      "step": 585
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2827896780647938,
      "learning_rate": 9.947144186797747e-06,
      "loss": 0.8218,
      "step": 586
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.375698542642964,
      "learning_rate": 9.946843120595415e-06,
      "loss": 0.9167,
      "step": 587
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4609679413811127,
      "learning_rate": 9.946541203973402e-06,
      "loss": 0.8808,
      "step": 588
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5348484940786171,
      "learning_rate": 9.94623843698361e-06,
      "loss": 0.8404,
      "step": 589
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4738676377736217,
      "learning_rate": 9.945934819678092e-06,
      "loss": 0.8859,
      "step": 590
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5194580435844762,
      "learning_rate": 9.945630352109039e-06,
      "loss": 0.9043,
      "step": 591
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9296148776919961,
      "learning_rate": 9.945325034328791e-06,
      "loss": 0.6174,
      "step": 592
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.458672476003813,
      "learning_rate": 9.945018866389844e-06,
      "loss": 0.9546,
      "step": 593
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.376185053713388,
      "learning_rate": 9.944711848344822e-06,
      "loss": 0.8506,
      "step": 594
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.664646367285724,
      "learning_rate": 9.94440398024651e-06,
      "loss": 0.8218,
      "step": 595
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1222335049939067,
      "learning_rate": 9.944095262147835e-06,
      "loss": 0.9259,
      "step": 596
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9599545669583864,
      "learning_rate": 9.943785694101865e-06,
      "loss": 0.6564,
      "step": 597
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4603898300109004,
      "learning_rate": 9.943475276161823e-06,
      "loss": 0.9017,
      "step": 598
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8565707259162203,
      "learning_rate": 9.943164008381072e-06,
      "loss": 0.7003,
      "step": 599
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9347256911323591,
      "learning_rate": 9.94285189081312e-06,
      "loss": 0.8642,
      "step": 600
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8745850982539625,
      "learning_rate": 9.942538923511627e-06,
      "loss": 0.6739,
      "step": 601
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8489025019819414,
      "learning_rate": 9.942225106530391e-06,
      "loss": 0.896,
      "step": 602
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7215887174122178,
      "learning_rate": 9.941910439923367e-06,
      "loss": 0.7768,
      "step": 603
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4701879085774872,
      "learning_rate": 9.941594923744647e-06,
      "loss": 0.7988,
      "step": 604
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6502024659871408,
      "learning_rate": 9.941278558048471e-06,
      "loss": 0.8468,
      "step": 605
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.580734754644296,
      "learning_rate": 9.940961342889225e-06,
      "loss": 0.9478,
      "step": 606
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4127809009436683,
      "learning_rate": 9.940643278321447e-06,
      "loss": 0.8885,
      "step": 607
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3277003825580165,
      "learning_rate": 9.94032436439981e-06,
      "loss": 0.7987,
      "step": 608
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.527872886248207,
      "learning_rate": 9.940004601179143e-06,
      "loss": 0.8567,
      "step": 609
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3766393386239812,
      "learning_rate": 9.939683988714415e-06,
      "loss": 0.8272,
      "step": 610
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9010621252535538,
      "learning_rate": 9.939362527060743e-06,
      "loss": 0.6556,
      "step": 611
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3017739493801015,
      "learning_rate": 9.939040216273392e-06,
      "loss": 0.8245,
      "step": 612
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.208355344955502,
      "learning_rate": 9.938717056407768e-06,
      "loss": 0.8517,
      "step": 613
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4172354670549505,
      "learning_rate": 9.938393047519428e-06,
      "loss": 0.8797,
      "step": 614
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5640082381683456,
      "learning_rate": 9.938068189664073e-06,
      "loss": 0.8983,
      "step": 615
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6722210945965905,
      "learning_rate": 9.937742482897549e-06,
      "loss": 0.9042,
      "step": 616
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.039856884043207,
      "learning_rate": 9.937415927275848e-06,
      "loss": 0.8627,
      "step": 617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5726533028259955,
      "learning_rate": 9.937088522855111e-06,
      "loss": 0.8492,
      "step": 618
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3556522170871306,
      "learning_rate": 9.936760269691621e-06,
      "loss": 0.7813,
      "step": 619
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8892900211369752,
      "learning_rate": 9.936431167841808e-06,
      "loss": 0.7557,
      "step": 620
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.488836657573666,
      "learning_rate": 9.93610121736225e-06,
      "loss": 0.9445,
      "step": 621
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5795022449048135,
      "learning_rate": 9.935770418309667e-06,
      "loss": 0.8413,
      "step": 622
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8915456605371566,
      "learning_rate": 9.93543877074093e-06,
      "loss": 0.6863,
      "step": 623
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4578377076981088,
      "learning_rate": 9.935106274713053e-06,
      "loss": 0.8787,
      "step": 624
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8422337816640346,
      "learning_rate": 9.934772930283194e-06,
      "loss": 0.6711,
      "step": 625
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.498667916382007,
      "learning_rate": 9.934438737508658e-06,
      "loss": 0.9583,
      "step": 626
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2782439398737684,
      "learning_rate": 9.9341036964469e-06,
      "loss": 0.8292,
      "step": 627
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.951615144889712,
      "learning_rate": 9.933767807155515e-06,
      "loss": 0.9361,
      "step": 628
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9324811554986812,
      "learning_rate": 9.933431069692245e-06,
      "loss": 0.9113,
      "step": 629
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5210897644375438,
      "learning_rate": 9.933093484114983e-06,
      "loss": 0.995,
      "step": 630
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5397177920161405,
      "learning_rate": 9.932755050481762e-06,
      "loss": 0.975,
      "step": 631
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.641956570010699,
      "learning_rate": 9.93241576885076e-06,
      "loss": 0.9845,
      "step": 632
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6456936657172132,
      "learning_rate": 9.932075639280308e-06,
      "loss": 0.8595,
      "step": 633
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4434330452025825,
      "learning_rate": 9.931734661828876e-06,
      "loss": 0.8475,
      "step": 634
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9378600588482606,
      "learning_rate": 9.931392836555081e-06,
      "loss": 0.705,
      "step": 635
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3765995574283285,
      "learning_rate": 9.931050163517688e-06,
      "loss": 0.9285,
      "step": 636
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3736004663973527,
      "learning_rate": 9.930706642775607e-06,
      "loss": 0.8063,
      "step": 637
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7154896888630713,
      "learning_rate": 9.930362274387892e-06,
      "loss": 0.8429,
      "step": 638
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9507201596601529,
      "learning_rate": 9.930017058413745e-06,
      "loss": 0.5987,
      "step": 639
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4516960679474098,
      "learning_rate": 9.92967099491251e-06,
      "loss": 0.8389,
      "step": 640
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.392341158458147,
      "learning_rate": 9.929324083943683e-06,
      "loss": 0.8998,
      "step": 641
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7900296498355857,
      "learning_rate": 9.9289763255669e-06,
      "loss": 0.8916,
      "step": 642
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5206773911765277,
      "learning_rate": 9.928627719841945e-06,
      "loss": 0.956,
      "step": 643
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5358019543853512,
      "learning_rate": 9.928278266828747e-06,
      "loss": 0.8204,
      "step": 644
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4170322411962841,
      "learning_rate": 9.927927966587383e-06,
      "loss": 0.813,
      "step": 645
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3952705558957943,
      "learning_rate": 9.927576819178071e-06,
      "loss": 0.9132,
      "step": 646
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.538678566572952,
      "learning_rate": 9.927224824661179e-06,
      "loss": 0.8087,
      "step": 647
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7085181283427815,
      "learning_rate": 9.926871983097217e-06,
      "loss": 0.9248,
      "step": 648
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4733745689124658,
      "learning_rate": 9.926518294546846e-06,
      "loss": 0.8412,
      "step": 649
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4925852395231978,
      "learning_rate": 9.926163759070863e-06,
      "loss": 0.7883,
      "step": 650
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6574700943681302,
      "learning_rate": 9.925808376730225e-06,
      "loss": 0.9255,
      "step": 651
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6641441995680464,
      "learning_rate": 9.925452147586022e-06,
      "loss": 0.9027,
      "step": 652
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5844052331543086,
      "learning_rate": 9.925095071699491e-06,
      "loss": 0.8321,
      "step": 653
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.448622391704303,
      "learning_rate": 9.924737149132022e-06,
      "loss": 0.8299,
      "step": 654
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5068981400168837,
      "learning_rate": 9.924378379945145e-06,
      "loss": 0.8044,
      "step": 655
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4934861006066042,
      "learning_rate": 9.924018764200538e-06,
      "loss": 0.8661,
      "step": 656
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4064512408824712,
      "learning_rate": 9.923658301960018e-06,
      "loss": 0.8415,
      "step": 657
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.488841159058093,
      "learning_rate": 9.923296993285558e-06,
      "loss": 0.8341,
      "step": 658
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.656004971083269,
      "learning_rate": 9.922934838239269e-06,
      "loss": 0.8251,
      "step": 659
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.48013645401219,
      "learning_rate": 9.922571836883408e-06,
      "loss": 0.7664,
      "step": 660
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3739369301205393,
      "learning_rate": 9.922207989280383e-06,
      "loss": 0.8437,
      "step": 661
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4991258115270556,
      "learning_rate": 9.92184329549274e-06,
      "loss": 0.8552,
      "step": 662
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7496590787063584,
      "learning_rate": 9.921477755583178e-06,
      "loss": 0.833,
      "step": 663
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.284248500998559,
      "learning_rate": 9.921111369614533e-06,
      "loss": 0.8045,
      "step": 664
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.579041999811659,
      "learning_rate": 9.920744137649793e-06,
      "loss": 0.847,
      "step": 665
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4680269226763274,
      "learning_rate": 9.920376059752091e-06,
      "loss": 0.8318,
      "step": 666
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3771789089457156,
      "learning_rate": 9.920007135984701e-06,
      "loss": 0.9274,
      "step": 667
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5487751937136376,
      "learning_rate": 9.919637366411047e-06,
      "loss": 0.7566,
      "step": 668
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6480742072669918,
      "learning_rate": 9.919266751094698e-06,
      "loss": 0.8612,
      "step": 669
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9745530200494898,
      "learning_rate": 9.918895290099364e-06,
      "loss": 0.9629,
      "step": 670
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9950714119926627,
      "learning_rate": 9.918522983488905e-06,
      "loss": 0.6529,
      "step": 671
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4951198724835026,
      "learning_rate": 9.918149831327323e-06,
      "loss": 0.8895,
      "step": 672
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3637598058619336,
      "learning_rate": 9.917775833678772e-06,
      "loss": 0.9286,
      "step": 673
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5412314337009732,
      "learning_rate": 9.917400990607542e-06,
      "loss": 0.877,
      "step": 674
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.335430653109575,
      "learning_rate": 9.917025302178074e-06,
      "loss": 0.7958,
      "step": 675
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6570771408995133,
      "learning_rate": 9.916648768454955e-06,
      "loss": 0.9068,
      "step": 676
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.692546187745389,
      "learning_rate": 9.916271389502911e-06,
      "loss": 0.8111,
      "step": 677
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9203671241732108,
      "learning_rate": 9.915893165386824e-06,
      "loss": 0.6277,
      "step": 678
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4924531344575378,
      "learning_rate": 9.915514096171711e-06,
      "loss": 0.8159,
      "step": 679
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8885284631189853,
      "learning_rate": 9.915134181922739e-06,
      "loss": 0.7669,
      "step": 680
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.44041481207171,
      "learning_rate": 9.91475342270522e-06,
      "loss": 0.841,
      "step": 681
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6550181256852081,
      "learning_rate": 9.914371818584612e-06,
      "loss": 0.8876,
      "step": 682
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.799370957703094,
      "learning_rate": 9.913989369626515e-06,
      "loss": 0.6867,
      "step": 683
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.569953968074743,
      "learning_rate": 9.913606075896677e-06,
      "loss": 0.8223,
      "step": 684
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9571040080181074,
      "learning_rate": 9.913221937460991e-06,
      "loss": 0.6964,
      "step": 685
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.745237822788137,
      "learning_rate": 9.912836954385496e-06,
      "loss": 0.8119,
      "step": 686
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3339035400692685,
      "learning_rate": 9.912451126736374e-06,
      "loss": 0.8125,
      "step": 687
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3836186341938692,
      "learning_rate": 9.912064454579953e-06,
      "loss": 0.8715,
      "step": 688
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3773273659995597,
      "learning_rate": 9.911676937982706e-06,
      "loss": 0.8899,
      "step": 689
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2856111527664957,
      "learning_rate": 9.911288577011254e-06,
      "loss": 0.9731,
      "step": 690
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8324550335222667,
      "learning_rate": 9.910899371732358e-06,
      "loss": 0.644,
      "step": 691
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6003230304058695,
      "learning_rate": 9.91050932221293e-06,
      "loss": 0.8769,
      "step": 692
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3278068810114632,
      "learning_rate": 9.910118428520019e-06,
      "loss": 0.8581,
      "step": 693
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2138076876041985,
      "learning_rate": 9.909726690720829e-06,
      "loss": 0.8178,
      "step": 694
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5353620913983101,
      "learning_rate": 9.9093341088827e-06,
      "loss": 0.85,
      "step": 695
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6771323516795542,
      "learning_rate": 9.908940683073127e-06,
      "loss": 0.8927,
      "step": 696
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5459154880939272,
      "learning_rate": 9.90854641335974e-06,
      "loss": 0.8092,
      "step": 697
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4019337028086536,
      "learning_rate": 9.908151299810319e-06,
      "loss": 0.9157,
      "step": 698
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.552556433545433,
      "learning_rate": 9.90775534249279e-06,
      "loss": 0.8437,
      "step": 699
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3556319440890268,
      "learning_rate": 9.907358541475223e-06,
      "loss": 0.9249,
      "step": 700
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6114301357617618,
      "learning_rate": 9.90696089682583e-06,
      "loss": 0.8754,
      "step": 701
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4832720553595964,
      "learning_rate": 9.906562408612973e-06,
      "loss": 0.8893,
      "step": 702
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.576786275663621,
      "learning_rate": 9.906163076905158e-06,
      "loss": 0.7671,
      "step": 703
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3466429428915698,
      "learning_rate": 9.90576290177103e-06,
      "loss": 0.8313,
      "step": 704
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4369838423605703,
      "learning_rate": 9.905361883279388e-06,
      "loss": 0.9417,
      "step": 705
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3643747455711062,
      "learning_rate": 9.90496002149917e-06,
      "loss": 0.8436,
      "step": 706
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3704644987336836,
      "learning_rate": 9.904557316499462e-06,
      "loss": 0.8729,
      "step": 707
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4684701732524132,
      "learning_rate": 9.904153768349494e-06,
      "loss": 0.864,
      "step": 708
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.609820225588423,
      "learning_rate": 9.903749377118637e-06,
      "loss": 0.8621,
      "step": 709
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.248239834504578,
      "learning_rate": 9.903344142876414e-06,
      "loss": 0.8686,
      "step": 710
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.429314963010538,
      "learning_rate": 9.902938065692488e-06,
      "loss": 0.8469,
      "step": 711
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.416176343759042,
      "learning_rate": 9.902531145636668e-06,
      "loss": 0.8379,
      "step": 712
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4321289587272663,
      "learning_rate": 9.902123382778911e-06,
      "loss": 0.8673,
      "step": 713
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3491953009238185,
      "learning_rate": 9.901714777189313e-06,
      "loss": 0.8596,
      "step": 714
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8658988687036724,
      "learning_rate": 9.90130532893812e-06,
      "loss": 0.7823,
      "step": 715
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5338586339576796,
      "learning_rate": 9.900895038095718e-06,
      "loss": 0.8329,
      "step": 716
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9363642182216414,
      "learning_rate": 9.900483904732645e-06,
      "loss": 0.7489,
      "step": 717
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3132336329361194,
      "learning_rate": 9.900071928919577e-06,
      "loss": 0.8137,
      "step": 718
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7101548677815281,
      "learning_rate": 9.899659110727339e-06,
      "loss": 1.0395,
      "step": 719
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4039193908796952,
      "learning_rate": 9.899245450226897e-06,
      "loss": 0.8363,
      "step": 720
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0409291822274638,
      "learning_rate": 9.898830947489365e-06,
      "loss": 0.643,
      "step": 721
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6221827386269123,
      "learning_rate": 9.898415602586e-06,
      "loss": 0.8696,
      "step": 722
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4674712744584608,
      "learning_rate": 9.89799941558821e-06,
      "loss": 0.9138,
      "step": 723
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5332614226409378,
      "learning_rate": 9.897582386567532e-06,
      "loss": 0.8336,
      "step": 724
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2322429515602236,
      "learning_rate": 9.897164515595667e-06,
      "loss": 0.8338,
      "step": 725
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0837388391935363,
      "learning_rate": 9.89674580274445e-06,
      "loss": 0.9391,
      "step": 726
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1142031112460025,
      "learning_rate": 9.896326248085862e-06,
      "loss": 0.6938,
      "step": 727
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.402250737204862,
      "learning_rate": 9.895905851692026e-06,
      "loss": 0.9391,
      "step": 728
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4567200185822768,
      "learning_rate": 9.895484613635217e-06,
      "loss": 0.9557,
      "step": 729
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3212631191908613,
      "learning_rate": 9.89506253398785e-06,
      "loss": 0.9809,
      "step": 730
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7054681084737362,
      "learning_rate": 9.894639612822486e-06,
      "loss": 0.8736,
      "step": 731
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7846339225737795,
      "learning_rate": 9.894215850211825e-06,
      "loss": 0.904,
      "step": 732
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4703119444350163,
      "learning_rate": 9.893791246228726e-06,
      "loss": 0.904,
      "step": 733
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4731258367090465,
      "learning_rate": 9.893365800946176e-06,
      "loss": 0.8994,
      "step": 734
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4036669819975356,
      "learning_rate": 9.892939514437314e-06,
      "loss": 0.8973,
      "step": 735
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.579992340664332,
      "learning_rate": 9.892512386775429e-06,
      "loss": 0.7782,
      "step": 736
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.592225277738906,
      "learning_rate": 9.892084418033942e-06,
      "loss": 0.9714,
      "step": 737
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.467138257249972,
      "learning_rate": 9.891655608286432e-06,
      "loss": 0.8043,
      "step": 738
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2604314859335095,
      "learning_rate": 9.891225957606613e-06,
      "loss": 0.9035,
      "step": 739
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7683225178956719,
      "learning_rate": 9.890795466068351e-06,
      "loss": 0.893,
      "step": 740
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.310403216421056,
      "learning_rate": 9.890364133745646e-06,
      "loss": 0.9249,
      "step": 741
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4523475735862401,
      "learning_rate": 9.889931960712653e-06,
      "loss": 0.9536,
      "step": 742
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4790068748048797,
      "learning_rate": 9.88949894704367e-06,
      "loss": 0.9081,
      "step": 743
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.708777260189117,
      "learning_rate": 9.889065092813131e-06,
      "loss": 0.7895,
      "step": 744
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5154150424173516,
      "learning_rate": 9.888630398095623e-06,
      "loss": 1.0235,
      "step": 745
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.73447888633648,
      "learning_rate": 9.888194862965877e-06,
      "loss": 0.8935,
      "step": 746
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4222287267531637,
      "learning_rate": 9.887758487498765e-06,
      "loss": 0.9265,
      "step": 747
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.350053050880964,
      "learning_rate": 9.887321271769305e-06,
      "loss": 0.8412,
      "step": 748
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5103343334655954,
      "learning_rate": 9.88688321585266e-06,
      "loss": 0.925,
      "step": 749
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9429570719245018,
      "learning_rate": 9.886444319824138e-06,
      "loss": 0.6805,
      "step": 750
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3745178740064528,
      "learning_rate": 9.886004583759187e-06,
      "loss": 0.9162,
      "step": 751
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.424972728528098,
      "learning_rate": 9.885564007733406e-06,
      "loss": 0.8749,
      "step": 752
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5689519708165216,
      "learning_rate": 9.885122591822533e-06,
      "loss": 0.8966,
      "step": 753
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.511032261639701,
      "learning_rate": 9.884680336102452e-06,
      "loss": 0.8827,
      "step": 754
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3965990630737286,
      "learning_rate": 9.884237240649195e-06,
      "loss": 0.9267,
      "step": 755
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3670411711511294,
      "learning_rate": 9.883793305538934e-06,
      "loss": 0.8706,
      "step": 756
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5149432656476183,
      "learning_rate": 9.883348530847985e-06,
      "loss": 0.8644,
      "step": 757
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7766157561393137,
      "learning_rate": 9.88290291665281e-06,
      "loss": 0.9082,
      "step": 758
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.330077990831086,
      "learning_rate": 9.882456463030017e-06,
      "loss": 0.8941,
      "step": 759
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.428311760946521,
      "learning_rate": 9.882009170056354e-06,
      "loss": 0.9368,
      "step": 760
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3430501258861223,
      "learning_rate": 9.88156103780872e-06,
      "loss": 0.824,
      "step": 761
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9964658245636102,
      "learning_rate": 9.881112066364151e-06,
      "loss": 0.6772,
      "step": 762
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9303816518890744,
      "learning_rate": 9.880662255799831e-06,
      "loss": 0.9218,
      "step": 763
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5781973302081018,
      "learning_rate": 9.880211606193089e-06,
      "loss": 0.977,
      "step": 764
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4201450907042772,
      "learning_rate": 9.879760117621393e-06,
      "loss": 0.829,
      "step": 765
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.767896008110221,
      "learning_rate": 9.879307790162364e-06,
      "loss": 1.0187,
      "step": 766
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3061386796203327,
      "learning_rate": 9.87885462389376e-06,
      "loss": 0.7575,
      "step": 767
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.477008730427403,
      "learning_rate": 9.878400618893483e-06,
      "loss": 0.8936,
      "step": 768
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5734494290755383,
      "learning_rate": 9.877945775239585e-06,
      "loss": 0.8404,
      "step": 769
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3973869308047526,
      "learning_rate": 9.87749009301026e-06,
      "loss": 0.8615,
      "step": 770
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.014872428761152,
      "learning_rate": 9.877033572283842e-06,
      "loss": 0.8313,
      "step": 771
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5130416040075685,
      "learning_rate": 9.876576213138815e-06,
      "loss": 0.8273,
      "step": 772
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4834005835709236,
      "learning_rate": 9.876118015653804e-06,
      "loss": 0.7743,
      "step": 773
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5289996659244764,
      "learning_rate": 9.875658979907574e-06,
      "loss": 0.8739,
      "step": 774
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.954795381529671,
      "learning_rate": 9.875199105979043e-06,
      "loss": 0.9338,
      "step": 775
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5686130109469965,
      "learning_rate": 9.874738393947268e-06,
      "loss": 0.8913,
      "step": 776
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.748351229866532,
      "learning_rate": 9.87427684389145e-06,
      "loss": 0.844,
      "step": 777
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6652579280282767,
      "learning_rate": 9.873814455890934e-06,
      "loss": 0.8508,
      "step": 778
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4305362456738273,
      "learning_rate": 9.873351230025212e-06,
      "loss": 0.8296,
      "step": 779
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6086471371608642,
      "learning_rate": 9.872887166373916e-06,
      "loss": 0.8478,
      "step": 780
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5757613432304625,
      "learning_rate": 9.872422265016827e-06,
      "loss": 0.8363,
      "step": 781
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6720460184118002,
      "learning_rate": 9.871956526033863e-06,
      "loss": 0.9327,
      "step": 782
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9146293303692931,
      "learning_rate": 9.871489949505091e-06,
      "loss": 0.7235,
      "step": 783
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5345894532680096,
      "learning_rate": 9.871022535510722e-06,
      "loss": 0.8875,
      "step": 784
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4179493337883022,
      "learning_rate": 9.870554284131111e-06,
      "loss": 0.8118,
      "step": 785
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3906348970861464,
      "learning_rate": 9.870085195446752e-06,
      "loss": 0.8602,
      "step": 786
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6705778417662676,
      "learning_rate": 9.86961526953829e-06,
      "loss": 0.835,
      "step": 787
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9652820350693408,
      "learning_rate": 9.86914450648651e-06,
      "loss": 0.6896,
      "step": 788
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9648061110151056,
      "learning_rate": 9.868672906372341e-06,
      "loss": 0.6643,
      "step": 789
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9641862302829514,
      "learning_rate": 9.868200469276858e-06,
      "loss": 0.9023,
      "step": 790
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7070245649493547,
      "learning_rate": 9.867727195281275e-06,
      "loss": 0.9437,
      "step": 791
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5084032762837243,
      "learning_rate": 9.867253084466957e-06,
      "loss": 0.8562,
      "step": 792
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.707299051150018,
      "learning_rate": 9.866778136915408e-06,
      "loss": 0.9179,
      "step": 793
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.102882961801543,
      "learning_rate": 9.866302352708276e-06,
      "loss": 0.8818,
      "step": 794
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.969827812864938,
      "learning_rate": 9.865825731927356e-06,
      "loss": 0.7666,
      "step": 795
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.111716589109969,
      "learning_rate": 9.865348274654582e-06,
      "loss": 0.8553,
      "step": 796
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3405844906139537,
      "learning_rate": 9.864869980972036e-06,
      "loss": 0.8772,
      "step": 797
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3639635149533027,
      "learning_rate": 9.86439085096194e-06,
      "loss": 0.8281,
      "step": 798
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4658298371908418,
      "learning_rate": 9.863910884706667e-06,
      "loss": 0.9095,
      "step": 799
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6613165085202914,
      "learning_rate": 9.863430082288724e-06,
      "loss": 0.8782,
      "step": 800
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.697276223446961,
      "learning_rate": 9.862948443790768e-06,
      "loss": 0.9894,
      "step": 801
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2889521409931344,
      "learning_rate": 9.862465969295599e-06,
      "loss": 0.9055,
      "step": 802
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5553614187203766,
      "learning_rate": 9.861982658886158e-06,
      "loss": 0.9249,
      "step": 803
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.01834224169436,
      "learning_rate": 9.861498512645532e-06,
      "loss": 0.8583,
      "step": 804
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8850165086064957,
      "learning_rate": 9.861013530656956e-06,
      "loss": 0.7058,
      "step": 805
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.00511275763958,
      "learning_rate": 9.860527713003797e-06,
      "loss": 0.8536,
      "step": 806
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5398524684267214,
      "learning_rate": 9.86004105976958e-06,
      "loss": 0.8179,
      "step": 807
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5001678687355027,
      "learning_rate": 9.85955357103796e-06,
      "loss": 0.8317,
      "step": 808
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6226770513354103,
      "learning_rate": 9.859065246892744e-06,
      "loss": 0.8918,
      "step": 809
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2997205917207006,
      "learning_rate": 9.858576087417881e-06,
      "loss": 0.8414,
      "step": 810
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3162408719406542,
      "learning_rate": 9.858086092697464e-06,
      "loss": 0.7935,
      "step": 811
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0407664963764045,
      "learning_rate": 9.857595262815726e-06,
      "loss": 0.7426,
      "step": 812
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.46292623919243,
      "learning_rate": 9.857103597857051e-06,
      "loss": 1.0004,
      "step": 813
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6355070758651977,
      "learning_rate": 9.856611097905956e-06,
      "loss": 0.8805,
      "step": 814
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9393407050942018,
      "learning_rate": 9.856117763047115e-06,
      "loss": 0.7562,
      "step": 815
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.50495605706704,
      "learning_rate": 9.855623593365332e-06,
      "loss": 0.8834,
      "step": 816
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3788004151606135,
      "learning_rate": 9.855128588945559e-06,
      "loss": 0.9155,
      "step": 817
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5247308449594232,
      "learning_rate": 9.8546327498729e-06,
      "loss": 0.8266,
      "step": 818
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4532678075209806,
      "learning_rate": 9.854136076232587e-06,
      "loss": 0.9612,
      "step": 819
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.376664473659083,
      "learning_rate": 9.853638568110013e-06,
      "loss": 0.8991,
      "step": 820
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7217265255462086,
      "learning_rate": 9.853140225590698e-06,
      "loss": 0.8531,
      "step": 821
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.689070809927971,
      "learning_rate": 9.852641048760319e-06,
      "loss": 0.9015,
      "step": 822
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4146176616941846,
      "learning_rate": 9.852141037704683e-06,
      "loss": 0.8995,
      "step": 823
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8667143772822771,
      "learning_rate": 9.851640192509755e-06,
      "loss": 0.7043,
      "step": 824
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6097882577742928,
      "learning_rate": 9.85113851326163e-06,
      "loss": 0.8749,
      "step": 825
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5012679365806136,
      "learning_rate": 9.850636000046558e-06,
      "loss": 0.874,
      "step": 826
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3381486012055865,
      "learning_rate": 9.850132652950923e-06,
      "loss": 0.9273,
      "step": 827
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4103579477495969,
      "learning_rate": 9.84962847206126e-06,
      "loss": 0.8867,
      "step": 828
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.769674190166834,
      "learning_rate": 9.849123457464238e-06,
      "loss": 0.9434,
      "step": 829
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7533912333137849,
      "learning_rate": 9.848617609246678e-06,
      "loss": 0.7706,
      "step": 830
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.449377111095686,
      "learning_rate": 9.84811092749554e-06,
      "loss": 0.8232,
      "step": 831
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0064039502261775,
      "learning_rate": 9.84760341229793e-06,
      "loss": 0.6668,
      "step": 832
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4305161594133005,
      "learning_rate": 9.847095063741097e-06,
      "loss": 0.8282,
      "step": 833
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4685242491321868,
      "learning_rate": 9.84658588191243e-06,
      "loss": 0.944,
      "step": 834
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2310830522722906,
      "learning_rate": 9.846075866899463e-06,
      "loss": 0.8123,
      "step": 835
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4755983316034802,
      "learning_rate": 9.845565018789873e-06,
      "loss": 0.7892,
      "step": 836
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.163619120641701,
      "learning_rate": 9.845053337671482e-06,
      "loss": 0.9024,
      "step": 837
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0410221701482754,
      "learning_rate": 9.844540823632254e-06,
      "loss": 0.9109,
      "step": 838
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5649973665787666,
      "learning_rate": 9.844027476760295e-06,
      "loss": 0.9219,
      "step": 839
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.702123981597596,
      "learning_rate": 9.843513297143856e-06,
      "loss": 0.8586,
      "step": 840
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3907376384788737,
      "learning_rate": 9.842998284871332e-06,
      "loss": 0.9188,
      "step": 841
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.637723010416902,
      "learning_rate": 9.842482440031256e-06,
      "loss": 0.8101,
      "step": 842
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3983264761678764,
      "learning_rate": 9.84196576271231e-06,
      "loss": 0.8689,
      "step": 843
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.445168950112961,
      "learning_rate": 9.841448253003319e-06,
      "loss": 0.9923,
      "step": 844
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4775484795601541,
      "learning_rate": 9.840929910993244e-06,
      "loss": 0.8957,
      "step": 845
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6263534398591193,
      "learning_rate": 9.840410736771196e-06,
      "loss": 0.8098,
      "step": 846
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9976475155905054,
      "learning_rate": 9.839890730426429e-06,
      "loss": 0.9612,
      "step": 847
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4087856184681267,
      "learning_rate": 9.839369892048337e-06,
      "loss": 0.8777,
      "step": 848
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.246891149286199,
      "learning_rate": 9.838848221726455e-06,
      "loss": 0.7264,
      "step": 849
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6479519590506346,
      "learning_rate": 9.838325719550469e-06,
      "loss": 0.8052,
      "step": 850
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.406946139580939,
      "learning_rate": 9.8378023856102e-06,
      "loss": 0.8455,
      "step": 851
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4660699343860122,
      "learning_rate": 9.837278219995618e-06,
      "loss": 0.8941,
      "step": 852
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.57894949374186,
      "learning_rate": 9.836753222796831e-06,
      "loss": 0.7698,
      "step": 853
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8061955599195283,
      "learning_rate": 9.836227394104093e-06,
      "loss": 0.8245,
      "step": 854
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8746285601148243,
      "learning_rate": 9.835700734007798e-06,
      "loss": 0.6819,
      "step": 855
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4004118103421537,
      "learning_rate": 9.835173242598486e-06,
      "loss": 0.8019,
      "step": 856
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6539085489732672,
      "learning_rate": 9.834644919966842e-06,
      "loss": 0.7773,
      "step": 857
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4637120652282516,
      "learning_rate": 9.834115766203687e-06,
      "loss": 0.8577,
      "step": 858
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4851620885781454,
      "learning_rate": 9.833585781399989e-06,
      "loss": 0.871,
      "step": 859
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6646466477146553,
      "learning_rate": 9.83305496564686e-06,
      "loss": 0.839,
      "step": 860
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4620168194075307,
      "learning_rate": 9.832523319035553e-06,
      "loss": 0.9286,
      "step": 861
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9493350940488248,
      "learning_rate": 9.831990841657465e-06,
      "loss": 0.6925,
      "step": 862
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4388237707576106,
      "learning_rate": 9.831457533604133e-06,
      "loss": 0.8507,
      "step": 863
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8395714603565492,
      "learning_rate": 9.83092339496724e-06,
      "loss": 0.9098,
      "step": 864
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.368869891376885,
      "learning_rate": 9.83038842583861e-06,
      "loss": 0.7911,
      "step": 865
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3901943204128402,
      "learning_rate": 9.829852626310213e-06,
      "loss": 0.9064,
      "step": 866
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5021129458365892,
      "learning_rate": 9.829315996474156e-06,
      "loss": 0.9019,
      "step": 867
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.442565737942164,
      "learning_rate": 9.828778536422692e-06,
      "loss": 0.8131,
      "step": 868
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5946413853893826,
      "learning_rate": 9.828240246248217e-06,
      "loss": 0.8519,
      "step": 869
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7104002979652446,
      "learning_rate": 9.827701126043272e-06,
      "loss": 0.966,
      "step": 870
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3729249077215655,
      "learning_rate": 9.827161175900535e-06,
      "loss": 0.8343,
      "step": 871
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.477200910228354,
      "learning_rate": 9.826620395912832e-06,
      "loss": 0.7674,
      "step": 872
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4349012731754378,
      "learning_rate": 9.826078786173126e-06,
      "loss": 0.8885,
      "step": 873
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8251343704827756,
      "learning_rate": 9.82553634677453e-06,
      "loss": 0.9371,
      "step": 874
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5882711215055496,
      "learning_rate": 9.82499307781029e-06,
      "loss": 0.8215,
      "step": 875
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.44551445507311,
      "learning_rate": 9.824448979373807e-06,
      "loss": 0.8273,
      "step": 876
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3885658711748836,
      "learning_rate": 9.823904051558613e-06,
      "loss": 0.8574,
      "step": 877
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8054737367656206,
      "learning_rate": 9.82335829445839e-06,
      "loss": 0.8606,
      "step": 878
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2726595599544606,
      "learning_rate": 9.822811708166958e-06,
      "loss": 0.7375,
      "step": 879
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4450003844214918,
      "learning_rate": 9.822264292778282e-06,
      "loss": 0.7643,
      "step": 880
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6293376470259795,
      "learning_rate": 9.821716048386472e-06,
      "loss": 0.8156,
      "step": 881
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3087863487465878,
      "learning_rate": 9.821166975085774e-06,
      "loss": 0.7353,
      "step": 882
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.455354119701131,
      "learning_rate": 9.820617072970583e-06,
      "loss": 0.8765,
      "step": 883
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.352536827231204,
      "learning_rate": 9.820066342135428e-06,
      "loss": 0.8069,
      "step": 884
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9374592786128569,
      "learning_rate": 9.819514782674992e-06,
      "loss": 0.6064,
      "step": 885
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9855973951015008,
      "learning_rate": 9.818962394684094e-06,
      "loss": 0.6545,
      "step": 886
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.250983872350101,
      "learning_rate": 9.818409178257691e-06,
      "loss": 0.8439,
      "step": 887
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3971383245819453,
      "learning_rate": 9.817855133490891e-06,
      "loss": 0.8801,
      "step": 888
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9311599951451889,
      "learning_rate": 9.817300260478942e-06,
      "loss": 0.6674,
      "step": 889
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3712577959777104,
      "learning_rate": 9.816744559317228e-06,
      "loss": 0.8507,
      "step": 890
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5370598584512734,
      "learning_rate": 9.816188030101287e-06,
      "loss": 0.9503,
      "step": 891
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3392560634335735,
      "learning_rate": 9.815630672926789e-06,
      "loss": 0.7936,
      "step": 892
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.573144486239607,
      "learning_rate": 9.81507248788955e-06,
      "loss": 0.947,
      "step": 893
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.304294940873105,
      "learning_rate": 9.814513475085528e-06,
      "loss": 0.8528,
      "step": 894
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5784567256156228,
      "learning_rate": 9.813953634610827e-06,
      "loss": 0.8515,
      "step": 895
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3450261281937768,
      "learning_rate": 9.813392966561688e-06,
      "loss": 0.889,
      "step": 896
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3833298332361326,
      "learning_rate": 9.812831471034495e-06,
      "loss": 0.872,
      "step": 897
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5390059134278307,
      "learning_rate": 9.812269148125779e-06,
      "loss": 0.9634,
      "step": 898
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.526702522703937,
      "learning_rate": 9.811705997932206e-06,
      "loss": 0.8543,
      "step": 899
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.579094540803675,
      "learning_rate": 9.81114202055059e-06,
      "loss": 0.9514,
      "step": 900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5473813938013712,
      "learning_rate": 9.810577216077886e-06,
      "loss": 0.8449,
      "step": 901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.522433532152599,
      "learning_rate": 9.810011584611189e-06,
      "loss": 0.8477,
      "step": 902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6622714791329791,
      "learning_rate": 9.80944512624774e-06,
      "loss": 0.8822,
      "step": 903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9406377293073815,
      "learning_rate": 9.808877841084915e-06,
      "loss": 0.6707,
      "step": 904
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2738102703123448,
      "learning_rate": 9.808309729220241e-06,
      "loss": 0.9276,
      "step": 905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4629005486114546,
      "learning_rate": 9.807740790751383e-06,
      "loss": 0.9794,
      "step": 906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5982927197504497,
      "learning_rate": 9.807171025776145e-06,
      "loss": 0.8448,
      "step": 907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2507667051957883,
      "learning_rate": 9.806600434392479e-06,
      "loss": 0.7324,
      "step": 908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4885054114172271,
      "learning_rate": 9.806029016698475e-06,
      "loss": 0.872,
      "step": 909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4676896710105487,
      "learning_rate": 9.805456772792367e-06,
      "loss": 0.9078,
      "step": 910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6183124635959873,
      "learning_rate": 9.804883702772532e-06,
      "loss": 0.7917,
      "step": 911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.340255008926344,
      "learning_rate": 9.804309806737482e-06,
      "loss": 0.8277,
      "step": 912
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.546100972406489,
      "learning_rate": 9.803735084785884e-06,
      "loss": 0.848,
      "step": 913
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4378514847061041,
      "learning_rate": 9.803159537016533e-06,
      "loss": 0.924,
      "step": 914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5728380975763252,
      "learning_rate": 9.802583163528374e-06,
      "loss": 0.805,
      "step": 915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5652933040496722,
      "learning_rate": 9.802005964420493e-06,
      "loss": 0.9576,
      "step": 916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7556481179375851,
      "learning_rate": 9.801427939792118e-06,
      "loss": 0.929,
      "step": 917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.564521424788632,
      "learning_rate": 9.800849089742617e-06,
      "loss": 0.8087,
      "step": 918
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5196227722709377,
      "learning_rate": 9.800269414371501e-06,
      "loss": 0.8941,
      "step": 919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9688928444374703,
      "learning_rate": 9.799688913778423e-06,
      "loss": 0.8461,
      "step": 920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4059206263931843,
      "learning_rate": 9.799107588063178e-06,
      "loss": 0.8055,
      "step": 921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3823325317095518,
      "learning_rate": 9.798525437325704e-06,
      "loss": 0.8708,
      "step": 922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5570819382127605,
      "learning_rate": 9.797942461666075e-06,
      "loss": 0.889,
      "step": 923
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.033086114980717,
      "learning_rate": 9.797358661184517e-06,
      "loss": 0.866,
      "step": 924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4883449392415014,
      "learning_rate": 9.796774035981388e-06,
      "loss": 0.8903,
      "step": 925
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.304889512320619,
      "learning_rate": 9.796188586157194e-06,
      "loss": 0.7663,
      "step": 926
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0290720079923137,
      "learning_rate": 9.79560231181258e-06,
      "loss": 0.9592,
      "step": 927
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.237341144932085,
      "learning_rate": 9.795015213048334e-06,
      "loss": 0.7622,
      "step": 928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4585296512579755,
      "learning_rate": 9.794427289965383e-06,
      "loss": 0.8151,
      "step": 929
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8635387966088134,
      "learning_rate": 9.793838542664797e-06,
      "loss": 0.6891,
      "step": 930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3939630578659188,
      "learning_rate": 9.793248971247792e-06,
      "loss": 0.786,
      "step": 931
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7766399757691904,
      "learning_rate": 9.792658575815724e-06,
      "loss": 0.8246,
      "step": 932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.337215539879449,
      "learning_rate": 9.792067356470083e-06,
      "loss": 0.9248,
      "step": 933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3497987547874373,
      "learning_rate": 9.791475313312508e-06,
      "loss": 0.8729,
      "step": 934
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8168511676860826,
      "learning_rate": 9.79088244644478e-06,
      "loss": 0.674,
      "step": 935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4988200558194544,
      "learning_rate": 9.79028875596882e-06,
      "loss": 0.8348,
      "step": 936
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.367326245413367,
      "learning_rate": 9.789694241986687e-06,
      "loss": 0.9044,
      "step": 937
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.676523330029347,
      "learning_rate": 9.789098904600588e-06,
      "loss": 0.8626,
      "step": 938
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3508992254871974,
      "learning_rate": 9.788502743912866e-06,
      "loss": 0.9035,
      "step": 939
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.601227895416428,
      "learning_rate": 9.787905760026011e-06,
      "loss": 0.9065,
      "step": 940
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9293672297504252,
      "learning_rate": 9.78730795304265e-06,
      "loss": 0.8126,
      "step": 941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4320616627736495,
      "learning_rate": 9.786709323065553e-06,
      "loss": 0.8109,
      "step": 942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6861855029364976,
      "learning_rate": 9.786109870197629e-06,
      "loss": 0.9245,
      "step": 943
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4671640778733255,
      "learning_rate": 9.785509594541934e-06,
      "loss": 0.877,
      "step": 944
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.895714710633759,
      "learning_rate": 9.78490849620166e-06,
      "loss": 0.6371,
      "step": 945
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.87239210382955,
      "learning_rate": 9.784306575280146e-06,
      "loss": 0.6258,
      "step": 946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2721367265211971,
      "learning_rate": 9.783703831880867e-06,
      "loss": 0.7592,
      "step": 947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4587631852109597,
      "learning_rate": 9.78310026610744e-06,
      "loss": 0.8971,
      "step": 948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5808183868659478,
      "learning_rate": 9.78249587806363e-06,
      "loss": 0.8739,
      "step": 949
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.64494956645838,
      "learning_rate": 9.78189066785333e-06,
      "loss": 0.9351,
      "step": 950
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5083985669050757,
      "learning_rate": 9.781284635580593e-06,
      "loss": 0.8163,
      "step": 951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7519258973896876,
      "learning_rate": 9.780677781349594e-06,
      "loss": 0.7895,
      "step": 952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.372061690509118,
      "learning_rate": 9.780070105264665e-06,
      "loss": 0.8497,
      "step": 953
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8195425846546356,
      "learning_rate": 9.779461607430268e-06,
      "loss": 0.9444,
      "step": 954
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.433092019515787,
      "learning_rate": 9.778852287951013e-06,
      "loss": 0.8746,
      "step": 955
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3928695879742747,
      "learning_rate": 9.77824214693165e-06,
      "loss": 0.9344,
      "step": 956
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8946202325668566,
      "learning_rate": 9.777631184477065e-06,
      "loss": 0.8717,
      "step": 957
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3929389048288083,
      "learning_rate": 9.777019400692292e-06,
      "loss": 0.7689,
      "step": 958
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4194158788878566,
      "learning_rate": 9.776406795682506e-06,
      "loss": 0.822,
      "step": 959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.23330765017546,
      "learning_rate": 9.775793369553019e-06,
      "loss": 0.7488,
      "step": 960
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3622456858612135,
      "learning_rate": 9.775179122409283e-06,
      "loss": 0.8042,
      "step": 961
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6855339293434124,
      "learning_rate": 9.7745640543569e-06,
      "loss": 0.8531,
      "step": 962
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5316801042345416,
      "learning_rate": 9.773948165501606e-06,
      "loss": 0.8426,
      "step": 963
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4135107074312543,
      "learning_rate": 9.773331455949275e-06,
      "loss": 0.78,
      "step": 964
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8816239004545099,
      "learning_rate": 9.772713925805932e-06,
      "loss": 0.7119,
      "step": 965
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.625447046448455,
      "learning_rate": 9.772095575177733e-06,
      "loss": 0.913,
      "step": 966
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.342309081294382,
      "learning_rate": 9.771476404170984e-06,
      "loss": 0.8924,
      "step": 967
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2234112577589318,
      "learning_rate": 9.770856412892122e-06,
      "loss": 0.8805,
      "step": 968
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4309649434726566,
      "learning_rate": 9.770235601447737e-06,
      "loss": 0.9397,
      "step": 969
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4150990744551932,
      "learning_rate": 9.769613969944549e-06,
      "loss": 0.8189,
      "step": 970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.689500659965123,
      "learning_rate": 9.768991518489427e-06,
      "loss": 0.7956,
      "step": 971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4681249889184187,
      "learning_rate": 9.768368247189375e-06,
      "loss": 0.8929,
      "step": 972
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.523513509759519,
      "learning_rate": 9.767744156151542e-06,
      "loss": 0.9001,
      "step": 973
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4737512009545564,
      "learning_rate": 9.767119245483217e-06,
      "loss": 0.9191,
      "step": 974
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9058932018847845,
      "learning_rate": 9.766493515291827e-06,
      "loss": 0.8669,
      "step": 975
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3529240897345813,
      "learning_rate": 9.765866965684944e-06,
      "loss": 0.8688,
      "step": 976
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5416091715395306,
      "learning_rate": 9.76523959677028e-06,
      "loss": 0.8156,
      "step": 977
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1172350338675843,
      "learning_rate": 9.764611408655687e-06,
      "loss": 0.8113,
      "step": 978
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8146816984691727,
      "learning_rate": 9.763982401449154e-06,
      "loss": 0.8728,
      "step": 979
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.049740599017007,
      "learning_rate": 9.763352575258817e-06,
      "loss": 0.942,
      "step": 980
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.846418020076748,
      "learning_rate": 9.762721930192953e-06,
      "loss": 0.6674,
      "step": 981
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1843462888761156,
      "learning_rate": 9.762090466359975e-06,
      "loss": 0.8324,
      "step": 982
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3726926194757016,
      "learning_rate": 9.761458183868437e-06,
      "loss": 0.7623,
      "step": 983
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5788692616116609,
      "learning_rate": 9.760825082827039e-06,
      "loss": 0.7579,
      "step": 984
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.575018108958696,
      "learning_rate": 9.760191163344617e-06,
      "loss": 0.7355,
      "step": 985
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4606613563383462,
      "learning_rate": 9.75955642553015e-06,
      "loss": 0.7687,
      "step": 986
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0159849771641305,
      "learning_rate": 9.758920869492755e-06,
      "loss": 0.838,
      "step": 987
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4146353562244873,
      "learning_rate": 9.758284495341692e-06,
      "loss": 0.9089,
      "step": 988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3893998380641266,
      "learning_rate": 9.757647303186361e-06,
      "loss": 0.8489,
      "step": 989
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.575045404730761,
      "learning_rate": 9.757009293136303e-06,
      "loss": 0.899,
      "step": 990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8686436320695952,
      "learning_rate": 9.756370465301202e-06,
      "loss": 0.8327,
      "step": 991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3253232301129068,
      "learning_rate": 9.755730819790876e-06,
      "loss": 0.8797,
      "step": 992
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3251119889448857,
      "learning_rate": 9.755090356715288e-06,
      "loss": 0.8514,
      "step": 993
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7243529540254818,
      "learning_rate": 9.754449076184544e-06,
      "loss": 0.8268,
      "step": 994
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7919399414670933,
      "learning_rate": 9.753806978308884e-06,
      "loss": 0.8572,
      "step": 995
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.30821337060461,
      "learning_rate": 9.753164063198693e-06,
      "loss": 0.8692,
      "step": 996
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8985045906836215,
      "learning_rate": 9.752520330964497e-06,
      "loss": 0.8419,
      "step": 997
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7132642515658194,
      "learning_rate": 9.75187578171696e-06,
      "loss": 0.9284,
      "step": 998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2499241824370424,
      "learning_rate": 9.751230415566888e-06,
      "loss": 0.8053,
      "step": 999
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9027668164867895,
      "learning_rate": 9.750584232625226e-06,
      "loss": 0.7133,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0997310192190357,
      "learning_rate": 9.749937233003062e-06,
      "loss": 0.748,
      "step": 1001
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4354324136541798,
      "learning_rate": 9.749289416811622e-06,
      "loss": 0.847,
      "step": 1002
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5105513673386837,
      "learning_rate": 9.74864078416227e-06,
      "loss": 0.902,
      "step": 1003
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5925325670475263,
      "learning_rate": 9.747991335166518e-06,
      "loss": 0.7996,
      "step": 1004
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4957466125997607,
      "learning_rate": 9.747341069936013e-06,
      "loss": 0.8568,
      "step": 1005
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.447056731067825,
      "learning_rate": 9.74668998858254e-06,
      "loss": 0.8628,
      "step": 1006
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.790634895810149,
      "learning_rate": 9.746038091218032e-06,
      "loss": 0.6998,
      "step": 1007
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4117920556253722,
      "learning_rate": 9.745385377954555e-06,
      "loss": 0.7972,
      "step": 1008
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8854257584054064,
      "learning_rate": 9.744731848904316e-06,
      "loss": 0.6823,
      "step": 1009
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4772446070753822,
      "learning_rate": 9.74407750417967e-06,
      "loss": 0.7754,
      "step": 1010
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7587248878964434,
      "learning_rate": 9.743422343893099e-06,
      "loss": 0.8719,
      "step": 1011
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3307972115634332,
      "learning_rate": 9.742766368157239e-06,
      "loss": 0.7973,
      "step": 1012
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4972106475129239,
      "learning_rate": 9.742109577084857e-06,
      "loss": 0.8058,
      "step": 1013
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3275079162786814,
      "learning_rate": 9.741451970788863e-06,
      "loss": 0.7563,
      "step": 1014
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.279486643397566,
      "learning_rate": 9.740793549382308e-06,
      "loss": 0.8849,
      "step": 1015
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9359103635250037,
      "learning_rate": 9.740134312978382e-06,
      "loss": 0.8212,
      "step": 1016
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5402497069086851,
      "learning_rate": 9.739474261690415e-06,
      "loss": 0.7763,
      "step": 1017
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0908515186421917,
      "learning_rate": 9.73881339563188e-06,
      "loss": 0.6646,
      "step": 1018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3823014633984938,
      "learning_rate": 9.738151714916382e-06,
      "loss": 0.7548,
      "step": 1019
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9636569658250401,
      "learning_rate": 9.737489219657676e-06,
      "loss": 0.6908,
      "step": 1020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6002739127161607,
      "learning_rate": 9.73682590996965e-06,
      "loss": 0.7666,
      "step": 1021
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3752045444639032,
      "learning_rate": 9.736161785966339e-06,
      "loss": 0.8318,
      "step": 1022
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4023255243925503,
      "learning_rate": 9.735496847761908e-06,
      "loss": 0.8216,
      "step": 1023
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.522452463158001,
      "learning_rate": 9.734831095470671e-06,
      "loss": 0.8844,
      "step": 1024
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6541697199068728,
      "learning_rate": 9.73416452920708e-06,
      "loss": 0.9182,
      "step": 1025
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6347679295369455,
      "learning_rate": 9.73349714908572e-06,
      "loss": 0.7878,
      "step": 1026
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7102873468232225,
      "learning_rate": 9.732828955221328e-06,
      "loss": 0.9314,
      "step": 1027
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.125523850412865,
      "learning_rate": 9.732159947728767e-06,
      "loss": 0.7367,
      "step": 1028
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.77880843384016,
      "learning_rate": 9.731490126723053e-06,
      "loss": 0.881,
      "step": 1029
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4674821017065425,
      "learning_rate": 9.730819492319336e-06,
      "loss": 0.8842,
      "step": 1030
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.326783727854924,
      "learning_rate": 9.730148044632902e-06,
      "loss": 0.9172,
      "step": 1031
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.369857628996545,
      "learning_rate": 9.729475783779182e-06,
      "loss": 0.8533,
      "step": 1032
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5207290381282965,
      "learning_rate": 9.728802709873747e-06,
      "loss": 0.9353,
      "step": 1033
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4019151358936326,
      "learning_rate": 9.728128823032305e-06,
      "loss": 0.8139,
      "step": 1034
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6407363744593775,
      "learning_rate": 9.727454123370705e-06,
      "loss": 0.8269,
      "step": 1035
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.020518237568515,
      "learning_rate": 9.726778611004936e-06,
      "loss": 0.6399,
      "step": 1036
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5304779170282379,
      "learning_rate": 9.726102286051126e-06,
      "loss": 0.9214,
      "step": 1037
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3965933180833214,
      "learning_rate": 9.725425148625544e-06,
      "loss": 0.8486,
      "step": 1038
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8753394100882117,
      "learning_rate": 9.724747198844597e-06,
      "loss": 0.6847,
      "step": 1039
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3978474310171685,
      "learning_rate": 9.724068436824831e-06,
      "loss": 0.8249,
      "step": 1040
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9331404920894577,
      "learning_rate": 9.723388862682936e-06,
      "loss": 0.6715,
      "step": 1041
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.541029208224505,
      "learning_rate": 9.722708476535736e-06,
      "loss": 0.8898,
      "step": 1042
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4701774102892993,
      "learning_rate": 9.722027278500201e-06,
      "loss": 1.0151,
      "step": 1043
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2541390018030916,
      "learning_rate": 9.721345268693433e-06,
      "loss": 0.877,
      "step": 1044
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6932895604208353,
      "learning_rate": 9.720662447232679e-06,
      "loss": 0.7852,
      "step": 1045
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8989133251965493,
      "learning_rate": 9.719978814235323e-06,
      "loss": 0.9273,
      "step": 1046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.482853417468595,
      "learning_rate": 9.71929436981889e-06,
      "loss": 0.8261,
      "step": 1047
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.400660625826224,
      "learning_rate": 9.718609114101045e-06,
      "loss": 0.7683,
      "step": 1048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4267350312979152,
      "learning_rate": 9.717923047199591e-06,
      "loss": 0.9091,
      "step": 1049
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2737956341141468,
      "learning_rate": 9.71723616923247e-06,
      "loss": 0.7878,
      "step": 1050
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4322843509049596,
      "learning_rate": 9.716548480317763e-06,
      "loss": 0.8266,
      "step": 1051
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8740324984367535,
      "learning_rate": 9.715859980573694e-06,
      "loss": 0.6777,
      "step": 1052
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4399102972148616,
      "learning_rate": 9.715170670118625e-06,
      "loss": 0.8684,
      "step": 1053
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.520654425197761,
      "learning_rate": 9.714480549071053e-06,
      "loss": 0.8769,
      "step": 1054
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3204601327769891,
      "learning_rate": 9.713789617549621e-06,
      "loss": 0.7892,
      "step": 1055
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4890007815707251,
      "learning_rate": 9.713097875673105e-06,
      "loss": 0.9051,
      "step": 1056
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3708332532223695,
      "learning_rate": 9.712405323560427e-06,
      "loss": 0.8286,
      "step": 1057
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7777541667794925,
      "learning_rate": 9.711711961330644e-06,
      "loss": 0.8431,
      "step": 1058
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.831180722385582,
      "learning_rate": 9.711017789102948e-06,
      "loss": 0.8706,
      "step": 1059
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9552062856682144,
      "learning_rate": 9.710322806996682e-06,
      "loss": 0.8127,
      "step": 1060
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3456666691478496,
      "learning_rate": 9.70962701513132e-06,
      "loss": 0.9132,
      "step": 1061
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8368915499705991,
      "learning_rate": 9.708930413626473e-06,
      "loss": 0.6425,
      "step": 1062
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7318745135936857,
      "learning_rate": 9.708233002601897e-06,
      "loss": 0.7071,
      "step": 1063
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5305877911512966,
      "learning_rate": 9.707534782177487e-06,
      "loss": 0.897,
      "step": 1064
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6173097684660085,
      "learning_rate": 9.706835752473273e-06,
      "loss": 0.8858,
      "step": 1065
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4102138750102562,
      "learning_rate": 9.706135913609426e-06,
      "loss": 0.8417,
      "step": 1066
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5440813306561145,
      "learning_rate": 9.70543526570626e-06,
      "loss": 0.9486,
      "step": 1067
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.510060968658177,
      "learning_rate": 9.704733808884219e-06,
      "loss": 0.8008,
      "step": 1068
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4872524739983397,
      "learning_rate": 9.704031543263893e-06,
      "loss": 0.928,
      "step": 1069
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5490832901690643,
      "learning_rate": 9.703328468966016e-06,
      "loss": 0.8629,
      "step": 1070
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.595354365037843,
      "learning_rate": 9.702624586111446e-06,
      "loss": 0.8367,
      "step": 1071
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4803398260329856,
      "learning_rate": 9.701919894821192e-06,
      "loss": 0.8666,
      "step": 1072
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8232214663065411,
      "learning_rate": 9.7012143952164e-06,
      "loss": 0.859,
      "step": 1073
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.577323097301772,
      "learning_rate": 9.70050808741835e-06,
      "loss": 0.8355,
      "step": 1074
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2441386368120413,
      "learning_rate": 9.69980097154847e-06,
      "loss": 0.8428,
      "step": 1075
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9527971642821473,
      "learning_rate": 9.699093047728317e-06,
      "loss": 0.6842,
      "step": 1076
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3606568472067784,
      "learning_rate": 9.698384316079592e-06,
      "loss": 0.842,
      "step": 1077
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4203365113104036,
      "learning_rate": 9.697674776724135e-06,
      "loss": 0.9067,
      "step": 1078
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2935627689086364,
      "learning_rate": 9.696964429783923e-06,
      "loss": 0.8385,
      "step": 1079
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4257994771059288,
      "learning_rate": 9.696253275381074e-06,
      "loss": 0.7988,
      "step": 1080
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7102934700168715,
      "learning_rate": 9.695541313637845e-06,
      "loss": 0.9195,
      "step": 1081
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3398983657266725,
      "learning_rate": 9.694828544676626e-06,
      "loss": 0.8023,
      "step": 1082
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.881286876006318,
      "learning_rate": 9.694114968619955e-06,
      "loss": 0.6184,
      "step": 1083
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.452434497838863,
      "learning_rate": 9.693400585590502e-06,
      "loss": 0.8405,
      "step": 1084
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.375828667148205,
      "learning_rate": 9.692685395711077e-06,
      "loss": 0.7232,
      "step": 1085
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7011641851267512,
      "learning_rate": 9.69196939910463e-06,
      "loss": 0.8748,
      "step": 1086
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3298627022723368,
      "learning_rate": 9.69125259589425e-06,
      "loss": 0.8336,
      "step": 1087
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.077742737883768,
      "learning_rate": 9.690534986203164e-06,
      "loss": 0.8751,
      "step": 1088
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3217363156888182,
      "learning_rate": 9.689816570154735e-06,
      "loss": 0.9577,
      "step": 1089
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3398505227293962,
      "learning_rate": 9.68909734787247e-06,
      "loss": 0.7843,
      "step": 1090
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3090045430020312,
      "learning_rate": 9.688377319480008e-06,
      "loss": 0.8613,
      "step": 1091
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9949994983224337,
      "learning_rate": 9.687656485101134e-06,
      "loss": 0.7926,
      "step": 1092
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2687555343776218,
      "learning_rate": 9.686934844859766e-06,
      "loss": 0.712,
      "step": 1093
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3415572998819973,
      "learning_rate": 9.686212398879963e-06,
      "loss": 0.8444,
      "step": 1094
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4613223103807664,
      "learning_rate": 9.68548914728592e-06,
      "loss": 0.8505,
      "step": 1095
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.542713314338235,
      "learning_rate": 9.684765090201975e-06,
      "loss": 0.8057,
      "step": 1096
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4503922764608232,
      "learning_rate": 9.684040227752601e-06,
      "loss": 0.8178,
      "step": 1097
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9835049952608671,
      "learning_rate": 9.683314560062409e-06,
      "loss": 0.6162,
      "step": 1098
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5587324764898038,
      "learning_rate": 9.68258808725615e-06,
      "loss": 0.923,
      "step": 1099
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4254247178684063,
      "learning_rate": 9.681860809458713e-06,
      "loss": 0.8047,
      "step": 1100
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.380260333129015,
      "learning_rate": 9.681132726795128e-06,
      "loss": 0.8572,
      "step": 1101
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4688003302260049,
      "learning_rate": 9.680403839390558e-06,
      "loss": 0.8862,
      "step": 1102
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5989099538703444,
      "learning_rate": 9.679674147370308e-06,
      "loss": 0.8213,
      "step": 1103
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5606527963121897,
      "learning_rate": 9.67894365085982e-06,
      "loss": 0.8499,
      "step": 1104
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8196205702065393,
      "learning_rate": 9.678212349984677e-06,
      "loss": 0.8591,
      "step": 1105
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5202225382914378,
      "learning_rate": 9.677480244870597e-06,
      "loss": 0.8918,
      "step": 1106
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.383933528486037,
      "learning_rate": 9.676747335643435e-06,
      "loss": 0.8547,
      "step": 1107
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6564455319980587,
      "learning_rate": 9.676013622429187e-06,
      "loss": 0.8945,
      "step": 1108
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2720135136307014,
      "learning_rate": 9.675279105353991e-06,
      "loss": 0.8068,
      "step": 1109
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3748946737342862,
      "learning_rate": 9.674543784544118e-06,
      "loss": 0.9062,
      "step": 1110
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.371584507002583,
      "learning_rate": 9.673807660125974e-06,
      "loss": 0.8882,
      "step": 1111
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5510272702580885,
      "learning_rate": 9.673070732226109e-06,
      "loss": 0.85,
      "step": 1112
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8299819101316426,
      "learning_rate": 9.67233300097121e-06,
      "loss": 0.6651,
      "step": 1113
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9376882584393623,
      "learning_rate": 9.671594466488104e-06,
      "loss": 0.6783,
      "step": 1114
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4726595096687631,
      "learning_rate": 9.670855128903752e-06,
      "loss": 0.8541,
      "step": 1115
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6994215701650677,
      "learning_rate": 9.670114988345252e-06,
      "loss": 0.8282,
      "step": 1116
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.373112156314513,
      "learning_rate": 9.669374044939846e-06,
      "loss": 0.9078,
      "step": 1117
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3039161475856154,
      "learning_rate": 9.668632298814908e-06,
      "loss": 0.8035,
      "step": 1118
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8110073023396642,
      "learning_rate": 9.667889750097956e-06,
      "loss": 0.8793,
      "step": 1119
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8468689646029608,
      "learning_rate": 9.66714639891664e-06,
      "loss": 0.8449,
      "step": 1120
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4270610202960454,
      "learning_rate": 9.666402245398753e-06,
      "loss": 0.9076,
      "step": 1121
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7028513478159408,
      "learning_rate": 9.665657289672222e-06,
      "loss": 0.9072,
      "step": 1122
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4751404077340504,
      "learning_rate": 9.664911531865115e-06,
      "loss": 0.8382,
      "step": 1123
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3979573082388097,
      "learning_rate": 9.664164972105634e-06,
      "loss": 0.8698,
      "step": 1124
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8646732586932233,
      "learning_rate": 9.663417610522124e-06,
      "loss": 0.9116,
      "step": 1125
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3438939986244987,
      "learning_rate": 9.66266944724306e-06,
      "loss": 0.9814,
      "step": 1126
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.415870869176456,
      "learning_rate": 9.661920482397069e-06,
      "loss": 0.8722,
      "step": 1127
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5231308800087855,
      "learning_rate": 9.661170716112897e-06,
      "loss": 0.871,
      "step": 1128
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6580610795036221,
      "learning_rate": 9.660420148519444e-06,
      "loss": 0.8991,
      "step": 1129
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9244979711687746,
      "learning_rate": 9.65966877974574e-06,
      "loss": 0.7905,
      "step": 1130
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4182508029300245,
      "learning_rate": 9.658916609920951e-06,
      "loss": 0.7782,
      "step": 1131
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5130050663199788,
      "learning_rate": 9.65816363917439e-06,
      "loss": 0.7884,
      "step": 1132
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9416163993086398,
      "learning_rate": 9.657409867635494e-06,
      "loss": 0.5991,
      "step": 1133
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4280239186481178,
      "learning_rate": 9.656655295433849e-06,
      "loss": 0.8476,
      "step": 1134
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4361486074413254,
      "learning_rate": 9.655899922699174e-06,
      "loss": 0.8426,
      "step": 1135
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5289030523433746,
      "learning_rate": 9.655143749561326e-06,
      "loss": 0.8639,
      "step": 1136
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.690166227276656,
      "learning_rate": 9.654386776150301e-06,
      "loss": 0.8444,
      "step": 1137
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6045647902017957,
      "learning_rate": 9.653629002596232e-06,
      "loss": 0.7702,
      "step": 1138
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8627614716332304,
      "learning_rate": 9.652870429029386e-06,
      "loss": 0.895,
      "step": 1139
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4652273040759825,
      "learning_rate": 9.652111055580175e-06,
      "loss": 0.8563,
      "step": 1140
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.569539336575209,
      "learning_rate": 9.65135088237914e-06,
      "loss": 0.8826,
      "step": 1141
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3732979933514144,
      "learning_rate": 9.650589909556964e-06,
      "loss": 0.922,
      "step": 1142
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5821818568445363,
      "learning_rate": 9.649828137244471e-06,
      "loss": 0.77,
      "step": 1143
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5366093892771826,
      "learning_rate": 9.649065565572615e-06,
      "loss": 0.9337,
      "step": 1144
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3919068905093355,
      "learning_rate": 9.64830219467249e-06,
      "loss": 0.8249,
      "step": 1145
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4719796017528046,
      "learning_rate": 9.647538024675331e-06,
      "loss": 0.7882,
      "step": 1146
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4349265127294617,
      "learning_rate": 9.646773055712508e-06,
      "loss": 0.9287,
      "step": 1147
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5038248787641195,
      "learning_rate": 9.646007287915524e-06,
      "loss": 0.8585,
      "step": 1148
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.431317273940488,
      "learning_rate": 9.64524072141603e-06,
      "loss": 0.7752,
      "step": 1149
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.021062085514492,
      "learning_rate": 9.6444733563458e-06,
      "loss": 0.823,
      "step": 1150
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6695567214557219,
      "learning_rate": 9.643705192836758e-06,
      "loss": 0.8289,
      "step": 1151
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.890158964233443,
      "learning_rate": 9.64293623102096e-06,
      "loss": 0.6912,
      "step": 1152
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6249361931593145,
      "learning_rate": 9.642166471030596e-06,
      "loss": 0.9161,
      "step": 1153
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9525965272082633,
      "learning_rate": 9.641395912998e-06,
      "loss": 0.6228,
      "step": 1154
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8678445890318419,
      "learning_rate": 9.64062455705564e-06,
      "loss": 0.7251,
      "step": 1155
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.29005515568079,
      "learning_rate": 9.639852403336118e-06,
      "loss": 0.8143,
      "step": 1156
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.80627429454135,
      "learning_rate": 9.63907945197218e-06,
      "loss": 0.7008,
      "step": 1157
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7445171676788644,
      "learning_rate": 9.638305703096702e-06,
      "loss": 0.9062,
      "step": 1158
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.390110580309052,
      "learning_rate": 9.637531156842702e-06,
      "loss": 0.7956,
      "step": 1159
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6810656205550831,
      "learning_rate": 9.636755813343334e-06,
      "loss": 0.8297,
      "step": 1160
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0207690361262634,
      "learning_rate": 9.635979672731888e-06,
      "loss": 0.7955,
      "step": 1161
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.349215594059627,
      "learning_rate": 9.635202735141792e-06,
      "loss": 0.8308,
      "step": 1162
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9061850843262143,
      "learning_rate": 9.634425000706607e-06,
      "loss": 0.8198,
      "step": 1163
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4495626336297494,
      "learning_rate": 9.633646469560039e-06,
      "loss": 0.8084,
      "step": 1164
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9507867165692312,
      "learning_rate": 9.632867141835926e-06,
      "loss": 0.8741,
      "step": 1165
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3774988259968433,
      "learning_rate": 9.632087017668242e-06,
      "loss": 0.8103,
      "step": 1166
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3889743931471514,
      "learning_rate": 9.6313060971911e-06,
      "loss": 0.8669,
      "step": 1167
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4742912883972097,
      "learning_rate": 9.630524380538748e-06,
      "loss": 0.9355,
      "step": 1168
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2859663814980355,
      "learning_rate": 9.629741867845574e-06,
      "loss": 0.8381,
      "step": 1169
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6574116312044913,
      "learning_rate": 9.628958559246101e-06,
      "loss": 0.8782,
      "step": 1170
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9797399949637846,
      "learning_rate": 9.628174454874988e-06,
      "loss": 0.8466,
      "step": 1171
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6703584016408708,
      "learning_rate": 9.627389554867032e-06,
      "loss": 0.9501,
      "step": 1172
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4186220000345902,
      "learning_rate": 9.626603859357165e-06,
      "loss": 0.8206,
      "step": 1173
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.462194804123181,
      "learning_rate": 9.625817368480459e-06,
      "loss": 0.9233,
      "step": 1174
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8335638037072298,
      "learning_rate": 9.625030082372122e-06,
      "loss": 0.6919,
      "step": 1175
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6615636940491365,
      "learning_rate": 9.624242001167493e-06,
      "loss": 0.7999,
      "step": 1176
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.614134197662985,
      "learning_rate": 9.623453125002056e-06,
      "loss": 0.8538,
      "step": 1177
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3988263621473023,
      "learning_rate": 9.622663454011429e-06,
      "loss": 0.772,
      "step": 1178
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4866726109540696,
      "learning_rate": 9.621872988331362e-06,
      "loss": 0.8798,
      "step": 1179
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0470375859326277,
      "learning_rate": 9.621081728097747e-06,
      "loss": 0.8397,
      "step": 1180
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6012619860621475,
      "learning_rate": 9.620289673446611e-06,
      "loss": 0.8701,
      "step": 1181
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.434934545406121,
      "learning_rate": 9.619496824514118e-06,
      "loss": 0.7913,
      "step": 1182
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7644986051376317,
      "learning_rate": 9.618703181436566e-06,
      "loss": 0.6748,
      "step": 1183
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4368354958381353,
      "learning_rate": 9.617908744350392e-06,
      "loss": 0.8342,
      "step": 1184
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5824929939366277,
      "learning_rate": 9.61711351339217e-06,
      "loss": 0.794,
      "step": 1185
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4870830001507789,
      "learning_rate": 9.61631748869861e-06,
      "loss": 0.7851,
      "step": 1186
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6310115835303758,
      "learning_rate": 9.615520670406555e-06,
      "loss": 0.777,
      "step": 1187
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.471816524392377,
      "learning_rate": 9.61472305865299e-06,
      "loss": 0.8472,
      "step": 1188
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3532732477014267,
      "learning_rate": 9.613924653575034e-06,
      "loss": 0.9157,
      "step": 1189
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.413182715381386,
      "learning_rate": 9.61312545530994e-06,
      "loss": 0.8833,
      "step": 1190
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.797585437461564,
      "learning_rate": 9.612325463995099e-06,
      "loss": 0.8705,
      "step": 1191
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.441573474171913,
      "learning_rate": 9.61152467976804e-06,
      "loss": 0.802,
      "step": 1192
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7925845267447098,
      "learning_rate": 9.610723102766429e-06,
      "loss": 0.6823,
      "step": 1193
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5661611020427673,
      "learning_rate": 9.609920733128064e-06,
      "loss": 0.7807,
      "step": 1194
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5128612129371597,
      "learning_rate": 9.609117570990882e-06,
      "loss": 0.7468,
      "step": 1195
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5888027563357894,
      "learning_rate": 9.608313616492954e-06,
      "loss": 0.8263,
      "step": 1196
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5336536224831219,
      "learning_rate": 9.607508869772495e-06,
      "loss": 0.8569,
      "step": 1197
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9715606166613975,
      "learning_rate": 9.606703330967843e-06,
      "loss": 0.787,
      "step": 1198
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.365076839093908,
      "learning_rate": 9.605897000217485e-06,
      "loss": 0.8772,
      "step": 1199
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.392139794183767,
      "learning_rate": 9.605089877660036e-06,
      "loss": 0.7482,
      "step": 1200
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2643307575941378,
      "learning_rate": 9.60428196343425e-06,
      "loss": 0.8661,
      "step": 1201
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5722644333351956,
      "learning_rate": 9.603473257679018e-06,
      "loss": 0.8171,
      "step": 1202
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4044180903601953,
      "learning_rate": 9.602663760533364e-06,
      "loss": 0.8559,
      "step": 1203
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5751313069225834,
      "learning_rate": 9.601853472136451e-06,
      "loss": 0.8321,
      "step": 1204
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2320892011306752,
      "learning_rate": 9.601042392627577e-06,
      "loss": 0.7491,
      "step": 1205
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.299026836212122,
      "learning_rate": 9.600230522146176e-06,
      "loss": 0.7955,
      "step": 1206
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7651489137869782,
      "learning_rate": 9.599417860831818e-06,
      "loss": 0.6453,
      "step": 1207
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.284023169273043,
      "learning_rate": 9.598604408824209e-06,
      "loss": 0.8052,
      "step": 1208
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4091856555400424,
      "learning_rate": 9.59779016626319e-06,
      "loss": 0.8718,
      "step": 1209
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.657043349153749,
      "learning_rate": 9.596975133288738e-06,
      "loss": 0.8192,
      "step": 1210
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.463631781594409,
      "learning_rate": 9.596159310040968e-06,
      "loss": 0.9115,
      "step": 1211
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4891093361270809,
      "learning_rate": 9.595342696660133e-06,
      "loss": 0.9045,
      "step": 1212
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.516537173455528,
      "learning_rate": 9.594525293286611e-06,
      "loss": 0.8601,
      "step": 1213
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8513553921727374,
      "learning_rate": 9.593707100060927e-06,
      "loss": 0.7097,
      "step": 1214
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3491308991979727,
      "learning_rate": 9.59288811712374e-06,
      "loss": 0.8306,
      "step": 1215
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3482038961041114,
      "learning_rate": 9.592068344615837e-06,
      "loss": 0.7884,
      "step": 1216
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4877257547283067,
      "learning_rate": 9.591247782678153e-06,
      "loss": 0.882,
      "step": 1217
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8283911676082057,
      "learning_rate": 9.590426431451748e-06,
      "loss": 0.6447,
      "step": 1218
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5201255422190887,
      "learning_rate": 9.58960429107782e-06,
      "loss": 0.792,
      "step": 1219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5073762378497364,
      "learning_rate": 9.58878136169771e-06,
      "loss": 0.7732,
      "step": 1220
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.456311510180172,
      "learning_rate": 9.587957643452886e-06,
      "loss": 0.8311,
      "step": 1221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4984222292718228,
      "learning_rate": 9.587133136484953e-06,
      "loss": 0.8759,
      "step": 1222
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.903672190930149,
      "learning_rate": 9.586307840935658e-06,
      "loss": 0.6799,
      "step": 1223
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4405978773450434,
      "learning_rate": 9.585481756946875e-06,
      "loss": 0.8314,
      "step": 1224
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4540340860308048,
      "learning_rate": 9.584654884660618e-06,
      "loss": 0.8758,
      "step": 1225
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9600188056009511,
      "learning_rate": 9.583827224219035e-06,
      "loss": 0.6281,
      "step": 1226
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9869279357526484,
      "learning_rate": 9.582998775764414e-06,
      "loss": 0.7868,
      "step": 1227
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4763967929039605,
      "learning_rate": 9.582169539439173e-06,
      "loss": 0.8763,
      "step": 1228
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.553248344992629,
      "learning_rate": 9.581339515385866e-06,
      "loss": 0.8512,
      "step": 1229
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9419817094258748,
      "learning_rate": 9.580508703747185e-06,
      "loss": 0.7604,
      "step": 1230
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4796553437276372,
      "learning_rate": 9.579677104665957e-06,
      "loss": 0.8334,
      "step": 1231
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5005499233982444,
      "learning_rate": 9.578844718285141e-06,
      "loss": 0.8378,
      "step": 1232
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7716387140971421,
      "learning_rate": 9.578011544747836e-06,
      "loss": 0.8668,
      "step": 1233
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4585028151998192,
      "learning_rate": 9.577177584197274e-06,
      "loss": 0.7736,
      "step": 1234
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4219700188092426,
      "learning_rate": 9.576342836776822e-06,
      "loss": 0.7865,
      "step": 1235
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4754405213604997,
      "learning_rate": 9.575507302629982e-06,
      "loss": 0.7917,
      "step": 1236
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5342942356638192,
      "learning_rate": 9.574670981900394e-06,
      "loss": 0.7426,
      "step": 1237
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9451586853496309,
      "learning_rate": 9.57383387473183e-06,
      "loss": 0.7447,
      "step": 1238
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8959828653698031,
      "learning_rate": 9.572995981268199e-06,
      "loss": 0.6486,
      "step": 1239
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.118541998189462,
      "learning_rate": 9.572157301653542e-06,
      "loss": 0.7884,
      "step": 1240
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.724632545376072,
      "learning_rate": 9.571317836032042e-06,
      "loss": 0.9332,
      "step": 1241
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4800704719892424,
      "learning_rate": 9.570477584548008e-06,
      "loss": 0.8798,
      "step": 1242
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5977659196013767,
      "learning_rate": 9.569636547345895e-06,
      "loss": 0.9154,
      "step": 1243
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9003304497856656,
      "learning_rate": 9.568794724570282e-06,
      "loss": 0.7559,
      "step": 1244
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3645625352505444,
      "learning_rate": 9.567952116365889e-06,
      "loss": 0.8686,
      "step": 1245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3974497057217423,
      "learning_rate": 9.567108722877572e-06,
      "loss": 0.8408,
      "step": 1246
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6453043313171705,
      "learning_rate": 9.566264544250319e-06,
      "loss": 0.7946,
      "step": 1247
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5389809735652809,
      "learning_rate": 9.565419580629254e-06,
      "loss": 0.9276,
      "step": 1248
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4391693460476158,
      "learning_rate": 9.564573832159638e-06,
      "loss": 0.8119,
      "step": 1249
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2695342571076675,
      "learning_rate": 9.56372729898686e-06,
      "loss": 0.9328,
      "step": 1250
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5795777328581693,
      "learning_rate": 9.562879981256455e-06,
      "loss": 0.7616,
      "step": 1251
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3326270605330675,
      "learning_rate": 9.562031879114082e-06,
      "loss": 0.8596,
      "step": 1252
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.345697589351044,
      "learning_rate": 9.561182992705541e-06,
      "loss": 0.831,
      "step": 1253
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6106272335571274,
      "learning_rate": 9.560333322176767e-06,
      "loss": 0.8497,
      "step": 1254
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4933041327507677,
      "learning_rate": 9.559482867673825e-06,
      "loss": 0.8495,
      "step": 1255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.91561217720066,
      "learning_rate": 9.558631629342922e-06,
      "loss": 0.6686,
      "step": 1256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4213278381254053,
      "learning_rate": 9.557779607330393e-06,
      "loss": 0.8521,
      "step": 1257
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4738697728816916,
      "learning_rate": 9.556926801782714e-06,
      "loss": 0.9726,
      "step": 1258
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8911403126583337,
      "learning_rate": 9.556073212846485e-06,
      "loss": 0.6811,
      "step": 1259
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8763985102962679,
      "learning_rate": 9.555218840668454e-06,
      "loss": 0.6887,
      "step": 1260
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4440264938140612,
      "learning_rate": 9.554363685395496e-06,
      "loss": 0.8045,
      "step": 1261
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.758562236796689,
      "learning_rate": 9.553507747174622e-06,
      "loss": 0.8298,
      "step": 1262
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.419044576466385,
      "learning_rate": 9.552651026152978e-06,
      "loss": 0.8528,
      "step": 1263
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4121734724299548,
      "learning_rate": 9.551793522477842e-06,
      "loss": 0.8607,
      "step": 1264
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3889818462563748,
      "learning_rate": 9.55093523629663e-06,
      "loss": 0.7098,
      "step": 1265
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8280407862286018,
      "learning_rate": 9.550076167756892e-06,
      "loss": 0.6469,
      "step": 1266
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.575070247732128,
      "learning_rate": 9.549216317006313e-06,
      "loss": 0.8318,
      "step": 1267
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5190208141121706,
      "learning_rate": 9.548355684192712e-06,
      "loss": 0.8782,
      "step": 1268
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7156900208597108,
      "learning_rate": 9.547494269464037e-06,
      "loss": 0.7904,
      "step": 1269
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5116409180271138,
      "learning_rate": 9.546632072968379e-06,
      "loss": 0.882,
      "step": 1270
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6734272548232894,
      "learning_rate": 9.545769094853958e-06,
      "loss": 0.8948,
      "step": 1271
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.328965318701487,
      "learning_rate": 9.54490533526913e-06,
      "loss": 0.8311,
      "step": 1272
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3835414945851021,
      "learning_rate": 9.544040794362389e-06,
      "loss": 0.9435,
      "step": 1273
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8503447653519499,
      "learning_rate": 9.543175472282353e-06,
      "loss": 0.6332,
      "step": 1274
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4097338465535723,
      "learning_rate": 9.542309369177785e-06,
      "loss": 0.8609,
      "step": 1275
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.417944915217198,
      "learning_rate": 9.541442485197577e-06,
      "loss": 0.7684,
      "step": 1276
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8115425203775175,
      "learning_rate": 9.540574820490759e-06,
      "loss": 0.6383,
      "step": 1277
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.436477575268394,
      "learning_rate": 9.539706375206487e-06,
      "loss": 0.8394,
      "step": 1278
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4645596090996107,
      "learning_rate": 9.538837149494065e-06,
      "loss": 0.8336,
      "step": 1279
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5264806318259914,
      "learning_rate": 9.537967143502915e-06,
      "loss": 0.8518,
      "step": 1280
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3532821106734263,
      "learning_rate": 9.537096357382606e-06,
      "loss": 0.812,
      "step": 1281
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8086381772183535,
      "learning_rate": 9.536224791282834e-06,
      "loss": 0.8545,
      "step": 1282
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3204806571868102,
      "learning_rate": 9.53535244535343e-06,
      "loss": 0.9138,
      "step": 1283
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3791697511364323,
      "learning_rate": 9.534479319744366e-06,
      "loss": 0.8127,
      "step": 1284
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7112930589901925,
      "learning_rate": 9.533605414605736e-06,
      "loss": 0.7958,
      "step": 1285
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7888799541880054,
      "learning_rate": 9.532730730087779e-06,
      "loss": 0.6488,
      "step": 1286
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3549327956026385,
      "learning_rate": 9.53185526634086e-06,
      "loss": 0.8839,
      "step": 1287
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4323276661548232,
      "learning_rate": 9.530979023515483e-06,
      "loss": 0.7915,
      "step": 1288
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8408652748700822,
      "learning_rate": 9.530102001762285e-06,
      "loss": 0.6966,
      "step": 1289
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.384477565829097,
      "learning_rate": 9.529224201232034e-06,
      "loss": 0.8078,
      "step": 1290
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7722387936137871,
      "learning_rate": 9.528345622075636e-06,
      "loss": 0.6007,
      "step": 1291
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6315146293502267,
      "learning_rate": 9.52746626444413e-06,
      "loss": 0.8013,
      "step": 1292
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4657495045501623,
      "learning_rate": 9.526586128488686e-06,
      "loss": 0.899,
      "step": 1293
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3807263254070279,
      "learning_rate": 9.52570521436061e-06,
      "loss": 0.873,
      "step": 1294
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3576260077529716,
      "learning_rate": 9.52482352221134e-06,
      "loss": 0.8053,
      "step": 1295
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3668893830016546,
      "learning_rate": 9.52394105219245e-06,
      "loss": 0.7559,
      "step": 1296
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6927901702788404,
      "learning_rate": 9.523057804455648e-06,
      "loss": 0.8469,
      "step": 1297
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3310190218087876,
      "learning_rate": 9.522173779152773e-06,
      "loss": 0.8701,
      "step": 1298
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5611345844097517,
      "learning_rate": 9.5212889764358e-06,
      "loss": 0.8407,
      "step": 1299
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4448195373365582,
      "learning_rate": 9.520403396456838e-06,
      "loss": 0.9316,
      "step": 1300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9159450487436673,
      "learning_rate": 9.519517039368127e-06,
      "loss": 0.7386,
      "step": 1301
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.819379919265649,
      "learning_rate": 9.518629905322041e-06,
      "loss": 0.7077,
      "step": 1302
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9291669909285484,
      "learning_rate": 9.517741994471091e-06,
      "loss": 0.6736,
      "step": 1303
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4407765841009155,
      "learning_rate": 9.516853306967917e-06,
      "loss": 0.8292,
      "step": 1304
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.575649216755863,
      "learning_rate": 9.515963842965294e-06,
      "loss": 0.8141,
      "step": 1305
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4276061478243738,
      "learning_rate": 9.515073602616135e-06,
      "loss": 0.8684,
      "step": 1306
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9204682828672184,
      "learning_rate": 9.51418258607348e-06,
      "loss": 0.6852,
      "step": 1307
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6610253262497747,
      "learning_rate": 9.513290793490506e-06,
      "loss": 0.9825,
      "step": 1308
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4018181577802729,
      "learning_rate": 9.512398225020523e-06,
      "loss": 0.8738,
      "step": 1309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2458919852629526,
      "learning_rate": 9.511504880816971e-06,
      "loss": 0.7355,
      "step": 1310
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4156049918778988,
      "learning_rate": 9.510610761033427e-06,
      "loss": 0.8233,
      "step": 1311
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.399966330965646,
      "learning_rate": 9.509715865823605e-06,
      "loss": 0.9047,
      "step": 1312
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.614439227651117,
      "learning_rate": 9.508820195341343e-06,
      "loss": 0.8269,
      "step": 1313
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7465047914957723,
      "learning_rate": 9.507923749740619e-06,
      "loss": 0.8777,
      "step": 1314
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6554128014896496,
      "learning_rate": 9.50702652917554e-06,
      "loss": 0.921,
      "step": 1315
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.378633082217697,
      "learning_rate": 9.506128533800353e-06,
      "loss": 0.7657,
      "step": 1316
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3580849087161728,
      "learning_rate": 9.505229763769432e-06,
      "loss": 0.7975,
      "step": 1317
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6525467573479102,
      "learning_rate": 9.504330219237284e-06,
      "loss": 0.8648,
      "step": 1318
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.449502300718978,
      "learning_rate": 9.503429900358554e-06,
      "loss": 0.7629,
      "step": 1319
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6033373150174055,
      "learning_rate": 9.502528807288014e-06,
      "loss": 0.8752,
      "step": 1320
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3922817247420218,
      "learning_rate": 9.501626940180574e-06,
      "loss": 0.8544,
      "step": 1321
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9171943779513686,
      "learning_rate": 9.50072429919128e-06,
      "loss": 0.6913,
      "step": 1322
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5513010924217387,
      "learning_rate": 9.499820884475296e-06,
      "loss": 0.9042,
      "step": 1323
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7376368961716342,
      "learning_rate": 9.498916696187942e-06,
      "loss": 0.7313,
      "step": 1324
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3767394003602296,
      "learning_rate": 9.498011734484647e-06,
      "loss": 0.7968,
      "step": 1325
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3352798551986358,
      "learning_rate": 9.49710599952099e-06,
      "loss": 0.8085,
      "step": 1326
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.871518250423332,
      "learning_rate": 9.49619949145268e-06,
      "loss": 0.7364,
      "step": 1327
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.674073174799157,
      "learning_rate": 9.49529221043555e-06,
      "loss": 0.9083,
      "step": 1328
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3506169314344352,
      "learning_rate": 9.494384156625575e-06,
      "loss": 0.8783,
      "step": 1329
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.03658386949232,
      "learning_rate": 9.49347533017886e-06,
      "loss": 0.8481,
      "step": 1330
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4180839280265254,
      "learning_rate": 9.492565731251645e-06,
      "loss": 0.8099,
      "step": 1331
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.890924393381307,
      "learning_rate": 9.491655360000298e-06,
      "loss": 0.7527,
      "step": 1332
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.577833122445516,
      "learning_rate": 9.490744216581323e-06,
      "loss": 0.8119,
      "step": 1333
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.518172178985552,
      "learning_rate": 9.489832301151354e-06,
      "loss": 0.8722,
      "step": 1334
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8603207079292843,
      "learning_rate": 9.488919613867162e-06,
      "loss": 0.7831,
      "step": 1335
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5920286194072992,
      "learning_rate": 9.48800615488565e-06,
      "loss": 0.9059,
      "step": 1336
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.483643394960124,
      "learning_rate": 9.487091924363852e-06,
      "loss": 0.9343,
      "step": 1337
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6682701935011446,
      "learning_rate": 9.486176922458929e-06,
      "loss": 0.7867,
      "step": 1338
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8994365444579201,
      "learning_rate": 9.485261149328189e-06,
      "loss": 0.7022,
      "step": 1339
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4029522699109764,
      "learning_rate": 9.484344605129057e-06,
      "loss": 0.8565,
      "step": 1340
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5240641860211555,
      "learning_rate": 9.4834272900191e-06,
      "loss": 0.8656,
      "step": 1341
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8708223877379924,
      "learning_rate": 9.482509204156019e-06,
      "loss": 0.7053,
      "step": 1342
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4457271159539233,
      "learning_rate": 9.481590347697638e-06,
      "loss": 0.855,
      "step": 1343
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6545456836125148,
      "learning_rate": 9.480670720801921e-06,
      "loss": 0.8576,
      "step": 1344
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4454623553573296,
      "learning_rate": 9.479750323626963e-06,
      "loss": 0.7949,
      "step": 1345
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9441816428878839,
      "learning_rate": 9.47882915633099e-06,
      "loss": 0.6537,
      "step": 1346
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4185751397609128,
      "learning_rate": 9.477907219072362e-06,
      "loss": 0.8632,
      "step": 1347
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2978234573822232,
      "learning_rate": 9.476984512009572e-06,
      "loss": 0.8219,
      "step": 1348
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4465610173680463,
      "learning_rate": 9.476061035301242e-06,
      "loss": 0.8436,
      "step": 1349
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3622855931029751,
      "learning_rate": 9.475136789106128e-06,
      "loss": 0.8275,
      "step": 1350
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3912301983839752,
      "learning_rate": 9.474211773583122e-06,
      "loss": 0.8867,
      "step": 1351
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7698601280634598,
      "learning_rate": 9.47328598889124e-06,
      "loss": 0.784,
      "step": 1352
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3399374581311065,
      "learning_rate": 9.47235943518964e-06,
      "loss": 0.8228,
      "step": 1353
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4027152960688,
      "learning_rate": 9.471432112637604e-06,
      "loss": 0.8699,
      "step": 1354
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4354170349990514,
      "learning_rate": 9.47050402139455e-06,
      "loss": 0.9251,
      "step": 1355
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6097673131104469,
      "learning_rate": 9.469575161620029e-06,
      "loss": 0.7845,
      "step": 1356
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.416252553499435,
      "learning_rate": 9.468645533473721e-06,
      "loss": 0.8048,
      "step": 1357
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.373216286634512,
      "learning_rate": 9.467715137115442e-06,
      "loss": 0.8868,
      "step": 1358
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6836327882869702,
      "learning_rate": 9.466783972705138e-06,
      "loss": 0.6836,
      "step": 1359
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.498557064853149,
      "learning_rate": 9.465852040402883e-06,
      "loss": 0.8022,
      "step": 1360
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3726209311555855,
      "learning_rate": 9.46491934036889e-06,
      "loss": 0.8273,
      "step": 1361
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.663417862718668,
      "learning_rate": 9.463985872763501e-06,
      "loss": 0.822,
      "step": 1362
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3375693654831564,
      "learning_rate": 9.463051637747191e-06,
      "loss": 0.8265,
      "step": 1363
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2822813101406407,
      "learning_rate": 9.462116635480562e-06,
      "loss": 0.821,
      "step": 1364
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6419064600735633,
      "learning_rate": 9.461180866124357e-06,
      "loss": 0.8292,
      "step": 1365
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4768275765766288,
      "learning_rate": 9.46024432983944e-06,
      "loss": 0.8612,
      "step": 1366
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4954381728641524,
      "learning_rate": 9.459307026786817e-06,
      "loss": 0.8343,
      "step": 1367
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4229687425962578,
      "learning_rate": 9.458368957127616e-06,
      "loss": 0.8175,
      "step": 1368
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4020772838320776,
      "learning_rate": 9.45743012102311e-06,
      "loss": 0.8585,
      "step": 1369
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.327482701575462,
      "learning_rate": 9.456490518634688e-06,
      "loss": 0.8314,
      "step": 1370
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9910282977275741,
      "learning_rate": 9.455550150123884e-06,
      "loss": 0.6826,
      "step": 1371
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5895674711005487,
      "learning_rate": 9.454609015652355e-06,
      "loss": 0.8316,
      "step": 1372
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4896357146985104,
      "learning_rate": 9.453667115381893e-06,
      "loss": 0.7986,
      "step": 1373
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.24015422552344,
      "learning_rate": 9.452724449474423e-06,
      "loss": 0.8126,
      "step": 1374
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4500664583613723,
      "learning_rate": 9.451781018092e-06,
      "loss": 0.7801,
      "step": 1375
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2472329617125737,
      "learning_rate": 9.45083682139681e-06,
      "loss": 0.7867,
      "step": 1376
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5326079303660405,
      "learning_rate": 9.449891859551172e-06,
      "loss": 0.7608,
      "step": 1377
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5888227393255756,
      "learning_rate": 9.448946132717536e-06,
      "loss": 0.9544,
      "step": 1378
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9286141896436222,
      "learning_rate": 9.447999641058486e-06,
      "loss": 0.6953,
      "step": 1379
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7526888422606637,
      "learning_rate": 9.447052384736729e-06,
      "loss": 0.8289,
      "step": 1380
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9466128761734843,
      "learning_rate": 9.446104363915112e-06,
      "loss": 0.6469,
      "step": 1381
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3596325851686482,
      "learning_rate": 9.445155578756612e-06,
      "loss": 0.7263,
      "step": 1382
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9536589945585402,
      "learning_rate": 9.444206029424334e-06,
      "loss": 0.6684,
      "step": 1383
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7136202895812793,
      "learning_rate": 9.443255716081522e-06,
      "loss": 0.8459,
      "step": 1384
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5721345648610496,
      "learning_rate": 9.442304638891538e-06,
      "loss": 0.8026,
      "step": 1385
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5745470120736478,
      "learning_rate": 9.441352798017886e-06,
      "loss": 0.828,
      "step": 1386
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.83055671411936,
      "learning_rate": 9.440400193624202e-06,
      "loss": 0.8339,
      "step": 1387
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8495890740705009,
      "learning_rate": 9.439446825874248e-06,
      "loss": 0.605,
      "step": 1388
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5411740114932593,
      "learning_rate": 9.438492694931916e-06,
      "loss": 0.8691,
      "step": 1389
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5080769115379908,
      "learning_rate": 9.437537800961235e-06,
      "loss": 0.843,
      "step": 1390
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3224099041658057,
      "learning_rate": 9.436582144126362e-06,
      "loss": 0.7639,
      "step": 1391
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8674856766666255,
      "learning_rate": 9.435625724591584e-06,
      "loss": 0.7418,
      "step": 1392
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.506661138105234,
      "learning_rate": 9.434668542521323e-06,
      "loss": 0.7973,
      "step": 1393
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3738647324388042,
      "learning_rate": 9.433710598080128e-06,
      "loss": 0.7901,
      "step": 1394
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3088984180037202,
      "learning_rate": 9.432751891432681e-06,
      "loss": 0.8552,
      "step": 1395
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3249267802041815,
      "learning_rate": 9.431792422743795e-06,
      "loss": 0.8378,
      "step": 1396
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7876312518910599,
      "learning_rate": 9.430832192178413e-06,
      "loss": 0.6516,
      "step": 1397
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4854959319317187,
      "learning_rate": 9.429871199901614e-06,
      "loss": 0.8556,
      "step": 1398
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.745199503349025,
      "learning_rate": 9.428909446078597e-06,
      "loss": 0.8092,
      "step": 1399
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4492520450497588,
      "learning_rate": 9.427946930874704e-06,
      "loss": 0.8908,
      "step": 1400
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3896897838158224,
      "learning_rate": 9.426983654455399e-06,
      "loss": 0.7747,
      "step": 1401
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4200938199517303,
      "learning_rate": 9.426019616986281e-06,
      "loss": 0.8433,
      "step": 1402
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9145171421554599,
      "learning_rate": 9.425054818633081e-06,
      "loss": 0.7167,
      "step": 1403
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8038877505232922,
      "learning_rate": 9.424089259561658e-06,
      "loss": 0.6144,
      "step": 1404
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5686584817761395,
      "learning_rate": 9.423122939938003e-06,
      "loss": 0.9529,
      "step": 1405
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5837459607460134,
      "learning_rate": 9.422155859928237e-06,
      "loss": 0.8588,
      "step": 1406
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7685902589120444,
      "learning_rate": 9.421188019698613e-06,
      "loss": 0.8363,
      "step": 1407
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7985771013689862,
      "learning_rate": 9.420219419415513e-06,
      "loss": 0.6492,
      "step": 1408
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7403497680584215,
      "learning_rate": 9.419250059245453e-06,
      "loss": 0.9004,
      "step": 1409
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7020361884416484,
      "learning_rate": 9.418279939355073e-06,
      "loss": 0.8587,
      "step": 1410
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4323548660196859,
      "learning_rate": 9.417309059911152e-06,
      "loss": 0.8749,
      "step": 1411
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.555445803079358,
      "learning_rate": 9.416337421080594e-06,
      "loss": 0.9079,
      "step": 1412
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8459473883650538,
      "learning_rate": 9.415365023030433e-06,
      "loss": 0.6888,
      "step": 1413
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3755122327853457,
      "learning_rate": 9.41439186592784e-06,
      "loss": 0.8678,
      "step": 1414
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9975346689522615,
      "learning_rate": 9.413417949940107e-06,
      "loss": 0.9467,
      "step": 1415
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3573156031615359,
      "learning_rate": 9.412443275234663e-06,
      "loss": 0.9052,
      "step": 1416
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4350722866192038,
      "learning_rate": 9.411467841979069e-06,
      "loss": 0.8143,
      "step": 1417
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6810797246712101,
      "learning_rate": 9.410491650341009e-06,
      "loss": 0.9019,
      "step": 1418
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9108461016437793,
      "learning_rate": 9.409514700488304e-06,
      "loss": 0.7145,
      "step": 1419
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.643018298480315,
      "learning_rate": 9.408536992588903e-06,
      "loss": 0.8674,
      "step": 1420
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4423304751511168,
      "learning_rate": 9.407558526810884e-06,
      "loss": 0.7958,
      "step": 1421
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7069021420994954,
      "learning_rate": 9.406579303322458e-06,
      "loss": 0.7894,
      "step": 1422
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4730845279985332,
      "learning_rate": 9.405599322291963e-06,
      "loss": 0.8348,
      "step": 1423
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.740970350524253,
      "learning_rate": 9.404618583887872e-06,
      "loss": 0.8142,
      "step": 1424
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2283250418108693,
      "learning_rate": 9.403637088278784e-06,
      "loss": 0.8773,
      "step": 1425
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6559369140705322,
      "learning_rate": 9.402654835633429e-06,
      "loss": 0.7678,
      "step": 1426
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8450667087648123,
      "learning_rate": 9.401671826120667e-06,
      "loss": 0.6135,
      "step": 1427
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5093895833361815,
      "learning_rate": 9.400688059909489e-06,
      "loss": 0.8696,
      "step": 1428
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5185979657486508,
      "learning_rate": 9.399703537169017e-06,
      "loss": 0.8369,
      "step": 1429
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3014076668244394,
      "learning_rate": 9.398718258068502e-06,
      "loss": 0.8636,
      "step": 1430
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.538122491540372,
      "learning_rate": 9.397732222777323e-06,
      "loss": 0.7724,
      "step": 1431
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6109407475394133,
      "learning_rate": 9.396745431464993e-06,
      "loss": 0.7776,
      "step": 1432
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5956736351898682,
      "learning_rate": 9.395757884301152e-06,
      "loss": 0.786,
      "step": 1433
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3789646328936986,
      "learning_rate": 9.394769581455569e-06,
      "loss": 0.8052,
      "step": 1434
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.586479907511523,
      "learning_rate": 9.393780523098148e-06,
      "loss": 0.8832,
      "step": 1435
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6369098030310327,
      "learning_rate": 9.392790709398916e-06,
      "loss": 0.7677,
      "step": 1436
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7080544537064029,
      "learning_rate": 9.391800140528038e-06,
      "loss": 0.8274,
      "step": 1437
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3460551067899975,
      "learning_rate": 9.390808816655801e-06,
      "loss": 0.828,
      "step": 1438
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.352811211414972,
      "learning_rate": 9.389816737952624e-06,
      "loss": 0.8706,
      "step": 1439
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3044732604652087,
      "learning_rate": 9.388823904589062e-06,
      "loss": 0.818,
      "step": 1440
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4694073473812241,
      "learning_rate": 9.387830316735789e-06,
      "loss": 0.8358,
      "step": 1441
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6369187457688288,
      "learning_rate": 9.386835974563616e-06,
      "loss": 0.8572,
      "step": 1442
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.896054013851866,
      "learning_rate": 9.385840878243482e-06,
      "loss": 0.6757,
      "step": 1443
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7814387357678279,
      "learning_rate": 9.384845027946458e-06,
      "loss": 0.8346,
      "step": 1444
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6255659644474831,
      "learning_rate": 9.38384842384374e-06,
      "loss": 0.8133,
      "step": 1445
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7937609916292766,
      "learning_rate": 9.382851066106655e-06,
      "loss": 0.595,
      "step": 1446
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4288486593434948,
      "learning_rate": 9.381852954906662e-06,
      "loss": 0.7911,
      "step": 1447
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7041308629838678,
      "learning_rate": 9.380854090415347e-06,
      "loss": 0.8571,
      "step": 1448
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3716201187139911,
      "learning_rate": 9.379854472804425e-06,
      "loss": 0.7713,
      "step": 1449
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2451034734214907,
      "learning_rate": 9.378854102245747e-06,
      "loss": 0.805,
      "step": 1450
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8043966213380126,
      "learning_rate": 9.377852978911282e-06,
      "loss": 0.6229,
      "step": 1451
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2214074918469968,
      "learning_rate": 9.376851102973139e-06,
      "loss": 0.8375,
      "step": 1452
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.9837798573674763,
      "learning_rate": 9.375848474603549e-06,
      "loss": 0.8116,
      "step": 1453
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.348796942215705,
      "learning_rate": 9.374845093974879e-06,
      "loss": 0.8008,
      "step": 1454
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7698651810440107,
      "learning_rate": 9.373840961259619e-06,
      "loss": 0.8229,
      "step": 1455
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4784829185967825,
      "learning_rate": 9.372836076630391e-06,
      "loss": 0.7928,
      "step": 1456
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4824710662975578,
      "learning_rate": 9.371830440259948e-06,
      "loss": 0.8673,
      "step": 1457
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4530818197246835,
      "learning_rate": 9.370824052321168e-06,
      "loss": 0.8286,
      "step": 1458
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8862560498571266,
      "learning_rate": 9.369816912987066e-06,
      "loss": 0.641,
      "step": 1459
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3918074961484168,
      "learning_rate": 9.368809022430773e-06,
      "loss": 0.839,
      "step": 1460
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6255032243896086,
      "learning_rate": 9.367800380825564e-06,
      "loss": 0.87,
      "step": 1461
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3910538438318303,
      "learning_rate": 9.366790988344835e-06,
      "loss": 0.792,
      "step": 1462
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.632559596057369,
      "learning_rate": 9.365780845162109e-06,
      "loss": 0.8228,
      "step": 1463
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6118868231285395,
      "learning_rate": 9.364769951451045e-06,
      "loss": 0.7983,
      "step": 1464
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8200259685653888,
      "learning_rate": 9.363758307385423e-06,
      "loss": 0.7163,
      "step": 1465
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6608269234519262,
      "learning_rate": 9.36274591313916e-06,
      "loss": 0.9472,
      "step": 1466
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4888597405958284,
      "learning_rate": 9.361732768886299e-06,
      "loss": 0.7835,
      "step": 1467
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7731801985433371,
      "learning_rate": 9.360718874801007e-06,
      "loss": 0.6214,
      "step": 1468
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9235635604418199,
      "learning_rate": 9.359704231057587e-06,
      "loss": 0.6596,
      "step": 1469
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.029660733437832,
      "learning_rate": 9.358688837830469e-06,
      "loss": 0.8359,
      "step": 1470
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9927575961094214,
      "learning_rate": 9.357672695294208e-06,
      "loss": 0.711,
      "step": 1471
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4943810164469942,
      "learning_rate": 9.356655803623493e-06,
      "loss": 0.7869,
      "step": 1472
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3824512367914903,
      "learning_rate": 9.355638162993139e-06,
      "loss": 0.8288,
      "step": 1473
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2611380131620327,
      "learning_rate": 9.354619773578088e-06,
      "loss": 0.8794,
      "step": 1474
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.404391618267962,
      "learning_rate": 9.353600635553418e-06,
      "loss": 0.8239,
      "step": 1475
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2788761961427928,
      "learning_rate": 9.352580749094325e-06,
      "loss": 0.7803,
      "step": 1476
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3888105511824322,
      "learning_rate": 9.351560114376142e-06,
      "loss": 0.8531,
      "step": 1477
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.932147733112067,
      "learning_rate": 9.350538731574329e-06,
      "loss": 0.7644,
      "step": 1478
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8322450579038998,
      "learning_rate": 9.349516600864473e-06,
      "loss": 0.776,
      "step": 1479
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3928440631898802,
      "learning_rate": 9.348493722422288e-06,
      "loss": 0.8313,
      "step": 1480
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4536791237325895,
      "learning_rate": 9.347470096423618e-06,
      "loss": 0.757,
      "step": 1481
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.384223803548264,
      "learning_rate": 9.346445723044441e-06,
      "loss": 0.8424,
      "step": 1482
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4071256819251592,
      "learning_rate": 9.345420602460856e-06,
      "loss": 0.7885,
      "step": 1483
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4842763431274435,
      "learning_rate": 9.344394734849092e-06,
      "loss": 0.8751,
      "step": 1484
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4372283755617272,
      "learning_rate": 9.343368120385508e-06,
      "loss": 0.8585,
      "step": 1485
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4008352245289315,
      "learning_rate": 9.342340759246591e-06,
      "loss": 0.8396,
      "step": 1486
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.488434892371641,
      "learning_rate": 9.341312651608957e-06,
      "loss": 0.8626,
      "step": 1487
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3884672308297759,
      "learning_rate": 9.340283797649352e-06,
      "loss": 0.7197,
      "step": 1488
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.602704751657422,
      "learning_rate": 9.339254197544642e-06,
      "loss": 0.7964,
      "step": 1489
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6457711767603862,
      "learning_rate": 9.338223851471833e-06,
      "loss": 0.8157,
      "step": 1490
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1890211131984403,
      "learning_rate": 9.33719275960805e-06,
      "loss": 0.8302,
      "step": 1491
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9574412014032925,
      "learning_rate": 9.33616092213055e-06,
      "loss": 0.6339,
      "step": 1492
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9681397599135848,
      "learning_rate": 9.335128339216719e-06,
      "loss": 0.7031,
      "step": 1493
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4676940148195283,
      "learning_rate": 9.33409501104407e-06,
      "loss": 0.8207,
      "step": 1494
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5347943457954947,
      "learning_rate": 9.333060937790243e-06,
      "loss": 0.8213,
      "step": 1495
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5080282548876804,
      "learning_rate": 9.332026119633007e-06,
      "loss": 0.8421,
      "step": 1496
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6100679910937241,
      "learning_rate": 9.33099055675026e-06,
      "loss": 0.8842,
      "step": 1497
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5352295266857718,
      "learning_rate": 9.329954249320028e-06,
      "loss": 0.8632,
      "step": 1498
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5184893104558386,
      "learning_rate": 9.328917197520461e-06,
      "loss": 0.7952,
      "step": 1499
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2593088854662036,
      "learning_rate": 9.327879401529846e-06,
      "loss": 0.7789,
      "step": 1500
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4466328014118095,
      "learning_rate": 9.326840861526587e-06,
      "loss": 0.8276,
      "step": 1501
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.984124168132179,
      "learning_rate": 9.325801577689224e-06,
      "loss": 0.6754,
      "step": 1502
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.264200501599543,
      "learning_rate": 9.32476155019642e-06,
      "loss": 0.6733,
      "step": 1503
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4594883367194795,
      "learning_rate": 9.323720779226969e-06,
      "loss": 0.8283,
      "step": 1504
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.662475776858313,
      "learning_rate": 9.322679264959789e-06,
      "loss": 0.8364,
      "step": 1505
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5482196241116102,
      "learning_rate": 9.321637007573932e-06,
      "loss": 0.8202,
      "step": 1506
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4735074013175313,
      "learning_rate": 9.320594007248573e-06,
      "loss": 0.7505,
      "step": 1507
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.578724039445685,
      "learning_rate": 9.319550264163015e-06,
      "loss": 0.7785,
      "step": 1508
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.404888965896916,
      "learning_rate": 9.318505778496692e-06,
      "loss": 0.7929,
      "step": 1509
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4361252972276766,
      "learning_rate": 9.317460550429159e-06,
      "loss": 0.8679,
      "step": 1510
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4601639104004995,
      "learning_rate": 9.316414580140105e-06,
      "loss": 0.8111,
      "step": 1511
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8569779727775972,
      "learning_rate": 9.315367867809346e-06,
      "loss": 0.678,
      "step": 1512
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.349238035066073,
      "learning_rate": 9.314320413616821e-06,
      "loss": 0.803,
      "step": 1513
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.437970354496224,
      "learning_rate": 9.313272217742604e-06,
      "loss": 0.7236,
      "step": 1514
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2644394310843543,
      "learning_rate": 9.312223280366886e-06,
      "loss": 0.8764,
      "step": 1515
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5828463779175634,
      "learning_rate": 9.311173601669996e-06,
      "loss": 0.8036,
      "step": 1516
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3103290153894704,
      "learning_rate": 9.310123181832382e-06,
      "loss": 0.7277,
      "step": 1517
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6486751108556454,
      "learning_rate": 9.309072021034629e-06,
      "loss": 0.8503,
      "step": 1518
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2038082066648548,
      "learning_rate": 9.308020119457439e-06,
      "loss": 0.7988,
      "step": 1519
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5153459450907092,
      "learning_rate": 9.306967477281648e-06,
      "loss": 0.8672,
      "step": 1520
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5539504342987445,
      "learning_rate": 9.305914094688216e-06,
      "loss": 0.8404,
      "step": 1521
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8250538117395476,
      "learning_rate": 9.304859971858233e-06,
      "loss": 0.7129,
      "step": 1522
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3213576613142213,
      "learning_rate": 9.303805108972916e-06,
      "loss": 0.8214,
      "step": 1523
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4374906546702115,
      "learning_rate": 9.302749506213607e-06,
      "loss": 0.7487,
      "step": 1524
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6713859728964162,
      "learning_rate": 9.301693163761774e-06,
      "loss": 0.6912,
      "step": 1525
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.557867023648295,
      "learning_rate": 9.300636081799017e-06,
      "loss": 0.7442,
      "step": 1526
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3976571554281052,
      "learning_rate": 9.299578260507061e-06,
      "loss": 0.7341,
      "step": 1527
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7150023686921525,
      "learning_rate": 9.298519700067757e-06,
      "loss": 0.822,
      "step": 1528
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6898433625229847,
      "learning_rate": 9.297460400663083e-06,
      "loss": 0.8839,
      "step": 1529
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.40548171286923,
      "learning_rate": 9.296400362475149e-06,
      "loss": 0.8596,
      "step": 1530
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.624245531734731,
      "learning_rate": 9.29533958568618e-06,
      "loss": 0.8819,
      "step": 1531
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4325173812264766,
      "learning_rate": 9.294278070478544e-06,
      "loss": 0.8046,
      "step": 1532
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3196194643821872,
      "learning_rate": 9.293215817034722e-06,
      "loss": 0.8911,
      "step": 1533
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3272250634899223,
      "learning_rate": 9.29215282553733e-06,
      "loss": 0.7921,
      "step": 1534
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5865921911109049,
      "learning_rate": 9.291089096169109e-06,
      "loss": 0.851,
      "step": 1535
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4974938296540417,
      "learning_rate": 9.290024629112927e-06,
      "loss": 0.8441,
      "step": 1536
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.503618858772292,
      "learning_rate": 9.288959424551775e-06,
      "loss": 0.8244,
      "step": 1537
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.329102581622713,
      "learning_rate": 9.287893482668774e-06,
      "loss": 0.7846,
      "step": 1538
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8402504777891284,
      "learning_rate": 9.286826803647178e-06,
      "loss": 0.6345,
      "step": 1539
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4246974045979317,
      "learning_rate": 9.285759387670357e-06,
      "loss": 0.8577,
      "step": 1540
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2906428281054885,
      "learning_rate": 9.28469123492181e-06,
      "loss": 0.8286,
      "step": 1541
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4544684234703262,
      "learning_rate": 9.283622345585169e-06,
      "loss": 0.8942,
      "step": 1542
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9323006964583083,
      "learning_rate": 9.282552719844186e-06,
      "loss": 0.8401,
      "step": 1543
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8523069696830382,
      "learning_rate": 9.281482357882743e-06,
      "loss": 0.6703,
      "step": 1544
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2940352456093138,
      "learning_rate": 9.280411259884847e-06,
      "loss": 0.8353,
      "step": 1545
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3610329666944065,
      "learning_rate": 9.279339426034634e-06,
      "loss": 0.8602,
      "step": 1546
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.246862267235372,
      "learning_rate": 9.278266856516363e-06,
      "loss": 0.8229,
      "step": 1547
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.519394043055703,
      "learning_rate": 9.27719355151442e-06,
      "loss": 0.8314,
      "step": 1548
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.276337705181428,
      "learning_rate": 9.276119511213321e-06,
      "loss": 0.8105,
      "step": 1549
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.387557640440583,
      "learning_rate": 9.275044735797705e-06,
      "loss": 0.8466,
      "step": 1550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.920816577415595,
      "learning_rate": 9.27396922545234e-06,
      "loss": 0.7423,
      "step": 1551
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3945268815287517,
      "learning_rate": 9.272892980362113e-06,
      "loss": 0.7588,
      "step": 1552
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4385378018454462,
      "learning_rate": 9.27181600071205e-06,
      "loss": 0.8752,
      "step": 1553
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4084828199777324,
      "learning_rate": 9.270738286687293e-06,
      "loss": 0.7981,
      "step": 1554
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.9243792517566325,
      "learning_rate": 9.269659838473113e-06,
      "loss": 0.8646,
      "step": 1555
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4570520057115581,
      "learning_rate": 9.26858065625491e-06,
      "loss": 0.8144,
      "step": 1556
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4877454600091735,
      "learning_rate": 9.267500740218205e-06,
      "loss": 0.9197,
      "step": 1557
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4590227336326853,
      "learning_rate": 9.266420090548652e-06,
      "loss": 0.8103,
      "step": 1558
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4022554968572727,
      "learning_rate": 9.265338707432024e-06,
      "loss": 0.8736,
      "step": 1559
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8934964783580404,
      "learning_rate": 9.264256591054222e-06,
      "loss": 0.7712,
      "step": 1560
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4371255528871187,
      "learning_rate": 9.263173741601278e-06,
      "loss": 0.9168,
      "step": 1561
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5319586263285612,
      "learning_rate": 9.262090159259345e-06,
      "loss": 0.8033,
      "step": 1562
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8615282145956448,
      "learning_rate": 9.261005844214703e-06,
      "loss": 0.7126,
      "step": 1563
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8512435155902547,
      "learning_rate": 9.25992079665376e-06,
      "loss": 0.6428,
      "step": 1564
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6367024956723983,
      "learning_rate": 9.258835016763044e-06,
      "loss": 0.8673,
      "step": 1565
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.358556240373955,
      "learning_rate": 9.257748504729218e-06,
      "loss": 0.7995,
      "step": 1566
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8296929108964768,
      "learning_rate": 9.256661260739065e-06,
      "loss": 0.6447,
      "step": 1567
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7047122536794719,
      "learning_rate": 9.255573284979491e-06,
      "loss": 0.7718,
      "step": 1568
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9396367322173059,
      "learning_rate": 9.254484577637534e-06,
      "loss": 0.7657,
      "step": 1569
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8911801148107342,
      "learning_rate": 9.253395138900359e-06,
      "loss": 0.6357,
      "step": 1570
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3947281290953017,
      "learning_rate": 9.252304968955248e-06,
      "loss": 0.8514,
      "step": 1571
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4047484882894534,
      "learning_rate": 9.251214067989617e-06,
      "loss": 0.8396,
      "step": 1572
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5380819747988403,
      "learning_rate": 9.250122436191002e-06,
      "loss": 0.8063,
      "step": 1573
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5575704057981308,
      "learning_rate": 9.249030073747069e-06,
      "loss": 0.8108,
      "step": 1574
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9753815955383522,
      "learning_rate": 9.247936980845608e-06,
      "loss": 0.619,
      "step": 1575
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9991213616915804,
      "learning_rate": 9.246843157674535e-06,
      "loss": 0.6393,
      "step": 1576
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5480782380209837,
      "learning_rate": 9.245748604421888e-06,
      "loss": 0.7717,
      "step": 1577
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3278293792975342,
      "learning_rate": 9.244653321275834e-06,
      "loss": 0.7799,
      "step": 1578
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.390609595973983,
      "learning_rate": 9.243557308424667e-06,
      "loss": 0.8719,
      "step": 1579
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.482991140397778,
      "learning_rate": 9.242460566056803e-06,
      "loss": 0.7794,
      "step": 1580
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5185218285710056,
      "learning_rate": 9.241363094360785e-06,
      "loss": 0.8591,
      "step": 1581
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2563512483032302,
      "learning_rate": 9.240264893525281e-06,
      "loss": 0.7275,
      "step": 1582
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.402099152369809,
      "learning_rate": 9.239165963739084e-06,
      "loss": 0.8488,
      "step": 1583
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7651270889512392,
      "learning_rate": 9.238066305191113e-06,
      "loss": 0.7409,
      "step": 1584
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8094122625600163,
      "learning_rate": 9.236965918070413e-06,
      "loss": 0.682,
      "step": 1585
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.370535270679577,
      "learning_rate": 9.235864802566153e-06,
      "loss": 0.7642,
      "step": 1586
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.459488889144908,
      "learning_rate": 9.234762958867627e-06,
      "loss": 0.6958,
      "step": 1587
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4104783949912043,
      "learning_rate": 9.233660387164254e-06,
      "loss": 0.6938,
      "step": 1588
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.488004308522009,
      "learning_rate": 9.23255708764558e-06,
      "loss": 0.7404,
      "step": 1589
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2783559599340988,
      "learning_rate": 9.231453060501274e-06,
      "loss": 0.8303,
      "step": 1590
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5536866910084064,
      "learning_rate": 9.230348305921131e-06,
      "loss": 0.8668,
      "step": 1591
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3656211112126495,
      "learning_rate": 9.229242824095074e-06,
      "loss": 0.8952,
      "step": 1592
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4917255249594688,
      "learning_rate": 9.228136615213144e-06,
      "loss": 0.7876,
      "step": 1593
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2907449514494735,
      "learning_rate": 9.227029679465515e-06,
      "loss": 0.8039,
      "step": 1594
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6405955345233785,
      "learning_rate": 9.225922017042478e-06,
      "loss": 0.8105,
      "step": 1595
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5186713734652615,
      "learning_rate": 9.224813628134457e-06,
      "loss": 0.7578,
      "step": 1596
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9512520289273545,
      "learning_rate": 9.223704512931994e-06,
      "loss": 0.638,
      "step": 1597
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.371959965467338,
      "learning_rate": 9.22259467162576e-06,
      "loss": 0.7868,
      "step": 1598
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4178502393129846,
      "learning_rate": 9.221484104406551e-06,
      "loss": 0.8284,
      "step": 1599
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2904761958914706,
      "learning_rate": 9.220372811465282e-06,
      "loss": 0.8027,
      "step": 1600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8271957603993723,
      "learning_rate": 9.219260792993001e-06,
      "loss": 0.6277,
      "step": 1601
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8379809101252315,
      "learning_rate": 9.218148049180878e-06,
      "loss": 0.653,
      "step": 1602
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.882316352319655,
      "learning_rate": 9.217034580220203e-06,
      "loss": 0.6687,
      "step": 1603
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4014489495349736,
      "learning_rate": 9.215920386302393e-06,
      "loss": 0.7805,
      "step": 1604
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5252203860134272,
      "learning_rate": 9.214805467618996e-06,
      "loss": 0.7963,
      "step": 1605
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3062259139189465,
      "learning_rate": 9.213689824361678e-06,
      "loss": 0.6916,
      "step": 1606
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1863408087541114,
      "learning_rate": 9.212573456722227e-06,
      "loss": 0.7274,
      "step": 1607
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4770104408029798,
      "learning_rate": 9.211456364892566e-06,
      "loss": 0.9023,
      "step": 1608
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8727841883005647,
      "learning_rate": 9.210338549064728e-06,
      "loss": 0.8778,
      "step": 1609
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3593733740461906,
      "learning_rate": 9.209220009430886e-06,
      "loss": 0.8883,
      "step": 1610
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4270472941404901,
      "learning_rate": 9.208100746183327e-06,
      "loss": 0.7896,
      "step": 1611
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5012874682912003,
      "learning_rate": 9.206980759514464e-06,
      "loss": 0.7934,
      "step": 1612
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4835038096116386,
      "learning_rate": 9.205860049616837e-06,
      "loss": 0.718,
      "step": 1613
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.901573057909016,
      "learning_rate": 9.20473861668311e-06,
      "loss": 0.9095,
      "step": 1614
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3717157478922046,
      "learning_rate": 9.203616460906068e-06,
      "loss": 0.8309,
      "step": 1615
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.49064478002632,
      "learning_rate": 9.202493582478625e-06,
      "loss": 0.7873,
      "step": 1616
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.389623599661092,
      "learning_rate": 9.201369981593816e-06,
      "loss": 0.8581,
      "step": 1617
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2294547884652745,
      "learning_rate": 9.200245658444799e-06,
      "loss": 0.801,
      "step": 1618
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5146218560625473,
      "learning_rate": 9.199120613224862e-06,
      "loss": 0.8639,
      "step": 1619
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4201122361873544,
      "learning_rate": 9.197994846127409e-06,
      "loss": 0.8253,
      "step": 1620
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3257125333254802,
      "learning_rate": 9.196868357345976e-06,
      "loss": 0.8031,
      "step": 1621
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5732589661736716,
      "learning_rate": 9.195741147074218e-06,
      "loss": 0.817,
      "step": 1622
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5244178009370912,
      "learning_rate": 9.194613215505916e-06,
      "loss": 0.8193,
      "step": 1623
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3274012721242354,
      "learning_rate": 9.193484562834973e-06,
      "loss": 0.8292,
      "step": 1624
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3854156793640466,
      "learning_rate": 9.192355189255418e-06,
      "loss": 0.7399,
      "step": 1625
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4245615224520232,
      "learning_rate": 9.191225094961407e-06,
      "loss": 0.8253,
      "step": 1626
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3452325878183005,
      "learning_rate": 9.19009428014721e-06,
      "loss": 0.7605,
      "step": 1627
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5882847963406144,
      "learning_rate": 9.188962745007233e-06,
      "loss": 0.8169,
      "step": 1628
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4415564837000048,
      "learning_rate": 9.187830489735996e-06,
      "loss": 0.8552,
      "step": 1629
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.960999637129685,
      "learning_rate": 9.18669751452815e-06,
      "loss": 0.9266,
      "step": 1630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3729725953951764,
      "learning_rate": 9.185563819578462e-06,
      "loss": 0.7486,
      "step": 1631
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.34337956646604,
      "learning_rate": 9.184429405081832e-06,
      "loss": 0.826,
      "step": 1632
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8193637945983593,
      "learning_rate": 9.183294271233278e-06,
      "loss": 0.6025,
      "step": 1633
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.492329996209424,
      "learning_rate": 9.18215841822794e-06,
      "loss": 0.9566,
      "step": 1634
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.421258522747985,
      "learning_rate": 9.181021846261088e-06,
      "loss": 0.8225,
      "step": 1635
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3132477563361296,
      "learning_rate": 9.17988455552811e-06,
      "loss": 0.8268,
      "step": 1636
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3404501820867802,
      "learning_rate": 9.178746546224517e-06,
      "loss": 0.8275,
      "step": 1637
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.431891107073225,
      "learning_rate": 9.177607818545951e-06,
      "loss": 0.7735,
      "step": 1638
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.286946167525775,
      "learning_rate": 9.176468372688168e-06,
      "loss": 0.7644,
      "step": 1639
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2869942000921135,
      "learning_rate": 9.175328208847056e-06,
      "loss": 0.7661,
      "step": 1640
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9044808333679064,
      "learning_rate": 9.17418732721862e-06,
      "loss": 0.6596,
      "step": 1641
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4526054425873984,
      "learning_rate": 9.17304572799899e-06,
      "loss": 0.7801,
      "step": 1642
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.67848505068573,
      "learning_rate": 9.171903411384422e-06,
      "loss": 0.8985,
      "step": 1643
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8512892838393947,
      "learning_rate": 9.170760377571293e-06,
      "loss": 0.6996,
      "step": 1644
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3003824754042255,
      "learning_rate": 9.169616626756103e-06,
      "loss": 0.8674,
      "step": 1645
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7215509168805279,
      "learning_rate": 9.168472159135477e-06,
      "loss": 0.5623,
      "step": 1646
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3118454033607612,
      "learning_rate": 9.167326974906161e-06,
      "loss": 0.8244,
      "step": 1647
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3587066071551583,
      "learning_rate": 9.166181074265027e-06,
      "loss": 0.8983,
      "step": 1648
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3917605749528073,
      "learning_rate": 9.165034457409066e-06,
      "loss": 0.6664,
      "step": 1649
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4872569406029736,
      "learning_rate": 9.163887124535398e-06,
      "loss": 0.7678,
      "step": 1650
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3867309860880663,
      "learning_rate": 9.162739075841263e-06,
      "loss": 0.9009,
      "step": 1651
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3579758737835685,
      "learning_rate": 9.16159031152402e-06,
      "loss": 0.8284,
      "step": 1652
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3185213354355032,
      "learning_rate": 9.160440831781157e-06,
      "loss": 0.7694,
      "step": 1653
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4532481049684678,
      "learning_rate": 9.159290636810284e-06,
      "loss": 0.77,
      "step": 1654
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.50967587638838,
      "learning_rate": 9.158139726809132e-06,
      "loss": 0.8608,
      "step": 1655
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4320860889494669,
      "learning_rate": 9.156988101975555e-06,
      "loss": 0.8991,
      "step": 1656
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2910967947513778,
      "learning_rate": 9.155835762507532e-06,
      "loss": 0.8401,
      "step": 1657
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5562930159373147,
      "learning_rate": 9.154682708603162e-06,
      "loss": 0.8554,
      "step": 1658
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3134852720326948,
      "learning_rate": 9.153528940460669e-06,
      "loss": 0.7673,
      "step": 1659
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.359608279157495,
      "learning_rate": 9.152374458278402e-06,
      "loss": 0.7872,
      "step": 1660
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.643010193226634,
      "learning_rate": 9.151219262254824e-06,
      "loss": 0.7947,
      "step": 1661
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2750258016432738,
      "learning_rate": 9.150063352588531e-06,
      "loss": 0.7935,
      "step": 1662
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8628051793133975,
      "learning_rate": 9.148906729478236e-06,
      "loss": 0.8586,
      "step": 1663
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6414909256873988,
      "learning_rate": 9.147749393122776e-06,
      "loss": 0.7483,
      "step": 1664
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.438958664168014,
      "learning_rate": 9.14659134372111e-06,
      "loss": 0.8113,
      "step": 1665
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8245664541012325,
      "learning_rate": 9.145432581472323e-06,
      "loss": 0.6671,
      "step": 1666
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.735170434376667,
      "learning_rate": 9.144273106575614e-06,
      "loss": 0.9147,
      "step": 1667
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.322937142111943,
      "learning_rate": 9.143112919230314e-06,
      "loss": 0.7349,
      "step": 1668
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3421623894271182,
      "learning_rate": 9.141952019635874e-06,
      "loss": 0.7434,
      "step": 1669
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.41641532941143,
      "learning_rate": 9.140790407991862e-06,
      "loss": 0.792,
      "step": 1670
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5772316726646423,
      "learning_rate": 9.139628084497975e-06,
      "loss": 0.8849,
      "step": 1671
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8470747101449194,
      "learning_rate": 9.138465049354031e-06,
      "loss": 0.6441,
      "step": 1672
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0419636847040596,
      "learning_rate": 9.137301302759968e-06,
      "loss": 0.6421,
      "step": 1673
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.381945292852732,
      "learning_rate": 9.136136844915847e-06,
      "loss": 0.7658,
      "step": 1674
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.21987899813404,
      "learning_rate": 9.134971676021851e-06,
      "loss": 1.0028,
      "step": 1675
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9126251973001418,
      "learning_rate": 9.133805796278288e-06,
      "loss": 0.6454,
      "step": 1676
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5340592626341696,
      "learning_rate": 9.132639205885584e-06,
      "loss": 0.8295,
      "step": 1677
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.31640783373053,
      "learning_rate": 9.131471905044294e-06,
      "loss": 0.7866,
      "step": 1678
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8815913083996119,
      "learning_rate": 9.130303893955084e-06,
      "loss": 0.7001,
      "step": 1679
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3838770489143022,
      "learning_rate": 9.129135172818754e-06,
      "loss": 0.8107,
      "step": 1680
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4185652947810021,
      "learning_rate": 9.127965741836218e-06,
      "loss": 0.7555,
      "step": 1681
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3400866893508576,
      "learning_rate": 9.126795601208516e-06,
      "loss": 0.8621,
      "step": 1682
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.412650029964292,
      "learning_rate": 9.125624751136809e-06,
      "loss": 0.7966,
      "step": 1683
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6162736577863366,
      "learning_rate": 9.124453191822376e-06,
      "loss": 0.787,
      "step": 1684
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.618798175707368,
      "learning_rate": 9.123280923466627e-06,
      "loss": 0.8744,
      "step": 1685
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5468513718622265,
      "learning_rate": 9.122107946271086e-06,
      "loss": 0.8622,
      "step": 1686
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3186469313571556,
      "learning_rate": 9.1209342604374e-06,
      "loss": 0.7829,
      "step": 1687
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4303367721359759,
      "learning_rate": 9.119759866167342e-06,
      "loss": 0.8136,
      "step": 1688
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4780002059503283,
      "learning_rate": 9.118584763662803e-06,
      "loss": 0.8667,
      "step": 1689
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.031995073320435,
      "learning_rate": 9.117408953125794e-06,
      "loss": 0.8087,
      "step": 1690
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.396618698353101,
      "learning_rate": 9.116232434758456e-06,
      "loss": 0.7637,
      "step": 1691
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4616059500123375,
      "learning_rate": 9.115055208763042e-06,
      "loss": 0.7626,
      "step": 1692
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2858650314343159,
      "learning_rate": 9.113877275341932e-06,
      "loss": 0.8779,
      "step": 1693
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3797164885723048,
      "learning_rate": 9.112698634697625e-06,
      "loss": 0.832,
      "step": 1694
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7138598006029846,
      "learning_rate": 9.111519287032745e-06,
      "loss": 0.8346,
      "step": 1695
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8625201642665632,
      "learning_rate": 9.110339232550034e-06,
      "loss": 0.653,
      "step": 1696
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9078217437449598,
      "learning_rate": 9.109158471452358e-06,
      "loss": 0.7037,
      "step": 1697
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7724267342854464,
      "learning_rate": 9.107977003942703e-06,
      "loss": 0.6841,
      "step": 1698
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4987471778450492,
      "learning_rate": 9.106794830224179e-06,
      "loss": 0.7712,
      "step": 1699
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.536172773949211,
      "learning_rate": 9.105611950500012e-06,
      "loss": 0.8802,
      "step": 1700
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8356811737046397,
      "learning_rate": 9.104428364973555e-06,
      "loss": 1.0352,
      "step": 1701
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.7697379415774566,
      "learning_rate": 9.10324407384828e-06,
      "loss": 0.7777,
      "step": 1702
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3508370161563354,
      "learning_rate": 9.102059077327779e-06,
      "loss": 0.8264,
      "step": 1703
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2407949447248978,
      "learning_rate": 9.100873375615767e-06,
      "loss": 0.7388,
      "step": 1704
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.508732202557496,
      "learning_rate": 9.099686968916081e-06,
      "loss": 0.84,
      "step": 1705
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7125969916324406,
      "learning_rate": 9.098499857432677e-06,
      "loss": 0.8654,
      "step": 1706
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8681706838139185,
      "learning_rate": 9.097312041369634e-06,
      "loss": 0.7477,
      "step": 1707
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.482780629056216,
      "learning_rate": 9.09612352093115e-06,
      "loss": 0.8025,
      "step": 1708
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6781854613699916,
      "learning_rate": 9.094934296321549e-06,
      "loss": 0.7729,
      "step": 1709
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.491344167304114,
      "learning_rate": 9.093744367745267e-06,
      "loss": 0.8174,
      "step": 1710
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.835954190238401,
      "learning_rate": 9.092553735406871e-06,
      "loss": 0.5526,
      "step": 1711
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4186621006060065,
      "learning_rate": 9.091362399511043e-06,
      "loss": 0.896,
      "step": 1712
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2070598404832118,
      "learning_rate": 9.090170360262587e-06,
      "loss": 0.7164,
      "step": 1713
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3959037033944732,
      "learning_rate": 9.088977617866428e-06,
      "loss": 0.8203,
      "step": 1714
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2923443356161057,
      "learning_rate": 9.087784172527614e-06,
      "loss": 0.803,
      "step": 1715
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2142835933028517,
      "learning_rate": 9.086590024451312e-06,
      "loss": 0.8931,
      "step": 1716
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6221780486738633,
      "learning_rate": 9.085395173842807e-06,
      "loss": 0.7677,
      "step": 1717
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4821937802814993,
      "learning_rate": 9.08419962090751e-06,
      "loss": 0.703,
      "step": 1718
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4364809741669253,
      "learning_rate": 9.08300336585095e-06,
      "loss": 0.8047,
      "step": 1719
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4062610600627934,
      "learning_rate": 9.081806408878778e-06,
      "loss": 0.7409,
      "step": 1720
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.549733648127868,
      "learning_rate": 9.080608750196764e-06,
      "loss": 0.7986,
      "step": 1721
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.480909396237745,
      "learning_rate": 9.079410390010798e-06,
      "loss": 0.8903,
      "step": 1722
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.406580880034901,
      "learning_rate": 9.078211328526896e-06,
      "loss": 0.8818,
      "step": 1723
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5377090171132284,
      "learning_rate": 9.077011565951186e-06,
      "loss": 0.8656,
      "step": 1724
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9115791701139706,
      "learning_rate": 9.075811102489923e-06,
      "loss": 0.7113,
      "step": 1725
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8468204384943577,
      "learning_rate": 9.074609938349481e-06,
      "loss": 0.6969,
      "step": 1726
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4204374830019713,
      "learning_rate": 9.073408073736355e-06,
      "loss": 0.7827,
      "step": 1727
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.5178647323662275,
      "learning_rate": 9.072205508857155e-06,
      "loss": 0.8529,
      "step": 1728
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2792033696817047,
      "learning_rate": 9.071002243918621e-06,
      "loss": 0.7746,
      "step": 1729
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6542472164230582,
      "learning_rate": 9.069798279127606e-06,
      "loss": 0.8595,
      "step": 1730
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8276665378471693,
      "learning_rate": 9.068593614691086e-06,
      "loss": 0.8138,
      "step": 1731
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6828028936792192,
      "learning_rate": 9.067388250816155e-06,
      "loss": 0.8234,
      "step": 1732
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5860346380031207,
      "learning_rate": 9.066182187710032e-06,
      "loss": 0.9717,
      "step": 1733
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8239623417956028,
      "learning_rate": 9.064975425580051e-06,
      "loss": 0.847,
      "step": 1734
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5562718412656453,
      "learning_rate": 9.06376796463367e-06,
      "loss": 0.7607,
      "step": 1735
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7414550395548878,
      "learning_rate": 9.062559805078463e-06,
      "loss": 0.7726,
      "step": 1736
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4583103187122792,
      "learning_rate": 9.061350947122129e-06,
      "loss": 0.9793,
      "step": 1737
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.340556656790762,
      "learning_rate": 9.060141390972486e-06,
      "loss": 0.7566,
      "step": 1738
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5383273199374312,
      "learning_rate": 9.058931136837465e-06,
      "loss": 0.8437,
      "step": 1739
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.533305675107989,
      "learning_rate": 9.05772018492513e-06,
      "loss": 0.7864,
      "step": 1740
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8689877911325486,
      "learning_rate": 9.056508535443652e-06,
      "loss": 0.6081,
      "step": 1741
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6697953639466863,
      "learning_rate": 9.05529618860133e-06,
      "loss": 0.8565,
      "step": 1742
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3901737699653522,
      "learning_rate": 9.05408314460658e-06,
      "loss": 0.797,
      "step": 1743
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2764909600836196,
      "learning_rate": 9.052869403667938e-06,
      "loss": 0.8191,
      "step": 1744
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5694343193651155,
      "learning_rate": 9.051654965994062e-06,
      "loss": 0.8197,
      "step": 1745
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7231194411713957,
      "learning_rate": 9.050439831793726e-06,
      "loss": 0.8565,
      "step": 1746
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3616435370760285,
      "learning_rate": 9.049224001275825e-06,
      "loss": 0.8254,
      "step": 1747
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4234299679392872,
      "learning_rate": 9.048007474649377e-06,
      "loss": 0.8145,
      "step": 1748
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.562902559465972,
      "learning_rate": 9.046790252123514e-06,
      "loss": 0.848,
      "step": 1749
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.401830638435909,
      "learning_rate": 9.045572333907495e-06,
      "loss": 0.8207,
      "step": 1750
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8341655302470341,
      "learning_rate": 9.04435372021069e-06,
      "loss": 0.706,
      "step": 1751
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2902650087013974,
      "learning_rate": 9.043134411242593e-06,
      "loss": 0.8575,
      "step": 1752
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.587098563966667,
      "learning_rate": 9.041914407212821e-06,
      "loss": 0.9321,
      "step": 1753
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.505495067428561,
      "learning_rate": 9.040693708331103e-06,
      "loss": 0.8826,
      "step": 1754
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2864988804736568,
      "learning_rate": 9.039472314807293e-06,
      "loss": 0.8144,
      "step": 1755
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9467539436537057,
      "learning_rate": 9.038250226851362e-06,
      "loss": 0.5892,
      "step": 1756
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2957071703285676,
      "learning_rate": 9.037027444673402e-06,
      "loss": 0.8192,
      "step": 1757
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.43766320885955,
      "learning_rate": 9.035803968483625e-06,
      "loss": 0.8584,
      "step": 1758
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.323708376143817,
      "learning_rate": 9.034579798492356e-06,
      "loss": 0.8083,
      "step": 1759
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4584204018318387,
      "learning_rate": 9.033354934910049e-06,
      "loss": 0.8124,
      "step": 1760
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8889634858773512,
      "learning_rate": 9.032129377947267e-06,
      "loss": 0.6957,
      "step": 1761
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.921652252782142,
      "learning_rate": 9.030903127814704e-06,
      "loss": 0.851,
      "step": 1762
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.402862687561734,
      "learning_rate": 9.029676184723161e-06,
      "loss": 0.8888,
      "step": 1763
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3514696662857797,
      "learning_rate": 9.028448548883566e-06,
      "loss": 0.7871,
      "step": 1764
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6303897076891425,
      "learning_rate": 9.027220220506964e-06,
      "loss": 0.9126,
      "step": 1765
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8817037820558099,
      "learning_rate": 9.025991199804518e-06,
      "loss": 0.6366,
      "step": 1766
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5483626757302735,
      "learning_rate": 9.024761486987512e-06,
      "loss": 0.8239,
      "step": 1767
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8655813297399173,
      "learning_rate": 9.023531082267347e-06,
      "loss": 0.8873,
      "step": 1768
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6425139879781758,
      "learning_rate": 9.022299985855544e-06,
      "loss": 0.722,
      "step": 1769
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4519266397969242,
      "learning_rate": 9.021068197963744e-06,
      "loss": 0.8669,
      "step": 1770
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2132299845656855,
      "learning_rate": 9.019835718803704e-06,
      "loss": 0.7768,
      "step": 1771
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3982004201453144,
      "learning_rate": 9.018602548587303e-06,
      "loss": 0.8256,
      "step": 1772
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3793387282022243,
      "learning_rate": 9.017368687526535e-06,
      "loss": 0.8412,
      "step": 1773
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3617315538896957,
      "learning_rate": 9.016134135833517e-06,
      "loss": 0.7596,
      "step": 1774
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.461063031939189,
      "learning_rate": 9.01489889372048e-06,
      "loss": 0.7904,
      "step": 1775
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8299735632560389,
      "learning_rate": 9.01366296139978e-06,
      "loss": 0.7007,
      "step": 1776
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3525672264668829,
      "learning_rate": 9.012426339083887e-06,
      "loss": 0.7861,
      "step": 1777
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6424506019921525,
      "learning_rate": 9.01118902698539e-06,
      "loss": 0.8375,
      "step": 1778
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3330952233018152,
      "learning_rate": 9.009951025316998e-06,
      "loss": 0.8302,
      "step": 1779
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.308633053036038,
      "learning_rate": 9.008712334291536e-06,
      "loss": 0.8285,
      "step": 1780
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5461948610782537,
      "learning_rate": 9.007472954121952e-06,
      "loss": 0.8348,
      "step": 1781
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.266872522053327,
      "learning_rate": 9.006232885021309e-06,
      "loss": 0.8313,
      "step": 1782
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3718960027475309,
      "learning_rate": 9.004992127202787e-06,
      "loss": 0.8502,
      "step": 1783
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8587394835159036,
      "learning_rate": 9.00375068087969e-06,
      "loss": 0.6146,
      "step": 1784
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4359374526927218,
      "learning_rate": 9.002508546265433e-06,
      "loss": 0.8715,
      "step": 1785
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5901075834523697,
      "learning_rate": 9.001265723573559e-06,
      "loss": 0.8086,
      "step": 1786
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3549946134269895,
      "learning_rate": 9.00002221301772e-06,
      "loss": 0.7288,
      "step": 1787
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4421173656378679,
      "learning_rate": 8.998778014811688e-06,
      "loss": 0.8566,
      "step": 1788
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3600784886987174,
      "learning_rate": 8.99753312916936e-06,
      "loss": 0.7489,
      "step": 1789
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6446495457547559,
      "learning_rate": 8.996287556304743e-06,
      "loss": 0.8225,
      "step": 1790
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3020548009798503,
      "learning_rate": 8.995041296431965e-06,
      "loss": 0.8848,
      "step": 1791
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2488351154659736,
      "learning_rate": 8.993794349765275e-06,
      "loss": 0.861,
      "step": 1792
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4410578193509902,
      "learning_rate": 8.992546716519034e-06,
      "loss": 0.844,
      "step": 1793
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5956598692765884,
      "learning_rate": 8.99129839690773e-06,
      "loss": 0.8234,
      "step": 1794
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.572815170871305,
      "learning_rate": 8.99004939114596e-06,
      "loss": 0.7568,
      "step": 1795
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6116388208465893,
      "learning_rate": 8.988799699448442e-06,
      "loss": 0.8896,
      "step": 1796
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5639797526007477,
      "learning_rate": 8.987549322030013e-06,
      "loss": 0.802,
      "step": 1797
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.901986868046541,
      "learning_rate": 8.986298259105629e-06,
      "loss": 0.6365,
      "step": 1798
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5366902993673421,
      "learning_rate": 8.98504651089036e-06,
      "loss": 0.8419,
      "step": 1799
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3582010456539226,
      "learning_rate": 8.983794077599398e-06,
      "loss": 0.8125,
      "step": 1800
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8073676886347951,
      "learning_rate": 8.98254095944805e-06,
      "loss": 0.8545,
      "step": 1801
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4935996453360647,
      "learning_rate": 8.981287156651741e-06,
      "loss": 0.8679,
      "step": 1802
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3910376671322755,
      "learning_rate": 8.980032669426015e-06,
      "loss": 0.839,
      "step": 1803
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2113793777509045,
      "learning_rate": 8.978777497986533e-06,
      "loss": 0.7436,
      "step": 1804
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8979119317335506,
      "learning_rate": 8.977521642549073e-06,
      "loss": 0.6279,
      "step": 1805
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.923658214953837,
      "learning_rate": 8.97626510332953e-06,
      "loss": 0.8304,
      "step": 1806
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8029171165849934,
      "learning_rate": 8.97500788054392e-06,
      "loss": 0.6955,
      "step": 1807
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2782332471764843,
      "learning_rate": 8.973749974408375e-06,
      "loss": 0.7598,
      "step": 1808
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7398975594212538,
      "learning_rate": 8.972491385139138e-06,
      "loss": 0.8186,
      "step": 1809
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7636045995214977,
      "learning_rate": 8.971232112952581e-06,
      "loss": 0.7634,
      "step": 1810
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.432376928598332,
      "learning_rate": 8.969972158065185e-06,
      "loss": 0.8102,
      "step": 1811
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.28692593670441,
      "learning_rate": 8.968711520693551e-06,
      "loss": 0.7801,
      "step": 1812
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3136543124563715,
      "learning_rate": 8.967450201054397e-06,
      "loss": 0.8765,
      "step": 1813
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4478528927869587,
      "learning_rate": 8.96618819936456e-06,
      "loss": 0.791,
      "step": 1814
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5932050974921095,
      "learning_rate": 8.964925515840991e-06,
      "loss": 0.8932,
      "step": 1815
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4244400435369342,
      "learning_rate": 8.963662150700761e-06,
      "loss": 0.8364,
      "step": 1816
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5894975497254284,
      "learning_rate": 8.962398104161055e-06,
      "loss": 0.8085,
      "step": 1817
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6491165854068026,
      "learning_rate": 8.96113337643918e-06,
      "loss": 0.8022,
      "step": 1818
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.446062413467632,
      "learning_rate": 8.959867967752556e-06,
      "loss": 0.8151,
      "step": 1819
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3390452211073214,
      "learning_rate": 8.95860187831872e-06,
      "loss": 0.8256,
      "step": 1820
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1902219429284393,
      "learning_rate": 8.957335108355332e-06,
      "loss": 0.7604,
      "step": 1821
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6061053426614753,
      "learning_rate": 8.956067658080158e-06,
      "loss": 0.8318,
      "step": 1822
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3655178383615059,
      "learning_rate": 8.954799527711094e-06,
      "loss": 0.8549,
      "step": 1823
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8836036250448795,
      "learning_rate": 8.953530717466143e-06,
      "loss": 0.6523,
      "step": 1824
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4863838057683818,
      "learning_rate": 8.952261227563427e-06,
      "loss": 0.8384,
      "step": 1825
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3049416951967874,
      "learning_rate": 8.950991058221187e-06,
      "loss": 0.8689,
      "step": 1826
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.369335116568156,
      "learning_rate": 8.949720209657781e-06,
      "loss": 0.8244,
      "step": 1827
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.286531383804481,
      "learning_rate": 8.948448682091684e-06,
      "loss": 0.7947,
      "step": 1828
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9753775700281524,
      "learning_rate": 8.947176475741482e-06,
      "loss": 0.7431,
      "step": 1829
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.379420799260217,
      "learning_rate": 8.945903590825884e-06,
      "loss": 0.9108,
      "step": 1830
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4020719797838435,
      "learning_rate": 8.944630027563714e-06,
      "loss": 0.8727,
      "step": 1831
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8903418777181296,
      "learning_rate": 8.943355786173913e-06,
      "loss": 0.7257,
      "step": 1832
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4254599109910524,
      "learning_rate": 8.942080866875538e-06,
      "loss": 0.8428,
      "step": 1833
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.269535126471229,
      "learning_rate": 8.94080526988776e-06,
      "loss": 0.8359,
      "step": 1834
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5813658341858108,
      "learning_rate": 8.93952899542987e-06,
      "loss": 0.8011,
      "step": 1835
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5901368927335156,
      "learning_rate": 8.938252043721275e-06,
      "loss": 0.7376,
      "step": 1836
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.231431439768134,
      "learning_rate": 8.936974414981498e-06,
      "loss": 0.7961,
      "step": 1837
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1953088537610912,
      "learning_rate": 8.935696109430178e-06,
      "loss": 0.7817,
      "step": 1838
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3616453945494484,
      "learning_rate": 8.93441712728707e-06,
      "loss": 0.825,
      "step": 1839
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2990126879303772,
      "learning_rate": 8.933137468772047e-06,
      "loss": 0.7796,
      "step": 1840
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1731931883693318,
      "learning_rate": 8.931857134105093e-06,
      "loss": 0.7705,
      "step": 1841
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3782915462768668,
      "learning_rate": 8.930576123506318e-06,
      "loss": 0.8235,
      "step": 1842
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8004876690377943,
      "learning_rate": 8.929294437195942e-06,
      "loss": 0.7138,
      "step": 1843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4556622549754754,
      "learning_rate": 8.928012075394297e-06,
      "loss": 0.7923,
      "step": 1844
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.935118542279019,
      "learning_rate": 8.92672903832184e-06,
      "loss": 0.6525,
      "step": 1845
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.334375308401218,
      "learning_rate": 8.925445326199138e-06,
      "loss": 0.8097,
      "step": 1846
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7226713804652454,
      "learning_rate": 8.924160939246877e-06,
      "loss": 0.8213,
      "step": 1847
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9702400488743288,
      "learning_rate": 8.922875877685859e-06,
      "loss": 0.6623,
      "step": 1848
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.435298678487028,
      "learning_rate": 8.921590141737e-06,
      "loss": 0.8326,
      "step": 1849
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8443773983165772,
      "learning_rate": 8.920303731621332e-06,
      "loss": 0.8165,
      "step": 1850
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.373342594283246,
      "learning_rate": 8.919016647560003e-06,
      "loss": 0.8194,
      "step": 1851
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9680021781397212,
      "learning_rate": 8.91772888977428e-06,
      "loss": 0.8483,
      "step": 1852
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5648446811576027,
      "learning_rate": 8.916440458485542e-06,
      "loss": 0.8123,
      "step": 1853
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4455303875209562,
      "learning_rate": 8.915151353915288e-06,
      "loss": 0.8477,
      "step": 1854
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2108045779988028,
      "learning_rate": 8.913861576285126e-06,
      "loss": 0.8203,
      "step": 1855
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4027943574697264,
      "learning_rate": 8.912571125816787e-06,
      "loss": 0.8705,
      "step": 1856
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2275272481157737,
      "learning_rate": 8.911280002732112e-06,
      "loss": 0.7767,
      "step": 1857
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.717119790321796,
      "learning_rate": 8.909988207253062e-06,
      "loss": 0.8675,
      "step": 1858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5734575150030203,
      "learning_rate": 8.90869573960171e-06,
      "loss": 0.8289,
      "step": 1859
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8821627535604097,
      "learning_rate": 8.907402600000248e-06,
      "loss": 0.7174,
      "step": 1860
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3279052355819654,
      "learning_rate": 8.906108788670979e-06,
      "loss": 0.8062,
      "step": 1861
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.249383726983393,
      "learning_rate": 8.904814305836326e-06,
      "loss": 0.7863,
      "step": 1862
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2722245709292035,
      "learning_rate": 8.903519151718826e-06,
      "loss": 0.7927,
      "step": 1863
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2858173286800958,
      "learning_rate": 8.90222332654113e-06,
      "loss": 0.7644,
      "step": 1864
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9247585904252018,
      "learning_rate": 8.900926830526006e-06,
      "loss": 0.8777,
      "step": 1865
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7017434301158243,
      "learning_rate": 8.899629663896336e-06,
      "loss": 0.7533,
      "step": 1866
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4282442834658569,
      "learning_rate": 8.898331826875119e-06,
      "loss": 0.736,
      "step": 1867
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.352436998329237,
      "learning_rate": 8.897033319685466e-06,
      "loss": 0.9645,
      "step": 1868
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.373215385190397,
      "learning_rate": 8.89573414255061e-06,
      "loss": 0.8792,
      "step": 1869
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.376591655294965,
      "learning_rate": 8.89443429569389e-06,
      "loss": 0.7174,
      "step": 1870
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4334854329470885,
      "learning_rate": 8.893133779338765e-06,
      "loss": 0.8263,
      "step": 1871
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4494586036172226,
      "learning_rate": 8.89183259370881e-06,
      "loss": 0.8628,
      "step": 1872
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3902465916095892,
      "learning_rate": 8.890530739027718e-06,
      "loss": 0.802,
      "step": 1873
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.795532954369307,
      "learning_rate": 8.889228215519286e-06,
      "loss": 0.7167,
      "step": 1874
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7661043152247664,
      "learning_rate": 8.887925023407437e-06,
      "loss": 0.6715,
      "step": 1875
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4863474013496565,
      "learning_rate": 8.886621162916204e-06,
      "loss": 0.856,
      "step": 1876
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4038439304855796,
      "learning_rate": 8.885316634269735e-06,
      "loss": 0.832,
      "step": 1877
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.367089486224512,
      "learning_rate": 8.884011437692295e-06,
      "loss": 0.69,
      "step": 1878
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5954486504513319,
      "learning_rate": 8.882705573408258e-06,
      "loss": 0.7592,
      "step": 1879
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6161286195919957,
      "learning_rate": 8.881399041642125e-06,
      "loss": 0.8335,
      "step": 1880
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4361016429342528,
      "learning_rate": 8.880091842618498e-06,
      "loss": 0.7719,
      "step": 1881
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4929844853374779,
      "learning_rate": 8.878783976562102e-06,
      "loss": 0.7083,
      "step": 1882
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5195169320000594,
      "learning_rate": 8.877475443697773e-06,
      "loss": 0.8195,
      "step": 1883
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4765237075320903,
      "learning_rate": 8.876166244250463e-06,
      "loss": 0.8689,
      "step": 1884
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4222340623939644,
      "learning_rate": 8.87485637844524e-06,
      "loss": 0.8572,
      "step": 1885
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8585831850428545,
      "learning_rate": 8.873545846507286e-06,
      "loss": 0.8021,
      "step": 1886
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4194589041403347,
      "learning_rate": 8.872234648661893e-06,
      "loss": 0.8753,
      "step": 1887
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8437688460893142,
      "learning_rate": 8.870922785134473e-06,
      "loss": 0.679,
      "step": 1888
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4015495658306234,
      "learning_rate": 8.869610256150552e-06,
      "loss": 0.8221,
      "step": 1889
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3180942363220305,
      "learning_rate": 8.868297061935767e-06,
      "loss": 0.8232,
      "step": 1890
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3331098814107427,
      "learning_rate": 8.86698320271587e-06,
      "loss": 0.8481,
      "step": 1891
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6753896350140252,
      "learning_rate": 8.865668678716734e-06,
      "loss": 0.7616,
      "step": 1892
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7990106282254628,
      "learning_rate": 8.864353490164335e-06,
      "loss": 0.6199,
      "step": 1893
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.396581224913173,
      "learning_rate": 8.863037637284773e-06,
      "loss": 0.8599,
      "step": 1894
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5812957644770211,
      "learning_rate": 8.861721120304257e-06,
      "loss": 0.838,
      "step": 1895
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.415857530377387,
      "learning_rate": 8.86040393944911e-06,
      "loss": 0.8962,
      "step": 1896
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4957791636317386,
      "learning_rate": 8.859086094945773e-06,
      "loss": 0.8366,
      "step": 1897
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2760484693060126,
      "learning_rate": 8.857767587020798e-06,
      "loss": 0.8131,
      "step": 1898
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.367666362783479,
      "learning_rate": 8.856448415900851e-06,
      "loss": 0.8257,
      "step": 1899
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5992048347061578,
      "learning_rate": 8.855128581812714e-06,
      "loss": 0.8454,
      "step": 1900
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5093782285727075,
      "learning_rate": 8.853808084983282e-06,
      "loss": 0.8038,
      "step": 1901
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8610904494280787,
      "learning_rate": 8.852486925639562e-06,
      "loss": 0.6632,
      "step": 1902
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3815011339944776,
      "learning_rate": 8.851165104008678e-06,
      "loss": 0.8481,
      "step": 1903
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.248356858942281,
      "learning_rate": 8.849842620317865e-06,
      "loss": 0.8784,
      "step": 1904
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6122446435193156,
      "learning_rate": 8.848519474794476e-06,
      "loss": 0.8174,
      "step": 1905
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3200879277206623,
      "learning_rate": 8.847195667665974e-06,
      "loss": 0.8634,
      "step": 1906
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9716740361854275,
      "learning_rate": 8.845871199159935e-06,
      "loss": 0.816,
      "step": 1907
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5187991041900595,
      "learning_rate": 8.844546069504051e-06,
      "loss": 0.8425,
      "step": 1908
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0643042438682295,
      "learning_rate": 8.843220278926128e-06,
      "loss": 0.8324,
      "step": 1909
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.657553067732471,
      "learning_rate": 8.841893827654087e-06,
      "loss": 0.7873,
      "step": 1910
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9514304991662225,
      "learning_rate": 8.840566715915955e-06,
      "loss": 0.6537,
      "step": 1911
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7901001713056635,
      "learning_rate": 8.839238943939883e-06,
      "loss": 0.761,
      "step": 1912
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7462611567644408,
      "learning_rate": 8.837910511954128e-06,
      "loss": 0.7872,
      "step": 1913
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5608027005068128,
      "learning_rate": 8.836581420187062e-06,
      "loss": 0.7975,
      "step": 1914
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3592414951093037,
      "learning_rate": 8.835251668867172e-06,
      "loss": 0.8357,
      "step": 1915
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2077064150327335,
      "learning_rate": 8.833921258223059e-06,
      "loss": 0.7398,
      "step": 1916
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3308475884180544,
      "learning_rate": 8.832590188483437e-06,
      "loss": 0.7677,
      "step": 1917
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.665783427336642,
      "learning_rate": 8.831258459877128e-06,
      "loss": 0.77,
      "step": 1918
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3732742987391715,
      "learning_rate": 8.829926072633075e-06,
      "loss": 0.7469,
      "step": 1919
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3700606863974332,
      "learning_rate": 8.828593026980328e-06,
      "loss": 0.8982,
      "step": 1920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4717726326297509,
      "learning_rate": 8.827259323148056e-06,
      "loss": 0.8114,
      "step": 1921
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.287780843847837,
      "learning_rate": 8.825924961365538e-06,
      "loss": 0.8061,
      "step": 1922
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4931748622778265,
      "learning_rate": 8.824589941862164e-06,
      "loss": 0.7338,
      "step": 1923
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3005591415392008,
      "learning_rate": 8.82325426486744e-06,
      "loss": 0.7762,
      "step": 1924
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3132261205311089,
      "learning_rate": 8.821917930610987e-06,
      "loss": 0.8124,
      "step": 1925
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3651362677305652,
      "learning_rate": 8.820580939322532e-06,
      "loss": 0.7809,
      "step": 1926
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2705082381784125,
      "learning_rate": 8.819243291231922e-06,
      "loss": 0.8258,
      "step": 1927
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7876243660605803,
      "learning_rate": 8.817904986569115e-06,
      "loss": 0.6448,
      "step": 1928
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3052308369232637,
      "learning_rate": 8.81656602556418e-06,
      "loss": 0.8388,
      "step": 1929
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3006739858717287,
      "learning_rate": 8.815226408447298e-06,
      "loss": 0.7538,
      "step": 1930
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4923531223532076,
      "learning_rate": 8.813886135448766e-06,
      "loss": 0.8959,
      "step": 1931
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3481018739671211,
      "learning_rate": 8.812545206798995e-06,
      "loss": 0.7761,
      "step": 1932
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2363884574757626,
      "learning_rate": 8.811203622728504e-06,
      "loss": 0.7283,
      "step": 1933
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6129363112252373,
      "learning_rate": 8.809861383467926e-06,
      "loss": 0.8676,
      "step": 1934
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.729024066736115,
      "learning_rate": 8.808518489248009e-06,
      "loss": 0.8298,
      "step": 1935
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.340773535008976,
      "learning_rate": 8.807174940299613e-06,
      "loss": 0.8056,
      "step": 1936
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.465677931533233,
      "learning_rate": 8.805830736853705e-06,
      "loss": 0.7809,
      "step": 1937
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.02831185889165,
      "learning_rate": 8.804485879141375e-06,
      "loss": 0.589,
      "step": 1938
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8459615765208223,
      "learning_rate": 8.803140367393815e-06,
      "loss": 0.7374,
      "step": 1939
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4483987020441786,
      "learning_rate": 8.801794201842337e-06,
      "loss": 0.8005,
      "step": 1940
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3096963687703005,
      "learning_rate": 8.800447382718362e-06,
      "loss": 0.787,
      "step": 1941
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.116924381375922,
      "learning_rate": 8.799099910253424e-06,
      "loss": 0.657,
      "step": 1942
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5596640654364413,
      "learning_rate": 8.797751784679167e-06,
      "loss": 0.8764,
      "step": 1943
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3122950451261808,
      "learning_rate": 8.796403006227352e-06,
      "loss": 0.7325,
      "step": 1944
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6068032009171496,
      "learning_rate": 8.795053575129846e-06,
      "loss": 0.7924,
      "step": 1945
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3585915248033738,
      "learning_rate": 8.793703491618638e-06,
      "loss": 0.8022,
      "step": 1946
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.360394625037298,
      "learning_rate": 8.792352755925817e-06,
      "loss": 0.8103,
      "step": 1947
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4018694803284213,
      "learning_rate": 8.791001368283593e-06,
      "loss": 0.7383,
      "step": 1948
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5317122198212973,
      "learning_rate": 8.789649328924286e-06,
      "loss": 0.9175,
      "step": 1949
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.379875623853965,
      "learning_rate": 8.788296638080325e-06,
      "loss": 0.7742,
      "step": 1950
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8249094919249307,
      "learning_rate": 8.786943295984254e-06,
      "loss": 0.8406,
      "step": 1951
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5063344659283886,
      "learning_rate": 8.785589302868729e-06,
      "loss": 0.7906,
      "step": 1952
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.339526180609535,
      "learning_rate": 8.784234658966514e-06,
      "loss": 0.8169,
      "step": 1953
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4244493468696786,
      "learning_rate": 8.782879364510494e-06,
      "loss": 0.7157,
      "step": 1954
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.442760648379215,
      "learning_rate": 8.781523419733655e-06,
      "loss": 0.7864,
      "step": 1955
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.277138866150591,
      "learning_rate": 8.7801668248691e-06,
      "loss": 0.7837,
      "step": 1956
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2854317575797967,
      "learning_rate": 8.778809580150044e-06,
      "loss": 0.7796,
      "step": 1957
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5599374189159232,
      "learning_rate": 8.777451685809817e-06,
      "loss": 0.8553,
      "step": 1958
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3519089111111169,
      "learning_rate": 8.77609314208185e-06,
      "loss": 0.8275,
      "step": 1959
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4544213157730461,
      "learning_rate": 8.774733949199696e-06,
      "loss": 0.7489,
      "step": 1960
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3317919191640948,
      "learning_rate": 8.773374107397017e-06,
      "loss": 0.816,
      "step": 1961
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.847314818871362,
      "learning_rate": 8.772013616907584e-06,
      "loss": 0.6078,
      "step": 1962
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.407113409981729,
      "learning_rate": 8.77065247796528e-06,
      "loss": 0.767,
      "step": 1963
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3487446522902828,
      "learning_rate": 8.769290690804104e-06,
      "loss": 0.8654,
      "step": 1964
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.467251475660194,
      "learning_rate": 8.767928255658158e-06,
      "loss": 0.7688,
      "step": 1965
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3131435357308283,
      "learning_rate": 8.766565172761666e-06,
      "loss": 0.8285,
      "step": 1966
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4889188826860151,
      "learning_rate": 8.765201442348953e-06,
      "loss": 0.8444,
      "step": 1967
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3792422574365932,
      "learning_rate": 8.763837064654464e-06,
      "loss": 0.8003,
      "step": 1968
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2302964612435368,
      "learning_rate": 8.762472039912748e-06,
      "loss": 0.7954,
      "step": 1969
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5213798689637974,
      "learning_rate": 8.76110636835847e-06,
      "loss": 0.7987,
      "step": 1970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8617085000012497,
      "learning_rate": 8.759740050226406e-06,
      "loss": 0.6576,
      "step": 1971
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.638351313334348,
      "learning_rate": 8.758373085751439e-06,
      "loss": 0.8018,
      "step": 1972
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.397558535741326,
      "learning_rate": 8.75700547516857e-06,
      "loss": 0.8104,
      "step": 1973
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.923457986008488,
      "learning_rate": 8.755637218712902e-06,
      "loss": 0.7607,
      "step": 1974
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0085629915422105,
      "learning_rate": 8.75426831661966e-06,
      "loss": 0.7941,
      "step": 1975
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.378561929090995,
      "learning_rate": 8.75289876912417e-06,
      "loss": 0.7909,
      "step": 1976
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5016724216083828,
      "learning_rate": 8.751528576461873e-06,
      "loss": 0.8033,
      "step": 1977
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6360272788632564,
      "learning_rate": 8.750157738868323e-06,
      "loss": 0.7988,
      "step": 1978
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7064654765628346,
      "learning_rate": 8.748786256579182e-06,
      "loss": 0.8005,
      "step": 1979
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3844179746963647,
      "learning_rate": 8.747414129830225e-06,
      "loss": 0.8468,
      "step": 1980
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2777868832560353,
      "learning_rate": 8.746041358857334e-06,
      "loss": 0.7774,
      "step": 1981
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7647241954991764,
      "learning_rate": 8.744667943896507e-06,
      "loss": 0.6609,
      "step": 1982
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9164235847846858,
      "learning_rate": 8.743293885183847e-06,
      "loss": 0.6297,
      "step": 1983
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8066665509958488,
      "learning_rate": 8.741919182955573e-06,
      "loss": 0.7932,
      "step": 1984
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.401552994923248,
      "learning_rate": 8.740543837448012e-06,
      "loss": 0.7851,
      "step": 1985
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3819700436180904,
      "learning_rate": 8.739167848897601e-06,
      "loss": 0.7769,
      "step": 1986
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.46751821502689,
      "learning_rate": 8.737791217540887e-06,
      "loss": 0.7873,
      "step": 1987
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.485179581240517,
      "learning_rate": 8.736413943614533e-06,
      "loss": 0.7106,
      "step": 1988
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.539301443666117,
      "learning_rate": 8.735036027355304e-06,
      "loss": 0.973,
      "step": 1989
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5552712871756353,
      "learning_rate": 8.733657469000081e-06,
      "loss": 0.7548,
      "step": 1990
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3421097210847488,
      "learning_rate": 8.732278268785856e-06,
      "loss": 0.7714,
      "step": 1991
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.00386117128086,
      "learning_rate": 8.730898426949728e-06,
      "loss": 0.8458,
      "step": 1992
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.523787892512217,
      "learning_rate": 8.729517943728909e-06,
      "loss": 0.8088,
      "step": 1993
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4362406933086376,
      "learning_rate": 8.728136819360717e-06,
      "loss": 0.7999,
      "step": 1994
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5174003559731257,
      "learning_rate": 8.726755054082589e-06,
      "loss": 0.8264,
      "step": 1995
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.421679782716405,
      "learning_rate": 8.72537264813206e-06,
      "loss": 0.7897,
      "step": 1996
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3655395550374068,
      "learning_rate": 8.723989601746785e-06,
      "loss": 0.7727,
      "step": 1997
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6208442517692827,
      "learning_rate": 8.722605915164526e-06,
      "loss": 0.765,
      "step": 1998
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5709331798272086,
      "learning_rate": 8.721221588623154e-06,
      "loss": 0.7818,
      "step": 1999
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3407427072150777,
      "learning_rate": 8.71983662236065e-06,
      "loss": 0.7455,
      "step": 2000
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5285969028704556,
      "learning_rate": 8.718451016615108e-06,
      "loss": 0.8342,
      "step": 2001
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3923695678508228,
      "learning_rate": 8.717064771624728e-06,
      "loss": 0.8412,
      "step": 2002
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3239438978710474,
      "learning_rate": 8.715677887627822e-06,
      "loss": 0.8569,
      "step": 2003
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8179888970137774,
      "learning_rate": 8.71429036486281e-06,
      "loss": 0.5652,
      "step": 2004
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.571588195688572,
      "learning_rate": 8.712902203568226e-06,
      "loss": 0.8615,
      "step": 2005
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6592525523075117,
      "learning_rate": 8.711513403982708e-06,
      "loss": 0.8348,
      "step": 2006
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8450682404957596,
      "learning_rate": 8.710123966345008e-06,
      "loss": 0.6183,
      "step": 2007
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8601081960447267,
      "learning_rate": 8.708733890893987e-06,
      "loss": 0.7263,
      "step": 2008
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3326045977702579,
      "learning_rate": 8.707343177868616e-06,
      "loss": 0.7461,
      "step": 2009
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.293327767105622,
      "learning_rate": 8.705951827507974e-06,
      "loss": 0.829,
      "step": 2010
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9622051364717408,
      "learning_rate": 8.704559840051249e-06,
      "loss": 0.7531,
      "step": 2011
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.40591596152254,
      "learning_rate": 8.70316721573774e-06,
      "loss": 0.7467,
      "step": 2012
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3915349390438254,
      "learning_rate": 8.701773954806856e-06,
      "loss": 0.7843,
      "step": 2013
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2685710722018244,
      "learning_rate": 8.700380057498115e-06,
      "loss": 0.889,
      "step": 2014
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3499545914830464,
      "learning_rate": 8.698985524051143e-06,
      "loss": 0.813,
      "step": 2015
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4069810223575974,
      "learning_rate": 8.697590354705679e-06,
      "loss": 0.7293,
      "step": 2016
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3408900585100652,
      "learning_rate": 8.696194549701564e-06,
      "loss": 0.7807,
      "step": 2017
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3572244777170233,
      "learning_rate": 8.694798109278758e-06,
      "loss": 0.8249,
      "step": 2018
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.382394287243468,
      "learning_rate": 8.693401033677322e-06,
      "loss": 0.9558,
      "step": 2019
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.344751711063741,
      "learning_rate": 8.69200332313743e-06,
      "loss": 0.7525,
      "step": 2020
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8670352532225181,
      "learning_rate": 8.690604977899369e-06,
      "loss": 0.6488,
      "step": 2021
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8403132295630269,
      "learning_rate": 8.689205998203522e-06,
      "loss": 0.705,
      "step": 2022
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2057777853410736,
      "learning_rate": 8.6878063842904e-06,
      "loss": 0.7079,
      "step": 2023
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8470173887964023,
      "learning_rate": 8.686406136400604e-06,
      "loss": 0.6806,
      "step": 2024
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7481981592205251,
      "learning_rate": 8.685005254774859e-06,
      "loss": 0.7644,
      "step": 2025
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7934237693076998,
      "learning_rate": 8.68360373965399e-06,
      "loss": 0.7893,
      "step": 2026
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3795962487092108,
      "learning_rate": 8.682201591278934e-06,
      "loss": 0.7688,
      "step": 2027
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2642211381897273,
      "learning_rate": 8.680798809890737e-06,
      "loss": 0.8657,
      "step": 2028
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3727132302805742,
      "learning_rate": 8.679395395730552e-06,
      "loss": 0.7978,
      "step": 2029
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3943959673479036,
      "learning_rate": 8.677991349039644e-06,
      "loss": 0.8368,
      "step": 2030
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.203315114950082,
      "learning_rate": 8.676586670059383e-06,
      "loss": 0.896,
      "step": 2031
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.321693387932457,
      "learning_rate": 8.675181359031253e-06,
      "loss": 0.7414,
      "step": 2032
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9809059895983547,
      "learning_rate": 8.673775416196838e-06,
      "loss": 0.9583,
      "step": 2033
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.031613337217689,
      "learning_rate": 8.672368841797842e-06,
      "loss": 0.8657,
      "step": 2034
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.86196289217309,
      "learning_rate": 8.670961636076067e-06,
      "loss": 0.7368,
      "step": 2035
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.962085929244615,
      "learning_rate": 8.669553799273429e-06,
      "loss": 0.7582,
      "step": 2036
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4126005105078536,
      "learning_rate": 8.668145331631953e-06,
      "loss": 0.8088,
      "step": 2037
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.473283325967228,
      "learning_rate": 8.666736233393769e-06,
      "loss": 0.7308,
      "step": 2038
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3114535908639657,
      "learning_rate": 8.665326504801117e-06,
      "loss": 0.8199,
      "step": 2039
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2181111597923857,
      "learning_rate": 8.663916146096348e-06,
      "loss": 0.7516,
      "step": 2040
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2420653764294556,
      "learning_rate": 8.662505157521918e-06,
      "loss": 0.7232,
      "step": 2041
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.415582800108884,
      "learning_rate": 8.66109353932039e-06,
      "loss": 0.7686,
      "step": 2042
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3223082613713042,
      "learning_rate": 8.659681291734441e-06,
      "loss": 0.8205,
      "step": 2043
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8936764230395716,
      "learning_rate": 8.658268415006853e-06,
      "loss": 0.7268,
      "step": 2044
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4707586096803964,
      "learning_rate": 8.656854909380512e-06,
      "loss": 0.8245,
      "step": 2045
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3379595779175772,
      "learning_rate": 8.655440775098418e-06,
      "loss": 0.8513,
      "step": 2046
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8868026737029349,
      "learning_rate": 8.654026012403678e-06,
      "loss": 0.8025,
      "step": 2047
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3332145959582613,
      "learning_rate": 8.652610621539505e-06,
      "loss": 0.8263,
      "step": 2048
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.340545358592753,
      "learning_rate": 8.651194602749223e-06,
      "loss": 0.7644,
      "step": 2049
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.339486694332928,
      "learning_rate": 8.649777956276257e-06,
      "loss": 0.8094,
      "step": 2050
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.533659476243655,
      "learning_rate": 8.648360682364153e-06,
      "loss": 0.8692,
      "step": 2051
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2450373503587893,
      "learning_rate": 8.646942781256548e-06,
      "loss": 0.7736,
      "step": 2052
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.805220734605671,
      "learning_rate": 8.645524253197202e-06,
      "loss": 0.8695,
      "step": 2053
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2579698732946454,
      "learning_rate": 8.644105098429975e-06,
      "loss": 0.8437,
      "step": 2054
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4540614746912002,
      "learning_rate": 8.642685317198833e-06,
      "loss": 0.7807,
      "step": 2055
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4923425196936864,
      "learning_rate": 8.64126490974786e-06,
      "loss": 0.8438,
      "step": 2056
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3995554321578103,
      "learning_rate": 8.639843876321232e-06,
      "loss": 0.7911,
      "step": 2057
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.26099346957006,
      "learning_rate": 8.638422217163244e-06,
      "loss": 0.818,
      "step": 2058
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4418937792753816,
      "learning_rate": 8.6369999325183e-06,
      "loss": 0.8102,
      "step": 2059
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3391561627032356,
      "learning_rate": 8.6355770226309e-06,
      "loss": 0.9386,
      "step": 2060
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8742186894880402,
      "learning_rate": 8.634153487745667e-06,
      "loss": 0.6978,
      "step": 2061
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.329040674145113,
      "learning_rate": 8.632729328107317e-06,
      "loss": 0.69,
      "step": 2062
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3374950320440318,
      "learning_rate": 8.63130454396068e-06,
      "loss": 0.781,
      "step": 2063
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6049331615945452,
      "learning_rate": 8.629879135550695e-06,
      "loss": 0.9486,
      "step": 2064
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7880870071662842,
      "learning_rate": 8.628453103122404e-06,
      "loss": 0.7703,
      "step": 2065
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.873258234696278,
      "learning_rate": 8.627026446920963e-06,
      "loss": 0.7794,
      "step": 2066
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3323596886651228,
      "learning_rate": 8.625599167191627e-06,
      "loss": 0.77,
      "step": 2067
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3276924109983663,
      "learning_rate": 8.624171264179761e-06,
      "loss": 0.8237,
      "step": 2068
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.441668042423098,
      "learning_rate": 8.62274273813084e-06,
      "loss": 0.803,
      "step": 2069
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4306904580533268,
      "learning_rate": 8.621313589290446e-06,
      "loss": 0.7723,
      "step": 2070
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3333653122377018,
      "learning_rate": 8.619883817904262e-06,
      "loss": 0.8105,
      "step": 2071
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4266172902592567,
      "learning_rate": 8.618453424218085e-06,
      "loss": 0.7413,
      "step": 2072
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6161436280293224,
      "learning_rate": 8.617022408477816e-06,
      "loss": 0.8356,
      "step": 2073
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7995044389101456,
      "learning_rate": 8.615590770929461e-06,
      "loss": 0.6068,
      "step": 2074
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3524448572967394,
      "learning_rate": 8.614158511819138e-06,
      "loss": 0.8187,
      "step": 2075
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.417287600114121,
      "learning_rate": 8.612725631393068e-06,
      "loss": 0.796,
      "step": 2076
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.346570671315767,
      "learning_rate": 8.611292129897581e-06,
      "loss": 0.7895,
      "step": 2077
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8086827182205027,
      "learning_rate": 8.60985800757911e-06,
      "loss": 0.7064,
      "step": 2078
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2953020530250894,
      "learning_rate": 8.608423264684198e-06,
      "loss": 0.7919,
      "step": 2079
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.479419431587533,
      "learning_rate": 8.606987901459497e-06,
      "loss": 0.7653,
      "step": 2080
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.418747991981527,
      "learning_rate": 8.605551918151755e-06,
      "loss": 0.7915,
      "step": 2081
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9276680285514827,
      "learning_rate": 8.604115315007844e-06,
      "loss": 0.6396,
      "step": 2082
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.393529823713013,
      "learning_rate": 8.602678092274725e-06,
      "loss": 0.8399,
      "step": 2083
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4730701450985138,
      "learning_rate": 8.601240250199476e-06,
      "loss": 0.7528,
      "step": 2084
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3192625878443756,
      "learning_rate": 8.599801789029281e-06,
      "loss": 0.8231,
      "step": 2085
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5292809877062419,
      "learning_rate": 8.598362709011425e-06,
      "loss": 0.7643,
      "step": 2086
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3342700563883634,
      "learning_rate": 8.596923010393303e-06,
      "loss": 0.7445,
      "step": 2087
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6067867628956298,
      "learning_rate": 8.595482693422416e-06,
      "loss": 0.82,
      "step": 2088
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5515929756759754,
      "learning_rate": 8.594041758346372e-06,
      "loss": 0.7821,
      "step": 2089
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.875489863585603,
      "learning_rate": 8.592600205412884e-06,
      "loss": 0.6683,
      "step": 2090
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.408687107991687,
      "learning_rate": 8.591158034869773e-06,
      "loss": 0.7733,
      "step": 2091
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4922325753731054,
      "learning_rate": 8.589715246964963e-06,
      "loss": 0.8825,
      "step": 2092
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.87428726428709,
      "learning_rate": 8.588271841946485e-06,
      "loss": 0.6994,
      "step": 2093
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4646981985850582,
      "learning_rate": 8.58682782006248e-06,
      "loss": 0.7845,
      "step": 2094
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9495643397817497,
      "learning_rate": 8.585383181561191e-06,
      "loss": 0.6523,
      "step": 2095
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4524548273851803,
      "learning_rate": 8.583937926690967e-06,
      "loss": 0.7897,
      "step": 2096
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6009476175145105,
      "learning_rate": 8.582492055700264e-06,
      "loss": 0.7584,
      "step": 2097
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.42706519737943,
      "learning_rate": 8.581045568837647e-06,
      "loss": 0.7909,
      "step": 2098
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2531673605194578,
      "learning_rate": 8.57959846635178e-06,
      "loss": 0.788,
      "step": 2099
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.315012783352777,
      "learning_rate": 8.578150748491438e-06,
      "loss": 0.8905,
      "step": 2100
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.558505569553684,
      "learning_rate": 8.576702415505501e-06,
      "loss": 0.701,
      "step": 2101
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4864193638001268,
      "learning_rate": 8.575253467642954e-06,
      "loss": 0.8228,
      "step": 2102
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.38577091984171,
      "learning_rate": 8.57380390515289e-06,
      "loss": 0.9015,
      "step": 2103
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3364219346384507,
      "learning_rate": 8.5723537282845e-06,
      "loss": 0.8094,
      "step": 2104
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.445040833238054,
      "learning_rate": 8.570902937287093e-06,
      "loss": 0.8201,
      "step": 2105
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4811997585111316,
      "learning_rate": 8.569451532410073e-06,
      "loss": 0.9602,
      "step": 2106
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4038302695407514,
      "learning_rate": 8.567999513902953e-06,
      "loss": 0.9254,
      "step": 2107
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.591495445087483,
      "learning_rate": 8.566546882015355e-06,
      "loss": 0.8258,
      "step": 2108
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9940064531994481,
      "learning_rate": 8.565093636996999e-06,
      "loss": 0.824,
      "step": 2109
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3318623919819284,
      "learning_rate": 8.563639779097717e-06,
      "loss": 0.8064,
      "step": 2110
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4398099913805404,
      "learning_rate": 8.562185308567443e-06,
      "loss": 0.8057,
      "step": 2111
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5964574496518646,
      "learning_rate": 8.56073022565622e-06,
      "loss": 0.7561,
      "step": 2112
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8616298574459632,
      "learning_rate": 8.55927453061419e-06,
      "loss": 0.6288,
      "step": 2113
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3147404433453298,
      "learning_rate": 8.557818223691607e-06,
      "loss": 0.7721,
      "step": 2114
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4065186207973388,
      "learning_rate": 8.556361305138825e-06,
      "loss": 0.8334,
      "step": 2115
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0039011090787695,
      "learning_rate": 8.554903775206305e-06,
      "loss": 0.8456,
      "step": 2116
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.484287482587944,
      "learning_rate": 8.553445634144614e-06,
      "loss": 0.7778,
      "step": 2117
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.841103463498516,
      "learning_rate": 8.551986882204424e-06,
      "loss": 0.6528,
      "step": 2118
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4028222215799893,
      "learning_rate": 8.550527519636511e-06,
      "loss": 0.8649,
      "step": 2119
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8550370564745459,
      "learning_rate": 8.549067546691756e-06,
      "loss": 0.6962,
      "step": 2120
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.498959004501231,
      "learning_rate": 8.547606963621146e-06,
      "loss": 0.7659,
      "step": 2121
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3373254302003472,
      "learning_rate": 8.54614577067577e-06,
      "loss": 0.818,
      "step": 2122
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.153273998923734,
      "learning_rate": 8.544683968106827e-06,
      "loss": 0.9044,
      "step": 2123
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6994741193597118,
      "learning_rate": 8.543221556165615e-06,
      "loss": 0.7807,
      "step": 2124
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3412835116887813,
      "learning_rate": 8.541758535103542e-06,
      "loss": 0.8777,
      "step": 2125
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2454214738789517,
      "learning_rate": 8.540294905172117e-06,
      "loss": 0.8451,
      "step": 2126
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.354179112925378,
      "learning_rate": 8.538830666622955e-06,
      "loss": 0.8328,
      "step": 2127
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5095739784713564,
      "learning_rate": 8.537365819707776e-06,
      "loss": 0.7343,
      "step": 2128
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3714317767487711,
      "learning_rate": 8.535900364678403e-06,
      "loss": 0.8036,
      "step": 2129
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.395114745239493,
      "learning_rate": 8.534434301786767e-06,
      "loss": 0.9104,
      "step": 2130
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.638903986519497,
      "learning_rate": 8.532967631284898e-06,
      "loss": 0.7227,
      "step": 2131
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8685710541065764,
      "learning_rate": 8.531500353424937e-06,
      "loss": 0.6561,
      "step": 2132
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8554456768072976,
      "learning_rate": 8.530032468459126e-06,
      "loss": 0.7529,
      "step": 2133
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4280969558025114,
      "learning_rate": 8.528563976639807e-06,
      "loss": 0.7659,
      "step": 2134
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6645518149161913,
      "learning_rate": 8.527094878219435e-06,
      "loss": 0.8608,
      "step": 2135
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4977247311552948,
      "learning_rate": 8.525625173450564e-06,
      "loss": 0.8323,
      "step": 2136
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0363859922681669,
      "learning_rate": 8.524154862585854e-06,
      "loss": 0.7017,
      "step": 2137
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4980181342207635,
      "learning_rate": 8.522683945878068e-06,
      "loss": 0.8063,
      "step": 2138
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.443639143391538,
      "learning_rate": 8.52121242358007e-06,
      "loss": 0.8354,
      "step": 2139
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3958751226776975,
      "learning_rate": 8.519740295944838e-06,
      "loss": 0.798,
      "step": 2140
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2906523453625156,
      "learning_rate": 8.518267563225443e-06,
      "loss": 0.9048,
      "step": 2141
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4008181246034257,
      "learning_rate": 8.51679422567507e-06,
      "loss": 0.8256,
      "step": 2142
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4170849208741305,
      "learning_rate": 8.515320283546996e-06,
      "loss": 0.8651,
      "step": 2143
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8931356788168775,
      "learning_rate": 8.513845737094613e-06,
      "loss": 0.6308,
      "step": 2144
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.406919416625592,
      "learning_rate": 8.51237058657141e-06,
      "loss": 0.8457,
      "step": 2145
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.567338128754516,
      "learning_rate": 8.510894832230988e-06,
      "loss": 0.7828,
      "step": 2146
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.581063066054542,
      "learning_rate": 8.50941847432704e-06,
      "loss": 0.8824,
      "step": 2147
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4983047362043331,
      "learning_rate": 8.507941513113372e-06,
      "loss": 0.7684,
      "step": 2148
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.423723908494687,
      "learning_rate": 8.506463948843891e-06,
      "loss": 0.801,
      "step": 2149
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8301507397558696,
      "learning_rate": 8.504985781772606e-06,
      "loss": 0.6811,
      "step": 2150
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8576236458647036,
      "learning_rate": 8.503507012153632e-06,
      "loss": 0.6465,
      "step": 2151
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3225509928710328,
      "learning_rate": 8.502027640241188e-06,
      "loss": 0.7799,
      "step": 2152
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.851873453670766,
      "learning_rate": 8.500547666289592e-06,
      "loss": 0.8467,
      "step": 2153
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5771310260386522,
      "learning_rate": 8.49906709055327e-06,
      "loss": 0.8655,
      "step": 2154
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2547210155233912,
      "learning_rate": 8.497585913286752e-06,
      "loss": 0.8557,
      "step": 2155
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2555502063397794,
      "learning_rate": 8.496104134744667e-06,
      "loss": 0.7823,
      "step": 2156
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7088925571368008,
      "learning_rate": 8.49462175518175e-06,
      "loss": 0.7888,
      "step": 2157
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3842346042590479,
      "learning_rate": 8.493138774852842e-06,
      "loss": 0.9183,
      "step": 2158
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3073521086061048,
      "learning_rate": 8.49165519401288e-06,
      "loss": 0.828,
      "step": 2159
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3400875493941502,
      "learning_rate": 8.490171012916915e-06,
      "loss": 0.8084,
      "step": 2160
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.360293007829938,
      "learning_rate": 8.488686231820089e-06,
      "loss": 0.8072,
      "step": 2161
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3032974748014712,
      "learning_rate": 8.487200850977657e-06,
      "loss": 0.919,
      "step": 2162
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.253353225521673,
      "learning_rate": 8.48571487064497e-06,
      "loss": 0.9067,
      "step": 2163
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7839073819028516,
      "learning_rate": 8.484228291077488e-06,
      "loss": 0.6522,
      "step": 2164
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3019859767502782,
      "learning_rate": 8.48274111253077e-06,
      "loss": 0.8716,
      "step": 2165
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.538377437419877,
      "learning_rate": 8.481253335260478e-06,
      "loss": 0.8521,
      "step": 2166
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.764172697031754,
      "learning_rate": 8.479764959522381e-06,
      "loss": 0.8182,
      "step": 2167
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8114596315995032,
      "learning_rate": 8.478275985572346e-06,
      "loss": 0.7255,
      "step": 2168
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8332640624702972,
      "learning_rate": 8.476786413666346e-06,
      "loss": 0.6298,
      "step": 2169
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.266179596273028,
      "learning_rate": 8.475296244060454e-06,
      "loss": 0.8403,
      "step": 2170
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.346688563152719,
      "learning_rate": 8.473805477010848e-06,
      "loss": 0.8471,
      "step": 2171
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.56909973891168,
      "learning_rate": 8.472314112773813e-06,
      "loss": 0.8112,
      "step": 2172
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3408328049637002,
      "learning_rate": 8.470822151605723e-06,
      "loss": 0.7469,
      "step": 2173
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4173437400451383,
      "learning_rate": 8.469329593763069e-06,
      "loss": 0.8956,
      "step": 2174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8585841378081065,
      "learning_rate": 8.467836439502439e-06,
      "loss": 0.6741,
      "step": 2175
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5006797358083321,
      "learning_rate": 8.46634268908052e-06,
      "loss": 0.6935,
      "step": 2176
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8500508112285095,
      "learning_rate": 8.46484834275411e-06,
      "loss": 0.7867,
      "step": 2177
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.6456487205350605,
      "learning_rate": 8.463353400780101e-06,
      "loss": 0.7437,
      "step": 2178
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5563209928382165,
      "learning_rate": 8.461857863415493e-06,
      "loss": 0.7398,
      "step": 2179
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8641013125263782,
      "learning_rate": 8.460361730917384e-06,
      "loss": 0.7373,
      "step": 2180
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9512417452562135,
      "learning_rate": 8.458865003542975e-06,
      "loss": 0.5906,
      "step": 2181
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4019121540258952,
      "learning_rate": 8.457367681549577e-06,
      "loss": 0.8015,
      "step": 2182
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8003861216788758,
      "learning_rate": 8.455869765194592e-06,
      "loss": 0.6533,
      "step": 2183
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3218167308406787,
      "learning_rate": 8.45437125473553e-06,
      "loss": 0.7471,
      "step": 2184
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5016384575316686,
      "learning_rate": 8.452872150430002e-06,
      "loss": 0.8304,
      "step": 2185
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.554310826535067,
      "learning_rate": 8.451372452535724e-06,
      "loss": 0.7203,
      "step": 2186
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1839239268691497,
      "learning_rate": 8.44987216131051e-06,
      "loss": 0.7599,
      "step": 2187
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2833948941338968,
      "learning_rate": 8.448371277012275e-06,
      "loss": 0.7286,
      "step": 2188
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3202791893653665,
      "learning_rate": 8.446869799899042e-06,
      "loss": 0.7488,
      "step": 2189
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4123339288740049,
      "learning_rate": 8.44536773022893e-06,
      "loss": 0.7893,
      "step": 2190
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7958402052113562,
      "learning_rate": 8.443865068260164e-06,
      "loss": 0.9079,
      "step": 2191
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.475965878529243,
      "learning_rate": 8.442361814251069e-06,
      "loss": 0.8121,
      "step": 2192
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4146622273838145,
      "learning_rate": 8.440857968460068e-06,
      "loss": 0.7801,
      "step": 2193
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2893442149435879,
      "learning_rate": 8.439353531145695e-06,
      "loss": 0.8511,
      "step": 2194
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8659090094894247,
      "learning_rate": 8.437848502566576e-06,
      "loss": 0.7469,
      "step": 2195
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8310002389228373,
      "learning_rate": 8.436342882981445e-06,
      "loss": 0.6425,
      "step": 2196
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2772922994647593,
      "learning_rate": 8.434836672649134e-06,
      "loss": 0.7402,
      "step": 2197
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7784663336550999,
      "learning_rate": 8.433329871828582e-06,
      "loss": 0.5792,
      "step": 2198
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3156778616292923,
      "learning_rate": 8.431822480778818e-06,
      "loss": 0.7762,
      "step": 2199
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3715449025310888,
      "learning_rate": 8.430314499758986e-06,
      "loss": 0.7692,
      "step": 2200
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6851540008911177,
      "learning_rate": 8.428805929028327e-06,
      "loss": 0.7309,
      "step": 2201
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.861973447216193,
      "learning_rate": 8.427296768846176e-06,
      "loss": 0.7471,
      "step": 2202
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.335267000713049,
      "learning_rate": 8.425787019471979e-06,
      "loss": 0.8535,
      "step": 2203
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7905235330338373,
      "learning_rate": 8.424276681165276e-06,
      "loss": 0.6137,
      "step": 2204
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.334265240920781,
      "learning_rate": 8.422765754185716e-06,
      "loss": 0.7947,
      "step": 2205
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6192215884194299,
      "learning_rate": 8.421254238793041e-06,
      "loss": 0.7501,
      "step": 2206
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5498012218257793,
      "learning_rate": 8.419742135247099e-06,
      "loss": 0.8146,
      "step": 2207
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5220904874657633,
      "learning_rate": 8.418229443807842e-06,
      "loss": 0.7169,
      "step": 2208
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2755277546005406,
      "learning_rate": 8.416716164735315e-06,
      "loss": 0.7756,
      "step": 2209
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8272482764589647,
      "learning_rate": 8.415202298289668e-06,
      "loss": 0.6503,
      "step": 2210
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.069285451387922,
      "learning_rate": 8.413687844731155e-06,
      "loss": 0.828,
      "step": 2211
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7652836532786267,
      "learning_rate": 8.412172804320127e-06,
      "loss": 0.6588,
      "step": 2212
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9703118916355005,
      "learning_rate": 8.410657177317035e-06,
      "loss": 0.7792,
      "step": 2213
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4716705126387164,
      "learning_rate": 8.409140963982436e-06,
      "loss": 0.7864,
      "step": 2214
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2212657222616334,
      "learning_rate": 8.407624164576982e-06,
      "loss": 0.8405,
      "step": 2215
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3509592327250102,
      "learning_rate": 8.406106779361429e-06,
      "loss": 0.8888,
      "step": 2216
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.528689181183049,
      "learning_rate": 8.404588808596635e-06,
      "loss": 0.7992,
      "step": 2217
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.33603589245025,
      "learning_rate": 8.403070252543555e-06,
      "loss": 0.7915,
      "step": 2218
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.273785287697819,
      "learning_rate": 8.401551111463246e-06,
      "loss": 0.7499,
      "step": 2219
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.564866924481155,
      "learning_rate": 8.400031385616868e-06,
      "loss": 0.8197,
      "step": 2220
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5081233028286267,
      "learning_rate": 8.398511075265677e-06,
      "loss": 0.8439,
      "step": 2221
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3445445470091473,
      "learning_rate": 8.396990180671034e-06,
      "loss": 0.8053,
      "step": 2222
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3537339528649606,
      "learning_rate": 8.395468702094399e-06,
      "loss": 0.8445,
      "step": 2223
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2355150400743984,
      "learning_rate": 8.393946639797328e-06,
      "loss": 0.8121,
      "step": 2224
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.26607350933799,
      "learning_rate": 8.392423994041486e-06,
      "loss": 0.8628,
      "step": 2225
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.43464418109761,
      "learning_rate": 8.39090076508863e-06,
      "loss": 0.7736,
      "step": 2226
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1030588569083135,
      "learning_rate": 8.389376953200622e-06,
      "loss": 0.7595,
      "step": 2227
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3842390698731692,
      "learning_rate": 8.387852558639422e-06,
      "loss": 0.7681,
      "step": 2228
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3386479388451238,
      "learning_rate": 8.386327581667091e-06,
      "loss": 0.8278,
      "step": 2229
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4600323931995043,
      "learning_rate": 8.384802022545793e-06,
      "loss": 0.8372,
      "step": 2230
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3112270840836995,
      "learning_rate": 8.383275881537786e-06,
      "loss": 0.8417,
      "step": 2231
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6363581513320695,
      "learning_rate": 8.381749158905433e-06,
      "loss": 0.857,
      "step": 2232
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4844655667662026,
      "learning_rate": 8.380221854911195e-06,
      "loss": 0.826,
      "step": 2233
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2067098007158685,
      "learning_rate": 8.378693969817633e-06,
      "loss": 0.8192,
      "step": 2234
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2828999698330872,
      "learning_rate": 8.37716550388741e-06,
      "loss": 0.8174,
      "step": 2235
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8435958081611732,
      "learning_rate": 8.375636457383282e-06,
      "loss": 0.7329,
      "step": 2236
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3667859312208288,
      "learning_rate": 8.374106830568117e-06,
      "loss": 0.7967,
      "step": 2237
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.316247964797869,
      "learning_rate": 8.372576623704872e-06,
      "loss": 0.7047,
      "step": 2238
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.542699997435124,
      "learning_rate": 8.371045837056603e-06,
      "loss": 0.8852,
      "step": 2239
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.344392672292014,
      "learning_rate": 8.369514470886478e-06,
      "loss": 0.7863,
      "step": 2240
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3930818002708996,
      "learning_rate": 8.36798252545775e-06,
      "loss": 0.7987,
      "step": 2241
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4262689288332742,
      "learning_rate": 8.366450001033784e-06,
      "loss": 0.8456,
      "step": 2242
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3621680718347369,
      "learning_rate": 8.364916897878033e-06,
      "loss": 0.8098,
      "step": 2243
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.368483216087356,
      "learning_rate": 8.363383216254058e-06,
      "loss": 0.7588,
      "step": 2244
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6466609948660043,
      "learning_rate": 8.361848956425516e-06,
      "loss": 0.7673,
      "step": 2245
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4909494179981935,
      "learning_rate": 8.360314118656165e-06,
      "loss": 0.8047,
      "step": 2246
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3723133237587288,
      "learning_rate": 8.358778703209862e-06,
      "loss": 0.7797,
      "step": 2247
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.882318186515981,
      "learning_rate": 8.357242710350561e-06,
      "loss": 0.6915,
      "step": 2248
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.770737349432316,
      "learning_rate": 8.355706140342317e-06,
      "loss": 0.7927,
      "step": 2249
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.440208034400605,
      "learning_rate": 8.354168993449285e-06,
      "loss": 0.8624,
      "step": 2250
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.659530550790231,
      "learning_rate": 8.352631269935719e-06,
      "loss": 0.7096,
      "step": 2251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4942868056932601,
      "learning_rate": 8.35109297006597e-06,
      "loss": 0.8834,
      "step": 2252
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4032792308760003,
      "learning_rate": 8.349554094104491e-06,
      "loss": 0.7491,
      "step": 2253
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7575110036443742,
      "learning_rate": 8.348014642315831e-06,
      "loss": 0.8891,
      "step": 2254
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9053602756097442,
      "learning_rate": 8.346474614964642e-06,
      "loss": 0.6139,
      "step": 2255
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6271582442344736,
      "learning_rate": 8.34493401231567e-06,
      "loss": 0.8614,
      "step": 2256
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2993157923323992,
      "learning_rate": 8.34339283463376e-06,
      "loss": 0.7415,
      "step": 2257
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.475110590335809,
      "learning_rate": 8.341851082183868e-06,
      "loss": 0.7072,
      "step": 2258
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.453941148991723,
      "learning_rate": 8.340308755231027e-06,
      "loss": 0.7579,
      "step": 2259
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3891276051584731,
      "learning_rate": 8.338765854040391e-06,
      "loss": 0.832,
      "step": 2260
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4262555987624206,
      "learning_rate": 8.337222378877196e-06,
      "loss": 0.7856,
      "step": 2261
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4238160758713136,
      "learning_rate": 8.33567833000679e-06,
      "loss": 0.7715,
      "step": 2262
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5312647221798439,
      "learning_rate": 8.334133707694603e-06,
      "loss": 0.8525,
      "step": 2263
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5053294623626297,
      "learning_rate": 8.33258851220618e-06,
      "loss": 0.8415,
      "step": 2264
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3474169413584651,
      "learning_rate": 8.33104274380716e-06,
      "loss": 0.8554,
      "step": 2265
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4490904469691668,
      "learning_rate": 8.329496402763275e-06,
      "loss": 0.8482,
      "step": 2266
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8663232048644132,
      "learning_rate": 8.327949489340359e-06,
      "loss": 0.6552,
      "step": 2267
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7290938811932373,
      "learning_rate": 8.326402003804344e-06,
      "loss": 0.7932,
      "step": 2268
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.341463161555033,
      "learning_rate": 8.324853946421261e-06,
      "loss": 0.7918,
      "step": 2269
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5979882620055466,
      "learning_rate": 8.323305317457241e-06,
      "loss": 0.8158,
      "step": 2270
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.238372310866884,
      "learning_rate": 8.32175611717851e-06,
      "loss": 0.8023,
      "step": 2271
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.601073109139352,
      "learning_rate": 8.320206345851393e-06,
      "loss": 0.8242,
      "step": 2272
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5338986594055843,
      "learning_rate": 8.318656003742314e-06,
      "loss": 0.7619,
      "step": 2273
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3552647790213428,
      "learning_rate": 8.317105091117795e-06,
      "loss": 0.793,
      "step": 2274
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3576807086050529,
      "learning_rate": 8.315553608244453e-06,
      "loss": 0.9265,
      "step": 2275
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.00776129906446,
      "learning_rate": 8.314001555389014e-06,
      "loss": 0.6449,
      "step": 2276
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3729343271114813,
      "learning_rate": 8.312448932818284e-06,
      "loss": 0.8168,
      "step": 2277
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4850749398160135,
      "learning_rate": 8.310895740799181e-06,
      "loss": 0.9149,
      "step": 2278
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3701137896486173,
      "learning_rate": 8.30934197959872e-06,
      "loss": 0.9044,
      "step": 2279
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9540839378713852,
      "learning_rate": 8.307787649484005e-06,
      "loss": 0.8538,
      "step": 2280
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4131001366144822,
      "learning_rate": 8.306232750722248e-06,
      "loss": 0.7859,
      "step": 2281
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.353495084379699,
      "learning_rate": 8.30467728358075e-06,
      "loss": 0.9145,
      "step": 2282
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5204121289543748,
      "learning_rate": 8.303121248326917e-06,
      "loss": 0.6965,
      "step": 2283
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3661576049397257,
      "learning_rate": 8.301564645228249e-06,
      "loss": 0.7315,
      "step": 2284
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4295743068297575,
      "learning_rate": 8.300007474552343e-06,
      "loss": 0.7949,
      "step": 2285
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4809831151352468,
      "learning_rate": 8.298449736566894e-06,
      "loss": 0.7827,
      "step": 2286
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9084174990007927,
      "learning_rate": 8.296891431539696e-06,
      "loss": 0.6984,
      "step": 2287
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.525400770322519,
      "learning_rate": 8.29533255973864e-06,
      "loss": 0.8184,
      "step": 2288
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.309604449830263,
      "learning_rate": 8.293773121431717e-06,
      "loss": 0.8108,
      "step": 2289
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3841391882209166,
      "learning_rate": 8.292213116887008e-06,
      "loss": 0.8011,
      "step": 2290
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4574719689831528,
      "learning_rate": 8.290652546372698e-06,
      "loss": 0.861,
      "step": 2291
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7873939156850784,
      "learning_rate": 8.289091410157067e-06,
      "loss": 0.8044,
      "step": 2292
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3480124854168425,
      "learning_rate": 8.28752970850849e-06,
      "loss": 0.8361,
      "step": 2293
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4054359203701163,
      "learning_rate": 8.285967441695445e-06,
      "loss": 0.7501,
      "step": 2294
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6143807917781223,
      "learning_rate": 8.284404609986505e-06,
      "loss": 0.8085,
      "step": 2295
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3600533437723443,
      "learning_rate": 8.282841213650334e-06,
      "loss": 0.8515,
      "step": 2296
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3177145306087716,
      "learning_rate": 8.281277252955699e-06,
      "loss": 0.7441,
      "step": 2297
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4123103192660296,
      "learning_rate": 8.279712728171468e-06,
      "loss": 0.7394,
      "step": 2298
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5809925276967514,
      "learning_rate": 8.278147639566596e-06,
      "loss": 0.6826,
      "step": 2299
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.404524365784919,
      "learning_rate": 8.27658198741014e-06,
      "loss": 0.925,
      "step": 2300
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2733895060316196,
      "learning_rate": 8.275015771971255e-06,
      "loss": 0.8174,
      "step": 2301
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3254270258959817,
      "learning_rate": 8.273448993519194e-06,
      "loss": 0.8534,
      "step": 2302
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6205585128468927,
      "learning_rate": 8.2718816523233e-06,
      "loss": 0.8851,
      "step": 2303
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8129954401331755,
      "learning_rate": 8.270313748653018e-06,
      "loss": 0.6912,
      "step": 2304
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7705557698889676,
      "learning_rate": 8.268745282777893e-06,
      "loss": 0.6661,
      "step": 2305
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2594245775715662,
      "learning_rate": 8.267176254967556e-06,
      "loss": 0.785,
      "step": 2306
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9673410584059314,
      "learning_rate": 8.265606665491746e-06,
      "loss": 0.7855,
      "step": 2307
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2777142251324713,
      "learning_rate": 8.264036514620292e-06,
      "loss": 0.8127,
      "step": 2308
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3456847464880508,
      "learning_rate": 8.262465802623122e-06,
      "loss": 0.7973,
      "step": 2309
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3426515179934966,
      "learning_rate": 8.260894529770258e-06,
      "loss": 0.8631,
      "step": 2310
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3118056264498354,
      "learning_rate": 8.25932269633182e-06,
      "loss": 0.7769,
      "step": 2311
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.451000262433385,
      "learning_rate": 8.257750302578027e-06,
      "loss": 0.7774,
      "step": 2312
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3546746542729942,
      "learning_rate": 8.256177348779188e-06,
      "loss": 0.8211,
      "step": 2313
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7350006325325924,
      "learning_rate": 8.254603835205715e-06,
      "loss": 0.9088,
      "step": 2314
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5305657894743274,
      "learning_rate": 8.253029762128111e-06,
      "loss": 0.7918,
      "step": 2315
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0678321583954211,
      "learning_rate": 8.25145512981698e-06,
      "loss": 0.7046,
      "step": 2316
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3354350008598652,
      "learning_rate": 8.249879938543017e-06,
      "loss": 0.7623,
      "step": 2317
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4000854189366303,
      "learning_rate": 8.248304188577018e-06,
      "loss": 0.893,
      "step": 2318
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4084229362490799,
      "learning_rate": 8.24672788018987e-06,
      "loss": 0.7764,
      "step": 2319
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4624191267727407,
      "learning_rate": 8.245151013652561e-06,
      "loss": 0.8739,
      "step": 2320
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4909844192304018,
      "learning_rate": 8.24357358923617e-06,
      "loss": 0.7712,
      "step": 2321
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6214151166370034,
      "learning_rate": 8.241995607211878e-06,
      "loss": 0.8104,
      "step": 2322
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.6991010027597886,
      "learning_rate": 8.240417067850957e-06,
      "loss": 0.8478,
      "step": 2323
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4172992074953856,
      "learning_rate": 8.238837971424776e-06,
      "loss": 0.7639,
      "step": 2324
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5394817876117703,
      "learning_rate": 8.2372583182048e-06,
      "loss": 0.7612,
      "step": 2325
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3941517800791539,
      "learning_rate": 8.235678108462589e-06,
      "loss": 0.7683,
      "step": 2326
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3756779682337164,
      "learning_rate": 8.2340973424698e-06,
      "loss": 0.7726,
      "step": 2327
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4339723837333798,
      "learning_rate": 8.232516020498184e-06,
      "loss": 0.7313,
      "step": 2328
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8410408911759287,
      "learning_rate": 8.230934142819588e-06,
      "loss": 0.6703,
      "step": 2329
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2800311993182953,
      "learning_rate": 8.229351709705961e-06,
      "loss": 0.8729,
      "step": 2330
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6414595965104202,
      "learning_rate": 8.227768721429334e-06,
      "loss": 0.8142,
      "step": 2331
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2227401595289666,
      "learning_rate": 8.226185178261846e-06,
      "loss": 0.7609,
      "step": 2332
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2212107313099823,
      "learning_rate": 8.224601080475723e-06,
      "loss": 0.7501,
      "step": 2333
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3160790254638686,
      "learning_rate": 8.223016428343294e-06,
      "loss": 0.7209,
      "step": 2334
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3891395047883954,
      "learning_rate": 8.221431222136976e-06,
      "loss": 0.7914,
      "step": 2335
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6548162664511488,
      "learning_rate": 8.219845462129284e-06,
      "loss": 0.9397,
      "step": 2336
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3941080418285021,
      "learning_rate": 8.218259148592828e-06,
      "loss": 0.8429,
      "step": 2337
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2919986605701397,
      "learning_rate": 8.216672281800317e-06,
      "loss": 0.7466,
      "step": 2338
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5647572847653552,
      "learning_rate": 8.21508486202455e-06,
      "loss": 0.8755,
      "step": 2339
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7872891885189686,
      "learning_rate": 8.213496889538422e-06,
      "loss": 0.6865,
      "step": 2340
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3265557580333336,
      "learning_rate": 8.211908364614924e-06,
      "loss": 0.8052,
      "step": 2341
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3193706346310319,
      "learning_rate": 8.210319287527143e-06,
      "loss": 0.763,
      "step": 2342
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3855944149313502,
      "learning_rate": 8.20872965854826e-06,
      "loss": 0.7201,
      "step": 2343
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5655460767589844,
      "learning_rate": 8.207139477951549e-06,
      "loss": 0.7723,
      "step": 2344
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.395487027587227,
      "learning_rate": 8.205548746010383e-06,
      "loss": 0.8218,
      "step": 2345
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7329790848018265,
      "learning_rate": 8.203957462998225e-06,
      "loss": 0.9405,
      "step": 2346
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4430310870181748,
      "learning_rate": 8.202365629188634e-06,
      "loss": 0.8389,
      "step": 2347
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.988766607942873,
      "learning_rate": 8.200773244855267e-06,
      "loss": 0.6851,
      "step": 2348
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3428067752896333,
      "learning_rate": 8.199180310271873e-06,
      "loss": 0.8519,
      "step": 2349
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8429976217227648,
      "learning_rate": 8.197586825712295e-06,
      "loss": 0.7536,
      "step": 2350
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4081939329655435,
      "learning_rate": 8.195992791450475e-06,
      "loss": 0.8687,
      "step": 2351
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4237842408900871,
      "learning_rate": 8.19439820776044e-06,
      "loss": 0.7486,
      "step": 2352
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2695400200852034,
      "learning_rate": 8.19280307491632e-06,
      "loss": 0.9126,
      "step": 2353
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4041100889779212,
      "learning_rate": 8.19120739319234e-06,
      "loss": 0.8253,
      "step": 2354
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3360557996911675,
      "learning_rate": 8.189611162862811e-06,
      "loss": 0.8009,
      "step": 2355
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.644334590261672,
      "learning_rate": 8.188014384202148e-06,
      "loss": 0.764,
      "step": 2356
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.539843340738506,
      "learning_rate": 8.186417057484851e-06,
      "loss": 0.6924,
      "step": 2357
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4625211502049331,
      "learning_rate": 8.184819182985524e-06,
      "loss": 0.6799,
      "step": 2358
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6729661868507695,
      "learning_rate": 8.183220760978858e-06,
      "loss": 0.8026,
      "step": 2359
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3351297444277919,
      "learning_rate": 8.18162179173964e-06,
      "loss": 0.8064,
      "step": 2360
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.430868064105706,
      "learning_rate": 8.18002227554275e-06,
      "loss": 0.8329,
      "step": 2361
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4886248932154864,
      "learning_rate": 8.178422212663166e-06,
      "loss": 0.825,
      "step": 2362
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4664814102051533,
      "learning_rate": 8.176821603375955e-06,
      "loss": 0.8318,
      "step": 2363
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3794809798660974,
      "learning_rate": 8.175220447956282e-06,
      "loss": 0.8365,
      "step": 2364
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9786539402694958,
      "learning_rate": 8.173618746679406e-06,
      "loss": 0.8196,
      "step": 2365
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7224702505032645,
      "learning_rate": 8.172016499820672e-06,
      "loss": 0.8195,
      "step": 2366
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6586270657304403,
      "learning_rate": 8.170413707655532e-06,
      "loss": 0.7315,
      "step": 2367
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.531152881307162,
      "learning_rate": 8.168810370459519e-06,
      "loss": 0.84,
      "step": 2368
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4678527963893673,
      "learning_rate": 8.167206488508268e-06,
      "loss": 0.7829,
      "step": 2369
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.399111038221551,
      "learning_rate": 8.165602062077502e-06,
      "loss": 0.8283,
      "step": 2370
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3008515588433913,
      "learning_rate": 8.163997091443046e-06,
      "loss": 0.8509,
      "step": 2371
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4461095704908558,
      "learning_rate": 8.162391576880808e-06,
      "loss": 0.7483,
      "step": 2372
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8429056661664451,
      "learning_rate": 8.160785518666795e-06,
      "loss": 0.6342,
      "step": 2373
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.267801407629536,
      "learning_rate": 8.159178917077112e-06,
      "loss": 0.8283,
      "step": 2374
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4664244628511351,
      "learning_rate": 8.157571772387947e-06,
      "loss": 0.7443,
      "step": 2375
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.235437899112933,
      "learning_rate": 8.155964084875587e-06,
      "loss": 0.7828,
      "step": 2376
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5128566247109114,
      "learning_rate": 8.154355854816416e-06,
      "loss": 0.7558,
      "step": 2377
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4184962446428724,
      "learning_rate": 8.152747082486905e-06,
      "loss": 0.7891,
      "step": 2378
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2766189061528594,
      "learning_rate": 8.15113776816362e-06,
      "loss": 0.7833,
      "step": 2379
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2698188034015445,
      "learning_rate": 8.14952791212322e-06,
      "loss": 0.7409,
      "step": 2380
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3498844385315727,
      "learning_rate": 8.147917514642462e-06,
      "loss": 0.71,
      "step": 2381
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2636497553097086,
      "learning_rate": 8.146306575998188e-06,
      "loss": 0.7985,
      "step": 2382
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6514404719075122,
      "learning_rate": 8.14469509646734e-06,
      "loss": 0.8055,
      "step": 2383
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.307602190216505,
      "learning_rate": 8.143083076326947e-06,
      "loss": 0.7873,
      "step": 2384
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.571853080215751,
      "learning_rate": 8.141470515854137e-06,
      "loss": 0.7336,
      "step": 2385
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3851547008449612,
      "learning_rate": 8.139857415326125e-06,
      "loss": 0.8914,
      "step": 2386
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2865399411633973,
      "learning_rate": 8.138243775020222e-06,
      "loss": 0.7932,
      "step": 2387
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5782658241458944,
      "learning_rate": 8.136629595213834e-06,
      "loss": 0.8254,
      "step": 2388
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3454524478734367,
      "learning_rate": 8.135014876184454e-06,
      "loss": 0.8804,
      "step": 2389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6614609301803263,
      "learning_rate": 8.133399618209675e-06,
      "loss": 0.8946,
      "step": 2390
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2934582769281522,
      "learning_rate": 8.131783821567175e-06,
      "loss": 0.8015,
      "step": 2391
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9525809811944814,
      "learning_rate": 8.13016748653473e-06,
      "loss": 0.766,
      "step": 2392
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8478732056971658,
      "learning_rate": 8.128550613390205e-06,
      "loss": 0.6706,
      "step": 2393
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4237938150570613,
      "learning_rate": 8.12693320241156e-06,
      "loss": 0.8047,
      "step": 2394
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.454049000291558,
      "learning_rate": 8.12531525387685e-06,
      "loss": 0.8521,
      "step": 2395
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2507920385122517,
      "learning_rate": 8.123696768064212e-06,
      "loss": 0.817,
      "step": 2396
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2951343205125252,
      "learning_rate": 8.122077745251888e-06,
      "loss": 0.7049,
      "step": 2397
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7984334512870773,
      "learning_rate": 8.120458185718206e-06,
      "loss": 0.6443,
      "step": 2398
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.290308506123364,
      "learning_rate": 8.118838089741585e-06,
      "loss": 0.8835,
      "step": 2399
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5189366605985557,
      "learning_rate": 8.117217457600541e-06,
      "loss": 0.7772,
      "step": 2400
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8289423472618078,
      "learning_rate": 8.11559628957368e-06,
      "loss": 0.604,
      "step": 2401
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1703861712568797,
      "learning_rate": 8.113974585939694e-06,
      "loss": 0.7804,
      "step": 2402
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5129455223220523,
      "learning_rate": 8.112352346977378e-06,
      "loss": 0.8283,
      "step": 2403
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4510410332646724,
      "learning_rate": 8.110729572965613e-06,
      "loss": 0.7437,
      "step": 2404
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8061906376001642,
      "learning_rate": 8.109106264183369e-06,
      "loss": 0.6277,
      "step": 2405
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5672629968721532,
      "learning_rate": 8.107482420909719e-06,
      "loss": 0.8956,
      "step": 2406
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2223347807094922,
      "learning_rate": 8.105858043423811e-06,
      "loss": 0.8136,
      "step": 2407
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5338279079623107,
      "learning_rate": 8.104233132004902e-06,
      "loss": 0.8056,
      "step": 2408
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5854941771156947,
      "learning_rate": 8.10260768693233e-06,
      "loss": 0.7762,
      "step": 2409
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4090773971392279,
      "learning_rate": 8.100981708485527e-06,
      "loss": 0.8024,
      "step": 2410
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.267257988651179,
      "learning_rate": 8.09935519694402e-06,
      "loss": 0.7733,
      "step": 2411
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2680637948340845,
      "learning_rate": 8.09772815258742e-06,
      "loss": 0.8015,
      "step": 2412
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2700525863262466,
      "learning_rate": 8.096100575695443e-06,
      "loss": 0.8177,
      "step": 2413
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8275860708813179,
      "learning_rate": 8.094472466547882e-06,
      "loss": 0.7359,
      "step": 2414
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8860345277215833,
      "learning_rate": 8.09284382542463e-06,
      "loss": 0.7325,
      "step": 2415
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.223048559903839,
      "learning_rate": 8.09121465260567e-06,
      "loss": 0.8411,
      "step": 2416
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3054132012797925,
      "learning_rate": 8.089584948371074e-06,
      "loss": 0.7622,
      "step": 2417
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2640249210441379,
      "learning_rate": 8.087954713001007e-06,
      "loss": 0.8058,
      "step": 2418
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.294794042779225,
      "learning_rate": 8.086323946775727e-06,
      "loss": 0.7709,
      "step": 2419
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.247525589737122,
      "learning_rate": 8.084692649975583e-06,
      "loss": 0.7605,
      "step": 2420
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.284246759812064,
      "learning_rate": 8.083060822881008e-06,
      "loss": 0.7785,
      "step": 2421
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.388033644573606,
      "learning_rate": 8.081428465772539e-06,
      "loss": 0.7909,
      "step": 2422
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3065208942814257,
      "learning_rate": 8.079795578930792e-06,
      "loss": 0.7381,
      "step": 2423
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.46629070027989,
      "learning_rate": 8.07816216263648e-06,
      "loss": 0.8301,
      "step": 2424
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.668745532338149,
      "learning_rate": 8.076528217170408e-06,
      "loss": 0.772,
      "step": 2425
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2957422311746603,
      "learning_rate": 8.07489374281347e-06,
      "loss": 0.7803,
      "step": 2426
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3706070772418233,
      "learning_rate": 8.07325873984665e-06,
      "loss": 0.8722,
      "step": 2427
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7799560346217005,
      "learning_rate": 8.071623208551023e-06,
      "loss": 0.7482,
      "step": 2428
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3203114121503383,
      "learning_rate": 8.069987149207759e-06,
      "loss": 0.81,
      "step": 2429
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3836823232748532,
      "learning_rate": 8.068350562098113e-06,
      "loss": 0.7631,
      "step": 2430
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3535206241573532,
      "learning_rate": 8.066713447503434e-06,
      "loss": 0.8715,
      "step": 2431
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5437911403902187,
      "learning_rate": 8.065075805705161e-06,
      "loss": 0.7839,
      "step": 2432
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.317499738257446,
      "learning_rate": 8.063437636984824e-06,
      "loss": 0.7845,
      "step": 2433
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.357458574776688,
      "learning_rate": 8.061798941624041e-06,
      "loss": 0.8832,
      "step": 2434
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3156318777462501,
      "learning_rate": 8.060159719904526e-06,
      "loss": 0.7348,
      "step": 2435
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4322672741482443,
      "learning_rate": 8.058519972108078e-06,
      "loss": 0.7581,
      "step": 2436
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4148777350943234,
      "learning_rate": 8.05687969851659e-06,
      "loss": 0.7691,
      "step": 2437
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.391817350916005,
      "learning_rate": 8.055238899412046e-06,
      "loss": 0.7763,
      "step": 2438
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5019109367279493,
      "learning_rate": 8.053597575076513e-06,
      "loss": 0.7585,
      "step": 2439
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5611324593753642,
      "learning_rate": 8.051955725792155e-06,
      "loss": 0.8742,
      "step": 2440
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4266793976428667,
      "learning_rate": 8.050313351841229e-06,
      "loss": 0.8308,
      "step": 2441
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9207020177980754,
      "learning_rate": 8.048670453506074e-06,
      "loss": 0.6511,
      "step": 2442
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.312304718650565,
      "learning_rate": 8.047027031069126e-06,
      "loss": 0.7538,
      "step": 2443
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2929957740970466,
      "learning_rate": 8.045383084812907e-06,
      "loss": 0.8114,
      "step": 2444
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6131396066869628,
      "learning_rate": 8.04373861502003e-06,
      "loss": 0.7582,
      "step": 2445
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3271479925702487,
      "learning_rate": 8.0420936219732e-06,
      "loss": 0.7785,
      "step": 2446
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1849186872417514,
      "learning_rate": 8.040448105955209e-06,
      "loss": 0.6865,
      "step": 2447
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4386217172875724,
      "learning_rate": 8.03880206724894e-06,
      "loss": 0.856,
      "step": 2448
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3033827917466068,
      "learning_rate": 8.037155506137367e-06,
      "loss": 0.8224,
      "step": 2449
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8763694588895505,
      "learning_rate": 8.035508422903554e-06,
      "loss": 0.706,
      "step": 2450
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4390956224647466,
      "learning_rate": 8.033860817830651e-06,
      "loss": 0.7844,
      "step": 2451
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9734706044015349,
      "learning_rate": 8.0322126912019e-06,
      "loss": 0.7556,
      "step": 2452
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2925115970927787,
      "learning_rate": 8.03056404330064e-06,
      "loss": 0.7353,
      "step": 2453
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.484521839336688,
      "learning_rate": 8.028914874410284e-06,
      "loss": 0.809,
      "step": 2454
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3238655025868467,
      "learning_rate": 8.027265184814349e-06,
      "loss": 0.754,
      "step": 2455
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7573570429335603,
      "learning_rate": 8.025614974796432e-06,
      "loss": 0.6642,
      "step": 2456
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5409957631790685,
      "learning_rate": 8.023964244640225e-06,
      "loss": 0.8798,
      "step": 2457
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8355570367990048,
      "learning_rate": 8.022312994629508e-06,
      "loss": 0.9302,
      "step": 2458
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3872651502292284,
      "learning_rate": 8.020661225048149e-06,
      "loss": 0.8093,
      "step": 2459
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3105599897918199,
      "learning_rate": 8.019008936180108e-06,
      "loss": 0.8563,
      "step": 2460
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8599480454704718,
      "learning_rate": 8.017356128309432e-06,
      "loss": 0.6047,
      "step": 2461
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3181099888272279,
      "learning_rate": 8.015702801720255e-06,
      "loss": 0.7571,
      "step": 2462
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.443322449528398,
      "learning_rate": 8.014048956696807e-06,
      "loss": 0.7977,
      "step": 2463
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7966377933166,
      "learning_rate": 8.012394593523403e-06,
      "loss": 0.6914,
      "step": 2464
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4585793134287277,
      "learning_rate": 8.010739712484443e-06,
      "loss": 0.818,
      "step": 2465
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3060842090460727,
      "learning_rate": 8.009084313864424e-06,
      "loss": 0.8687,
      "step": 2466
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2832373671549033,
      "learning_rate": 8.007428397947926e-06,
      "loss": 0.762,
      "step": 2467
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5266495968775964,
      "learning_rate": 8.005771965019622e-06,
      "loss": 0.8291,
      "step": 2468
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3040481419622332,
      "learning_rate": 8.004115015364273e-06,
      "loss": 0.8052,
      "step": 2469
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8974651675433751,
      "learning_rate": 8.002457549266725e-06,
      "loss": 0.633,
      "step": 2470
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3620511225513778,
      "learning_rate": 8.000799567011916e-06,
      "loss": 0.8206,
      "step": 2471
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2736058983986858,
      "learning_rate": 7.999141068884873e-06,
      "loss": 0.7279,
      "step": 2472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3519279964451028,
      "learning_rate": 7.997482055170712e-06,
      "loss": 0.8085,
      "step": 2473
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.328694632381772,
      "learning_rate": 7.995822526154636e-06,
      "loss": 0.8331,
      "step": 2474
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3841152701205575,
      "learning_rate": 7.994162482121936e-06,
      "loss": 0.7334,
      "step": 2475
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7652992716319524,
      "learning_rate": 7.992501923357996e-06,
      "loss": 0.8305,
      "step": 2476
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4745918741677086,
      "learning_rate": 7.990840850148283e-06,
      "loss": 0.7569,
      "step": 2477
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3998438556870758,
      "learning_rate": 7.989179262778356e-06,
      "loss": 0.7244,
      "step": 2478
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2867407773833939,
      "learning_rate": 7.987517161533858e-06,
      "loss": 0.7345,
      "step": 2479
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3484418107014653,
      "learning_rate": 7.985854546700526e-06,
      "loss": 0.7368,
      "step": 2480
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3021984501812445,
      "learning_rate": 7.984191418564183e-06,
      "loss": 0.6826,
      "step": 2481
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9060408009562222,
      "learning_rate": 7.982527777410738e-06,
      "loss": 0.6498,
      "step": 2482
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5547128412254032,
      "learning_rate": 7.980863623526195e-06,
      "loss": 0.7933,
      "step": 2483
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8841395292815575,
      "learning_rate": 7.979198957196634e-06,
      "loss": 0.6807,
      "step": 2484
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2082016069024861,
      "learning_rate": 7.977533778708237e-06,
      "loss": 0.7713,
      "step": 2485
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5702569374984396,
      "learning_rate": 7.975868088347263e-06,
      "loss": 0.8099,
      "step": 2486
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5187043092426669,
      "learning_rate": 7.974201886400066e-06,
      "loss": 0.7772,
      "step": 2487
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8589953217348618,
      "learning_rate": 7.972535173153087e-06,
      "loss": 0.6133,
      "step": 2488
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8384473064437474,
      "learning_rate": 7.970867948892849e-06,
      "loss": 0.7422,
      "step": 2489
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6028302277537334,
      "learning_rate": 7.96920021390597e-06,
      "loss": 0.7768,
      "step": 2490
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.009565691661408,
      "learning_rate": 7.967531968479152e-06,
      "loss": 0.6836,
      "step": 2491
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3695536289615784,
      "learning_rate": 7.965863212899185e-06,
      "loss": 0.6929,
      "step": 2492
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4747515945001828,
      "learning_rate": 7.96419394745295e-06,
      "loss": 0.7225,
      "step": 2493
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4865403204932544,
      "learning_rate": 7.962524172427412e-06,
      "loss": 0.7976,
      "step": 2494
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1686088004241175,
      "learning_rate": 7.960853888109624e-06,
      "loss": 0.7872,
      "step": 2495
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.39884546212422,
      "learning_rate": 7.959183094786726e-06,
      "loss": 0.7503,
      "step": 2496
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8326433958484779,
      "learning_rate": 7.957511792745948e-06,
      "loss": 0.673,
      "step": 2497
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8492428402294998,
      "learning_rate": 7.955839982274608e-06,
      "loss": 0.6977,
      "step": 2498
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2996259799644951,
      "learning_rate": 7.954167663660108e-06,
      "loss": 0.8184,
      "step": 2499
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.220574705443864,
      "learning_rate": 7.952494837189937e-06,
      "loss": 0.7653,
      "step": 2500
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.194037518176866,
      "learning_rate": 7.950821503151678e-06,
      "loss": 0.7639,
      "step": 2501
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3187656339290277,
      "learning_rate": 7.949147661832992e-06,
      "loss": 0.8823,
      "step": 2502
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4837473727152088,
      "learning_rate": 7.947473313521635e-06,
      "loss": 0.7526,
      "step": 2503
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.842011476448487,
      "learning_rate": 7.945798458505442e-06,
      "loss": 0.6442,
      "step": 2504
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3578003695364824,
      "learning_rate": 7.944123097072348e-06,
      "loss": 0.8445,
      "step": 2505
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6202172015885639,
      "learning_rate": 7.942447229510359e-06,
      "loss": 0.7487,
      "step": 2506
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3892987503748506,
      "learning_rate": 7.94077085610758e-06,
      "loss": 0.765,
      "step": 2507
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0601577293513915,
      "learning_rate": 7.9390939771522e-06,
      "loss": 0.7271,
      "step": 2508
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1484940348769217,
      "learning_rate": 7.93741659293249e-06,
      "loss": 0.7813,
      "step": 2509
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7315217796712008,
      "learning_rate": 7.935738703736817e-06,
      "loss": 0.7675,
      "step": 2510
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3348242587003323,
      "learning_rate": 7.934060309853627e-06,
      "loss": 0.8178,
      "step": 2511
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8814026924035542,
      "learning_rate": 7.932381411571453e-06,
      "loss": 0.6877,
      "step": 2512
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.922907823533116,
      "learning_rate": 7.930702009178919e-06,
      "loss": 0.6643,
      "step": 2513
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4308714477972333,
      "learning_rate": 7.929022102964736e-06,
      "loss": 0.7827,
      "step": 2514
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3804038045215252,
      "learning_rate": 7.927341693217697e-06,
      "loss": 0.7871,
      "step": 2515
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3118649344325721,
      "learning_rate": 7.925660780226683e-06,
      "loss": 0.7863,
      "step": 2516
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2597844690571556,
      "learning_rate": 7.923979364280664e-06,
      "loss": 0.7331,
      "step": 2517
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3972177191774393,
      "learning_rate": 7.922297445668695e-06,
      "loss": 0.7848,
      "step": 2518
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.233541675651328,
      "learning_rate": 7.920615024679917e-06,
      "loss": 0.7813,
      "step": 2519
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3475597002890425,
      "learning_rate": 7.918932101603556e-06,
      "loss": 0.8103,
      "step": 2520
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8239027452197456,
      "learning_rate": 7.917248676728929e-06,
      "loss": 0.6535,
      "step": 2521
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6096612229169798,
      "learning_rate": 7.915564750345436e-06,
      "loss": 0.7899,
      "step": 2522
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3690072372380548,
      "learning_rate": 7.91388032274256e-06,
      "loss": 0.8117,
      "step": 2523
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7822429522025822,
      "learning_rate": 7.912195394209877e-06,
      "loss": 0.839,
      "step": 2524
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3172772370786268,
      "learning_rate": 7.910509965037045e-06,
      "loss": 0.8209,
      "step": 2525
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.421282773223067,
      "learning_rate": 7.90882403551381e-06,
      "loss": 0.8785,
      "step": 2526
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4303513676655772,
      "learning_rate": 7.907137605929998e-06,
      "loss": 0.7208,
      "step": 2527
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5951509739273086,
      "learning_rate": 7.90545067657553e-06,
      "loss": 0.7793,
      "step": 2528
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3106791722226099,
      "learning_rate": 7.903763247740407e-06,
      "loss": 0.8445,
      "step": 2529
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5154092787329116,
      "learning_rate": 7.902075319714722e-06,
      "loss": 0.826,
      "step": 2530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8289383164489466,
      "learning_rate": 7.900386892788644e-06,
      "loss": 0.6469,
      "step": 2531
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3685351778833266,
      "learning_rate": 7.898697967252433e-06,
      "loss": 0.7518,
      "step": 2532
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8193050096729133,
      "learning_rate": 7.897008543396438e-06,
      "loss": 0.6537,
      "step": 2533
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3654940834368963,
      "learning_rate": 7.89531862151109e-06,
      "loss": 0.8174,
      "step": 2534
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4656213712470216,
      "learning_rate": 7.893628201886906e-06,
      "loss": 0.7766,
      "step": 2535
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.33533073247034,
      "learning_rate": 7.891937284814489e-06,
      "loss": 0.8254,
      "step": 2536
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8642329223043608,
      "learning_rate": 7.890245870584523e-06,
      "loss": 0.6171,
      "step": 2537
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3491850887264336,
      "learning_rate": 7.888553959487788e-06,
      "loss": 0.8139,
      "step": 2538
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5452496762366188,
      "learning_rate": 7.886861551815139e-06,
      "loss": 0.8404,
      "step": 2539
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3755040313035545,
      "learning_rate": 7.885168647857523e-06,
      "loss": 0.7756,
      "step": 2540
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6480079881630756,
      "learning_rate": 7.88347524790597e-06,
      "loss": 0.7933,
      "step": 2541
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.234304513478104,
      "learning_rate": 7.881781352251591e-06,
      "loss": 0.852,
      "step": 2542
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5488921991957139,
      "learning_rate": 7.88008696118559e-06,
      "loss": 0.7389,
      "step": 2543
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.610977410395722,
      "learning_rate": 7.878392074999252e-06,
      "loss": 0.8074,
      "step": 2544
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5201183499712871,
      "learning_rate": 7.876696693983947e-06,
      "loss": 0.8459,
      "step": 2545
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8826582053260605,
      "learning_rate": 7.875000818431132e-06,
      "loss": 0.8071,
      "step": 2546
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2849151421015528,
      "learning_rate": 7.873304448632345e-06,
      "loss": 0.7261,
      "step": 2547
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7633415817283917,
      "learning_rate": 7.871607584879211e-06,
      "loss": 0.6306,
      "step": 2548
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2667179306790277,
      "learning_rate": 7.869910227463447e-06,
      "loss": 0.868,
      "step": 2549
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8093714881346654,
      "learning_rate": 7.86821237667684e-06,
      "loss": 0.8214,
      "step": 2550
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3055963709580145,
      "learning_rate": 7.866514032811276e-06,
      "loss": 0.807,
      "step": 2551
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2941000739178512,
      "learning_rate": 7.864815196158718e-06,
      "loss": 0.6962,
      "step": 2552
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3693179799534207,
      "learning_rate": 7.863115867011217e-06,
      "loss": 0.9161,
      "step": 2553
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.3712053832607056,
      "learning_rate": 7.861416045660906e-06,
      "loss": 0.9407,
      "step": 2554
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3599250208391005,
      "learning_rate": 7.859715732400004e-06,
      "loss": 0.9313,
      "step": 2555
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.479530102898811,
      "learning_rate": 7.858014927520815e-06,
      "loss": 0.8152,
      "step": 2556
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.981566314135474,
      "learning_rate": 7.856313631315726e-06,
      "loss": 0.753,
      "step": 2557
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2683111681668042,
      "learning_rate": 7.854611844077213e-06,
      "loss": 0.8411,
      "step": 2558
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0495828262424243,
      "learning_rate": 7.852909566097828e-06,
      "loss": 0.6468,
      "step": 2559
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3108012356509606,
      "learning_rate": 7.851206797670217e-06,
      "loss": 0.8565,
      "step": 2560
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.522110213640304,
      "learning_rate": 7.849503539087102e-06,
      "loss": 0.8139,
      "step": 2561
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8643225764653253,
      "learning_rate": 7.847799790641295e-06,
      "loss": 0.8352,
      "step": 2562
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9137235205430112,
      "learning_rate": 7.846095552625688e-06,
      "loss": 0.7701,
      "step": 2563
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2652408072658456,
      "learning_rate": 7.844390825333264e-06,
      "loss": 0.8445,
      "step": 2564
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3542726874107802,
      "learning_rate": 7.842685609057078e-06,
      "loss": 0.713,
      "step": 2565
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4929474581755864,
      "learning_rate": 7.840979904090282e-06,
      "loss": 0.8648,
      "step": 2566
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5599471361476551,
      "learning_rate": 7.839273710726107e-06,
      "loss": 0.7728,
      "step": 2567
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8739114804223594,
      "learning_rate": 7.837567029257864e-06,
      "loss": 0.7026,
      "step": 2568
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.291471582979695,
      "learning_rate": 7.835859859978952e-06,
      "loss": 0.8876,
      "step": 2569
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7507047296767237,
      "learning_rate": 7.834152203182854e-06,
      "loss": 0.8077,
      "step": 2570
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2086962944300401,
      "learning_rate": 7.832444059163137e-06,
      "loss": 0.7751,
      "step": 2571
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8096736565494184,
      "learning_rate": 7.830735428213451e-06,
      "loss": 0.6915,
      "step": 2572
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.347254821315814,
      "learning_rate": 7.829026310627526e-06,
      "loss": 0.8033,
      "step": 2573
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4379046919730947,
      "learning_rate": 7.827316706699182e-06,
      "loss": 0.8107,
      "step": 2574
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5740997545672855,
      "learning_rate": 7.825606616722319e-06,
      "loss": 0.8247,
      "step": 2575
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4422266360035558,
      "learning_rate": 7.823896040990922e-06,
      "loss": 0.88,
      "step": 2576
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.313336500803788,
      "learning_rate": 7.822184979799057e-06,
      "loss": 0.8286,
      "step": 2577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4504424105908393,
      "learning_rate": 7.820473433440876e-06,
      "loss": 0.8345,
      "step": 2578
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.257638855114575,
      "learning_rate": 7.818761402210616e-06,
      "loss": 0.8085,
      "step": 2579
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2943226297114787,
      "learning_rate": 7.817048886402591e-06,
      "loss": 0.75,
      "step": 2580
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5535605227834366,
      "learning_rate": 7.815335886311205e-06,
      "loss": 0.8613,
      "step": 2581
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7961326230121428,
      "learning_rate": 7.813622402230943e-06,
      "loss": 0.8089,
      "step": 2582
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2628497451693124,
      "learning_rate": 7.811908434456372e-06,
      "loss": 0.7148,
      "step": 2583
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1653161336844402,
      "learning_rate": 7.810193983282142e-06,
      "loss": 0.708,
      "step": 2584
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3635602649144842,
      "learning_rate": 7.808479049002988e-06,
      "loss": 0.8037,
      "step": 2585
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4037981453008341,
      "learning_rate": 7.806763631913726e-06,
      "loss": 0.8232,
      "step": 2586
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3624653261778439,
      "learning_rate": 7.805047732309257e-06,
      "loss": 0.7292,
      "step": 2587
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.391511491653691,
      "learning_rate": 7.803331350484563e-06,
      "loss": 0.8303,
      "step": 2588
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5752289435155462,
      "learning_rate": 7.801614486734712e-06,
      "loss": 0.7784,
      "step": 2589
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2472102091264177,
      "learning_rate": 7.799897141354854e-06,
      "loss": 0.8023,
      "step": 2590
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5298022688131925,
      "learning_rate": 7.798179314640214e-06,
      "loss": 0.7921,
      "step": 2591
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.78853162366274,
      "learning_rate": 7.79646100688611e-06,
      "loss": 0.7861,
      "step": 2592
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3832931970700169,
      "learning_rate": 7.794742218387943e-06,
      "loss": 0.7813,
      "step": 2593
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.962580879643208,
      "learning_rate": 7.793022949441189e-06,
      "loss": 0.8447,
      "step": 2594
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3571853093733401,
      "learning_rate": 7.79130320034141e-06,
      "loss": 0.7822,
      "step": 2595
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3786046419397737,
      "learning_rate": 7.789582971384252e-06,
      "loss": 0.7511,
      "step": 2596
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3039475687065594,
      "learning_rate": 7.787862262865438e-06,
      "loss": 0.8292,
      "step": 2597
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5238794911097016,
      "learning_rate": 7.786141075080786e-06,
      "loss": 0.8029,
      "step": 2598
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2915899945671914,
      "learning_rate": 7.784419408326181e-06,
      "loss": 0.7862,
      "step": 2599
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5118318585701165,
      "learning_rate": 7.782697262897601e-06,
      "loss": 0.8205,
      "step": 2600
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4559666012468626,
      "learning_rate": 7.7809746390911e-06,
      "loss": 0.8127,
      "step": 2601
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6964182291659207,
      "learning_rate": 7.77925153720282e-06,
      "loss": 0.7201,
      "step": 2602
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2800896112580686,
      "learning_rate": 7.777527957528982e-06,
      "loss": 0.8071,
      "step": 2603
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2049402649499097,
      "learning_rate": 7.775803900365885e-06,
      "loss": 0.8661,
      "step": 2604
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.453587334275302,
      "learning_rate": 7.77407936600992e-06,
      "loss": 0.8853,
      "step": 2605
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.423982757783704,
      "learning_rate": 7.77235435475755e-06,
      "loss": 0.75,
      "step": 2606
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2999481424943307,
      "learning_rate": 7.77062886690533e-06,
      "loss": 0.8482,
      "step": 2607
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.343781407945469,
      "learning_rate": 7.768902902749885e-06,
      "loss": 0.7976,
      "step": 2608
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.414265830878972,
      "learning_rate": 7.767176462587932e-06,
      "loss": 0.6335,
      "step": 2609
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3919545886362967,
      "learning_rate": 7.765449546716266e-06,
      "loss": 0.8485,
      "step": 2610
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.288613430978132,
      "learning_rate": 7.76372215543176e-06,
      "loss": 0.7891,
      "step": 2611
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1884306061037715,
      "learning_rate": 7.76199428903138e-06,
      "loss": 0.7907,
      "step": 2612
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2553861542136568,
      "learning_rate": 7.76026594781216e-06,
      "loss": 0.7644,
      "step": 2613
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6566168007489483,
      "learning_rate": 7.758537132071224e-06,
      "loss": 0.8499,
      "step": 2614
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7921162538545408,
      "learning_rate": 7.756807842105774e-06,
      "loss": 0.7976,
      "step": 2615
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4826449519272835,
      "learning_rate": 7.755078078213099e-06,
      "loss": 0.8139,
      "step": 2616
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3993042685901773,
      "learning_rate": 7.75334784069056e-06,
      "loss": 0.9168,
      "step": 2617
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4735226009922178,
      "learning_rate": 7.75161712983561e-06,
      "loss": 0.7449,
      "step": 2618
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.358666867679418,
      "learning_rate": 7.749885945945774e-06,
      "loss": 0.8619,
      "step": 2619
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5123924660761328,
      "learning_rate": 7.748154289318666e-06,
      "loss": 0.7854,
      "step": 2620
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.34392262525492,
      "learning_rate": 7.746422160251975e-06,
      "loss": 0.8383,
      "step": 2621
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3757490556601364,
      "learning_rate": 7.744689559043475e-06,
      "loss": 0.83,
      "step": 2622
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.40939218715108,
      "learning_rate": 7.742956485991022e-06,
      "loss": 0.7771,
      "step": 2623
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3502136827597329,
      "learning_rate": 7.74122294139255e-06,
      "loss": 0.7512,
      "step": 2624
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2409138850058221,
      "learning_rate": 7.739488925546074e-06,
      "loss": 0.7149,
      "step": 2625
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9951864142201992,
      "learning_rate": 7.737754438749694e-06,
      "loss": 0.7314,
      "step": 2626
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5336375501548856,
      "learning_rate": 7.736019481301587e-06,
      "loss": 0.7024,
      "step": 2627
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8522127613711472,
      "learning_rate": 7.734284053500015e-06,
      "loss": 0.6837,
      "step": 2628
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1963752349175167,
      "learning_rate": 7.732548155643311e-06,
      "loss": 0.7921,
      "step": 2629
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.30434147812285,
      "learning_rate": 7.730811788029903e-06,
      "loss": 0.8221,
      "step": 2630
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4107048326284262,
      "learning_rate": 7.72907495095829e-06,
      "loss": 0.7348,
      "step": 2631
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4843864128997097,
      "learning_rate": 7.727337644727055e-06,
      "loss": 0.6996,
      "step": 2632
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6161187458411492,
      "learning_rate": 7.725599869634861e-06,
      "loss": 0.7572,
      "step": 2633
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.940015285576448,
      "learning_rate": 7.72386162598045e-06,
      "loss": 0.665,
      "step": 2634
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.360262631093949,
      "learning_rate": 7.722122914062649e-06,
      "loss": 0.8196,
      "step": 2635
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3230928173962788,
      "learning_rate": 7.720383734180362e-06,
      "loss": 0.857,
      "step": 2636
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3750502526592714,
      "learning_rate": 7.718644086632575e-06,
      "loss": 0.7471,
      "step": 2637
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.145582468947622,
      "learning_rate": 7.71690397171835e-06,
      "loss": 0.7774,
      "step": 2638
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.469776664495418,
      "learning_rate": 7.715163389736834e-06,
      "loss": 0.8584,
      "step": 2639
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4363699073021485,
      "learning_rate": 7.713422340987258e-06,
      "loss": 0.8328,
      "step": 2640
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.83709526616732,
      "learning_rate": 7.711680825768922e-06,
      "loss": 0.7143,
      "step": 2641
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3448372432686415,
      "learning_rate": 7.709938844381217e-06,
      "loss": 0.7696,
      "step": 2642
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4980951489150436,
      "learning_rate": 7.708196397123607e-06,
      "loss": 0.8084,
      "step": 2643
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9315807255767126,
      "learning_rate": 7.706453484295643e-06,
      "loss": 0.6492,
      "step": 2644
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4173654368537827,
      "learning_rate": 7.704710106196947e-06,
      "loss": 0.7581,
      "step": 2645
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3709889835487932,
      "learning_rate": 7.702966263127227e-06,
      "loss": 0.847,
      "step": 2646
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2840466982662084,
      "learning_rate": 7.701221955386273e-06,
      "loss": 0.7889,
      "step": 2647
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3146853192895191,
      "learning_rate": 7.699477183273949e-06,
      "loss": 0.8207,
      "step": 2648
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2570955318815227,
      "learning_rate": 7.6977319470902e-06,
      "loss": 0.6999,
      "step": 2649
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3920144859273351,
      "learning_rate": 7.695986247135058e-06,
      "loss": 0.7868,
      "step": 2650
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5046415714909036,
      "learning_rate": 7.694240083708623e-06,
      "loss": 0.8145,
      "step": 2651
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3992862581687648,
      "learning_rate": 7.692493457111085e-06,
      "loss": 0.8181,
      "step": 2652
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3632840178291863,
      "learning_rate": 7.690746367642707e-06,
      "loss": 0.7992,
      "step": 2653
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6516990901261386,
      "learning_rate": 7.688998815603837e-06,
      "loss": 0.7269,
      "step": 2654
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3458854567235512,
      "learning_rate": 7.687250801294893e-06,
      "loss": 0.7801,
      "step": 2655
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5702824806357452,
      "learning_rate": 7.685502325016384e-06,
      "loss": 0.8365,
      "step": 2656
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3614017139153671,
      "learning_rate": 7.683753387068894e-06,
      "loss": 0.7957,
      "step": 2657
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3585683096489432,
      "learning_rate": 7.682003987753082e-06,
      "loss": 0.7729,
      "step": 2658
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4239156777977313,
      "learning_rate": 7.680254127369695e-06,
      "loss": 0.883,
      "step": 2659
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3175394624042118,
      "learning_rate": 7.67850380621955e-06,
      "loss": 0.7983,
      "step": 2660
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3347781453372818,
      "learning_rate": 7.676753024603546e-06,
      "loss": 0.8369,
      "step": 2661
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5102444165337823,
      "learning_rate": 7.675001782822669e-06,
      "loss": 0.8782,
      "step": 2662
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6906374716838413,
      "learning_rate": 7.673250081177974e-06,
      "loss": 0.7539,
      "step": 2663
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4379852664430612,
      "learning_rate": 7.671497919970598e-06,
      "loss": 0.7955,
      "step": 2664
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.39860731535261,
      "learning_rate": 7.669745299501758e-06,
      "loss": 0.8467,
      "step": 2665
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2971934948017614,
      "learning_rate": 7.667992220072751e-06,
      "loss": 0.929,
      "step": 2666
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.358999722427478,
      "learning_rate": 7.666238681984952e-06,
      "loss": 0.8403,
      "step": 2667
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4635752938519133,
      "learning_rate": 7.664484685539814e-06,
      "loss": 0.6695,
      "step": 2668
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3702395929137607,
      "learning_rate": 7.662730231038869e-06,
      "loss": 0.6915,
      "step": 2669
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3373576492149077,
      "learning_rate": 7.66097531878373e-06,
      "loss": 0.7332,
      "step": 2670
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2506161580990536,
      "learning_rate": 7.659219949076082e-06,
      "loss": 0.7959,
      "step": 2671
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5717718394199873,
      "learning_rate": 7.657464122217699e-06,
      "loss": 0.7891,
      "step": 2672
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4137192517693364,
      "learning_rate": 7.655707838510423e-06,
      "loss": 0.7733,
      "step": 2673
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.537491510518454,
      "learning_rate": 7.653951098256184e-06,
      "loss": 0.8199,
      "step": 2674
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.275039603839594,
      "learning_rate": 7.652193901756983e-06,
      "loss": 0.8385,
      "step": 2675
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2285040956240811,
      "learning_rate": 7.650436249314902e-06,
      "loss": 0.8006,
      "step": 2676
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2940683000038526,
      "learning_rate": 7.648678141232107e-06,
      "loss": 0.715,
      "step": 2677
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5517649171721506,
      "learning_rate": 7.646919577810831e-06,
      "loss": 0.8784,
      "step": 2678
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6736874234161685,
      "learning_rate": 7.645160559353392e-06,
      "loss": 0.7608,
      "step": 2679
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3559227319165188,
      "learning_rate": 7.64340108616219e-06,
      "loss": 0.7261,
      "step": 2680
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2402720453656275,
      "learning_rate": 7.641641158539696e-06,
      "loss": 0.7527,
      "step": 2681
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4864552058235931,
      "learning_rate": 7.639880776788464e-06,
      "loss": 0.8089,
      "step": 2682
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5474996688885747,
      "learning_rate": 7.63811994121112e-06,
      "loss": 0.8339,
      "step": 2683
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2642682440300765,
      "learning_rate": 7.636358652110375e-06,
      "loss": 0.8579,
      "step": 2684
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3262169530313292,
      "learning_rate": 7.634596909789014e-06,
      "loss": 0.863,
      "step": 2685
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2347838743022286,
      "learning_rate": 7.632834714549902e-06,
      "loss": 0.8097,
      "step": 2686
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6563398519175758,
      "learning_rate": 7.63107206669598e-06,
      "loss": 0.8341,
      "step": 2687
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2010958863286425,
      "learning_rate": 7.629308966530268e-06,
      "loss": 0.8262,
      "step": 2688
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7065450177523407,
      "learning_rate": 7.6275454143558614e-06,
      "loss": 0.8281,
      "step": 2689
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.474303536738022,
      "learning_rate": 7.6257814104759385e-06,
      "loss": 0.8095,
      "step": 2690
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3615718757330622,
      "learning_rate": 7.62401695519375e-06,
      "loss": 0.7574,
      "step": 2691
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3088618591919665,
      "learning_rate": 7.6222520488126286e-06,
      "loss": 0.7854,
      "step": 2692
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.282076028888767,
      "learning_rate": 7.620486691635978e-06,
      "loss": 0.859,
      "step": 2693
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3261017154216344,
      "learning_rate": 7.618720883967285e-06,
      "loss": 0.7877,
      "step": 2694
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3673968025921208,
      "learning_rate": 7.616954626110116e-06,
      "loss": 0.7808,
      "step": 2695
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3198516907565039,
      "learning_rate": 7.615187918368107e-06,
      "loss": 0.8362,
      "step": 2696
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3937371061200992,
      "learning_rate": 7.613420761044976e-06,
      "loss": 0.854,
      "step": 2697
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8377261308499286,
      "learning_rate": 7.611653154444521e-06,
      "loss": 0.6478,
      "step": 2698
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3362615636215605,
      "learning_rate": 7.60988509887061e-06,
      "loss": 0.8155,
      "step": 2699
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5708889031555866,
      "learning_rate": 7.608116594627195e-06,
      "loss": 0.842,
      "step": 2700
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2773238968421725,
      "learning_rate": 7.606347642018301e-06,
      "loss": 0.6493,
      "step": 2701
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9129743775179182,
      "learning_rate": 7.604578241348033e-06,
      "loss": 0.7136,
      "step": 2702
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5630727270922578,
      "learning_rate": 7.602808392920568e-06,
      "loss": 0.8322,
      "step": 2703
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4211501567056481,
      "learning_rate": 7.601038097040169e-06,
      "loss": 0.6946,
      "step": 2704
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4100743297504477,
      "learning_rate": 7.599267354011166e-06,
      "loss": 0.7814,
      "step": 2705
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.438949922404308,
      "learning_rate": 7.597496164137972e-06,
      "loss": 0.7146,
      "step": 2706
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5928412806011,
      "learning_rate": 7.595724527725074e-06,
      "loss": 0.8424,
      "step": 2707
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4112120438266333,
      "learning_rate": 7.593952445077035e-06,
      "loss": 0.7753,
      "step": 2708
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5580248388245863,
      "learning_rate": 7.592179916498503e-06,
      "loss": 0.8466,
      "step": 2709
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6202498362022926,
      "learning_rate": 7.590406942294191e-06,
      "loss": 0.7396,
      "step": 2710
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3993643390974473,
      "learning_rate": 7.588633522768896e-06,
      "loss": 0.8611,
      "step": 2711
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4026739198067244,
      "learning_rate": 7.586859658227487e-06,
      "loss": 0.7739,
      "step": 2712
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4528974216547816,
      "learning_rate": 7.585085348974912e-06,
      "loss": 0.8459,
      "step": 2713
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3774831879825211,
      "learning_rate": 7.5833105953161986e-06,
      "loss": 0.8478,
      "step": 2714
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3854023971059013,
      "learning_rate": 7.581535397556445e-06,
      "loss": 0.8448,
      "step": 2715
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7932890349017354,
      "learning_rate": 7.579759756000829e-06,
      "loss": 0.76,
      "step": 2716
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8185601686700074,
      "learning_rate": 7.577983670954604e-06,
      "loss": 0.732,
      "step": 2717
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4524146170596026,
      "learning_rate": 7.5762071427231e-06,
      "loss": 0.7496,
      "step": 2718
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3933920566394766,
      "learning_rate": 7.5744301716117206e-06,
      "loss": 0.7603,
      "step": 2719
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3868433725160612,
      "learning_rate": 7.572652757925951e-06,
      "loss": 0.7837,
      "step": 2720
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.347645333429927,
      "learning_rate": 7.570874901971347e-06,
      "loss": 0.7712,
      "step": 2721
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.319534590931268,
      "learning_rate": 7.569096604053543e-06,
      "loss": 0.7921,
      "step": 2722
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.468584013829871,
      "learning_rate": 7.5673178644782475e-06,
      "loss": 0.8599,
      "step": 2723
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1391123359625566,
      "learning_rate": 7.565538683551251e-06,
      "loss": 0.6699,
      "step": 2724
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8449329112275483,
      "learning_rate": 7.563759061578412e-06,
      "loss": 0.6701,
      "step": 2725
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4441077785379635,
      "learning_rate": 7.561978998865667e-06,
      "loss": 0.8049,
      "step": 2726
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4501187920013845,
      "learning_rate": 7.560198495719032e-06,
      "loss": 0.7897,
      "step": 2727
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.285080610398691,
      "learning_rate": 7.558417552444595e-06,
      "loss": 0.7856,
      "step": 2728
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.72667478133205,
      "learning_rate": 7.556636169348524e-06,
      "loss": 0.813,
      "step": 2729
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2760902410741346,
      "learning_rate": 7.554854346737053e-06,
      "loss": 0.8278,
      "step": 2730
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3469978767895392,
      "learning_rate": 7.5530720849165e-06,
      "loss": 0.7641,
      "step": 2731
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5384899319231031,
      "learning_rate": 7.551289384193261e-06,
      "loss": 0.8031,
      "step": 2732
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.832633155078563,
      "learning_rate": 7.549506244873799e-06,
      "loss": 0.6728,
      "step": 2733
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8296003528925884,
      "learning_rate": 7.547722667264656e-06,
      "loss": 0.833,
      "step": 2734
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5701383069460266,
      "learning_rate": 7.545938651672453e-06,
      "loss": 0.8517,
      "step": 2735
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8741248725785307,
      "learning_rate": 7.5441541984038795e-06,
      "loss": 0.6055,
      "step": 2736
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3364964514936892,
      "learning_rate": 7.542369307765705e-06,
      "loss": 0.8369,
      "step": 2737
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8547633863407703,
      "learning_rate": 7.540583980064772e-06,
      "loss": 0.7134,
      "step": 2738
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6593662725224787,
      "learning_rate": 7.538798215608001e-06,
      "loss": 0.8789,
      "step": 2739
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2022506553791499,
      "learning_rate": 7.537012014702383e-06,
      "loss": 0.7805,
      "step": 2740
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6726701173136305,
      "learning_rate": 7.535225377654989e-06,
      "loss": 0.7773,
      "step": 2741
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6493721040658742,
      "learning_rate": 7.533438304772962e-06,
      "loss": 0.7806,
      "step": 2742
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8906396489320912,
      "learning_rate": 7.531650796363521e-06,
      "loss": 0.58,
      "step": 2743
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3557036225985246,
      "learning_rate": 7.529862852733958e-06,
      "loss": 0.814,
      "step": 2744
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9869267977348017,
      "learning_rate": 7.52807447419164e-06,
      "loss": 0.657,
      "step": 2745
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5124324831132023,
      "learning_rate": 7.526285661044015e-06,
      "loss": 0.7754,
      "step": 2746
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4846193532789336,
      "learning_rate": 7.524496413598596e-06,
      "loss": 0.8573,
      "step": 2747
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3393033193693786,
      "learning_rate": 7.522706732162976e-06,
      "loss": 0.8823,
      "step": 2748
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9680798715911815,
      "learning_rate": 7.520916617044826e-06,
      "loss": 0.6816,
      "step": 2749
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9237172287435872,
      "learning_rate": 7.519126068551882e-06,
      "loss": 0.7209,
      "step": 2750
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4776895785039315,
      "learning_rate": 7.517335086991963e-06,
      "loss": 0.6553,
      "step": 2751
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.262760779255803,
      "learning_rate": 7.51554367267296e-06,
      "loss": 0.7669,
      "step": 2752
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.26857808477974,
      "learning_rate": 7.513751825902837e-06,
      "loss": 0.7939,
      "step": 2753
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.617150259007356,
      "learning_rate": 7.5119595469896325e-06,
      "loss": 0.8202,
      "step": 2754
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.385263442910664,
      "learning_rate": 7.51016683624146e-06,
      "loss": 0.7847,
      "step": 2755
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4929425587542704,
      "learning_rate": 7.50837369396651e-06,
      "loss": 0.7971,
      "step": 2756
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2509471443114306,
      "learning_rate": 7.506580120473042e-06,
      "loss": 0.7545,
      "step": 2757
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5918408097933694,
      "learning_rate": 7.504786116069391e-06,
      "loss": 0.7841,
      "step": 2758
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.227561765664,
      "learning_rate": 7.502991681063969e-06,
      "loss": 0.7643,
      "step": 2759
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2219572496156457,
      "learning_rate": 7.5011968157652615e-06,
      "loss": 0.7767,
      "step": 2760
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2592479124738465,
      "learning_rate": 7.499401520481824e-06,
      "loss": 0.8053,
      "step": 2761
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6317953208705072,
      "learning_rate": 7.497605795522289e-06,
      "loss": 0.7578,
      "step": 2762
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6495848112324087,
      "learning_rate": 7.495809641195364e-06,
      "loss": 0.7642,
      "step": 2763
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3303773793318243,
      "learning_rate": 7.494013057809827e-06,
      "loss": 0.7971,
      "step": 2764
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4094149722654281,
      "learning_rate": 7.492216045674532e-06,
      "loss": 0.8766,
      "step": 2765
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.66866859747718,
      "learning_rate": 7.490418605098408e-06,
      "loss": 0.7337,
      "step": 2766
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.597592730727451,
      "learning_rate": 7.488620736390454e-06,
      "loss": 0.7491,
      "step": 2767
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6087297777271132,
      "learning_rate": 7.486822439859744e-06,
      "loss": 0.7572,
      "step": 2768
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9912417128512782,
      "learning_rate": 7.485023715815427e-06,
      "loss": 0.7262,
      "step": 2769
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.76773853714651,
      "learning_rate": 7.483224564566725e-06,
      "loss": 0.7185,
      "step": 2770
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4870863138822958,
      "learning_rate": 7.481424986422933e-06,
      "loss": 0.759,
      "step": 2771
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8334333231271462,
      "learning_rate": 7.479624981693416e-06,
      "loss": 0.8402,
      "step": 2772
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3336264300443321,
      "learning_rate": 7.47782455068762e-06,
      "loss": 0.8362,
      "step": 2773
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.321973230424981,
      "learning_rate": 7.476023693715059e-06,
      "loss": 0.7895,
      "step": 2774
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3051692158098027,
      "learning_rate": 7.47422241108532e-06,
      "loss": 0.7682,
      "step": 2775
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4237735286773547,
      "learning_rate": 7.4724207031080645e-06,
      "loss": 0.915,
      "step": 2776
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3662324358282412,
      "learning_rate": 7.4706185700930266e-06,
      "loss": 0.8318,
      "step": 2777
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2530544533564063,
      "learning_rate": 7.4688160123500165e-06,
      "loss": 0.831,
      "step": 2778
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4201820329269212,
      "learning_rate": 7.467013030188912e-06,
      "loss": 0.6787,
      "step": 2779
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8104160157484473,
      "learning_rate": 7.465209623919668e-06,
      "loss": 0.6248,
      "step": 2780
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.45846008710992,
      "learning_rate": 7.46340579385231e-06,
      "loss": 0.8294,
      "step": 2781
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2020077051696347,
      "learning_rate": 7.4616015402969375e-06,
      "loss": 0.8097,
      "step": 2782
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4670430704730684,
      "learning_rate": 7.459796863563723e-06,
      "loss": 0.7989,
      "step": 2783
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3298550130622109,
      "learning_rate": 7.457991763962912e-06,
      "loss": 0.7549,
      "step": 2784
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8132212770319753,
      "learning_rate": 7.456186241804821e-06,
      "loss": 0.6945,
      "step": 2785
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.407508793366392,
      "learning_rate": 7.45438029739984e-06,
      "loss": 0.7394,
      "step": 2786
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1351208570843379,
      "learning_rate": 7.4525739310584326e-06,
      "loss": 0.7251,
      "step": 2787
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3942274141840465,
      "learning_rate": 7.450767143091133e-06,
      "loss": 0.7924,
      "step": 2788
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.244457344534875,
      "learning_rate": 7.448959933808552e-06,
      "loss": 0.7576,
      "step": 2789
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2815505634549909,
      "learning_rate": 7.447152303521366e-06,
      "loss": 0.7844,
      "step": 2790
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6830425562986535,
      "learning_rate": 7.445344252540327e-06,
      "loss": 0.8497,
      "step": 2791
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4385998720718598,
      "learning_rate": 7.443535781176265e-06,
      "loss": 0.802,
      "step": 2792
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3383510753210204,
      "learning_rate": 7.441726889740075e-06,
      "loss": 0.7578,
      "step": 2793
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8643732056680729,
      "learning_rate": 7.439917578542723e-06,
      "loss": 0.7219,
      "step": 2794
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6556147745710408,
      "learning_rate": 7.438107847895256e-06,
      "loss": 0.7949,
      "step": 2795
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8135266628080475,
      "learning_rate": 7.436297698108783e-06,
      "loss": 0.7015,
      "step": 2796
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2906767219084772,
      "learning_rate": 7.434487129494494e-06,
      "loss": 0.6693,
      "step": 2797
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.368939053451159,
      "learning_rate": 7.4326761423636415e-06,
      "loss": 0.7528,
      "step": 2798
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3317501067434347,
      "learning_rate": 7.430864737027561e-06,
      "loss": 0.8018,
      "step": 2799
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3510497397260757,
      "learning_rate": 7.429052913797649e-06,
      "loss": 0.7445,
      "step": 2800
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3656335416459469,
      "learning_rate": 7.4272406729853805e-06,
      "loss": 0.7861,
      "step": 2801
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3971211960458714,
      "learning_rate": 7.425428014902304e-06,
      "loss": 0.6695,
      "step": 2802
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2742692064293397,
      "learning_rate": 7.423614939860033e-06,
      "loss": 0.8128,
      "step": 2803
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2657087407126122,
      "learning_rate": 7.421801448170256e-06,
      "loss": 0.7592,
      "step": 2804
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5062756706671137,
      "learning_rate": 7.419987540144733e-06,
      "loss": 0.8467,
      "step": 2805
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4308363286813564,
      "learning_rate": 7.4181732160952995e-06,
      "loss": 0.8235,
      "step": 2806
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.460342126847787,
      "learning_rate": 7.416358476333856e-06,
      "loss": 0.8091,
      "step": 2807
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5175855740186053,
      "learning_rate": 7.414543321172376e-06,
      "loss": 0.7291,
      "step": 2808
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2894702731902616,
      "learning_rate": 7.412727750922908e-06,
      "loss": 0.85,
      "step": 2809
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.273175065700348,
      "learning_rate": 7.4109117658975685e-06,
      "loss": 0.731,
      "step": 2810
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7822048945847033,
      "learning_rate": 7.409095366408547e-06,
      "loss": 0.6945,
      "step": 2811
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2629126651274274,
      "learning_rate": 7.407278552768103e-06,
      "loss": 0.7673,
      "step": 2812
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2485840063840625,
      "learning_rate": 7.40546132528857e-06,
      "loss": 0.7652,
      "step": 2813
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3123594764955744,
      "learning_rate": 7.403643684282347e-06,
      "loss": 0.795,
      "step": 2814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8623619151164373,
      "learning_rate": 7.40182563006191e-06,
      "loss": 0.7202,
      "step": 2815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7737464047986586,
      "learning_rate": 7.4000071629398015e-06,
      "loss": 0.717,
      "step": 2816
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5021580081460406,
      "learning_rate": 7.398188283228641e-06,
      "loss": 0.7525,
      "step": 2817
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.662390816636065,
      "learning_rate": 7.39636899124111e-06,
      "loss": 0.7093,
      "step": 2818
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3419964739616979,
      "learning_rate": 7.39454928728997e-06,
      "loss": 0.7732,
      "step": 2819
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4850861369182482,
      "learning_rate": 7.392729171688047e-06,
      "loss": 0.8165,
      "step": 2820
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2237587095693858,
      "learning_rate": 7.39090864474824e-06,
      "loss": 0.783,
      "step": 2821
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3125498946352598,
      "learning_rate": 7.389087706783518e-06,
      "loss": 0.8235,
      "step": 2822
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0389218166502343,
      "learning_rate": 7.387266358106925e-06,
      "loss": 0.7011,
      "step": 2823
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4496998502270166,
      "learning_rate": 7.385444599031568e-06,
      "loss": 0.8078,
      "step": 2824
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.578342029727924,
      "learning_rate": 7.38362242987063e-06,
      "loss": 0.7311,
      "step": 2825
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.288622006501817,
      "learning_rate": 7.381799850937363e-06,
      "loss": 0.8504,
      "step": 2826
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4282800941201794,
      "learning_rate": 7.379976862545091e-06,
      "loss": 0.774,
      "step": 2827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.789034328475477,
      "learning_rate": 7.378153465007203e-06,
      "loss": 0.698,
      "step": 2828
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4293932986737166,
      "learning_rate": 7.376329658637167e-06,
      "loss": 0.8583,
      "step": 2829
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2894836690704343,
      "learning_rate": 7.374505443748512e-06,
      "loss": 0.7319,
      "step": 2830
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4075871757984022,
      "learning_rate": 7.372680820654844e-06,
      "loss": 0.7562,
      "step": 2831
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.392927229294536,
      "learning_rate": 7.3708557896698375e-06,
      "loss": 0.7572,
      "step": 2832
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2674358215519625,
      "learning_rate": 7.3690303511072324e-06,
      "loss": 0.8251,
      "step": 2833
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5362542760114404,
      "learning_rate": 7.367204505280848e-06,
      "loss": 0.7435,
      "step": 2834
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.38591785685084,
      "learning_rate": 7.365378252504567e-06,
      "loss": 0.8896,
      "step": 2835
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8264500933462445,
      "learning_rate": 7.363551593092342e-06,
      "loss": 0.7165,
      "step": 2836
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3585581988767155,
      "learning_rate": 7.361724527358195e-06,
      "loss": 0.7361,
      "step": 2837
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7878541564048683,
      "learning_rate": 7.359897055616225e-06,
      "loss": 0.6419,
      "step": 2838
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.227283392462127,
      "learning_rate": 7.358069178180592e-06,
      "loss": 0.7749,
      "step": 2839
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4053173958107126,
      "learning_rate": 7.356240895365531e-06,
      "loss": 0.793,
      "step": 2840
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4399444284747167,
      "learning_rate": 7.354412207485343e-06,
      "loss": 0.8063,
      "step": 2841
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0250554965937124,
      "learning_rate": 7.352583114854402e-06,
      "loss": 0.8276,
      "step": 2842
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0253969221948573,
      "learning_rate": 7.350753617787151e-06,
      "loss": 0.7046,
      "step": 2843
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5958383708427837,
      "learning_rate": 7.348923716598099e-06,
      "loss": 0.7423,
      "step": 2844
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.347707195828423,
      "learning_rate": 7.347093411601831e-06,
      "loss": 0.766,
      "step": 2845
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3815159640831118,
      "learning_rate": 7.3452627031129964e-06,
      "loss": 0.7507,
      "step": 2846
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.6185553058713014,
      "learning_rate": 7.3434315914463125e-06,
      "loss": 0.7971,
      "step": 2847
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3281647022979628,
      "learning_rate": 7.341600076916571e-06,
      "loss": 0.9269,
      "step": 2848
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.413750164265087,
      "learning_rate": 7.33976815983863e-06,
      "loss": 0.9121,
      "step": 2849
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3128316097094066,
      "learning_rate": 7.3379358405274195e-06,
      "loss": 0.8132,
      "step": 2850
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3351319467362932,
      "learning_rate": 7.33610311929793e-06,
      "loss": 0.7777,
      "step": 2851
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3968598362722822,
      "learning_rate": 7.334269996465236e-06,
      "loss": 0.7212,
      "step": 2852
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2985190657826249,
      "learning_rate": 7.332436472344468e-06,
      "loss": 0.7952,
      "step": 2853
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9119864556378003,
      "learning_rate": 7.330602547250828e-06,
      "loss": 0.6391,
      "step": 2854
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8507805488086754,
      "learning_rate": 7.3287682214995934e-06,
      "loss": 0.6685,
      "step": 2855
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.408540374519398,
      "learning_rate": 7.326933495406103e-06,
      "loss": 0.8152,
      "step": 2856
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.167992999262074,
      "learning_rate": 7.32509836928577e-06,
      "loss": 0.7815,
      "step": 2857
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.755471120972661,
      "learning_rate": 7.323262843454071e-06,
      "loss": 0.5817,
      "step": 2858
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2670667352130922,
      "learning_rate": 7.321426918226557e-06,
      "loss": 0.7275,
      "step": 2859
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2440925068743154,
      "learning_rate": 7.319590593918844e-06,
      "loss": 0.8603,
      "step": 2860
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2343785959267346,
      "learning_rate": 7.317753870846615e-06,
      "loss": 0.7686,
      "step": 2861
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2614444758866599,
      "learning_rate": 7.315916749325626e-06,
      "loss": 0.7239,
      "step": 2862
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4108077287693557,
      "learning_rate": 7.3140792296717004e-06,
      "loss": 0.8578,
      "step": 2863
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.456055156350077,
      "learning_rate": 7.312241312200727e-06,
      "loss": 0.8051,
      "step": 2864
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3497914506886883,
      "learning_rate": 7.310402997228667e-06,
      "loss": 0.8653,
      "step": 2865
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9346466336092977,
      "learning_rate": 7.308564285071547e-06,
      "loss": 0.682,
      "step": 2866
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.876895572488699,
      "learning_rate": 7.306725176045464e-06,
      "loss": 0.6233,
      "step": 2867
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.212971167450217,
      "learning_rate": 7.304885670466581e-06,
      "loss": 0.8439,
      "step": 2868
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.305245632101986,
      "learning_rate": 7.3030457686511305e-06,
      "loss": 0.9036,
      "step": 2869
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4650840131141696,
      "learning_rate": 7.3012054709154124e-06,
      "loss": 0.849,
      "step": 2870
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.9077421791835505,
      "learning_rate": 7.299364777575797e-06,
      "loss": 0.7995,
      "step": 2871
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.344953164610426,
      "learning_rate": 7.29752368894872e-06,
      "loss": 0.8912,
      "step": 2872
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3419160556752494,
      "learning_rate": 7.295682205350685e-06,
      "loss": 0.7923,
      "step": 2873
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4606890798617183,
      "learning_rate": 7.293840327098265e-06,
      "loss": 0.7945,
      "step": 2874
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.372875150101601,
      "learning_rate": 7.291998054508102e-06,
      "loss": 0.7726,
      "step": 2875
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3080816408730076,
      "learning_rate": 7.2901553878969e-06,
      "loss": 0.7046,
      "step": 2876
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1467533933245857,
      "learning_rate": 7.288312327581439e-06,
      "loss": 0.8234,
      "step": 2877
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.244012902277846,
      "learning_rate": 7.286468873878559e-06,
      "loss": 0.859,
      "step": 2878
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4194121264291406,
      "learning_rate": 7.2846250271051735e-06,
      "loss": 0.7417,
      "step": 2879
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2593590318925547,
      "learning_rate": 7.282780787578258e-06,
      "loss": 0.7319,
      "step": 2880
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4557700117271266,
      "learning_rate": 7.280936155614864e-06,
      "loss": 0.8261,
      "step": 2881
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3009742198393521,
      "learning_rate": 7.2790911315321015e-06,
      "loss": 0.868,
      "step": 2882
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.443934250845444,
      "learning_rate": 7.2772457156471496e-06,
      "loss": 0.7458,
      "step": 2883
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6170764074014612,
      "learning_rate": 7.275399908277261e-06,
      "loss": 0.8896,
      "step": 2884
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4919887828260632,
      "learning_rate": 7.273553709739749e-06,
      "loss": 0.7798,
      "step": 2885
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6325562125751827,
      "learning_rate": 7.271707120351997e-06,
      "loss": 0.7795,
      "step": 2886
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2759083455909637,
      "learning_rate": 7.269860140431455e-06,
      "loss": 0.888,
      "step": 2887
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4160522843045773,
      "learning_rate": 7.268012770295641e-06,
      "loss": 0.7795,
      "step": 2888
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4675544151496895,
      "learning_rate": 7.266165010262138e-06,
      "loss": 0.7723,
      "step": 2889
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.394325463374443,
      "learning_rate": 7.264316860648598e-06,
      "loss": 0.8143,
      "step": 2890
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6730563148174427,
      "learning_rate": 7.26246832177274e-06,
      "loss": 0.7216,
      "step": 2891
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3949962210462108,
      "learning_rate": 7.2606193939523496e-06,
      "loss": 0.7629,
      "step": 2892
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.487360501812043,
      "learning_rate": 7.258770077505276e-06,
      "loss": 0.8116,
      "step": 2893
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8054991548026821,
      "learning_rate": 7.256920372749441e-06,
      "loss": 0.6415,
      "step": 2894
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2139651852628333,
      "learning_rate": 7.255070280002829e-06,
      "loss": 0.8073,
      "step": 2895
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3503125219497423,
      "learning_rate": 7.253219799583495e-06,
      "loss": 0.7886,
      "step": 2896
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8225125179268782,
      "learning_rate": 7.251368931809554e-06,
      "loss": 0.6648,
      "step": 2897
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3809929563525023,
      "learning_rate": 7.249517676999192e-06,
      "loss": 0.8264,
      "step": 2898
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4832224158572254,
      "learning_rate": 7.247666035470666e-06,
      "loss": 0.8611,
      "step": 2899
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.537624017183132,
      "learning_rate": 7.24581400754229e-06,
      "loss": 0.761,
      "step": 2900
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5250316048400938,
      "learning_rate": 7.24396159353245e-06,
      "loss": 0.7808,
      "step": 2901
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6985632243935838,
      "learning_rate": 7.242108793759597e-06,
      "loss": 0.8416,
      "step": 2902
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3155482041234545,
      "learning_rate": 7.240255608542252e-06,
      "loss": 0.7449,
      "step": 2903
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5598229534351793,
      "learning_rate": 7.238402038198995e-06,
      "loss": 0.7938,
      "step": 2904
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2390675855539803,
      "learning_rate": 7.236548083048478e-06,
      "loss": 0.7894,
      "step": 2905
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2880499310517075,
      "learning_rate": 7.234693743409418e-06,
      "loss": 0.7621,
      "step": 2906
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3651426580556538,
      "learning_rate": 7.232839019600595e-06,
      "loss": 0.8652,
      "step": 2907
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.303239641688054,
      "learning_rate": 7.230983911940861e-06,
      "loss": 0.7322,
      "step": 2908
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3921751787676122,
      "learning_rate": 7.229128420749127e-06,
      "loss": 0.8292,
      "step": 2909
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3660755913247333,
      "learning_rate": 7.227272546344377e-06,
      "loss": 0.721,
      "step": 2910
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2880561767643413,
      "learning_rate": 7.225416289045655e-06,
      "loss": 0.7456,
      "step": 2911
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.824542427566471,
      "learning_rate": 7.2235596491720724e-06,
      "loss": 0.7079,
      "step": 2912
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3289667985518363,
      "learning_rate": 7.22170262704281e-06,
      "loss": 0.8564,
      "step": 2913
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3546606243071395,
      "learning_rate": 7.21984522297711e-06,
      "loss": 0.7673,
      "step": 2914
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8223893843320795,
      "learning_rate": 7.217987437294281e-06,
      "loss": 0.664,
      "step": 2915
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4638266408815608,
      "learning_rate": 7.216129270313698e-06,
      "loss": 0.8855,
      "step": 2916
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8225126887252734,
      "learning_rate": 7.214270722354802e-06,
      "loss": 0.6287,
      "step": 2917
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3643998058339974,
      "learning_rate": 7.2124117937371e-06,
      "loss": 0.7868,
      "step": 2918
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4843915620321484,
      "learning_rate": 7.21055248478016e-06,
      "loss": 0.8356,
      "step": 2919
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8703203536978158,
      "learning_rate": 7.208692795803622e-06,
      "loss": 0.6487,
      "step": 2920
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3895855060771534,
      "learning_rate": 7.206832727127186e-06,
      "loss": 0.7797,
      "step": 2921
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4168085772528578,
      "learning_rate": 7.204972279070623e-06,
      "loss": 0.8273,
      "step": 2922
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1928941738047478,
      "learning_rate": 7.203111451953761e-06,
      "loss": 0.7012,
      "step": 2923
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4962377892988292,
      "learning_rate": 7.201250246096501e-06,
      "loss": 0.8389,
      "step": 2924
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.59182275322572,
      "learning_rate": 7.1993886618188025e-06,
      "loss": 0.8185,
      "step": 2925
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5177082051763788,
      "learning_rate": 7.1975266994406965e-06,
      "loss": 0.7022,
      "step": 2926
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.661773870759736,
      "learning_rate": 7.195664359282275e-06,
      "loss": 0.7987,
      "step": 2927
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8728725227536229,
      "learning_rate": 7.193801641663697e-06,
      "loss": 0.621,
      "step": 2928
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2714588917464948,
      "learning_rate": 7.191938546905183e-06,
      "loss": 0.6733,
      "step": 2929
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4295600707700618,
      "learning_rate": 7.190075075327021e-06,
      "loss": 0.7983,
      "step": 2930
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.373724373949766,
      "learning_rate": 7.188211227249565e-06,
      "loss": 0.8406,
      "step": 2931
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.394512950589737,
      "learning_rate": 7.186347002993233e-06,
      "loss": 0.8026,
      "step": 2932
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.541250954126412,
      "learning_rate": 7.184482402878501e-06,
      "loss": 0.7757,
      "step": 2933
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9301945029787748,
      "learning_rate": 7.182617427225922e-06,
      "loss": 0.7841,
      "step": 2934
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4128612087815924,
      "learning_rate": 7.180752076356102e-06,
      "loss": 0.7575,
      "step": 2935
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8383271810179368,
      "learning_rate": 7.178886350589721e-06,
      "loss": 0.6548,
      "step": 2936
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4011085054554762,
      "learning_rate": 7.177020250247515e-06,
      "loss": 0.6653,
      "step": 2937
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2578012300303243,
      "learning_rate": 7.17515377565029e-06,
      "loss": 0.7696,
      "step": 2938
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7937725304837782,
      "learning_rate": 7.173286927118914e-06,
      "loss": 0.6179,
      "step": 2939
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5897564671446047,
      "learning_rate": 7.171419704974321e-06,
      "loss": 0.7813,
      "step": 2940
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3646291082512894,
      "learning_rate": 7.169552109537507e-06,
      "loss": 0.8242,
      "step": 2941
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3036222416417873,
      "learning_rate": 7.167684141129536e-06,
      "loss": 0.7409,
      "step": 2942
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2376359791292044,
      "learning_rate": 7.165815800071529e-06,
      "loss": 0.7756,
      "step": 2943
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.273549120617412,
      "learning_rate": 7.163947086684677e-06,
      "loss": 0.7202,
      "step": 2944
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9656648634310926,
      "learning_rate": 7.1620780012902356e-06,
      "loss": 0.7011,
      "step": 2945
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8489515875669478,
      "learning_rate": 7.160208544209521e-06,
      "loss": 0.6005,
      "step": 2946
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6013156431342757,
      "learning_rate": 7.158338715763912e-06,
      "loss": 0.8225,
      "step": 2947
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5191983994148905,
      "learning_rate": 7.156468516274859e-06,
      "loss": 0.709,
      "step": 2948
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3080844338619149,
      "learning_rate": 7.154597946063867e-06,
      "loss": 0.7354,
      "step": 2949
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5766582794050272,
      "learning_rate": 7.152727005452511e-06,
      "loss": 0.7964,
      "step": 2950
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6529589783576437,
      "learning_rate": 7.1508556947624245e-06,
      "loss": 0.8651,
      "step": 2951
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.372995918099208,
      "learning_rate": 7.14898401431531e-06,
      "loss": 0.7343,
      "step": 2952
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5433053596369382,
      "learning_rate": 7.14711196443293e-06,
      "loss": 0.7204,
      "step": 2953
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4547332981716983,
      "learning_rate": 7.145239545437113e-06,
      "loss": 0.7434,
      "step": 2954
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5522003317282644,
      "learning_rate": 7.143366757649746e-06,
      "loss": 0.7259,
      "step": 2955
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.27095181983738,
      "learning_rate": 7.141493601392787e-06,
      "loss": 0.7428,
      "step": 2956
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3430654504815485,
      "learning_rate": 7.139620076988252e-06,
      "loss": 0.7643,
      "step": 2957
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7607662321389268,
      "learning_rate": 7.137746184758218e-06,
      "loss": 0.8409,
      "step": 2958
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.322792169012627,
      "learning_rate": 7.135871925024835e-06,
      "loss": 0.818,
      "step": 2959
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8340041400294794,
      "learning_rate": 7.133997298110308e-06,
      "loss": 0.6863,
      "step": 2960
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.350593873585702,
      "learning_rate": 7.1321223043369034e-06,
      "loss": 0.7604,
      "step": 2961
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.699348541461064,
      "learning_rate": 7.130246944026958e-06,
      "loss": 0.6617,
      "step": 2962
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3398765974963533,
      "learning_rate": 7.128371217502868e-06,
      "loss": 0.7974,
      "step": 2963
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4128030282398474,
      "learning_rate": 7.12649512508709e-06,
      "loss": 0.8001,
      "step": 2964
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6198545661522024,
      "learning_rate": 7.1246186671021475e-06,
      "loss": 0.8545,
      "step": 2965
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3322415602372728,
      "learning_rate": 7.122741843870626e-06,
      "loss": 0.7922,
      "step": 2966
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3855082562104215,
      "learning_rate": 7.120864655715172e-06,
      "loss": 0.8524,
      "step": 2967
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.485342031089243,
      "learning_rate": 7.118987102958498e-06,
      "loss": 0.8098,
      "step": 2968
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3767814116683306,
      "learning_rate": 7.117109185923374e-06,
      "loss": 0.7763,
      "step": 2969
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4696395662706925,
      "learning_rate": 7.115230904932639e-06,
      "loss": 0.8771,
      "step": 2970
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3995425051708126,
      "learning_rate": 7.113352260309189e-06,
      "loss": 0.8392,
      "step": 2971
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3705932165063222,
      "learning_rate": 7.111473252375983e-06,
      "loss": 0.705,
      "step": 2972
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7048078809802156,
      "learning_rate": 7.109593881456048e-06,
      "loss": 0.8021,
      "step": 2973
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3547134101854457,
      "learning_rate": 7.10771414787247e-06,
      "loss": 0.918,
      "step": 2974
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5713233186523536,
      "learning_rate": 7.105834051948395e-06,
      "loss": 0.8837,
      "step": 2975
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.6113894381917016,
      "learning_rate": 7.1039535940070305e-06,
      "loss": 0.9116,
      "step": 2976
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6806658892456898,
      "learning_rate": 7.102072774371654e-06,
      "loss": 0.7491,
      "step": 2977
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3457642593099246,
      "learning_rate": 7.1001915933655994e-06,
      "loss": 0.7629,
      "step": 2978
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6014011401629356,
      "learning_rate": 7.098310051312261e-06,
      "loss": 0.812,
      "step": 2979
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4609594955547358,
      "learning_rate": 7.096428148535101e-06,
      "loss": 0.7419,
      "step": 2980
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5792756640186838,
      "learning_rate": 7.094545885357636e-06,
      "loss": 0.8722,
      "step": 2981
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5019917942044587,
      "learning_rate": 7.092663262103452e-06,
      "loss": 0.8058,
      "step": 2982
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.523398695719254,
      "learning_rate": 7.0907802790961925e-06,
      "loss": 0.6945,
      "step": 2983
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2417978939817818,
      "learning_rate": 7.088896936659566e-06,
      "loss": 0.8394,
      "step": 2984
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3807833541945669,
      "learning_rate": 7.087013235117339e-06,
      "loss": 0.7676,
      "step": 2985
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.399812452613616,
      "learning_rate": 7.0851291747933415e-06,
      "loss": 0.837,
      "step": 2986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8612743283081867,
      "learning_rate": 7.083244756011466e-06,
      "loss": 0.675,
      "step": 2987
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3893120270302424,
      "learning_rate": 7.081359979095667e-06,
      "loss": 0.7726,
      "step": 2988
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.478399712993501,
      "learning_rate": 7.079474844369958e-06,
      "loss": 0.7846,
      "step": 2989
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.472845727337586,
      "learning_rate": 7.077589352158415e-06,
      "loss": 0.7952,
      "step": 2990
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3898322579213074,
      "learning_rate": 7.075703502785178e-06,
      "loss": 0.8453,
      "step": 2991
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3760165117787564,
      "learning_rate": 7.073817296574444e-06,
      "loss": 0.8111,
      "step": 2992
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5297222657162632,
      "learning_rate": 7.071930733850476e-06,
      "loss": 0.8612,
      "step": 2993
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3613598263466,
      "learning_rate": 7.070043814937595e-06,
      "loss": 0.7733,
      "step": 2994
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2639176355872153,
      "learning_rate": 7.068156540160182e-06,
      "loss": 0.7892,
      "step": 2995
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5089979149862938,
      "learning_rate": 7.066268909842687e-06,
      "loss": 0.8247,
      "step": 2996
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5381195600743696,
      "learning_rate": 7.06438092430961e-06,
      "loss": 0.8502,
      "step": 2997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8389689054106236,
      "learning_rate": 7.062492583885521e-06,
      "loss": 0.6898,
      "step": 2998
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4973862232812698,
      "learning_rate": 7.060603888895046e-06,
      "loss": 0.8618,
      "step": 2999
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5262982532544802,
      "learning_rate": 7.058714839662874e-06,
      "loss": 0.8206,
      "step": 3000
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4151303264300952,
      "learning_rate": 7.056825436513754e-06,
      "loss": 0.87,
      "step": 3001
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.458138875379286,
      "learning_rate": 7.054935679772497e-06,
      "loss": 0.7741,
      "step": 3002
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.399996746783685,
      "learning_rate": 7.053045569763973e-06,
      "loss": 0.6973,
      "step": 3003
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.950535172102551,
      "learning_rate": 7.051155106813114e-06,
      "loss": 0.7896,
      "step": 3004
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.002925662964554,
      "learning_rate": 7.049264291244915e-06,
      "loss": 0.7549,
      "step": 3005
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5879659545191795,
      "learning_rate": 7.047373123384426e-06,
      "loss": 0.7894,
      "step": 3006
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4994744138977196,
      "learning_rate": 7.045481603556763e-06,
      "loss": 0.7973,
      "step": 3007
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.356604735408256,
      "learning_rate": 7.043589732087098e-06,
      "loss": 0.8007,
      "step": 3008
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4990771425751621,
      "learning_rate": 7.041697509300667e-06,
      "loss": 0.7064,
      "step": 3009
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8630648499579789,
      "learning_rate": 7.039804935522766e-06,
      "loss": 0.7032,
      "step": 3010
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3033538447173896,
      "learning_rate": 7.037912011078749e-06,
      "loss": 0.7443,
      "step": 3011
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6823532206885405,
      "learning_rate": 7.03601873629403e-06,
      "loss": 0.8053,
      "step": 3012
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.353690163189192,
      "learning_rate": 7.0341251114940864e-06,
      "loss": 0.8297,
      "step": 3013
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.310278686525061,
      "learning_rate": 7.032231137004457e-06,
      "loss": 0.6523,
      "step": 3014
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6292137968082359,
      "learning_rate": 7.030336813150734e-06,
      "loss": 0.803,
      "step": 3015
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3105833239038374,
      "learning_rate": 7.028442140258576e-06,
      "loss": 0.8485,
      "step": 3016
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2964085955793991,
      "learning_rate": 7.026547118653697e-06,
      "loss": 0.8044,
      "step": 3017
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.309932415743668,
      "learning_rate": 7.024651748661875e-06,
      "loss": 0.7394,
      "step": 3018
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6930973551644892,
      "learning_rate": 7.022756030608946e-06,
      "loss": 0.873,
      "step": 3019
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4871283192230733,
      "learning_rate": 7.0208599648208054e-06,
      "loss": 0.7471,
      "step": 3020
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9516180286824515,
      "learning_rate": 7.01896355162341e-06,
      "loss": 0.7077,
      "step": 3021
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4809495557839605,
      "learning_rate": 7.017066791342773e-06,
      "loss": 0.7834,
      "step": 3022
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5024580973589337,
      "learning_rate": 7.01516968430497e-06,
      "loss": 0.7925,
      "step": 3023
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8652342169532162,
      "learning_rate": 7.013272230836139e-06,
      "loss": 0.6126,
      "step": 3024
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3797830093674905,
      "learning_rate": 7.01137443126247e-06,
      "loss": 0.8344,
      "step": 3025
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4769479550609648,
      "learning_rate": 7.009476285910218e-06,
      "loss": 0.8421,
      "step": 3026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9216084619645708,
      "learning_rate": 7.007577795105697e-06,
      "loss": 0.664,
      "step": 3027
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3836430044516506,
      "learning_rate": 7.005678959175279e-06,
      "loss": 0.8325,
      "step": 3028
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3049826371039341,
      "learning_rate": 7.003779778445398e-06,
      "loss": 0.8184,
      "step": 3029
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5161371817164628,
      "learning_rate": 7.001880253242541e-06,
      "loss": 0.7873,
      "step": 3030
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4697354041882844,
      "learning_rate": 6.999980383893261e-06,
      "loss": 0.8226,
      "step": 3031
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2737267746730678,
      "learning_rate": 6.998080170724167e-06,
      "loss": 0.7657,
      "step": 3032
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5486036467872242,
      "learning_rate": 6.996179614061929e-06,
      "loss": 0.7598,
      "step": 3033
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5356501004002134,
      "learning_rate": 6.9942787142332735e-06,
      "loss": 0.8275,
      "step": 3034
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3700815927212042,
      "learning_rate": 6.992377471564987e-06,
      "loss": 0.7863,
      "step": 3035
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4201434137182016,
      "learning_rate": 6.990475886383915e-06,
      "loss": 0.8804,
      "step": 3036
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.300847364211352,
      "learning_rate": 6.988573959016963e-06,
      "loss": 0.8782,
      "step": 3037
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.393564283692947,
      "learning_rate": 6.9866716897910945e-06,
      "loss": 0.7037,
      "step": 3038
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5973141267690107,
      "learning_rate": 6.984769079033331e-06,
      "loss": 0.8553,
      "step": 3039
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.798218345295093,
      "learning_rate": 6.982866127070753e-06,
      "loss": 0.6408,
      "step": 3040
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5856296323025356,
      "learning_rate": 6.9809628342305e-06,
      "loss": 0.8199,
      "step": 3041
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0691066565639744,
      "learning_rate": 6.9790592008397705e-06,
      "loss": 0.7033,
      "step": 3042
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3304780257456292,
      "learning_rate": 6.9771552272258226e-06,
      "loss": 0.8896,
      "step": 3043
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.356331012093158,
      "learning_rate": 6.975250913715968e-06,
      "loss": 0.8021,
      "step": 3044
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4768695332383421,
      "learning_rate": 6.973346260637583e-06,
      "loss": 0.866,
      "step": 3045
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3562584613508788,
      "learning_rate": 6.9714412683181e-06,
      "loss": 0.764,
      "step": 3046
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4151868848274183,
      "learning_rate": 6.969535937085006e-06,
      "loss": 0.8326,
      "step": 3047
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3811952060630959,
      "learning_rate": 6.967630267265854e-06,
      "loss": 0.748,
      "step": 3048
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.835897994259201,
      "learning_rate": 6.965724259188246e-06,
      "loss": 0.691,
      "step": 3049
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5331220717143763,
      "learning_rate": 6.96381791317985e-06,
      "loss": 0.796,
      "step": 3050
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2193811843140505,
      "learning_rate": 6.961911229568388e-06,
      "loss": 0.8331,
      "step": 3051
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.330314421279358,
      "learning_rate": 6.9600042086816424e-06,
      "loss": 0.7287,
      "step": 3052
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2554294115774036,
      "learning_rate": 6.958096850847451e-06,
      "loss": 0.7923,
      "step": 3053
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1625022501384796,
      "learning_rate": 6.956189156393709e-06,
      "loss": 0.7032,
      "step": 3054
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.2255501172358985,
      "learning_rate": 6.954281125648373e-06,
      "loss": 0.8497,
      "step": 3055
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9816228845523667,
      "learning_rate": 6.952372758939457e-06,
      "loss": 0.645,
      "step": 3056
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3289565731169704,
      "learning_rate": 6.9504640565950295e-06,
      "loss": 0.8322,
      "step": 3057
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.471748051709676,
      "learning_rate": 6.948555018943219e-06,
      "loss": 0.7064,
      "step": 3058
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.365493937148685,
      "learning_rate": 6.946645646312212e-06,
      "loss": 0.6879,
      "step": 3059
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8362112229361596,
      "learning_rate": 6.944735939030249e-06,
      "loss": 0.7431,
      "step": 3060
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4446362011104892,
      "learning_rate": 6.942825897425633e-06,
      "loss": 0.7488,
      "step": 3061
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2671847409547992,
      "learning_rate": 6.940915521826723e-06,
      "loss": 0.7117,
      "step": 3062
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.452138464850027,
      "learning_rate": 6.939004812561934e-06,
      "loss": 0.809,
      "step": 3063
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5534414285507514,
      "learning_rate": 6.937093769959737e-06,
      "loss": 0.8321,
      "step": 3064
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7909481278336998,
      "learning_rate": 6.9351823943486654e-06,
      "loss": 0.6531,
      "step": 3065
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3401218388416434,
      "learning_rate": 6.933270686057308e-06,
      "loss": 0.8142,
      "step": 3066
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9239435197520599,
      "learning_rate": 6.931358645414304e-06,
      "loss": 0.6262,
      "step": 3067
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.718109003822597,
      "learning_rate": 6.929446272748361e-06,
      "loss": 0.7698,
      "step": 3068
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3851979559443155,
      "learning_rate": 6.927533568388232e-06,
      "loss": 0.8087,
      "step": 3069
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8726136563446462,
      "learning_rate": 6.92562053266274e-06,
      "loss": 0.6834,
      "step": 3070
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.414729697629926,
      "learning_rate": 6.923707165900753e-06,
      "loss": 0.7983,
      "step": 3071
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4763499830950462,
      "learning_rate": 6.921793468431204e-06,
      "loss": 0.74,
      "step": 3072
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.355166594435281,
      "learning_rate": 6.9198794405830776e-06,
      "loss": 0.8104,
      "step": 3073
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3781373586576926,
      "learning_rate": 6.917965082685418e-06,
      "loss": 0.7981,
      "step": 3074
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.353199693462041,
      "learning_rate": 6.9160503950673276e-06,
      "loss": 0.7046,
      "step": 3075
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9650086058133799,
      "learning_rate": 6.914135378057959e-06,
      "loss": 0.6366,
      "step": 3076
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3586224437541525,
      "learning_rate": 6.912220031986531e-06,
      "loss": 0.7923,
      "step": 3077
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5210625330109915,
      "learning_rate": 6.910304357182308e-06,
      "loss": 0.7926,
      "step": 3078
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2229045389908146,
      "learning_rate": 6.908388353974622e-06,
      "loss": 0.7395,
      "step": 3079
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5309456299029478,
      "learning_rate": 6.906472022692854e-06,
      "loss": 0.7994,
      "step": 3080
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.50155305178557,
      "learning_rate": 6.904555363666443e-06,
      "loss": 0.8111,
      "step": 3081
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.829759019508545,
      "learning_rate": 6.902638377224886e-06,
      "loss": 0.778,
      "step": 3082
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6034814986699468,
      "learning_rate": 6.900721063697733e-06,
      "loss": 0.7848,
      "step": 3083
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9152330161352363,
      "learning_rate": 6.898803423414595e-06,
      "loss": 0.72,
      "step": 3084
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3376080543506716,
      "learning_rate": 6.896885456705137e-06,
      "loss": 0.843,
      "step": 3085
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.391161129530749,
      "learning_rate": 6.894967163899077e-06,
      "loss": 0.8487,
      "step": 3086
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2795984350929677,
      "learning_rate": 6.893048545326193e-06,
      "loss": 0.8011,
      "step": 3087
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5033376773262284,
      "learning_rate": 6.891129601316319e-06,
      "loss": 0.7803,
      "step": 3088
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3893035293710696,
      "learning_rate": 6.889210332199343e-06,
      "loss": 0.7473,
      "step": 3089
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.23177326124834,
      "learning_rate": 6.887290738305208e-06,
      "loss": 0.8084,
      "step": 3090
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6444928233340947,
      "learning_rate": 6.885370819963917e-06,
      "loss": 0.7862,
      "step": 3091
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.430199790546697,
      "learning_rate": 6.883450577505524e-06,
      "loss": 0.8056,
      "step": 3092
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8709429797972174,
      "learning_rate": 6.881530011260142e-06,
      "loss": 0.5986,
      "step": 3093
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8181683132717353,
      "learning_rate": 6.879609121557938e-06,
      "loss": 0.6337,
      "step": 3094
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.692149990159472,
      "learning_rate": 6.877687908729137e-06,
      "loss": 0.8998,
      "step": 3095
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5292369771610568,
      "learning_rate": 6.875766373104016e-06,
      "loss": 0.8186,
      "step": 3096
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2639979373422423,
      "learning_rate": 6.873844515012909e-06,
      "loss": 0.8582,
      "step": 3097
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6655029467609999,
      "learning_rate": 6.871922334786206e-06,
      "loss": 0.7663,
      "step": 3098
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3477142168546845,
      "learning_rate": 6.8699998327543545e-06,
      "loss": 0.8189,
      "step": 3099
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5195357393843543,
      "learning_rate": 6.868077009247852e-06,
      "loss": 0.7564,
      "step": 3100
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.390312598999224,
      "learning_rate": 6.866153864597254e-06,
      "loss": 0.8119,
      "step": 3101
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6082058938197135,
      "learning_rate": 6.864230399133172e-06,
      "loss": 0.9332,
      "step": 3102
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.27010455539516,
      "learning_rate": 6.862306613186275e-06,
      "loss": 0.6793,
      "step": 3103
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6579280059465873,
      "learning_rate": 6.86038250708728e-06,
      "loss": 0.8261,
      "step": 3104
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3097115191434896,
      "learning_rate": 6.858458081166964e-06,
      "loss": 0.7665,
      "step": 3105
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.948381781820529,
      "learning_rate": 6.856533335756159e-06,
      "loss": 0.6368,
      "step": 3106
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.438762334723869,
      "learning_rate": 6.854608271185752e-06,
      "loss": 0.7607,
      "step": 3107
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2453185594822123,
      "learning_rate": 6.852682887786681e-06,
      "loss": 0.6998,
      "step": 3108
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3968379943975038,
      "learning_rate": 6.850757185889945e-06,
      "loss": 0.8122,
      "step": 3109
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.553204119274465,
      "learning_rate": 6.848831165826591e-06,
      "loss": 0.7147,
      "step": 3110
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4070436753966775,
      "learning_rate": 6.846904827927728e-06,
      "loss": 0.8345,
      "step": 3111
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5684576176682106,
      "learning_rate": 6.844978172524514e-06,
      "loss": 0.8057,
      "step": 3112
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2546644457708904,
      "learning_rate": 6.843051199948162e-06,
      "loss": 0.7135,
      "step": 3113
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3335213082831328,
      "learning_rate": 6.841123910529943e-06,
      "loss": 0.7651,
      "step": 3114
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.204170824620502,
      "learning_rate": 6.839196304601179e-06,
      "loss": 0.7816,
      "step": 3115
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.446153658910147,
      "learning_rate": 6.837268382493248e-06,
      "loss": 0.8131,
      "step": 3116
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7196205122507684,
      "learning_rate": 6.835340144537584e-06,
      "loss": 0.6205,
      "step": 3117
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4242087993732246,
      "learning_rate": 6.833411591065673e-06,
      "loss": 0.8333,
      "step": 3118
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6205948815562057,
      "learning_rate": 6.831482722409053e-06,
      "loss": 0.7471,
      "step": 3119
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3523819610839236,
      "learning_rate": 6.829553538899321e-06,
      "loss": 0.758,
      "step": 3120
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8331005140687985,
      "learning_rate": 6.8276240408681245e-06,
      "loss": 0.7095,
      "step": 3121
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.413004343450755,
      "learning_rate": 6.825694228647169e-06,
      "loss": 0.816,
      "step": 3122
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5583920999722138,
      "learning_rate": 6.823764102568211e-06,
      "loss": 0.746,
      "step": 3123
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7789916635711501,
      "learning_rate": 6.82183366296306e-06,
      "loss": 0.604,
      "step": 3124
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6582139395494808,
      "learning_rate": 6.819902910163582e-06,
      "loss": 0.7586,
      "step": 3125
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6819112490707377,
      "learning_rate": 6.817971844501695e-06,
      "loss": 0.8446,
      "step": 3126
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5763260227089335,
      "learning_rate": 6.8160404663093725e-06,
      "loss": 0.6619,
      "step": 3127
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2532671763105188,
      "learning_rate": 6.814108775918642e-06,
      "loss": 0.7682,
      "step": 3128
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2927987051434158,
      "learning_rate": 6.812176773661579e-06,
      "loss": 0.8696,
      "step": 3129
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3664434579122107,
      "learning_rate": 6.810244459870322e-06,
      "loss": 0.73,
      "step": 3130
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4248176511208308,
      "learning_rate": 6.808311834877057e-06,
      "loss": 0.7922,
      "step": 3131
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4333687653313276,
      "learning_rate": 6.806378899014023e-06,
      "loss": 0.8207,
      "step": 3132
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.354118855650509,
      "learning_rate": 6.804445652613514e-06,
      "loss": 0.7328,
      "step": 3133
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3336422092090836,
      "learning_rate": 6.802512096007879e-06,
      "loss": 0.769,
      "step": 3134
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4924243664922416,
      "learning_rate": 6.800578229529519e-06,
      "loss": 0.7858,
      "step": 3135
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.466114380509059,
      "learning_rate": 6.798644053510886e-06,
      "loss": 0.7641,
      "step": 3136
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2908699602477678,
      "learning_rate": 6.796709568284488e-06,
      "loss": 0.8042,
      "step": 3137
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6249376250144223,
      "learning_rate": 6.794774774182887e-06,
      "loss": 0.8294,
      "step": 3138
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5042723950013521,
      "learning_rate": 6.792839671538696e-06,
      "loss": 0.7657,
      "step": 3139
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8470846196900474,
      "learning_rate": 6.790904260684581e-06,
      "loss": 0.6614,
      "step": 3140
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3703487386571525,
      "learning_rate": 6.788968541953262e-06,
      "loss": 0.8182,
      "step": 3141
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4256163411746963,
      "learning_rate": 6.787032515677509e-06,
      "loss": 0.7543,
      "step": 3142
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4168860313513036,
      "learning_rate": 6.785096182190152e-06,
      "loss": 0.7592,
      "step": 3143
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4195494384990086,
      "learning_rate": 6.783159541824065e-06,
      "loss": 0.7697,
      "step": 3144
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.293635442763303,
      "learning_rate": 6.781222594912182e-06,
      "loss": 0.7624,
      "step": 3145
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6686406851317768,
      "learning_rate": 6.779285341787484e-06,
      "loss": 0.8003,
      "step": 3146
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.312996420519856,
      "learning_rate": 6.7773477827830085e-06,
      "loss": 0.7698,
      "step": 3147
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3635891480313547,
      "learning_rate": 6.775409918231843e-06,
      "loss": 0.7576,
      "step": 3148
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3679040607556345,
      "learning_rate": 6.773471748467133e-06,
      "loss": 0.75,
      "step": 3149
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.596077440469715,
      "learning_rate": 6.771533273822069e-06,
      "loss": 0.8188,
      "step": 3150
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3745839464830654,
      "learning_rate": 6.769594494629898e-06,
      "loss": 0.6894,
      "step": 3151
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5773437353221365,
      "learning_rate": 6.767655411223917e-06,
      "loss": 0.8307,
      "step": 3152
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3687816830439445,
      "learning_rate": 6.76571602393748e-06,
      "loss": 0.7271,
      "step": 3153
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8540383552842126,
      "learning_rate": 6.7637763331039885e-06,
      "loss": 0.6854,
      "step": 3154
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3282430478817555,
      "learning_rate": 6.761836339056896e-06,
      "loss": 0.7002,
      "step": 3155
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7161593342947634,
      "learning_rate": 6.759896042129713e-06,
      "loss": 0.8023,
      "step": 3156
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8894115421757239,
      "learning_rate": 6.757955442655998e-06,
      "loss": 0.6514,
      "step": 3157
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5943825786452086,
      "learning_rate": 6.756014540969362e-06,
      "loss": 0.77,
      "step": 3158
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8586613453374173,
      "learning_rate": 6.754073337403469e-06,
      "loss": 0.7267,
      "step": 3159
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2836055002058921,
      "learning_rate": 6.752131832292036e-06,
      "loss": 0.7525,
      "step": 3160
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5780139528826802,
      "learning_rate": 6.750190025968827e-06,
      "loss": 0.7696,
      "step": 3161
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3816557159199,
      "learning_rate": 6.748247918767662e-06,
      "loss": 0.7478,
      "step": 3162
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3499708757724143,
      "learning_rate": 6.746305511022414e-06,
      "loss": 0.8136,
      "step": 3163
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3925427278450082,
      "learning_rate": 6.744362803067003e-06,
      "loss": 0.7917,
      "step": 3164
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.411964606200237,
      "learning_rate": 6.742419795235403e-06,
      "loss": 0.7752,
      "step": 3165
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.202111962843992,
      "learning_rate": 6.74047648786164e-06,
      "loss": 0.7432,
      "step": 3166
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0153622153394788,
      "learning_rate": 6.738532881279794e-06,
      "loss": 0.7252,
      "step": 3167
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8821526504790512,
      "learning_rate": 6.73658897582399e-06,
      "loss": 0.667,
      "step": 3168
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.391563124515629,
      "learning_rate": 6.734644771828407e-06,
      "loss": 0.8063,
      "step": 3169
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3475112968326532,
      "learning_rate": 6.7327002696272795e-06,
      "loss": 0.7302,
      "step": 3170
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.407675069849111,
      "learning_rate": 6.730755469554888e-06,
      "loss": 0.8447,
      "step": 3171
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3860770293983593,
      "learning_rate": 6.728810371945567e-06,
      "loss": 0.729,
      "step": 3172
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7113718744752586,
      "learning_rate": 6.7268649771337e-06,
      "loss": 0.781,
      "step": 3173
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3626249890623723,
      "learning_rate": 6.724919285453726e-06,
      "loss": 0.78,
      "step": 3174
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3415619668917453,
      "learning_rate": 6.722973297240128e-06,
      "loss": 0.7935,
      "step": 3175
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2964353961958117,
      "learning_rate": 6.721027012827444e-06,
      "loss": 0.7774,
      "step": 3176
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4992079769787414,
      "learning_rate": 6.719080432550269e-06,
      "loss": 0.8839,
      "step": 3177
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9493861701194997,
      "learning_rate": 6.7171335567432365e-06,
      "loss": 0.8342,
      "step": 3178
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.741485265499993,
      "learning_rate": 6.7151863857410375e-06,
      "loss": 0.8639,
      "step": 3179
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3962510946764333,
      "learning_rate": 6.713238919878415e-06,
      "loss": 0.835,
      "step": 3180
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3818768698083819,
      "learning_rate": 6.711291159490162e-06,
      "loss": 0.8821,
      "step": 3181
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5065179370150614,
      "learning_rate": 6.709343104911119e-06,
      "loss": 0.8265,
      "step": 3182
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0524008231634476,
      "learning_rate": 6.70739475647618e-06,
      "loss": 0.8732,
      "step": 3183
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6534001142315775,
      "learning_rate": 6.705446114520289e-06,
      "loss": 0.8087,
      "step": 3184
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.8047507720477016,
      "learning_rate": 6.7034971793784385e-06,
      "loss": 0.7563,
      "step": 3185
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3532074338167634,
      "learning_rate": 6.701547951385675e-06,
      "loss": 0.8078,
      "step": 3186
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.258272573543155,
      "learning_rate": 6.699598430877092e-06,
      "loss": 0.8063,
      "step": 3187
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3683526286837044,
      "learning_rate": 6.697648618187836e-06,
      "loss": 0.8081,
      "step": 3188
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.532634678142958,
      "learning_rate": 6.6956985136531015e-06,
      "loss": 0.7736,
      "step": 3189
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8732739997997641,
      "learning_rate": 6.693748117608134e-06,
      "loss": 0.6078,
      "step": 3190
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1430987007459499,
      "learning_rate": 6.69179743038823e-06,
      "loss": 0.7505,
      "step": 3191
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3471035514873366,
      "learning_rate": 6.6898464523287354e-06,
      "loss": 0.7506,
      "step": 3192
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2652208654593766,
      "learning_rate": 6.687895183765043e-06,
      "loss": 0.8181,
      "step": 3193
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3037507180581478,
      "learning_rate": 6.685943625032602e-06,
      "loss": 0.7341,
      "step": 3194
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6272217726159408,
      "learning_rate": 6.683991776466907e-06,
      "loss": 0.8641,
      "step": 3195
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.555213051690505,
      "learning_rate": 6.682039638403503e-06,
      "loss": 0.7928,
      "step": 3196
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.484165203126098,
      "learning_rate": 6.6800872111779854e-06,
      "loss": 0.7429,
      "step": 3197
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4619963565494447,
      "learning_rate": 6.678134495125999e-06,
      "loss": 0.8943,
      "step": 3198
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.40232749945398,
      "learning_rate": 6.676181490583238e-06,
      "loss": 0.7943,
      "step": 3199
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5741775041564308,
      "learning_rate": 6.674228197885448e-06,
      "loss": 0.7371,
      "step": 3200
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6588042899259505,
      "learning_rate": 6.67227461736842e-06,
      "loss": 0.8524,
      "step": 3201
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2290480795966012,
      "learning_rate": 6.670320749367998e-06,
      "loss": 0.7955,
      "step": 3202
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.899391206486452,
      "learning_rate": 6.668366594220076e-06,
      "loss": 0.7998,
      "step": 3203
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1571739329397661,
      "learning_rate": 6.666412152260595e-06,
      "loss": 0.6858,
      "step": 3204
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6180099861147765,
      "learning_rate": 6.664457423825547e-06,
      "loss": 0.8017,
      "step": 3205
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3988105973326252,
      "learning_rate": 6.662502409250971e-06,
      "loss": 0.8231,
      "step": 3206
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2866523762117525,
      "learning_rate": 6.660547108872959e-06,
      "loss": 0.7475,
      "step": 3207
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4251348613128347,
      "learning_rate": 6.6585915230276445e-06,
      "loss": 0.8208,
      "step": 3208
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2458546609051808,
      "learning_rate": 6.656635652051222e-06,
      "loss": 0.8337,
      "step": 3209
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.433056392868669,
      "learning_rate": 6.654679496279925e-06,
      "loss": 0.8638,
      "step": 3210
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4205970351016262,
      "learning_rate": 6.6527230560500415e-06,
      "loss": 0.7971,
      "step": 3211
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.421447710993833,
      "learning_rate": 6.6507663316979025e-06,
      "loss": 0.7242,
      "step": 3212
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2736124878842643,
      "learning_rate": 6.648809323559895e-06,
      "loss": 0.7199,
      "step": 3213
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.344055280839174,
      "learning_rate": 6.64685203197245e-06,
      "loss": 0.9096,
      "step": 3214
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.184094964013448,
      "learning_rate": 6.644894457272048e-06,
      "loss": 0.7556,
      "step": 3215
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6507878505774898,
      "learning_rate": 6.64293659979522e-06,
      "loss": 0.7598,
      "step": 3216
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.23710319235816,
      "learning_rate": 6.640978459878543e-06,
      "loss": 0.7157,
      "step": 3217
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3499840172434194,
      "learning_rate": 6.639020037858647e-06,
      "loss": 0.7798,
      "step": 3218
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.792607460063827,
      "learning_rate": 6.637061334072204e-06,
      "loss": 0.7278,
      "step": 3219
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.588793432647185,
      "learning_rate": 6.635102348855939e-06,
      "loss": 0.9349,
      "step": 3220
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5374638575822457,
      "learning_rate": 6.6331430825466245e-06,
      "loss": 0.8749,
      "step": 3221
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3649247516935423,
      "learning_rate": 6.631183535481082e-06,
      "loss": 0.8635,
      "step": 3222
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.296007591725777,
      "learning_rate": 6.629223707996177e-06,
      "loss": 0.833,
      "step": 3223
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.200795975720326,
      "learning_rate": 6.627263600428833e-06,
      "loss": 0.7979,
      "step": 3224
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4587012674793756,
      "learning_rate": 6.625303213116009e-06,
      "loss": 0.7272,
      "step": 3225
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.5870540463656715,
      "learning_rate": 6.62334254639472e-06,
      "loss": 0.8005,
      "step": 3226
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2760461393826101,
      "learning_rate": 6.621381600602028e-06,
      "loss": 0.6996,
      "step": 3227
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3064086178858991,
      "learning_rate": 6.619420376075043e-06,
      "loss": 0.7522,
      "step": 3228
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6232410123272891,
      "learning_rate": 6.617458873150922e-06,
      "loss": 0.8071,
      "step": 3229
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6367067527881523,
      "learning_rate": 6.6154970921668685e-06,
      "loss": 0.7272,
      "step": 3230
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.769027714876402,
      "learning_rate": 6.613535033460135e-06,
      "loss": 0.6259,
      "step": 3231
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2456821383998893,
      "learning_rate": 6.611572697368026e-06,
      "loss": 0.7666,
      "step": 3232
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9334510511947236,
      "learning_rate": 6.6096100842278865e-06,
      "loss": 0.7068,
      "step": 3233
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5412635761895868,
      "learning_rate": 6.607647194377113e-06,
      "loss": 0.8308,
      "step": 3234
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6351231151588643,
      "learning_rate": 6.60568402815315e-06,
      "loss": 0.8065,
      "step": 3235
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.44204953761995,
      "learning_rate": 6.603720585893487e-06,
      "loss": 0.8381,
      "step": 3236
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2799481521259986,
      "learning_rate": 6.601756867935664e-06,
      "loss": 0.6856,
      "step": 3237
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7756725637708257,
      "learning_rate": 6.599792874617267e-06,
      "loss": 0.6027,
      "step": 3238
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4900754597625578,
      "learning_rate": 6.597828606275928e-06,
      "loss": 0.8373,
      "step": 3239
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.394629438206963,
      "learning_rate": 6.595864063249326e-06,
      "loss": 0.8084,
      "step": 3240
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7877539597924887,
      "learning_rate": 6.593899245875193e-06,
      "loss": 0.761,
      "step": 3241
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2293553647279816,
      "learning_rate": 6.591934154491301e-06,
      "loss": 0.7947,
      "step": 3242
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5183140158083595,
      "learning_rate": 6.589968789435472e-06,
      "loss": 0.8762,
      "step": 3243
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8275905178894394,
      "learning_rate": 6.588003151045577e-06,
      "loss": 0.6645,
      "step": 3244
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.498544243642036,
      "learning_rate": 6.586037239659529e-06,
      "loss": 0.8192,
      "step": 3245
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2776887609164909,
      "learning_rate": 6.584071055615293e-06,
      "loss": 0.6757,
      "step": 3246
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4616214226743656,
      "learning_rate": 6.582104599250878e-06,
      "loss": 0.8232,
      "step": 3247
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5876102877599256,
      "learning_rate": 6.580137870904342e-06,
      "loss": 0.8774,
      "step": 3248
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.503808966372145,
      "learning_rate": 6.578170870913787e-06,
      "loss": 0.7895,
      "step": 3249
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4435541223771184,
      "learning_rate": 6.576203599617363e-06,
      "loss": 0.7857,
      "step": 3250
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2831748023834983,
      "learning_rate": 6.574236057353268e-06,
      "loss": 0.7836,
      "step": 3251
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3974681051383482,
      "learning_rate": 6.572268244459745e-06,
      "loss": 0.8306,
      "step": 3252
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.370515164870129,
      "learning_rate": 6.570300161275084e-06,
      "loss": 0.8014,
      "step": 3253
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3017491981960232,
      "learning_rate": 6.568331808137619e-06,
      "loss": 0.7981,
      "step": 3254
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3060967802185843,
      "learning_rate": 6.5663631853857355e-06,
      "loss": 0.7521,
      "step": 3255
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2175439860945876,
      "learning_rate": 6.564394293357861e-06,
      "loss": 0.8127,
      "step": 3256
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2644324723679632,
      "learning_rate": 6.562425132392473e-06,
      "loss": 0.8031,
      "step": 3257
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1510606965523233,
      "learning_rate": 6.560455702828089e-06,
      "loss": 0.7109,
      "step": 3258
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4143123635874542,
      "learning_rate": 6.558486005003281e-06,
      "loss": 0.7587,
      "step": 3259
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3144288704419949,
      "learning_rate": 6.556516039256662e-06,
      "loss": 0.8116,
      "step": 3260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9035323449997001,
      "learning_rate": 6.554545805926891e-06,
      "loss": 0.6124,
      "step": 3261
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3436550325528764,
      "learning_rate": 6.552575305352672e-06,
      "loss": 0.7462,
      "step": 3262
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4093923277125728,
      "learning_rate": 6.55060453787276e-06,
      "loss": 0.8219,
      "step": 3263
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.682636307727386,
      "learning_rate": 6.548633503825953e-06,
      "loss": 0.8132,
      "step": 3264
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4989858929023883,
      "learning_rate": 6.546662203551092e-06,
      "loss": 0.8413,
      "step": 3265
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2602518426287985,
      "learning_rate": 6.544690637387068e-06,
      "loss": 0.7961,
      "step": 3266
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2013467251533732,
      "learning_rate": 6.5427188056728165e-06,
      "loss": 0.8175,
      "step": 3267
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3558388462880429,
      "learning_rate": 6.540746708747317e-06,
      "loss": 0.8077,
      "step": 3268
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3121016382102328,
      "learning_rate": 6.538774346949597e-06,
      "loss": 0.8338,
      "step": 3269
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4861747724529373,
      "learning_rate": 6.536801720618727e-06,
      "loss": 0.663,
      "step": 3270
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6633687463616065,
      "learning_rate": 6.534828830093827e-06,
      "loss": 0.7359,
      "step": 3271
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3649174729589963,
      "learning_rate": 6.532855675714055e-06,
      "loss": 0.7465,
      "step": 3272
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2738259154091858,
      "learning_rate": 6.5308822578186225e-06,
      "loss": 0.7864,
      "step": 3273
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7336042052333912,
      "learning_rate": 6.528908576746784e-06,
      "loss": 0.6775,
      "step": 3274
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5350299669159022,
      "learning_rate": 6.526934632837835e-06,
      "loss": 0.8079,
      "step": 3275
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9406902393643806,
      "learning_rate": 6.5249604264311216e-06,
      "loss": 0.8103,
      "step": 3276
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5169745201534603,
      "learning_rate": 6.522985957866032e-06,
      "loss": 0.8126,
      "step": 3277
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8245160845857376,
      "learning_rate": 6.521011227482e-06,
      "loss": 0.7164,
      "step": 3278
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3554678931711093,
      "learning_rate": 6.519036235618505e-06,
      "loss": 0.7992,
      "step": 3279
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2755962860363925,
      "learning_rate": 6.517060982615071e-06,
      "loss": 0.8207,
      "step": 3280
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2945863689406125,
      "learning_rate": 6.5150854688112686e-06,
      "loss": 0.6944,
      "step": 3281
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.425358770174467,
      "learning_rate": 6.513109694546707e-06,
      "loss": 0.7771,
      "step": 3282
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.230315095937363,
      "learning_rate": 6.511133660161047e-06,
      "loss": 0.7283,
      "step": 3283
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6057651422927102,
      "learning_rate": 6.5091573659939945e-06,
      "loss": 0.8421,
      "step": 3284
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5897603039697308,
      "learning_rate": 6.507180812385295e-06,
      "loss": 0.7716,
      "step": 3285
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4132000100724844,
      "learning_rate": 6.50520399967474e-06,
      "loss": 0.6897,
      "step": 3286
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5737891369009636,
      "learning_rate": 6.503226928202167e-06,
      "loss": 0.886,
      "step": 3287
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2764749562442057,
      "learning_rate": 6.5012495983074605e-06,
      "loss": 0.8778,
      "step": 3288
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8588273931670032,
      "learning_rate": 6.499272010330543e-06,
      "loss": 0.6527,
      "step": 3289
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5316369506690957,
      "learning_rate": 6.497294164611385e-06,
      "loss": 0.8098,
      "step": 3290
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8349565532122232,
      "learning_rate": 6.495316061490003e-06,
      "loss": 0.6665,
      "step": 3291
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8292037250414643,
      "learning_rate": 6.493337701306454e-06,
      "loss": 0.7299,
      "step": 3292
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4454820724573103,
      "learning_rate": 6.4913590844008436e-06,
      "loss": 0.7595,
      "step": 3293
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4169156254415785,
      "learning_rate": 6.489380211113316e-06,
      "loss": 0.7471,
      "step": 3294
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.360531345227435,
      "learning_rate": 6.487401081784063e-06,
      "loss": 0.7755,
      "step": 3295
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4688512976913681,
      "learning_rate": 6.485421696753321e-06,
      "loss": 0.7232,
      "step": 3296
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3093848416385159,
      "learning_rate": 6.4834420563613685e-06,
      "loss": 0.7463,
      "step": 3297
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2982038572603978,
      "learning_rate": 6.481462160948531e-06,
      "loss": 0.8616,
      "step": 3298
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.327013158489367,
      "learning_rate": 6.479482010855172e-06,
      "loss": 0.8656,
      "step": 3299
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3779716756017233,
      "learning_rate": 6.477501606421703e-06,
      "loss": 0.713,
      "step": 3300
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3373252269259797,
      "learning_rate": 6.4755209479885806e-06,
      "loss": 0.6758,
      "step": 3301
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9770603872346664,
      "learning_rate": 6.473540035896301e-06,
      "loss": 0.6467,
      "step": 3302
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3228874451877337,
      "learning_rate": 6.471558870485407e-06,
      "loss": 0.7451,
      "step": 3303
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3080465429348955,
      "learning_rate": 6.469577452096483e-06,
      "loss": 0.7711,
      "step": 3304
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9297077332174744,
      "learning_rate": 6.467595781070158e-06,
      "loss": 0.7276,
      "step": 3305
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8043776524949859,
      "learning_rate": 6.4656138577471036e-06,
      "loss": 0.6869,
      "step": 3306
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2621516037883733,
      "learning_rate": 6.463631682468038e-06,
      "loss": 0.773,
      "step": 3307
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5481719612279212,
      "learning_rate": 6.461649255573716e-06,
      "loss": 0.721,
      "step": 3308
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3840025522336434,
      "learning_rate": 6.459666577404944e-06,
      "loss": 0.7869,
      "step": 3309
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9747846274122969,
      "learning_rate": 6.457683648302565e-06,
      "loss": 0.6824,
      "step": 3310
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3484810715952553,
      "learning_rate": 6.455700468607469e-06,
      "loss": 0.8259,
      "step": 3311
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4028508151177757,
      "learning_rate": 6.453717038660584e-06,
      "loss": 0.7657,
      "step": 3312
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3693154261650466,
      "learning_rate": 6.451733358802889e-06,
      "loss": 0.8005,
      "step": 3313
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.359044203066375,
      "learning_rate": 6.449749429375398e-06,
      "loss": 0.8127,
      "step": 3314
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2176873333270493,
      "learning_rate": 6.4477652507191744e-06,
      "loss": 0.8175,
      "step": 3315
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.324799613584454,
      "learning_rate": 6.44578082317532e-06,
      "loss": 0.7045,
      "step": 3316
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.525176977481468,
      "learning_rate": 6.443796147084982e-06,
      "loss": 0.7841,
      "step": 3317
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3093866106251892,
      "learning_rate": 6.441811222789347e-06,
      "loss": 0.737,
      "step": 3318
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8325898206275221,
      "learning_rate": 6.439826050629649e-06,
      "loss": 0.6064,
      "step": 3319
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7907215353633897,
      "learning_rate": 6.4378406309471605e-06,
      "loss": 0.5929,
      "step": 3320
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3762719743272613,
      "learning_rate": 6.435854964083199e-06,
      "loss": 0.7859,
      "step": 3321
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.497693361275597,
      "learning_rate": 6.433869050379122e-06,
      "loss": 0.8544,
      "step": 3322
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3072152599611782,
      "learning_rate": 6.431882890176334e-06,
      "loss": 0.8976,
      "step": 3323
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2012952794909812,
      "learning_rate": 6.429896483816277e-06,
      "loss": 0.7797,
      "step": 3324
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8487272320063785,
      "learning_rate": 6.427909831640438e-06,
      "loss": 0.6384,
      "step": 3325
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6723086309253543,
      "learning_rate": 6.425922933990344e-06,
      "loss": 0.8233,
      "step": 3326
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2546353791786404,
      "learning_rate": 6.423935791207568e-06,
      "loss": 0.8396,
      "step": 3327
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3197876842212404,
      "learning_rate": 6.421948403633721e-06,
      "loss": 0.6956,
      "step": 3328
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7404457454532725,
      "learning_rate": 6.4199607716104605e-06,
      "loss": 0.6129,
      "step": 3329
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.463073096791234,
      "learning_rate": 6.41797289547948e-06,
      "loss": 0.7946,
      "step": 3330
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8071002223257775,
      "learning_rate": 6.415984775582521e-06,
      "loss": 0.7588,
      "step": 3331
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.251407611175826,
      "learning_rate": 6.413996412261363e-06,
      "loss": 0.7855,
      "step": 3332
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.904799198306487,
      "learning_rate": 6.412007805857828e-06,
      "loss": 0.7922,
      "step": 3333
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7809063920223698,
      "learning_rate": 6.410018956713784e-06,
      "loss": 0.6334,
      "step": 3334
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2673773606432472,
      "learning_rate": 6.408029865171135e-06,
      "loss": 0.748,
      "step": 3335
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3979900926290048,
      "learning_rate": 6.406040531571828e-06,
      "loss": 0.7037,
      "step": 3336
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2989493753141628,
      "learning_rate": 6.404050956257853e-06,
      "loss": 0.6809,
      "step": 3337
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3625183609595959,
      "learning_rate": 6.402061139571243e-06,
      "loss": 0.8241,
      "step": 3338
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4979752225181016,
      "learning_rate": 6.400071081854068e-06,
      "loss": 0.8967,
      "step": 3339
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.330040847858628,
      "learning_rate": 6.398080783448443e-06,
      "loss": 0.713,
      "step": 3340
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6180468257176386,
      "learning_rate": 6.396090244696523e-06,
      "loss": 0.7805,
      "step": 3341
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.268091836719546,
      "learning_rate": 6.394099465940505e-06,
      "loss": 0.7562,
      "step": 3342
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7370554867455632,
      "learning_rate": 6.3921084475226295e-06,
      "loss": 0.7212,
      "step": 3343
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2773432961640345,
      "learning_rate": 6.39011718978517e-06,
      "loss": 0.7805,
      "step": 3344
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2132419766734504,
      "learning_rate": 6.388125693070452e-06,
      "loss": 0.7645,
      "step": 3345
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4273933757599988,
      "learning_rate": 6.386133957720833e-06,
      "loss": 0.844,
      "step": 3346
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.421395461737626,
      "learning_rate": 6.384141984078719e-06,
      "loss": 0.7955,
      "step": 3347
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3079257911419315,
      "learning_rate": 6.38214977248655e-06,
      "loss": 0.8584,
      "step": 3348
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.312727227623789,
      "learning_rate": 6.380157323286813e-06,
      "loss": 0.7508,
      "step": 3349
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6212278085301812,
      "learning_rate": 6.378164636822033e-06,
      "loss": 0.8388,
      "step": 3350
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6418623977370754,
      "learning_rate": 6.376171713434771e-06,
      "loss": 0.8834,
      "step": 3351
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4084635909876975,
      "learning_rate": 6.3741785534676404e-06,
      "loss": 0.8262,
      "step": 3352
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4121803795608112,
      "learning_rate": 6.372185157263287e-06,
      "loss": 0.8258,
      "step": 3353
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3975578976836796,
      "learning_rate": 6.370191525164394e-06,
      "loss": 0.7885,
      "step": 3354
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2963401555381266,
      "learning_rate": 6.368197657513695e-06,
      "loss": 0.7852,
      "step": 3355
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5366107579094197,
      "learning_rate": 6.366203554653957e-06,
      "loss": 0.7583,
      "step": 3356
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9504775404487475,
      "learning_rate": 6.36420921692799e-06,
      "loss": 0.6131,
      "step": 3357
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4320650117688034,
      "learning_rate": 6.362214644678641e-06,
      "loss": 0.7476,
      "step": 3358
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4899112922086204,
      "learning_rate": 6.360219838248806e-06,
      "loss": 0.7642,
      "step": 3359
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5476548466703355,
      "learning_rate": 6.358224797981409e-06,
      "loss": 0.7944,
      "step": 3360
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7706949422791032,
      "learning_rate": 6.356229524219425e-06,
      "loss": 0.8134,
      "step": 3361
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3303164147272495,
      "learning_rate": 6.354234017305863e-06,
      "loss": 0.7763,
      "step": 3362
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4378947392178727,
      "learning_rate": 6.352238277583773e-06,
      "loss": 0.7379,
      "step": 3363
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5342448195933194,
      "learning_rate": 6.350242305396247e-06,
      "loss": 0.7941,
      "step": 3364
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8730794378179783,
      "learning_rate": 6.348246101086414e-06,
      "loss": 0.8644,
      "step": 3365
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3442989517228425,
      "learning_rate": 6.346249664997448e-06,
      "loss": 0.8812,
      "step": 3366
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4840207567966004,
      "learning_rate": 6.344252997472556e-06,
      "loss": 0.8161,
      "step": 3367
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4263133788905256,
      "learning_rate": 6.342256098854992e-06,
      "loss": 0.7898,
      "step": 3368
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.336390082714888,
      "learning_rate": 6.34025896948804e-06,
      "loss": 0.7271,
      "step": 3369
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3005609194104708,
      "learning_rate": 6.338261609715037e-06,
      "loss": 0.7434,
      "step": 3370
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3000540598790977,
      "learning_rate": 6.336264019879348e-06,
      "loss": 0.8302,
      "step": 3371
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8556822497406659,
      "learning_rate": 6.334266200324381e-06,
      "loss": 0.7497,
      "step": 3372
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8715959412578023,
      "learning_rate": 6.332268151393589e-06,
      "loss": 0.8569,
      "step": 3373
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.416615061440876,
      "learning_rate": 6.330269873430455e-06,
      "loss": 0.8231,
      "step": 3374
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3125489319013885,
      "learning_rate": 6.3282713667785086e-06,
      "loss": 0.7002,
      "step": 3375
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8921564023639657,
      "learning_rate": 6.326272631781314e-06,
      "loss": 0.7163,
      "step": 3376
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3906571143494373,
      "learning_rate": 6.32427366878248e-06,
      "loss": 0.7915,
      "step": 3377
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4924335742649724,
      "learning_rate": 6.322274478125651e-06,
      "loss": 0.7063,
      "step": 3378
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6764349743420668,
      "learning_rate": 6.320275060154508e-06,
      "loss": 0.8313,
      "step": 3379
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6589893801874054,
      "learning_rate": 6.318275415212777e-06,
      "loss": 0.7873,
      "step": 3380
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7888197332476622,
      "learning_rate": 6.316275543644219e-06,
      "loss": 0.6536,
      "step": 3381
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8097938817074637,
      "learning_rate": 6.314275445792637e-06,
      "loss": 0.7321,
      "step": 3382
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2365323861379274,
      "learning_rate": 6.312275122001867e-06,
      "loss": 0.8068,
      "step": 3383
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8428495920844651,
      "learning_rate": 6.310274572615792e-06,
      "loss": 0.6809,
      "step": 3384
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4879531333274842,
      "learning_rate": 6.308273797978328e-06,
      "loss": 0.7787,
      "step": 3385
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1873218632047957,
      "learning_rate": 6.30627279843343e-06,
      "loss": 0.7965,
      "step": 3386
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0196552380414605,
      "learning_rate": 6.304271574325096e-06,
      "loss": 0.5852,
      "step": 3387
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2940781404321626,
      "learning_rate": 6.3022701259973565e-06,
      "loss": 0.832,
      "step": 3388
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.561494368755599,
      "learning_rate": 6.300268453794287e-06,
      "loss": 0.8164,
      "step": 3389
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2799174223176188,
      "learning_rate": 6.298266558059995e-06,
      "loss": 0.7756,
      "step": 3390
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.541126387222446,
      "learning_rate": 6.296264439138631e-06,
      "loss": 0.7945,
      "step": 3391
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5661478991334215,
      "learning_rate": 6.294262097374383e-06,
      "loss": 0.832,
      "step": 3392
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3667283552576568,
      "learning_rate": 6.292259533111474e-06,
      "loss": 0.8464,
      "step": 3393
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4322941599009131,
      "learning_rate": 6.2902567466941725e-06,
      "loss": 0.8414,
      "step": 3394
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4131586794450106,
      "learning_rate": 6.288253738466777e-06,
      "loss": 0.6707,
      "step": 3395
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3313688262667909,
      "learning_rate": 6.286250508773631e-06,
      "loss": 0.7025,
      "step": 3396
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3898712344290451,
      "learning_rate": 6.284247057959107e-06,
      "loss": 0.7853,
      "step": 3397
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3051868034772696,
      "learning_rate": 6.282243386367628e-06,
      "loss": 0.847,
      "step": 3398
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3400252976970768,
      "learning_rate": 6.280239494343647e-06,
      "loss": 0.7681,
      "step": 3399
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7846103736590437,
      "learning_rate": 6.278235382231654e-06,
      "loss": 0.7479,
      "step": 3400
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2503510844118022,
      "learning_rate": 6.27623105037618e-06,
      "loss": 0.7334,
      "step": 3401
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3748107034714643,
      "learning_rate": 6.274226499121793e-06,
      "loss": 0.7986,
      "step": 3402
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3179850229139143,
      "learning_rate": 6.272221728813099e-06,
      "loss": 0.6597,
      "step": 3403
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4095214358867527,
      "learning_rate": 6.270216739794741e-06,
      "loss": 0.6913,
      "step": 3404
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.480945844922635,
      "learning_rate": 6.2682115324114e-06,
      "loss": 0.8225,
      "step": 3405
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3844873567294016,
      "learning_rate": 6.266206107007793e-06,
      "loss": 0.8196,
      "step": 3406
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3354755014464808,
      "learning_rate": 6.264200463928679e-06,
      "loss": 0.6883,
      "step": 3407
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4209027752409829,
      "learning_rate": 6.2621946035188474e-06,
      "loss": 0.7975,
      "step": 3408
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3831151167299216,
      "learning_rate": 6.260188526123133e-06,
      "loss": 0.7212,
      "step": 3409
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2415699048833795,
      "learning_rate": 6.2581822320864e-06,
      "loss": 0.8174,
      "step": 3410
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5634956728202936,
      "learning_rate": 6.256175721753556e-06,
      "loss": 0.827,
      "step": 3411
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7142877302552844,
      "learning_rate": 6.254168995469541e-06,
      "loss": 0.8676,
      "step": 3412
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2754371660625647,
      "learning_rate": 6.252162053579338e-06,
      "loss": 0.7632,
      "step": 3413
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3566654905900069,
      "learning_rate": 6.250154896427962e-06,
      "loss": 0.8237,
      "step": 3414
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.362909323990357,
      "learning_rate": 6.2481475243604654e-06,
      "loss": 0.7553,
      "step": 3415
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4392897046905697,
      "learning_rate": 6.246139937721939e-06,
      "loss": 0.8298,
      "step": 3416
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1793667877901661,
      "learning_rate": 6.244132136857511e-06,
      "loss": 0.7474,
      "step": 3417
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3005214140117864,
      "learning_rate": 6.242124122112347e-06,
      "loss": 0.6118,
      "step": 3418
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.26676225510164,
      "learning_rate": 6.240115893831644e-06,
      "loss": 0.7666,
      "step": 3419
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3678478665638805,
      "learning_rate": 6.238107452360643e-06,
      "loss": 0.8159,
      "step": 3420
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.265012981107556,
      "learning_rate": 6.236098798044619e-06,
      "loss": 0.7809,
      "step": 3421
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3613892934289054,
      "learning_rate": 6.2340899312288795e-06,
      "loss": 0.797,
      "step": 3422
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8226400530942369,
      "learning_rate": 6.232080852258776e-06,
      "loss": 0.6087,
      "step": 3423
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8737756612048824,
      "learning_rate": 6.23007156147969e-06,
      "loss": 0.659,
      "step": 3424
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4674021100939127,
      "learning_rate": 6.228062059237041e-06,
      "loss": 0.8033,
      "step": 3425
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9433665492333199,
      "learning_rate": 6.226052345876288e-06,
      "loss": 0.7057,
      "step": 3426
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.371126428662626,
      "learning_rate": 6.224042421742924e-06,
      "loss": 0.8,
      "step": 3427
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3016988948852204,
      "learning_rate": 6.222032287182477e-06,
      "loss": 0.7652,
      "step": 3428
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4150530913579615,
      "learning_rate": 6.220021942540512e-06,
      "loss": 0.8247,
      "step": 3429
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.710439383024829,
      "learning_rate": 6.21801138816263e-06,
      "loss": 0.8863,
      "step": 3430
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2914586081799833,
      "learning_rate": 6.2160006243944726e-06,
      "loss": 0.7595,
      "step": 3431
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4076091897001295,
      "learning_rate": 6.213989651581711e-06,
      "loss": 0.8004,
      "step": 3432
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2768087869056637,
      "learning_rate": 6.211978470070052e-06,
      "loss": 0.7541,
      "step": 3433
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6453167744651336,
      "learning_rate": 6.209967080205244e-06,
      "loss": 0.7838,
      "step": 3434
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2337495184694696,
      "learning_rate": 6.20795548233307e-06,
      "loss": 0.7452,
      "step": 3435
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9738209001677972,
      "learning_rate": 6.205943676799344e-06,
      "loss": 0.6967,
      "step": 3436
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3953537460082757,
      "learning_rate": 6.203931663949918e-06,
      "loss": 0.8497,
      "step": 3437
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7913748187068765,
      "learning_rate": 6.201919444130684e-06,
      "loss": 0.7336,
      "step": 3438
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5506856121199406,
      "learning_rate": 6.199907017687562e-06,
      "loss": 0.8826,
      "step": 3439
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6568096627951765,
      "learning_rate": 6.197894384966513e-06,
      "loss": 0.7388,
      "step": 3440
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3790114442673644,
      "learning_rate": 6.195881546313533e-06,
      "loss": 0.7644,
      "step": 3441
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9703314695926225,
      "learning_rate": 6.193868502074651e-06,
      "loss": 0.7565,
      "step": 3442
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4046209232148175,
      "learning_rate": 6.191855252595933e-06,
      "loss": 0.8014,
      "step": 3443
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4703500683088453,
      "learning_rate": 6.189841798223479e-06,
      "loss": 0.8168,
      "step": 3444
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3185093446639764,
      "learning_rate": 6.1878281393034275e-06,
      "loss": 0.7611,
      "step": 3445
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.331086405612798,
      "learning_rate": 6.1858142761819484e-06,
      "loss": 0.7373,
      "step": 3446
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.376596790078534,
      "learning_rate": 6.1838002092052465e-06,
      "loss": 0.8458,
      "step": 3447
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7775368693523647,
      "learning_rate": 6.181785938719566e-06,
      "loss": 0.6685,
      "step": 3448
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.396255929914059,
      "learning_rate": 6.179771465071182e-06,
      "loss": 0.7577,
      "step": 3449
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8202617053213997,
      "learning_rate": 6.177756788606406e-06,
      "loss": 0.6765,
      "step": 3450
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4059192791579065,
      "learning_rate": 6.175741909671584e-06,
      "loss": 0.7989,
      "step": 3451
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3084980098500696,
      "learning_rate": 6.173726828613098e-06,
      "loss": 0.6942,
      "step": 3452
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3151059824169984,
      "learning_rate": 6.171711545777363e-06,
      "loss": 0.7698,
      "step": 3453
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1957142442091375,
      "learning_rate": 6.169696061510831e-06,
      "loss": 0.7529,
      "step": 3454
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9224599496142176,
      "learning_rate": 6.167680376159983e-06,
      "loss": 0.6727,
      "step": 3455
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.391675085909392,
      "learning_rate": 6.165664490071343e-06,
      "loss": 0.8245,
      "step": 3456
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4624866732018762,
      "learning_rate": 6.163648403591462e-06,
      "loss": 0.7429,
      "step": 3457
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.703209388522409,
      "learning_rate": 6.161632117066932e-06,
      "loss": 0.776,
      "step": 3458
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.379168966680437,
      "learning_rate": 6.1596156308443746e-06,
      "loss": 0.7693,
      "step": 3459
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4267606086691498,
      "learning_rate": 6.157598945270447e-06,
      "loss": 0.7211,
      "step": 3460
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4599137768102741,
      "learning_rate": 6.1555820606918384e-06,
      "loss": 0.7528,
      "step": 3461
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4645133376398183,
      "learning_rate": 6.153564977455278e-06,
      "loss": 0.8725,
      "step": 3462
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3836830770843291,
      "learning_rate": 6.151547695907525e-06,
      "loss": 0.8359,
      "step": 3463
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3738270870520148,
      "learning_rate": 6.149530216395374e-06,
      "loss": 0.7345,
      "step": 3464
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.258939967701337,
      "learning_rate": 6.1475125392656506e-06,
      "loss": 0.6953,
      "step": 3465
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3737895322464895,
      "learning_rate": 6.1454946648652204e-06,
      "loss": 0.7495,
      "step": 3466
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.307409081795984,
      "learning_rate": 6.143476593540976e-06,
      "loss": 0.7195,
      "step": 3467
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3923583374392972,
      "learning_rate": 6.1414583256398494e-06,
      "loss": 0.8061,
      "step": 3468
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3190406461666282,
      "learning_rate": 6.139439861508804e-06,
      "loss": 0.7243,
      "step": 3469
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3401651454105932,
      "learning_rate": 6.137421201494837e-06,
      "loss": 0.7688,
      "step": 3470
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4144037088935806,
      "learning_rate": 6.135402345944979e-06,
      "loss": 0.7482,
      "step": 3471
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.298060962300958,
      "learning_rate": 6.1333832952062945e-06,
      "loss": 0.8141,
      "step": 3472
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.337335170060034,
      "learning_rate": 6.1313640496258834e-06,
      "loss": 0.7436,
      "step": 3473
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9515372010472691,
      "learning_rate": 6.129344609550876e-06,
      "loss": 0.7238,
      "step": 3474
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.378058548265461,
      "learning_rate": 6.127324975328437e-06,
      "loss": 0.8561,
      "step": 3475
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5104395449306458,
      "learning_rate": 6.125305147305764e-06,
      "loss": 0.7527,
      "step": 3476
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3101277324275138,
      "learning_rate": 6.1232851258300944e-06,
      "loss": 0.7554,
      "step": 3477
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5522330420967867,
      "learning_rate": 6.121264911248688e-06,
      "loss": 0.8188,
      "step": 3478
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8764865473305153,
      "learning_rate": 6.1192445039088435e-06,
      "loss": 0.6354,
      "step": 3479
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.343192230734996,
      "learning_rate": 6.117223904157893e-06,
      "loss": 0.7645,
      "step": 3480
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2982245334321676,
      "learning_rate": 6.115203112343203e-06,
      "loss": 0.7915,
      "step": 3481
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3758231252572621,
      "learning_rate": 6.11318212881217e-06,
      "loss": 0.8511,
      "step": 3482
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5899000482798684,
      "learning_rate": 6.111160953912222e-06,
      "loss": 0.8387,
      "step": 3483
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4756121349272742,
      "learning_rate": 6.1091395879908255e-06,
      "loss": 0.7548,
      "step": 3484
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4434785756654174,
      "learning_rate": 6.107118031395475e-06,
      "loss": 0.7884,
      "step": 3485
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4283360730456836,
      "learning_rate": 6.1050962844737005e-06,
      "loss": 0.894,
      "step": 3486
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2905783544163694,
      "learning_rate": 6.103074347573062e-06,
      "loss": 0.8584,
      "step": 3487
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7517881203838129,
      "learning_rate": 6.1010522210411575e-06,
      "loss": 0.6976,
      "step": 3488
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.302814973849657,
      "learning_rate": 6.0990299052256105e-06,
      "loss": 0.7173,
      "step": 3489
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.491286282107372,
      "learning_rate": 6.097007400474081e-06,
      "loss": 0.7911,
      "step": 3490
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3866939541812608,
      "learning_rate": 6.094984707134263e-06,
      "loss": 0.769,
      "step": 3491
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3361118673623826,
      "learning_rate": 6.092961825553881e-06,
      "loss": 0.8833,
      "step": 3492
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8605792075314249,
      "learning_rate": 6.090938756080688e-06,
      "loss": 0.6634,
      "step": 3493
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.28999749977783,
      "learning_rate": 6.088915499062475e-06,
      "loss": 0.8266,
      "step": 3494
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.481920142989902,
      "learning_rate": 6.0868920548470654e-06,
      "loss": 0.7522,
      "step": 3495
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4547847902704654,
      "learning_rate": 6.084868423782312e-06,
      "loss": 0.9107,
      "step": 3496
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9394439680251558,
      "learning_rate": 6.082844606216098e-06,
      "loss": 0.7146,
      "step": 3497
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6109452444052244,
      "learning_rate": 6.080820602496345e-06,
      "loss": 0.8086,
      "step": 3498
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2944488400686964,
      "learning_rate": 6.078796412970997e-06,
      "loss": 0.7321,
      "step": 3499
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3542940212794885,
      "learning_rate": 6.076772037988042e-06,
      "loss": 0.8293,
      "step": 3500
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7899823948293236,
      "learning_rate": 6.07474747789549e-06,
      "loss": 0.8583,
      "step": 3501
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6846587094861278,
      "learning_rate": 6.072722733041387e-06,
      "loss": 0.7577,
      "step": 3502
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.309286079802697,
      "learning_rate": 6.07069780377381e-06,
      "loss": 0.7807,
      "step": 3503
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3757799486641489,
      "learning_rate": 6.068672690440868e-06,
      "loss": 0.8157,
      "step": 3504
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9682791002367884,
      "learning_rate": 6.066647393390701e-06,
      "loss": 0.6081,
      "step": 3505
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7159974988679891,
      "learning_rate": 6.064621912971483e-06,
      "loss": 0.7897,
      "step": 3506
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6430701567347097,
      "learning_rate": 6.062596249531414e-06,
      "loss": 0.9458,
      "step": 3507
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.339638370838527,
      "learning_rate": 6.060570403418731e-06,
      "loss": 0.7659,
      "step": 3508
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5654234075169613,
      "learning_rate": 6.058544374981701e-06,
      "loss": 0.824,
      "step": 3509
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4867034213847714,
      "learning_rate": 6.056518164568622e-06,
      "loss": 0.772,
      "step": 3510
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9055334407434442,
      "learning_rate": 6.054491772527822e-06,
      "loss": 0.7004,
      "step": 3511
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4237008924265226,
      "learning_rate": 6.052465199207661e-06,
      "loss": 0.8049,
      "step": 3512
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.899323366869118,
      "learning_rate": 6.050438444956531e-06,
      "loss": 0.7128,
      "step": 3513
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.436574259813067,
      "learning_rate": 6.048411510122855e-06,
      "loss": 0.6973,
      "step": 3514
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.281111206172992,
      "learning_rate": 6.046384395055086e-06,
      "loss": 0.7625,
      "step": 3515
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2396231320829671,
      "learning_rate": 6.04435710010171e-06,
      "loss": 0.7592,
      "step": 3516
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.30007190709943,
      "learning_rate": 6.042329625611239e-06,
      "loss": 0.6895,
      "step": 3517
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8441930064967941,
      "learning_rate": 6.040301971932223e-06,
      "loss": 0.6615,
      "step": 3518
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5872064737676426,
      "learning_rate": 6.038274139413238e-06,
      "loss": 0.7124,
      "step": 3519
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2127010780976846,
      "learning_rate": 6.036246128402892e-06,
      "loss": 0.7344,
      "step": 3520
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8819698272115345,
      "learning_rate": 6.034217939249823e-06,
      "loss": 0.6664,
      "step": 3521
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3145778604605525,
      "learning_rate": 6.0321895723027e-06,
      "loss": 0.8114,
      "step": 3522
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4169631741047894,
      "learning_rate": 6.030161027910223e-06,
      "loss": 0.8007,
      "step": 3523
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1815266001515496,
      "learning_rate": 6.028132306421124e-06,
      "loss": 0.7554,
      "step": 3524
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.346294675837105,
      "learning_rate": 6.026103408184162e-06,
      "loss": 0.8093,
      "step": 3525
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5470398830216026,
      "learning_rate": 6.0240743335481265e-06,
      "loss": 0.8194,
      "step": 3526
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.474176564668479,
      "learning_rate": 6.0220450828618424e-06,
      "loss": 0.8664,
      "step": 3527
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5573777757513785,
      "learning_rate": 6.0200156564741606e-06,
      "loss": 0.7741,
      "step": 3528
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0087710436665476,
      "learning_rate": 6.017986054733962e-06,
      "loss": 0.7682,
      "step": 3529
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5339520812701992,
      "learning_rate": 6.0159562779901605e-06,
      "loss": 0.8736,
      "step": 3530
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5410083868054973,
      "learning_rate": 6.013926326591695e-06,
      "loss": 0.737,
      "step": 3531
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3365903548114562,
      "learning_rate": 6.0118962008875395e-06,
      "loss": 0.75,
      "step": 3532
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2741160413779586,
      "learning_rate": 6.009865901226697e-06,
      "loss": 0.7478,
      "step": 3533
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3768616288456046,
      "learning_rate": 6.007835427958199e-06,
      "loss": 0.6602,
      "step": 3534
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.340874476045425,
      "learning_rate": 6.005804781431106e-06,
      "loss": 0.7563,
      "step": 3535
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2331480011933498,
      "learning_rate": 6.0037739619945114e-06,
      "loss": 0.7585,
      "step": 3536
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2533501490824257,
      "learning_rate": 6.001742969997535e-06,
      "loss": 0.7699,
      "step": 3537
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1635176263567606,
      "learning_rate": 5.99971180578933e-06,
      "loss": 0.816,
      "step": 3538
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3050005845261363,
      "learning_rate": 5.997680469719076e-06,
      "loss": 0.8178,
      "step": 3539
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.35921434110093,
      "learning_rate": 5.995648962135983e-06,
      "loss": 0.6808,
      "step": 3540
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3688527741295726,
      "learning_rate": 5.993617283389289e-06,
      "loss": 0.7921,
      "step": 3541
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4401611375466423,
      "learning_rate": 5.991585433828267e-06,
      "loss": 0.7706,
      "step": 3542
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4304274238891665,
      "learning_rate": 5.9895534138022136e-06,
      "loss": 0.8036,
      "step": 3543
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6323451741850972,
      "learning_rate": 5.9875212236604564e-06,
      "loss": 0.6987,
      "step": 3544
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3703797072457855,
      "learning_rate": 5.985488863752351e-06,
      "loss": 0.724,
      "step": 3545
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3688389270817949,
      "learning_rate": 5.983456334427286e-06,
      "loss": 0.7122,
      "step": 3546
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4836364574390641,
      "learning_rate": 5.9814236360346765e-06,
      "loss": 0.8044,
      "step": 3547
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4717966203891735,
      "learning_rate": 5.9793907689239675e-06,
      "loss": 0.7762,
      "step": 3548
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6297317348706657,
      "learning_rate": 5.97735773344463e-06,
      "loss": 0.7876,
      "step": 3549
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5120490818166867,
      "learning_rate": 5.975324529946166e-06,
      "loss": 0.8178,
      "step": 3550
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4378495699524203,
      "learning_rate": 5.973291158778109e-06,
      "loss": 0.7797,
      "step": 3551
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3352419451254876,
      "learning_rate": 5.97125762029002e-06,
      "loss": 0.8324,
      "step": 3552
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.231266832969438,
      "learning_rate": 5.969223914831485e-06,
      "loss": 0.7403,
      "step": 3553
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3505805208241635,
      "learning_rate": 5.967190042752123e-06,
      "loss": 0.7497,
      "step": 3554
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.430613702798791,
      "learning_rate": 5.965156004401581e-06,
      "loss": 0.7895,
      "step": 3555
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2980989017627675,
      "learning_rate": 5.9631218001295325e-06,
      "loss": 0.8818,
      "step": 3556
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.71865571605958,
      "learning_rate": 5.961087430285681e-06,
      "loss": 0.7849,
      "step": 3557
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4246806513738335,
      "learning_rate": 5.959052895219758e-06,
      "loss": 0.8283,
      "step": 3558
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4487232076908108,
      "learning_rate": 5.957018195281523e-06,
      "loss": 0.7228,
      "step": 3559
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3920191923791603,
      "learning_rate": 5.954983330820767e-06,
      "loss": 0.6958,
      "step": 3560
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3406163807000135,
      "learning_rate": 5.9529483021873055e-06,
      "loss": 0.7981,
      "step": 3561
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3973550535582642,
      "learning_rate": 5.950913109730983e-06,
      "loss": 0.8282,
      "step": 3562
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4797905003485576,
      "learning_rate": 5.948877753801673e-06,
      "loss": 0.7467,
      "step": 3563
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1153603841601996,
      "learning_rate": 5.946842234749275e-06,
      "loss": 0.7737,
      "step": 3564
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.377230231655705,
      "learning_rate": 5.944806552923722e-06,
      "loss": 0.8015,
      "step": 3565
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.897770452768819,
      "learning_rate": 5.942770708674969e-06,
      "loss": 0.6303,
      "step": 3566
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.318087664334874,
      "learning_rate": 5.940734702353002e-06,
      "loss": 0.7568,
      "step": 3567
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.598484939435941,
      "learning_rate": 5.938698534307833e-06,
      "loss": 0.7944,
      "step": 3568
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8844843743700298,
      "learning_rate": 5.936662204889504e-06,
      "loss": 0.6653,
      "step": 3569
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5091089934374442,
      "learning_rate": 5.934625714448084e-06,
      "loss": 0.7403,
      "step": 3570
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4173907406185493,
      "learning_rate": 5.932589063333668e-06,
      "loss": 0.7535,
      "step": 3571
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4656440398649688,
      "learning_rate": 5.9305522518963795e-06,
      "loss": 0.7279,
      "step": 3572
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8995730717995356,
      "learning_rate": 5.928515280486372e-06,
      "loss": 0.6636,
      "step": 3573
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3821914886135431,
      "learning_rate": 5.9264781494538235e-06,
      "loss": 0.7217,
      "step": 3574
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5217145124322116,
      "learning_rate": 5.924440859148941e-06,
      "loss": 0.8399,
      "step": 3575
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3515375469724147,
      "learning_rate": 5.922403409921957e-06,
      "loss": 0.6976,
      "step": 3576
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.38181804484533,
      "learning_rate": 5.9203658021231335e-06,
      "loss": 0.8453,
      "step": 3577
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8252627632235815,
      "learning_rate": 5.918328036102758e-06,
      "loss": 0.6199,
      "step": 3578
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2988422410340912,
      "learning_rate": 5.916290112211149e-06,
      "loss": 0.8067,
      "step": 3579
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.84664855993425,
      "learning_rate": 5.9142520307986455e-06,
      "loss": 0.6481,
      "step": 3580
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.568147938638796,
      "learning_rate": 5.91221379221562e-06,
      "loss": 0.7969,
      "step": 3581
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.916699079433933,
      "learning_rate": 5.910175396812468e-06,
      "loss": 0.8679,
      "step": 3582
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8507000613149521,
      "learning_rate": 5.908136844939612e-06,
      "loss": 0.7097,
      "step": 3583
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4513718550685861,
      "learning_rate": 5.906098136947506e-06,
      "loss": 0.7323,
      "step": 3584
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3625592661891364,
      "learning_rate": 5.904059273186627e-06,
      "loss": 0.7827,
      "step": 3585
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3299791004495474,
      "learning_rate": 5.9020202540074755e-06,
      "loss": 0.765,
      "step": 3586
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8605308633698483,
      "learning_rate": 5.899981079760586e-06,
      "loss": 0.7448,
      "step": 3587
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.20893104241787,
      "learning_rate": 5.897941750796517e-06,
      "loss": 0.7779,
      "step": 3588
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.231421304366071,
      "learning_rate": 5.895902267465851e-06,
      "loss": 0.7121,
      "step": 3589
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3861248955087215,
      "learning_rate": 5.893862630119197e-06,
      "loss": 0.8481,
      "step": 3590
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8483672985258934,
      "learning_rate": 5.891822839107195e-06,
      "loss": 0.6689,
      "step": 3591
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3057496282477743,
      "learning_rate": 5.8897828947805094e-06,
      "loss": 0.7911,
      "step": 3592
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4565480914772784,
      "learning_rate": 5.887742797489828e-06,
      "loss": 0.7274,
      "step": 3593
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4287425985225592,
      "learning_rate": 5.8857025475858676e-06,
      "loss": 0.8407,
      "step": 3594
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9510689385162426,
      "learning_rate": 5.883662145419373e-06,
      "loss": 0.7026,
      "step": 3595
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.355128458548463,
      "learning_rate": 5.881621591341109e-06,
      "loss": 0.7726,
      "step": 3596
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.982504665966904,
      "learning_rate": 5.879580885701874e-06,
      "loss": 0.6635,
      "step": 3597
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.294747446094127,
      "learning_rate": 5.877540028852489e-06,
      "loss": 0.7946,
      "step": 3598
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5348264619775085,
      "learning_rate": 5.875499021143799e-06,
      "loss": 0.7307,
      "step": 3599
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5813186973968274,
      "learning_rate": 5.873457862926677e-06,
      "loss": 0.7844,
      "step": 3600
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3813693275756873,
      "learning_rate": 5.871416554552021e-06,
      "loss": 0.8092,
      "step": 3601
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9345598108195158,
      "learning_rate": 5.869375096370759e-06,
      "loss": 0.6325,
      "step": 3602
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2905896437828879,
      "learning_rate": 5.86733348873384e-06,
      "loss": 0.715,
      "step": 3603
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8255979800671772,
      "learning_rate": 5.8652917319922374e-06,
      "loss": 0.6426,
      "step": 3604
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2321643498829342,
      "learning_rate": 5.863249826496955e-06,
      "loss": 0.7417,
      "step": 3605
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.609150029986657,
      "learning_rate": 5.8612077725990206e-06,
      "loss": 0.9233,
      "step": 3606
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4568395179079872,
      "learning_rate": 5.859165570649485e-06,
      "loss": 0.7706,
      "step": 3607
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2307746253247476,
      "learning_rate": 5.857123220999429e-06,
      "loss": 0.8297,
      "step": 3608
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4420984186167698,
      "learning_rate": 5.855080723999954e-06,
      "loss": 0.8306,
      "step": 3609
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4655971624834614,
      "learning_rate": 5.853038080002189e-06,
      "loss": 0.7521,
      "step": 3610
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5865141172782646,
      "learning_rate": 5.85099528935729e-06,
      "loss": 0.8286,
      "step": 3611
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2399706065457947,
      "learning_rate": 5.848952352416434e-06,
      "loss": 0.7536,
      "step": 3612
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.493831477900626,
      "learning_rate": 5.8469092695308274e-06,
      "loss": 0.8006,
      "step": 3613
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6335027324435778,
      "learning_rate": 5.844866041051699e-06,
      "loss": 0.758,
      "step": 3614
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2524032136677699,
      "learning_rate": 5.8428226673303026e-06,
      "loss": 0.7425,
      "step": 3615
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.591947559356277,
      "learning_rate": 5.84077914871792e-06,
      "loss": 0.8819,
      "step": 3616
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.229509935618177,
      "learning_rate": 5.838735485565855e-06,
      "loss": 0.762,
      "step": 3617
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3111574782361628,
      "learning_rate": 5.8366916782254345e-06,
      "loss": 0.776,
      "step": 3618
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4556659671446295,
      "learning_rate": 5.834647727048016e-06,
      "loss": 0.7577,
      "step": 3619
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3790534743780507,
      "learning_rate": 5.832603632384978e-06,
      "loss": 0.8198,
      "step": 3620
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.473736385935436,
      "learning_rate": 5.8305593945877236e-06,
      "loss": 0.7221,
      "step": 3621
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4600731967640228,
      "learning_rate": 5.828515014007678e-06,
      "loss": 0.8388,
      "step": 3622
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4836491600828983,
      "learning_rate": 5.826470490996299e-06,
      "loss": 0.7191,
      "step": 3623
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5099727224713928,
      "learning_rate": 5.82442582590506e-06,
      "loss": 0.7542,
      "step": 3624
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2434062362091394,
      "learning_rate": 5.822381019085466e-06,
      "loss": 0.7183,
      "step": 3625
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2652452718635565,
      "learning_rate": 5.820336070889038e-06,
      "loss": 0.682,
      "step": 3626
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7185445916672738,
      "learning_rate": 5.8182909816673316e-06,
      "loss": 0.6121,
      "step": 3627
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.6074383416421583,
      "learning_rate": 5.816245751771917e-06,
      "loss": 0.8361,
      "step": 3628
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4890496625852359,
      "learning_rate": 5.814200381554397e-06,
      "loss": 0.8148,
      "step": 3629
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9398238626515917,
      "learning_rate": 5.81215487136639e-06,
      "loss": 0.7164,
      "step": 3630
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3854782133272598,
      "learning_rate": 5.810109221559548e-06,
      "loss": 0.8011,
      "step": 3631
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7924753726217812,
      "learning_rate": 5.808063432485538e-06,
      "loss": 0.7002,
      "step": 3632
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8866234018340926,
      "learning_rate": 5.806017504496055e-06,
      "loss": 0.7206,
      "step": 3633
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3668218111224089,
      "learning_rate": 5.803971437942819e-06,
      "loss": 0.674,
      "step": 3634
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8438343953070953,
      "learning_rate": 5.801925233177574e-06,
      "loss": 0.6893,
      "step": 3635
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.903398989258631,
      "learning_rate": 5.799878890552083e-06,
      "loss": 0.7238,
      "step": 3636
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3983066970938849,
      "learning_rate": 5.79783241041814e-06,
      "loss": 0.7115,
      "step": 3637
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6406155245083773,
      "learning_rate": 5.795785793127554e-06,
      "loss": 0.7671,
      "step": 3638
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3534449435427,
      "learning_rate": 5.793739039032166e-06,
      "loss": 0.7077,
      "step": 3639
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4269814420470626,
      "learning_rate": 5.791692148483834e-06,
      "loss": 0.7825,
      "step": 3640
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3710985369732598,
      "learning_rate": 5.789645121834445e-06,
      "loss": 0.8674,
      "step": 3641
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8842561181704464,
      "learning_rate": 5.7875979594359045e-06,
      "loss": 0.6466,
      "step": 3642
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.447213347430707,
      "learning_rate": 5.785550661640145e-06,
      "loss": 0.84,
      "step": 3643
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6613405581604193,
      "learning_rate": 5.783503228799119e-06,
      "loss": 0.8686,
      "step": 3644
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3858172283162098,
      "learning_rate": 5.781455661264805e-06,
      "loss": 0.766,
      "step": 3645
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4868433061963975,
      "learning_rate": 5.779407959389205e-06,
      "loss": 0.727,
      "step": 3646
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3699033594043357,
      "learning_rate": 5.777360123524338e-06,
      "loss": 0.8031,
      "step": 3647
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2931283300456016,
      "learning_rate": 5.775312154022256e-06,
      "loss": 0.7404,
      "step": 3648
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.56942514844127,
      "learning_rate": 5.773264051235026e-06,
      "loss": 0.7467,
      "step": 3649
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3376000093611065,
      "learning_rate": 5.771215815514741e-06,
      "loss": 0.7338,
      "step": 3650
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5803879558572724,
      "learning_rate": 5.7691674472135175e-06,
      "loss": 0.8336,
      "step": 3651
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.373805380715851,
      "learning_rate": 5.767118946683491e-06,
      "loss": 0.7442,
      "step": 3652
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8966368295824654,
      "learning_rate": 5.765070314276826e-06,
      "loss": 0.6988,
      "step": 3653
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3319837500779426,
      "learning_rate": 5.763021550345703e-06,
      "loss": 0.7157,
      "step": 3654
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5173139570154914,
      "learning_rate": 5.760972655242332e-06,
      "loss": 0.7822,
      "step": 3655
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2746911754053492,
      "learning_rate": 5.758923629318938e-06,
      "loss": 0.7168,
      "step": 3656
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3466600187660291,
      "learning_rate": 5.756874472927775e-06,
      "loss": 0.8048,
      "step": 3657
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7800803736938056,
      "learning_rate": 5.7548251864211135e-06,
      "loss": 0.6747,
      "step": 3658
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4781930225681577,
      "learning_rate": 5.752775770151254e-06,
      "loss": 0.714,
      "step": 3659
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5501721255797853,
      "learning_rate": 5.750726224470513e-06,
      "loss": 0.8333,
      "step": 3660
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7112088136864363,
      "learning_rate": 5.748676549731229e-06,
      "loss": 0.7792,
      "step": 3661
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5839492349404038,
      "learning_rate": 5.746626746285767e-06,
      "loss": 0.8152,
      "step": 3662
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3824831005729201,
      "learning_rate": 5.744576814486512e-06,
      "loss": 0.7356,
      "step": 3663
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.205497008017004,
      "learning_rate": 5.742526754685872e-06,
      "loss": 0.6802,
      "step": 3664
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3348789643199233,
      "learning_rate": 5.740476567236273e-06,
      "loss": 0.7361,
      "step": 3665
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3705845098202103,
      "learning_rate": 5.738426252490168e-06,
      "loss": 0.7573,
      "step": 3666
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6631482014575503,
      "learning_rate": 5.73637581080003e-06,
      "loss": 0.7916,
      "step": 3667
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.443650735385371,
      "learning_rate": 5.734325242518355e-06,
      "loss": 0.8126,
      "step": 3668
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4014074584522365,
      "learning_rate": 5.732274547997655e-06,
      "loss": 0.7776,
      "step": 3669
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2567142084977057,
      "learning_rate": 5.730223727590472e-06,
      "loss": 0.7293,
      "step": 3670
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4317133360079581,
      "learning_rate": 5.728172781649365e-06,
      "loss": 0.8252,
      "step": 3671
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9101820443237335,
      "learning_rate": 5.726121710526914e-06,
      "loss": 0.6981,
      "step": 3672
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2553932029629977,
      "learning_rate": 5.7240705145757244e-06,
      "loss": 0.7913,
      "step": 3673
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.403959584345201,
      "learning_rate": 5.722019194148419e-06,
      "loss": 0.8468,
      "step": 3674
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4713840689124953,
      "learning_rate": 5.719967749597644e-06,
      "loss": 0.8292,
      "step": 3675
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7432061968275405,
      "learning_rate": 5.717916181276065e-06,
      "loss": 0.8478,
      "step": 3676
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4311579818705618,
      "learning_rate": 5.715864489536373e-06,
      "loss": 0.6989,
      "step": 3677
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.536929798387681,
      "learning_rate": 5.713812674731276e-06,
      "loss": 0.8838,
      "step": 3678
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.19032534223644,
      "learning_rate": 5.711760737213504e-06,
      "loss": 0.7916,
      "step": 3679
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3839939060290332,
      "learning_rate": 5.70970867733581e-06,
      "loss": 0.782,
      "step": 3680
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.5259654886958054,
      "learning_rate": 5.707656495450969e-06,
      "loss": 0.7258,
      "step": 3681
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4288128940679607,
      "learning_rate": 5.70560419191177e-06,
      "loss": 0.7775,
      "step": 3682
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3303819717487613,
      "learning_rate": 5.7035517670710304e-06,
      "loss": 0.7975,
      "step": 3683
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3757365161910704,
      "learning_rate": 5.701499221281587e-06,
      "loss": 0.855,
      "step": 3684
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2992197764697624,
      "learning_rate": 5.699446554896294e-06,
      "loss": 0.7262,
      "step": 3685
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2919822045836629,
      "learning_rate": 5.697393768268031e-06,
      "loss": 0.8219,
      "step": 3686
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2719031539352939,
      "learning_rate": 5.695340861749693e-06,
      "loss": 0.7608,
      "step": 3687
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3391859161837893,
      "learning_rate": 5.6932878356942005e-06,
      "loss": 0.676,
      "step": 3688
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3006279094361701,
      "learning_rate": 5.691234690454491e-06,
      "loss": 0.7514,
      "step": 3689
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3690719429123641,
      "learning_rate": 5.6891814263835245e-06,
      "loss": 0.8071,
      "step": 3690
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8421170547492164,
      "learning_rate": 5.687128043834282e-06,
      "loss": 0.6535,
      "step": 3691
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.365210655683294,
      "learning_rate": 5.685074543159763e-06,
      "loss": 0.7564,
      "step": 3692
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3068103052126334,
      "learning_rate": 5.683020924712987e-06,
      "loss": 0.8141,
      "step": 3693
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.205448864600369,
      "learning_rate": 5.680967188846996e-06,
      "loss": 0.7325,
      "step": 3694
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7318633458711359,
      "learning_rate": 5.678913335914852e-06,
      "loss": 0.8214,
      "step": 3695
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7873824046830982,
      "learning_rate": 5.676859366269635e-06,
      "loss": 0.6598,
      "step": 3696
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4601825034360547,
      "learning_rate": 5.6748052802644445e-06,
      "loss": 0.8114,
      "step": 3697
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3443417087680765,
      "learning_rate": 5.672751078252403e-06,
      "loss": 0.7367,
      "step": 3698
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3668261526643646,
      "learning_rate": 5.670696760586654e-06,
      "loss": 0.7697,
      "step": 3699
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2643114205275565,
      "learning_rate": 5.668642327620356e-06,
      "loss": 0.832,
      "step": 3700
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3280835258840833,
      "learning_rate": 5.66658777970669e-06,
      "loss": 0.7232,
      "step": 3701
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9107153212919231,
      "learning_rate": 5.664533117198856e-06,
      "loss": 0.6326,
      "step": 3702
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3182979585229804,
      "learning_rate": 5.662478340450076e-06,
      "loss": 0.814,
      "step": 3703
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7792261245429098,
      "learning_rate": 5.660423449813588e-06,
      "loss": 0.6019,
      "step": 3704
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4078597615588673,
      "learning_rate": 5.658368445642655e-06,
      "loss": 0.7398,
      "step": 3705
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6282111132912533,
      "learning_rate": 5.656313328290552e-06,
      "loss": 0.7467,
      "step": 3706
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7689115506801896,
      "learning_rate": 5.65425809811058e-06,
      "loss": 0.8482,
      "step": 3707
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8877238811846926,
      "learning_rate": 5.652202755456055e-06,
      "loss": 0.7035,
      "step": 3708
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3729819240336407,
      "learning_rate": 5.650147300680318e-06,
      "loss": 0.6868,
      "step": 3709
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2398295774781172,
      "learning_rate": 5.6480917341367205e-06,
      "loss": 0.7599,
      "step": 3710
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2722439340504637,
      "learning_rate": 5.646036056178641e-06,
      "loss": 0.7494,
      "step": 3711
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3638967301569105,
      "learning_rate": 5.643980267159474e-06,
      "loss": 0.7926,
      "step": 3712
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3523278657333115,
      "learning_rate": 5.641924367432634e-06,
      "loss": 0.7756,
      "step": 3713
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.313192232321905,
      "learning_rate": 5.639868357351554e-06,
      "loss": 0.796,
      "step": 3714
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8731196511603724,
      "learning_rate": 5.637812237269683e-06,
      "loss": 0.7037,
      "step": 3715
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2345314521303388,
      "learning_rate": 5.635756007540495e-06,
      "loss": 0.8448,
      "step": 3716
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.659347158922595,
      "learning_rate": 5.6336996685174795e-06,
      "loss": 0.8256,
      "step": 3717
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4657220187791242,
      "learning_rate": 5.631643220554146e-06,
      "loss": 0.7869,
      "step": 3718
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8123735775789782,
      "learning_rate": 5.629586664004018e-06,
      "loss": 0.6129,
      "step": 3719
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3796614087967767,
      "learning_rate": 5.6275299992206444e-06,
      "loss": 0.7919,
      "step": 3720
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2555466347205315,
      "learning_rate": 5.625473226557588e-06,
      "loss": 0.7664,
      "step": 3721
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2948606082009144,
      "learning_rate": 5.623416346368434e-06,
      "loss": 0.7009,
      "step": 3722
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.269629636496542,
      "learning_rate": 5.621359359006784e-06,
      "loss": 0.7991,
      "step": 3723
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4467083630708253,
      "learning_rate": 5.6193022648262555e-06,
      "loss": 0.8045,
      "step": 3724
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5594634760289146,
      "learning_rate": 5.617245064180488e-06,
      "loss": 0.7588,
      "step": 3725
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2583202433160805,
      "learning_rate": 5.615187757423137e-06,
      "loss": 0.8428,
      "step": 3726
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3178063661452855,
      "learning_rate": 5.613130344907881e-06,
      "loss": 0.806,
      "step": 3727
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2042107031175713,
      "learning_rate": 5.611072826988409e-06,
      "loss": 0.7632,
      "step": 3728
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.512120143646466,
      "learning_rate": 5.609015204018435e-06,
      "loss": 0.883,
      "step": 3729
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2867347527475463,
      "learning_rate": 5.6069574763516866e-06,
      "loss": 0.7273,
      "step": 3730
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9286076602284072,
      "learning_rate": 5.604899644341911e-06,
      "loss": 0.7025,
      "step": 3731
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2816335809041994,
      "learning_rate": 5.602841708342876e-06,
      "loss": 0.7304,
      "step": 3732
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3801502478421275,
      "learning_rate": 5.60078366870836e-06,
      "loss": 0.8197,
      "step": 3733
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8590057769040071,
      "learning_rate": 5.598725525792168e-06,
      "loss": 0.7509,
      "step": 3734
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.96474032539184,
      "learning_rate": 5.596667279948117e-06,
      "loss": 0.8102,
      "step": 3735
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4705839413356725,
      "learning_rate": 5.594608931530044e-06,
      "loss": 0.7591,
      "step": 3736
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5307446058167624,
      "learning_rate": 5.592550480891801e-06,
      "loss": 0.7418,
      "step": 3737
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3395737030999484,
      "learning_rate": 5.590491928387261e-06,
      "loss": 0.765,
      "step": 3738
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9496816058485831,
      "learning_rate": 5.588433274370314e-06,
      "loss": 0.6502,
      "step": 3739
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8868547243460706,
      "learning_rate": 5.586374519194863e-06,
      "loss": 0.6461,
      "step": 3740
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5259614637514687,
      "learning_rate": 5.584315663214836e-06,
      "loss": 0.8337,
      "step": 3741
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3888667605023972,
      "learning_rate": 5.582256706784174e-06,
      "loss": 0.6818,
      "step": 3742
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4814755319016681,
      "learning_rate": 5.580197650256832e-06,
      "loss": 0.8448,
      "step": 3743
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3996247415340608,
      "learning_rate": 5.578138493986788e-06,
      "loss": 0.7887,
      "step": 3744
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1119172775458925,
      "learning_rate": 5.576079238328035e-06,
      "loss": 0.7325,
      "step": 3745
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2722951281797974,
      "learning_rate": 5.574019883634582e-06,
      "loss": 0.7553,
      "step": 3746
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3639928350392452,
      "learning_rate": 5.571960430260457e-06,
      "loss": 0.7848,
      "step": 3747
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.320453836377715,
      "learning_rate": 5.569900878559704e-06,
      "loss": 0.8108,
      "step": 3748
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8782867265296518,
      "learning_rate": 5.567841228886381e-06,
      "loss": 0.6869,
      "step": 3749
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5105497855851944,
      "learning_rate": 5.56578148159457e-06,
      "loss": 0.7512,
      "step": 3750
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2399186628334893,
      "learning_rate": 5.5637216370383615e-06,
      "loss": 0.708,
      "step": 3751
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1857614435977988,
      "learning_rate": 5.561661695571869e-06,
      "loss": 0.8205,
      "step": 3752
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.133024373391581,
      "learning_rate": 5.559601657549219e-06,
      "loss": 0.8411,
      "step": 3753
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.56474876586458,
      "learning_rate": 5.557541523324558e-06,
      "loss": 0.7414,
      "step": 3754
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.953918084915043,
      "learning_rate": 5.555481293252044e-06,
      "loss": 0.8194,
      "step": 3755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7851502127178667,
      "learning_rate": 5.553420967685857e-06,
      "loss": 0.6231,
      "step": 3756
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3545817209187272,
      "learning_rate": 5.551360546980191e-06,
      "loss": 0.7612,
      "step": 3757
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4069985428457317,
      "learning_rate": 5.549300031489252e-06,
      "loss": 0.8494,
      "step": 3758
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3110562511673522,
      "learning_rate": 5.547239421567272e-06,
      "loss": 0.8009,
      "step": 3759
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.404135306055435,
      "learning_rate": 5.545178717568491e-06,
      "loss": 0.7647,
      "step": 3760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8636656136074444,
      "learning_rate": 5.5431179198471674e-06,
      "loss": 0.6597,
      "step": 3761
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3225359747755872,
      "learning_rate": 5.5410570287575775e-06,
      "loss": 0.7452,
      "step": 3762
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3822359364790353,
      "learning_rate": 5.538996044654011e-06,
      "loss": 0.7718,
      "step": 3763
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5035878272612297,
      "learning_rate": 5.536934967890778e-06,
      "loss": 0.8229,
      "step": 3764
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.360452535259453,
      "learning_rate": 5.5348737988221964e-06,
      "loss": 0.7709,
      "step": 3765
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.275993586841153,
      "learning_rate": 5.532812537802611e-06,
      "loss": 0.7834,
      "step": 3766
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3911540695372688,
      "learning_rate": 5.530751185186372e-06,
      "loss": 0.7807,
      "step": 3767
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3076235038971888,
      "learning_rate": 5.528689741327851e-06,
      "loss": 0.7435,
      "step": 3768
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8971067467664953,
      "learning_rate": 5.526628206581434e-06,
      "loss": 0.6642,
      "step": 3769
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8314812476209352,
      "learning_rate": 5.5245665813015245e-06,
      "loss": 0.6404,
      "step": 3770
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3979402741333424,
      "learning_rate": 5.522504865842537e-06,
      "loss": 0.7666,
      "step": 3771
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4583685387161391,
      "learning_rate": 5.520443060558907e-06,
      "loss": 0.8322,
      "step": 3772
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.40781925688334,
      "learning_rate": 5.51838116580508e-06,
      "loss": 0.8079,
      "step": 3773
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8276485632372166,
      "learning_rate": 5.5163191819355225e-06,
      "loss": 0.6572,
      "step": 3774
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.418729256305684,
      "learning_rate": 5.514257109304712e-06,
      "loss": 0.8977,
      "step": 3775
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5790640327038383,
      "learning_rate": 5.51219494826714e-06,
      "loss": 0.7495,
      "step": 3776
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3787366204735056,
      "learning_rate": 5.51013269917732e-06,
      "loss": 0.7404,
      "step": 3777
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5485356271004698,
      "learning_rate": 5.5080703623897754e-06,
      "loss": 0.7957,
      "step": 3778
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8199597704698911,
      "learning_rate": 5.506007938259045e-06,
      "loss": 0.6625,
      "step": 3779
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.191964234678387,
      "learning_rate": 5.5039454271396856e-06,
      "loss": 0.7837,
      "step": 3780
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1643596742029336,
      "learning_rate": 5.501882829386262e-06,
      "loss": 0.7892,
      "step": 3781
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4819161706507697,
      "learning_rate": 5.499820145353364e-06,
      "loss": 0.774,
      "step": 3782
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2645381119666173,
      "learning_rate": 5.497757375395588e-06,
      "loss": 0.7734,
      "step": 3783
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3298546865273464,
      "learning_rate": 5.49569451986755e-06,
      "loss": 0.8553,
      "step": 3784
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.394242869122923,
      "learning_rate": 5.493631579123876e-06,
      "loss": 0.8098,
      "step": 3785
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.856358594335527,
      "learning_rate": 5.49156855351921e-06,
      "loss": 0.6465,
      "step": 3786
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5573543831944108,
      "learning_rate": 5.4895054434082115e-06,
      "loss": 0.7288,
      "step": 3787
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.475224999417481,
      "learning_rate": 5.487442249145554e-06,
      "loss": 0.7075,
      "step": 3788
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.862576597691678,
      "learning_rate": 5.485378971085921e-06,
      "loss": 0.7418,
      "step": 3789
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9584132320637201,
      "learning_rate": 5.4833156095840155e-06,
      "loss": 0.708,
      "step": 3790
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1932780797817995,
      "learning_rate": 5.481252164994555e-06,
      "loss": 0.8102,
      "step": 3791
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2261847038694544,
      "learning_rate": 5.479188637672269e-06,
      "loss": 0.7931,
      "step": 3792
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3234412996259832,
      "learning_rate": 5.4771250279719e-06,
      "loss": 0.8238,
      "step": 3793
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.658451206506786,
      "learning_rate": 5.475061336248208e-06,
      "loss": 0.8243,
      "step": 3794
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9244632140582292,
      "learning_rate": 5.4729975628559626e-06,
      "loss": 0.6672,
      "step": 3795
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7954918809647372,
      "learning_rate": 5.470933708149955e-06,
      "loss": 0.686,
      "step": 3796
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.275645628521064,
      "learning_rate": 5.468869772484982e-06,
      "loss": 0.7804,
      "step": 3797
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8146231244032173,
      "learning_rate": 5.4668057562158604e-06,
      "loss": 0.6319,
      "step": 3798
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4287935851863556,
      "learning_rate": 5.464741659697419e-06,
      "loss": 0.812,
      "step": 3799
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4281264141975962,
      "learning_rate": 5.462677483284496e-06,
      "loss": 0.8298,
      "step": 3800
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3815097723410426,
      "learning_rate": 5.46061322733195e-06,
      "loss": 0.7403,
      "step": 3801
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6207213906001299,
      "learning_rate": 5.458548892194652e-06,
      "loss": 0.7186,
      "step": 3802
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.102024167454047,
      "learning_rate": 5.4564844782274826e-06,
      "loss": 0.7247,
      "step": 3803
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.262286666500523,
      "learning_rate": 5.45441998578534e-06,
      "loss": 0.853,
      "step": 3804
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3459254148863693,
      "learning_rate": 5.452355415223132e-06,
      "loss": 0.8079,
      "step": 3805
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4957598012242346,
      "learning_rate": 5.450290766895786e-06,
      "loss": 0.7725,
      "step": 3806
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2711162577611121,
      "learning_rate": 5.4482260411582365e-06,
      "loss": 0.7315,
      "step": 3807
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3250200406869928,
      "learning_rate": 5.446161238365433e-06,
      "loss": 0.8044,
      "step": 3808
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3130445189210584,
      "learning_rate": 5.44409635887234e-06,
      "loss": 0.855,
      "step": 3809
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2488516892307697,
      "learning_rate": 5.4420314030339375e-06,
      "loss": 0.7979,
      "step": 3810
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3874502919552643,
      "learning_rate": 5.439966371205209e-06,
      "loss": 0.8016,
      "step": 3811
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2645651214005897,
      "learning_rate": 5.437901263741163e-06,
      "loss": 0.6945,
      "step": 3812
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.308891580527468,
      "learning_rate": 5.435836080996812e-06,
      "loss": 0.7695,
      "step": 3813
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8556205087388389,
      "learning_rate": 5.433770823327187e-06,
      "loss": 0.6881,
      "step": 3814
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3086654335332484,
      "learning_rate": 5.431705491087327e-06,
      "loss": 0.7838,
      "step": 3815
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4879408121398712,
      "learning_rate": 5.42964008463229e-06,
      "loss": 0.8075,
      "step": 3816
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3136314924149066,
      "learning_rate": 5.427574604317142e-06,
      "loss": 0.7633,
      "step": 3817
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2837788921940063,
      "learning_rate": 5.4255090504969606e-06,
      "loss": 0.8924,
      "step": 3818
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3349142897127713,
      "learning_rate": 5.423443423526842e-06,
      "loss": 0.7093,
      "step": 3819
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4634929368447782,
      "learning_rate": 5.42137772376189e-06,
      "loss": 0.7953,
      "step": 3820
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2185624211115778,
      "learning_rate": 5.419311951557222e-06,
      "loss": 0.7547,
      "step": 3821
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4149117114932148,
      "learning_rate": 5.417246107267968e-06,
      "loss": 0.7279,
      "step": 3822
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5899206448794183,
      "learning_rate": 5.415180191249271e-06,
      "loss": 0.8122,
      "step": 3823
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3727696214032046,
      "learning_rate": 5.413114203856287e-06,
      "loss": 0.7853,
      "step": 3824
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.459196871949327,
      "learning_rate": 5.411048145444182e-06,
      "loss": 0.7214,
      "step": 3825
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.615715771139039,
      "learning_rate": 5.408982016368134e-06,
      "loss": 0.7406,
      "step": 3826
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4161377874023495,
      "learning_rate": 5.406915816983337e-06,
      "loss": 0.7977,
      "step": 3827
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2874627938449243,
      "learning_rate": 5.404849547644993e-06,
      "loss": 0.776,
      "step": 3828
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4213643943116876,
      "learning_rate": 5.40278320870832e-06,
      "loss": 0.8137,
      "step": 3829
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8073503539729701,
      "learning_rate": 5.400716800528542e-06,
      "loss": 0.6604,
      "step": 3830
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.325121128191264,
      "learning_rate": 5.398650323460904e-06,
      "loss": 0.7898,
      "step": 3831
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2793288121914133,
      "learning_rate": 5.39658377786065e-06,
      "loss": 0.7909,
      "step": 3832
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8607162427305738,
      "learning_rate": 5.394517164083047e-06,
      "loss": 0.6603,
      "step": 3833
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5832574390003953,
      "learning_rate": 5.392450482483372e-06,
      "loss": 0.7344,
      "step": 3834
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2906882559212274,
      "learning_rate": 5.39038373341691e-06,
      "loss": 0.8348,
      "step": 3835
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4132328208452432,
      "learning_rate": 5.388316917238956e-06,
      "loss": 0.7241,
      "step": 3836
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.43283675843317,
      "learning_rate": 5.386250034304823e-06,
      "loss": 0.7314,
      "step": 3837
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6919808556654372,
      "learning_rate": 5.384183084969832e-06,
      "loss": 0.7769,
      "step": 3838
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2666335338834562,
      "learning_rate": 5.382116069589315e-06,
      "loss": 0.839,
      "step": 3839
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3692122569708294,
      "learning_rate": 5.380048988518613e-06,
      "loss": 0.7772,
      "step": 3840
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.367210285123109,
      "learning_rate": 5.3779818421130865e-06,
      "loss": 0.8699,
      "step": 3841
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2881472452809215,
      "learning_rate": 5.3759146307281e-06,
      "loss": 0.8099,
      "step": 3842
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3614958748540829,
      "learning_rate": 5.37384735471903e-06,
      "loss": 0.7755,
      "step": 3843
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3249407726487208,
      "learning_rate": 5.371780014441265e-06,
      "loss": 0.8103,
      "step": 3844
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2684673930600316,
      "learning_rate": 5.369712610250206e-06,
      "loss": 0.7225,
      "step": 3845
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2263212272604165,
      "learning_rate": 5.367645142501262e-06,
      "loss": 0.8328,
      "step": 3846
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.585503877648328,
      "learning_rate": 5.365577611549856e-06,
      "loss": 0.7984,
      "step": 3847
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.796381561511074,
      "learning_rate": 5.363510017751422e-06,
      "loss": 0.6863,
      "step": 3848
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8479379235778962,
      "learning_rate": 5.3614423614614016e-06,
      "loss": 0.6321,
      "step": 3849
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4445790386073794,
      "learning_rate": 5.359374643035248e-06,
      "loss": 0.766,
      "step": 3850
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.733393865804452,
      "learning_rate": 5.357306862828427e-06,
      "loss": 0.6285,
      "step": 3851
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5955928982193166,
      "learning_rate": 5.355239021196416e-06,
      "loss": 0.7867,
      "step": 3852
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3761335956026612,
      "learning_rate": 5.353171118494698e-06,
      "loss": 0.7479,
      "step": 3853
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4990839027344884,
      "learning_rate": 5.351103155078768e-06,
      "loss": 0.7865,
      "step": 3854
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8329682227219574,
      "learning_rate": 5.349035131304138e-06,
      "loss": 0.6317,
      "step": 3855
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3678275593576639,
      "learning_rate": 5.346967047526321e-06,
      "loss": 0.8457,
      "step": 3856
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.251563307563721,
      "learning_rate": 5.344898904100848e-06,
      "loss": 0.7888,
      "step": 3857
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.49679127643013,
      "learning_rate": 5.342830701383254e-06,
      "loss": 0.7455,
      "step": 3858
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2654958739107274,
      "learning_rate": 5.340762439729088e-06,
      "loss": 0.854,
      "step": 3859
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.326907507906043,
      "learning_rate": 5.338694119493908e-06,
      "loss": 0.7882,
      "step": 3860
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7799295048885434,
      "learning_rate": 5.336625741033283e-06,
      "loss": 0.6567,
      "step": 3861
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3625998787555123,
      "learning_rate": 5.33455730470279e-06,
      "loss": 0.7543,
      "step": 3862
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4554999684158973,
      "learning_rate": 5.332488810858017e-06,
      "loss": 0.7805,
      "step": 3863
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4704621481056948,
      "learning_rate": 5.330420259854564e-06,
      "loss": 0.7373,
      "step": 3864
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3138537630956302,
      "learning_rate": 5.328351652048037e-06,
      "loss": 0.8306,
      "step": 3865
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4789256484304718,
      "learning_rate": 5.326282987794056e-06,
      "loss": 0.8065,
      "step": 3866
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6861126415613765,
      "learning_rate": 5.3242142674482456e-06,
      "loss": 0.7998,
      "step": 3867
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.266878500520973,
      "learning_rate": 5.322145491366244e-06,
      "loss": 0.7452,
      "step": 3868
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3083288513758005,
      "learning_rate": 5.320076659903698e-06,
      "loss": 0.8218,
      "step": 3869
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7383682954245278,
      "learning_rate": 5.318007773416264e-06,
      "loss": 0.5908,
      "step": 3870
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2425782757845418,
      "learning_rate": 5.315938832259606e-06,
      "loss": 0.6941,
      "step": 3871
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8044711473422284,
      "learning_rate": 5.3138698367894e-06,
      "loss": 0.7993,
      "step": 3872
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8835500378041079,
      "learning_rate": 5.311800787361332e-06,
      "loss": 0.6814,
      "step": 3873
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4113696780433216,
      "learning_rate": 5.309731684331093e-06,
      "loss": 0.8443,
      "step": 3874
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1584343302278715,
      "learning_rate": 5.307662528054387e-06,
      "loss": 0.8037,
      "step": 3875
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3972783934110677,
      "learning_rate": 5.305593318886925e-06,
      "loss": 0.7321,
      "step": 3876
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1518591727002956,
      "learning_rate": 5.303524057184429e-06,
      "loss": 0.712,
      "step": 3877
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5715854081620626,
      "learning_rate": 5.301454743302629e-06,
      "loss": 0.7947,
      "step": 3878
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4636143855494805,
      "learning_rate": 5.299385377597265e-06,
      "loss": 0.8355,
      "step": 3879
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.887012770097883,
      "learning_rate": 5.297315960424081e-06,
      "loss": 0.7767,
      "step": 3880
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3870824969209608,
      "learning_rate": 5.295246492138839e-06,
      "loss": 0.7854,
      "step": 3881
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3125040822355807,
      "learning_rate": 5.293176973097303e-06,
      "loss": 0.7731,
      "step": 3882
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8280938851060915,
      "learning_rate": 5.2911074036552426e-06,
      "loss": 0.6174,
      "step": 3883
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8600046731179459,
      "learning_rate": 5.289037784168448e-06,
      "loss": 0.6832,
      "step": 3884
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4636117319794248,
      "learning_rate": 5.286968114992706e-06,
      "loss": 0.8231,
      "step": 3885
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3888685197197732,
      "learning_rate": 5.28489839648382e-06,
      "loss": 0.8436,
      "step": 3886
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3597998933537843,
      "learning_rate": 5.282828628997595e-06,
      "loss": 0.7973,
      "step": 3887
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.415978477644527,
      "learning_rate": 5.28075881288985e-06,
      "loss": 0.7383,
      "step": 3888
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2310134354114712,
      "learning_rate": 5.2786889485164115e-06,
      "loss": 0.7358,
      "step": 3889
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.483025007052184,
      "learning_rate": 5.276619036233111e-06,
      "loss": 0.7684,
      "step": 3890
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2863267940180807,
      "learning_rate": 5.274549076395792e-06,
      "loss": 0.7835,
      "step": 3891
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3810360046701151,
      "learning_rate": 5.2724790693603025e-06,
      "loss": 0.7869,
      "step": 3892
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3367522946871604,
      "learning_rate": 5.270409015482504e-06,
      "loss": 0.7324,
      "step": 3893
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.431205835890476,
      "learning_rate": 5.268338915118258e-06,
      "loss": 0.7909,
      "step": 3894
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2965946768209298,
      "learning_rate": 5.266268768623445e-06,
      "loss": 0.7156,
      "step": 3895
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8774810219350532,
      "learning_rate": 5.264198576353942e-06,
      "loss": 0.6737,
      "step": 3896
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2968119421276703,
      "learning_rate": 5.262128338665641e-06,
      "loss": 0.8177,
      "step": 3897
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.235439783696868,
      "learning_rate": 5.26005805591444e-06,
      "loss": 0.8567,
      "step": 3898
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3399731501428025,
      "learning_rate": 5.257987728456244e-06,
      "loss": 0.7973,
      "step": 3899
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.757672816247089,
      "learning_rate": 5.255917356646968e-06,
      "loss": 0.8369,
      "step": 3900
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2913187833592343,
      "learning_rate": 5.2538469408425284e-06,
      "loss": 0.6295,
      "step": 3901
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8701715302112696,
      "learning_rate": 5.25177648139886e-06,
      "loss": 0.6951,
      "step": 3902
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3680947982355238,
      "learning_rate": 5.2497059786718955e-06,
      "loss": 0.8665,
      "step": 3903
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2882944386792512,
      "learning_rate": 5.247635433017576e-06,
      "loss": 0.7608,
      "step": 3904
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3752523637244136,
      "learning_rate": 5.245564844791857e-06,
      "loss": 0.8822,
      "step": 3905
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3253330267089873,
      "learning_rate": 5.243494214350693e-06,
      "loss": 0.7807,
      "step": 3906
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9229384738953146,
      "learning_rate": 5.2414235420500516e-06,
      "loss": 0.6848,
      "step": 3907
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3610012314114264,
      "learning_rate": 5.2393528282459036e-06,
      "loss": 0.714,
      "step": 3908
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8215269193197297,
      "learning_rate": 5.237282073294231e-06,
      "loss": 0.6478,
      "step": 3909
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9991899718598495,
      "learning_rate": 5.235211277551019e-06,
      "loss": 0.6692,
      "step": 3910
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2661869952821903,
      "learning_rate": 5.2331404413722595e-06,
      "loss": 0.807,
      "step": 3911
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9422157924777924,
      "learning_rate": 5.231069565113957e-06,
      "loss": 0.6454,
      "step": 3912
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3581447871202785,
      "learning_rate": 5.2289986491321176e-06,
      "loss": 0.719,
      "step": 3913
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4516935287521242,
      "learning_rate": 5.226927693782755e-06,
      "loss": 0.701,
      "step": 3914
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.314894343670687,
      "learning_rate": 5.22485669942189e-06,
      "loss": 0.7636,
      "step": 3915
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4347132501872553,
      "learning_rate": 5.222785666405553e-06,
      "loss": 0.7484,
      "step": 3916
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.957282868153587,
      "learning_rate": 5.220714595089777e-06,
      "loss": 0.7431,
      "step": 3917
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1854232700771603,
      "learning_rate": 5.218643485830604e-06,
      "loss": 0.7456,
      "step": 3918
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4559337928068945,
      "learning_rate": 5.216572338984081e-06,
      "loss": 0.757,
      "step": 3919
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3998679511951677,
      "learning_rate": 5.21450115490626e-06,
      "loss": 0.7738,
      "step": 3920
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8351875941465996,
      "learning_rate": 5.212429933953207e-06,
      "loss": 0.5931,
      "step": 3921
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5007220163465596,
      "learning_rate": 5.210358676480983e-06,
      "loss": 0.7544,
      "step": 3922
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4256337183425136,
      "learning_rate": 5.208287382845666e-06,
      "loss": 0.8607,
      "step": 3923
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0923649013894636,
      "learning_rate": 5.206216053403333e-06,
      "loss": 0.7462,
      "step": 3924
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.352828754778774,
      "learning_rate": 5.204144688510069e-06,
      "loss": 0.8222,
      "step": 3925
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3908592794113501,
      "learning_rate": 5.202073288521965e-06,
      "loss": 0.7006,
      "step": 3926
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3964793241702809,
      "learning_rate": 5.2000018537951226e-06,
      "loss": 0.8492,
      "step": 3927
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3309834307485404,
      "learning_rate": 5.197930384685642e-06,
      "loss": 0.7339,
      "step": 3928
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.332617434146456,
      "learning_rate": 5.195858881549633e-06,
      "loss": 0.7005,
      "step": 3929
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8054288965627372,
      "learning_rate": 5.19378734474321e-06,
      "loss": 0.6645,
      "step": 3930
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.277847312972719,
      "learning_rate": 5.1917157746225e-06,
      "loss": 0.8048,
      "step": 3931
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3377701802666573,
      "learning_rate": 5.189644171543624e-06,
      "loss": 0.8725,
      "step": 3932
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3318592636399598,
      "learning_rate": 5.187572535862715e-06,
      "loss": 0.8377,
      "step": 3933
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.639509983819673,
      "learning_rate": 5.185500867935913e-06,
      "loss": 0.7887,
      "step": 3934
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5368224856654962,
      "learning_rate": 5.183429168119362e-06,
      "loss": 0.8241,
      "step": 3935
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.817065545865941,
      "learning_rate": 5.181357436769211e-06,
      "loss": 0.6438,
      "step": 3936
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5173154770830923,
      "learning_rate": 5.1792856742416145e-06,
      "loss": 0.6687,
      "step": 3937
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.707724107499714,
      "learning_rate": 5.17721388089273e-06,
      "loss": 0.7311,
      "step": 3938
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.562214108569005,
      "learning_rate": 5.175142057078727e-06,
      "loss": 0.7537,
      "step": 3939
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4033838704444683,
      "learning_rate": 5.173070203155772e-06,
      "loss": 0.803,
      "step": 3940
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2844641831599453,
      "learning_rate": 5.170998319480044e-06,
      "loss": 0.7512,
      "step": 3941
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2679616388242874,
      "learning_rate": 5.168926406407723e-06,
      "loss": 0.7576,
      "step": 3942
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4865573378978245,
      "learning_rate": 5.166854464294993e-06,
      "loss": 0.6935,
      "step": 3943
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3536372547498852,
      "learning_rate": 5.164782493498046e-06,
      "loss": 0.8366,
      "step": 3944
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.062379739736857,
      "learning_rate": 5.16271049437308e-06,
      "loss": 0.6779,
      "step": 3945
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1811360953879388,
      "learning_rate": 5.160638467276293e-06,
      "loss": 0.8068,
      "step": 3946
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6423378880718762,
      "learning_rate": 5.15856641256389e-06,
      "loss": 0.7744,
      "step": 3947
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3618031368719041,
      "learning_rate": 5.156494330592084e-06,
      "loss": 0.739,
      "step": 3948
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3253187542012732,
      "learning_rate": 5.154422221717087e-06,
      "loss": 0.811,
      "step": 3949
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3917094119664755,
      "learning_rate": 5.152350086295121e-06,
      "loss": 0.7517,
      "step": 3950
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8276705474392033,
      "learning_rate": 5.150277924682408e-06,
      "loss": 0.6396,
      "step": 3951
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.472339515001979,
      "learning_rate": 5.148205737235178e-06,
      "loss": 0.6457,
      "step": 3952
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7555839329796308,
      "learning_rate": 5.146133524309664e-06,
      "loss": 0.6806,
      "step": 3953
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9446053932613895,
      "learning_rate": 5.144061286262102e-06,
      "loss": 0.6741,
      "step": 3954
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.374077304222365,
      "learning_rate": 5.141989023448736e-06,
      "loss": 0.6949,
      "step": 3955
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3268102307522474,
      "learning_rate": 5.139916736225811e-06,
      "loss": 0.8196,
      "step": 3956
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4078490210266035,
      "learning_rate": 5.137844424949576e-06,
      "loss": 0.7824,
      "step": 3957
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9772230841095253,
      "learning_rate": 5.135772089976285e-06,
      "loss": 0.7434,
      "step": 3958
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7395351841123954,
      "learning_rate": 5.133699731662201e-06,
      "loss": 0.7774,
      "step": 3959
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5885830247239137,
      "learning_rate": 5.131627350363582e-06,
      "loss": 0.7437,
      "step": 3960
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3484461448200844,
      "learning_rate": 5.1295549464366944e-06,
      "loss": 0.7788,
      "step": 3961
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5828277214649173,
      "learning_rate": 5.1274825202378085e-06,
      "loss": 0.8241,
      "step": 3962
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.477629368254742,
      "learning_rate": 5.125410072123203e-06,
      "loss": 0.9143,
      "step": 3963
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3030465678802368,
      "learning_rate": 5.12333760244915e-06,
      "loss": 0.7547,
      "step": 3964
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.446605197929685,
      "learning_rate": 5.121265111571933e-06,
      "loss": 0.8397,
      "step": 3965
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.312410929682147,
      "learning_rate": 5.119192599847838e-06,
      "loss": 0.7982,
      "step": 3966
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3927627615073985,
      "learning_rate": 5.1171200676331535e-06,
      "loss": 0.7124,
      "step": 3967
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9980391833454736,
      "learning_rate": 5.115047515284173e-06,
      "loss": 0.6725,
      "step": 3968
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6585759482355273,
      "learning_rate": 5.112974943157188e-06,
      "loss": 0.7722,
      "step": 3969
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2277735458088168,
      "learning_rate": 5.110902351608504e-06,
      "loss": 0.7571,
      "step": 3970
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3500814446321079,
      "learning_rate": 5.108829740994417e-06,
      "loss": 0.7726,
      "step": 3971
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4752911702572522,
      "learning_rate": 5.106757111671235e-06,
      "loss": 0.8021,
      "step": 3972
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2077937445136575,
      "learning_rate": 5.104684463995271e-06,
      "loss": 0.8195,
      "step": 3973
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3394841434804643,
      "learning_rate": 5.102611798322833e-06,
      "loss": 0.7993,
      "step": 3974
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.333777913884664,
      "learning_rate": 5.100539115010237e-06,
      "loss": 0.7592,
      "step": 3975
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.500042767434039,
      "learning_rate": 5.098466414413801e-06,
      "loss": 0.6893,
      "step": 3976
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7456354288221099,
      "learning_rate": 5.096393696889848e-06,
      "loss": 0.8694,
      "step": 3977
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5804223652437215,
      "learning_rate": 5.094320962794701e-06,
      "loss": 0.7987,
      "step": 3978
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5548920871352225,
      "learning_rate": 5.092248212484686e-06,
      "loss": 0.824,
      "step": 3979
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4458718076733954,
      "learning_rate": 5.090175446316133e-06,
      "loss": 0.7827,
      "step": 3980
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3183326205876162,
      "learning_rate": 5.088102664645378e-06,
      "loss": 0.8052,
      "step": 3981
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4014272543742186,
      "learning_rate": 5.086029867828752e-06,
      "loss": 0.7008,
      "step": 3982
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8683966856337827,
      "learning_rate": 5.083957056222595e-06,
      "loss": 0.6097,
      "step": 3983
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2318106052946562,
      "learning_rate": 5.081884230183247e-06,
      "loss": 0.7677,
      "step": 3984
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7879243761279723,
      "learning_rate": 5.079811390067049e-06,
      "loss": 0.6628,
      "step": 3985
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5898217312298593,
      "learning_rate": 5.07773853623035e-06,
      "loss": 0.8882,
      "step": 3986
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3666928712847408,
      "learning_rate": 5.075665669029494e-06,
      "loss": 0.7587,
      "step": 3987
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3165718493824539,
      "learning_rate": 5.073592788820832e-06,
      "loss": 0.7469,
      "step": 3988
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5501606685341585,
      "learning_rate": 5.071519895960717e-06,
      "loss": 0.8478,
      "step": 3989
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2680276465696092,
      "learning_rate": 5.069446990805501e-06,
      "loss": 0.7779,
      "step": 3990
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2299905287868416,
      "learning_rate": 5.067374073711545e-06,
      "loss": 0.7052,
      "step": 3991
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4107605791944295,
      "learning_rate": 5.065301145035204e-06,
      "loss": 0.8165,
      "step": 3992
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.386148715387468,
      "learning_rate": 5.063228205132838e-06,
      "loss": 0.7687,
      "step": 3993
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7737347222300871,
      "learning_rate": 5.061155254360811e-06,
      "loss": 0.6367,
      "step": 3994
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.512425672762187,
      "learning_rate": 5.059082293075487e-06,
      "loss": 0.7882,
      "step": 3995
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3347966185713707,
      "learning_rate": 5.057009321633233e-06,
      "loss": 0.7967,
      "step": 3996
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8329176303727095,
      "learning_rate": 5.054936340390414e-06,
      "loss": 0.734,
      "step": 3997
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.36148295087241,
      "learning_rate": 5.052863349703402e-06,
      "loss": 0.7676,
      "step": 3998
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.480079504221475,
      "learning_rate": 5.050790349928566e-06,
      "loss": 0.7526,
      "step": 3999
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4057777031984857,
      "learning_rate": 5.048717341422282e-06,
      "loss": 0.8467,
      "step": 4000
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4420263550763122,
      "learning_rate": 5.0466443245409215e-06,
      "loss": 0.8129,
      "step": 4001
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2350849428570958,
      "learning_rate": 5.04457129964086e-06,
      "loss": 0.7776,
      "step": 4002
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7783274694495372,
      "learning_rate": 5.042498267078476e-06,
      "loss": 0.6738,
      "step": 4003
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2607854504581317,
      "learning_rate": 5.040425227210148e-06,
      "loss": 0.6696,
      "step": 4004
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.465099409861707,
      "learning_rate": 5.038352180392254e-06,
      "loss": 0.7908,
      "step": 4005
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3638292950342101,
      "learning_rate": 5.036279126981176e-06,
      "loss": 0.8148,
      "step": 4006
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3501340743829038,
      "learning_rate": 5.034206067333296e-06,
      "loss": 0.7243,
      "step": 4007
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3748970384102988,
      "learning_rate": 5.032133001804994e-06,
      "loss": 0.7912,
      "step": 4008
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3075764461825266,
      "learning_rate": 5.030059930752659e-06,
      "loss": 0.6779,
      "step": 4009
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2617375219589237,
      "learning_rate": 5.027986854532673e-06,
      "loss": 0.7928,
      "step": 4010
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3437803904577283,
      "learning_rate": 5.025913773501421e-06,
      "loss": 0.7975,
      "step": 4011
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2543148795391974,
      "learning_rate": 5.023840688015291e-06,
      "loss": 0.7918,
      "step": 4012
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.500559562766042,
      "learning_rate": 5.021767598430672e-06,
      "loss": 0.7356,
      "step": 4013
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8092492366607545,
      "learning_rate": 5.019694505103949e-06,
      "loss": 0.7385,
      "step": 4014
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4036293641129138,
      "learning_rate": 5.017621408391513e-06,
      "loss": 0.7463,
      "step": 4015
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4856012072043536,
      "learning_rate": 5.015548308649752e-06,
      "loss": 0.7654,
      "step": 4016
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3651614190941974,
      "learning_rate": 5.013475206235056e-06,
      "loss": 0.7725,
      "step": 4017
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.382258823876229,
      "learning_rate": 5.0114021015038185e-06,
      "loss": 0.6906,
      "step": 4018
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.521184455125855,
      "learning_rate": 5.0093289948124244e-06,
      "loss": 0.8582,
      "step": 4019
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.28639975080556,
      "learning_rate": 5.00725588651727e-06,
      "loss": 0.7036,
      "step": 4020
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3599733594340684,
      "learning_rate": 5.005182776974745e-06,
      "loss": 0.8904,
      "step": 4021
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2382302389172057,
      "learning_rate": 5.0031096665412385e-06,
      "loss": 0.8342,
      "step": 4022
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3383251661239226,
      "learning_rate": 5.0010365555731455e-06,
      "loss": 0.7388,
      "step": 4023
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5325024780473044,
      "learning_rate": 4.998963444426855e-06,
      "loss": 0.7793,
      "step": 4024
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4009550855720323,
      "learning_rate": 4.996890333458762e-06,
      "loss": 0.8181,
      "step": 4025
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8237868442301204,
      "learning_rate": 4.994817223025256e-06,
      "loss": 0.6925,
      "step": 4026
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8497149632811752,
      "learning_rate": 4.992744113482731e-06,
      "loss": 0.6113,
      "step": 4027
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.538679997093105,
      "learning_rate": 4.990671005187576e-06,
      "loss": 0.7568,
      "step": 4028
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3998226363107422,
      "learning_rate": 4.988597898496183e-06,
      "loss": 0.7622,
      "step": 4029
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2188752218567334,
      "learning_rate": 4.9865247937649445e-06,
      "loss": 0.7263,
      "step": 4030
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3248372162197253,
      "learning_rate": 4.98445169135025e-06,
      "loss": 0.7806,
      "step": 4031
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0669736588477647,
      "learning_rate": 4.982378591608489e-06,
      "loss": 0.8429,
      "step": 4032
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2827133863016558,
      "learning_rate": 4.980305494896052e-06,
      "loss": 0.7818,
      "step": 4033
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4494748349764046,
      "learning_rate": 4.978232401569329e-06,
      "loss": 0.8059,
      "step": 4034
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5053848957868927,
      "learning_rate": 4.97615931198471e-06,
      "loss": 0.8162,
      "step": 4035
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4330258325392582,
      "learning_rate": 4.974086226498581e-06,
      "loss": 0.6851,
      "step": 4036
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3310048322576291,
      "learning_rate": 4.972013145467329e-06,
      "loss": 0.6935,
      "step": 4037
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2836028161073136,
      "learning_rate": 4.969940069247343e-06,
      "loss": 0.7359,
      "step": 4038
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4647966588477723,
      "learning_rate": 4.9678669981950076e-06,
      "loss": 0.799,
      "step": 4039
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3540229831123145,
      "learning_rate": 4.965793932666707e-06,
      "loss": 0.8306,
      "step": 4040
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2162580551434965,
      "learning_rate": 4.963720873018826e-06,
      "loss": 0.6887,
      "step": 4041
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4265145865822701,
      "learning_rate": 4.961647819607749e-06,
      "loss": 0.7799,
      "step": 4042
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.488186626292265,
      "learning_rate": 4.959574772789853e-06,
      "loss": 0.8794,
      "step": 4043
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.364015166425402,
      "learning_rate": 4.957501732921524e-06,
      "loss": 0.7998,
      "step": 4044
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.621669860660813,
      "learning_rate": 4.955428700359141e-06,
      "loss": 0.8114,
      "step": 4045
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.278227748071624,
      "learning_rate": 4.953355675459081e-06,
      "loss": 0.8253,
      "step": 4046
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4288966773367864,
      "learning_rate": 4.951282658577719e-06,
      "loss": 0.7654,
      "step": 4047
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4736685495342052,
      "learning_rate": 4.9492096500714346e-06,
      "loss": 0.8109,
      "step": 4048
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.465417386588883,
      "learning_rate": 4.947136650296601e-06,
      "loss": 0.7117,
      "step": 4049
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3024542037887055,
      "learning_rate": 4.945063659609588e-06,
      "loss": 0.8231,
      "step": 4050
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4572207130528638,
      "learning_rate": 4.94299067836677e-06,
      "loss": 0.8085,
      "step": 4051
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.240539968033293,
      "learning_rate": 4.940917706924513e-06,
      "loss": 0.8998,
      "step": 4052
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3868732923298672,
      "learning_rate": 4.93884474563919e-06,
      "loss": 0.7454,
      "step": 4053
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3724862032154252,
      "learning_rate": 4.936771794867163e-06,
      "loss": 0.7267,
      "step": 4054
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0338936197763853,
      "learning_rate": 4.9346988549647974e-06,
      "loss": 0.7964,
      "step": 4055
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3658255736370921,
      "learning_rate": 4.9326259262884565e-06,
      "loss": 0.8273,
      "step": 4056
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3613507601027766,
      "learning_rate": 4.9305530091945e-06,
      "loss": 0.7843,
      "step": 4057
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.554873263331186,
      "learning_rate": 4.928480104039285e-06,
      "loss": 0.8367,
      "step": 4058
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7934201592978966,
      "learning_rate": 4.92640721117917e-06,
      "loss": 0.6414,
      "step": 4059
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4752551856951104,
      "learning_rate": 4.924334330970509e-06,
      "loss": 0.7734,
      "step": 4060
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3466344831283599,
      "learning_rate": 4.922261463769652e-06,
      "loss": 0.8295,
      "step": 4061
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3039038147307802,
      "learning_rate": 4.920188609932951e-06,
      "loss": 0.7657,
      "step": 4062
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3471401606464264,
      "learning_rate": 4.918115769816754e-06,
      "loss": 0.7981,
      "step": 4063
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.294675270126517,
      "learning_rate": 4.9160429437774065e-06,
      "loss": 0.814,
      "step": 4064
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.345472124223205,
      "learning_rate": 4.913970132171248e-06,
      "loss": 0.8563,
      "step": 4065
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5903620805718535,
      "learning_rate": 4.911897335354624e-06,
      "loss": 0.7284,
      "step": 4066
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.328440400484838,
      "learning_rate": 4.909824553683868e-06,
      "loss": 0.8061,
      "step": 4067
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2901421229763133,
      "learning_rate": 4.907751787515316e-06,
      "loss": 0.7704,
      "step": 4068
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2265862842171205,
      "learning_rate": 4.905679037205302e-06,
      "loss": 0.7433,
      "step": 4069
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.313562550971135,
      "learning_rate": 4.903606303110153e-06,
      "loss": 0.7446,
      "step": 4070
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.266284096280134,
      "learning_rate": 4.9015335855862e-06,
      "loss": 0.6812,
      "step": 4071
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4886484833380844,
      "learning_rate": 4.899460884989764e-06,
      "loss": 0.7608,
      "step": 4072
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1552813335415382,
      "learning_rate": 4.897388201677169e-06,
      "loss": 0.744,
      "step": 4073
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2852114281473836,
      "learning_rate": 4.895315536004731e-06,
      "loss": 0.8433,
      "step": 4074
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3379546749962172,
      "learning_rate": 4.893242888328765e-06,
      "loss": 0.7662,
      "step": 4075
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7789889928830476,
      "learning_rate": 4.891170259005585e-06,
      "loss": 0.6238,
      "step": 4076
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0036775113606318,
      "learning_rate": 4.8890976483915e-06,
      "loss": 0.8655,
      "step": 4077
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6161555807900214,
      "learning_rate": 4.887025056842815e-06,
      "loss": 0.8217,
      "step": 4078
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4137260141002805,
      "learning_rate": 4.88495248471583e-06,
      "loss": 0.7378,
      "step": 4079
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3336711011139817,
      "learning_rate": 4.8828799323668465e-06,
      "loss": 0.7463,
      "step": 4080
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4003195327855227,
      "learning_rate": 4.880807400152162e-06,
      "loss": 0.7177,
      "step": 4081
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1576597878544959,
      "learning_rate": 4.878734888428068e-06,
      "loss": 0.7604,
      "step": 4082
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.369196925349239,
      "learning_rate": 4.876662397550851e-06,
      "loss": 0.9106,
      "step": 4083
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2731653604168374,
      "learning_rate": 4.874589927876799e-06,
      "loss": 0.814,
      "step": 4084
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2172552891158825,
      "learning_rate": 4.872517479762192e-06,
      "loss": 0.6609,
      "step": 4085
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3601809401847775,
      "learning_rate": 4.870445053563307e-06,
      "loss": 0.7633,
      "step": 4086
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3368938569282671,
      "learning_rate": 4.868372649636421e-06,
      "loss": 0.7229,
      "step": 4087
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2209552197729803,
      "learning_rate": 4.866300268337802e-06,
      "loss": 0.7454,
      "step": 4088
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3761631635563025,
      "learning_rate": 4.864227910023714e-06,
      "loss": 0.7826,
      "step": 4089
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3683629453486483,
      "learning_rate": 4.862155575050426e-06,
      "loss": 0.8093,
      "step": 4090
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3590401271268895,
      "learning_rate": 4.86008326377419e-06,
      "loss": 0.7828,
      "step": 4091
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.44252774616604,
      "learning_rate": 4.858010976551265e-06,
      "loss": 0.8231,
      "step": 4092
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8385752076240705,
      "learning_rate": 4.855938713737899e-06,
      "loss": 0.622,
      "step": 4093
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2832910337266301,
      "learning_rate": 4.853866475690338e-06,
      "loss": 0.7539,
      "step": 4094
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3374419199207819,
      "learning_rate": 4.851794262764824e-06,
      "loss": 0.8598,
      "step": 4095
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2832339784977222,
      "learning_rate": 4.849722075317594e-06,
      "loss": 0.7756,
      "step": 4096
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4649389409717475,
      "learning_rate": 4.847649913704881e-06,
      "loss": 0.6916,
      "step": 4097
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4125910595832218,
      "learning_rate": 4.845577778282913e-06,
      "loss": 0.7228,
      "step": 4098
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3259192517238068,
      "learning_rate": 4.843505669407917e-06,
      "loss": 0.7048,
      "step": 4099
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3544047976446596,
      "learning_rate": 4.841433587436111e-06,
      "loss": 0.7659,
      "step": 4100
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4452605217303127,
      "learning_rate": 4.839361532723708e-06,
      "loss": 0.8436,
      "step": 4101
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2316649048261545,
      "learning_rate": 4.837289505626921e-06,
      "loss": 0.6907,
      "step": 4102
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2818060617918956,
      "learning_rate": 4.835217506501955e-06,
      "loss": 0.8031,
      "step": 4103
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6205221876160276,
      "learning_rate": 4.8331455357050084e-06,
      "loss": 0.7745,
      "step": 4104
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5191085817652445,
      "learning_rate": 4.831073593592279e-06,
      "loss": 0.6845,
      "step": 4105
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.795772539685145,
      "learning_rate": 4.829001680519958e-06,
      "loss": 0.7261,
      "step": 4106
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8341844983927582,
      "learning_rate": 4.826929796844228e-06,
      "loss": 0.8112,
      "step": 4107
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3440131593969968,
      "learning_rate": 4.824857942921274e-06,
      "loss": 0.7086,
      "step": 4108
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5590490549114915,
      "learning_rate": 4.822786119107271e-06,
      "loss": 0.8053,
      "step": 4109
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3981239429004242,
      "learning_rate": 4.820714325758388e-06,
      "loss": 0.7733,
      "step": 4110
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2015190540182594,
      "learning_rate": 4.818642563230791e-06,
      "loss": 0.7908,
      "step": 4111
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2627812577310389,
      "learning_rate": 4.8165708318806385e-06,
      "loss": 0.7973,
      "step": 4112
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3802579881516592,
      "learning_rate": 4.814499132064088e-06,
      "loss": 0.8278,
      "step": 4113
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.436195823268203,
      "learning_rate": 4.8124274641372875e-06,
      "loss": 0.8773,
      "step": 4114
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3207911369153809,
      "learning_rate": 4.810355828456379e-06,
      "loss": 0.728,
      "step": 4115
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3897238739006577,
      "learning_rate": 4.808284225377503e-06,
      "loss": 0.8371,
      "step": 4116
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.273075133944343,
      "learning_rate": 4.8062126552567896e-06,
      "loss": 0.7556,
      "step": 4117
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6571475545043015,
      "learning_rate": 4.804141118450369e-06,
      "loss": 0.7939,
      "step": 4118
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.368086743381335,
      "learning_rate": 4.80206961531436e-06,
      "loss": 0.6831,
      "step": 4119
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3227493587182078,
      "learning_rate": 4.799998146204879e-06,
      "loss": 0.7446,
      "step": 4120
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2810912962299135,
      "learning_rate": 4.797926711478037e-06,
      "loss": 0.7771,
      "step": 4121
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9021555270145285,
      "learning_rate": 4.7958553114899335e-06,
      "loss": 0.6389,
      "step": 4122
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4482903155048148,
      "learning_rate": 4.793783946596669e-06,
      "loss": 0.6684,
      "step": 4123
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.376796005962722,
      "learning_rate": 4.791712617154336e-06,
      "loss": 0.7505,
      "step": 4124
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4038753772200692,
      "learning_rate": 4.789641323519018e-06,
      "loss": 0.8069,
      "step": 4125
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4510544099770162,
      "learning_rate": 4.787570066046794e-06,
      "loss": 0.8583,
      "step": 4126
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6081546643788969,
      "learning_rate": 4.785498845093739e-06,
      "loss": 0.79,
      "step": 4127
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2217158178025316,
      "learning_rate": 4.78342766101592e-06,
      "loss": 0.6835,
      "step": 4128
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6108803751750171,
      "learning_rate": 4.7813565141693965e-06,
      "loss": 0.84,
      "step": 4129
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7146770591409002,
      "learning_rate": 4.7792854049102234e-06,
      "loss": 0.7983,
      "step": 4130
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3094620335156701,
      "learning_rate": 4.7772143335944486e-06,
      "loss": 0.8093,
      "step": 4131
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.493570931118132,
      "learning_rate": 4.775143300578112e-06,
      "loss": 0.8482,
      "step": 4132
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8120832749983746,
      "learning_rate": 4.773072306217247e-06,
      "loss": 0.5985,
      "step": 4133
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1559589629709164,
      "learning_rate": 4.771001350867886e-06,
      "loss": 0.8048,
      "step": 4134
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3485359987615395,
      "learning_rate": 4.768930434886044e-06,
      "loss": 0.7901,
      "step": 4135
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8764581116942768,
      "learning_rate": 4.766859558627741e-06,
      "loss": 0.6951,
      "step": 4136
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3558241339784425,
      "learning_rate": 4.7647887224489834e-06,
      "loss": 0.8566,
      "step": 4137
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2870816583498124,
      "learning_rate": 4.762717926705771e-06,
      "loss": 0.7038,
      "step": 4138
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4783096981529689,
      "learning_rate": 4.760647171754098e-06,
      "loss": 0.7141,
      "step": 4139
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.392590866092221,
      "learning_rate": 4.75857645794995e-06,
      "loss": 0.7441,
      "step": 4140
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4605212064019812,
      "learning_rate": 4.756505785649309e-06,
      "loss": 0.7107,
      "step": 4141
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3508055351231985,
      "learning_rate": 4.754435155208146e-06,
      "loss": 0.7165,
      "step": 4142
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4566899279960952,
      "learning_rate": 4.752364566982425e-06,
      "loss": 0.8369,
      "step": 4143
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2350724365509855,
      "learning_rate": 4.750294021328105e-06,
      "loss": 0.7865,
      "step": 4144
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3188981891337626,
      "learning_rate": 4.748223518601141e-06,
      "loss": 0.7097,
      "step": 4145
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3878746877152028,
      "learning_rate": 4.7461530591574715e-06,
      "loss": 0.8115,
      "step": 4146
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.24124561849257,
      "learning_rate": 4.744082643353034e-06,
      "loss": 0.8297,
      "step": 4147
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5469310093359605,
      "learning_rate": 4.742012271543757e-06,
      "loss": 0.7387,
      "step": 4148
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4392751212946813,
      "learning_rate": 4.739941944085561e-06,
      "loss": 0.6919,
      "step": 4149
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8729258651696843,
      "learning_rate": 4.73787166133436e-06,
      "loss": 0.636,
      "step": 4150
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7264716189876234,
      "learning_rate": 4.735801423646059e-06,
      "loss": 0.6903,
      "step": 4151
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3610952239807859,
      "learning_rate": 4.733731231376557e-06,
      "loss": 0.8264,
      "step": 4152
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4116173388815292,
      "learning_rate": 4.731661084881744e-06,
      "loss": 0.7986,
      "step": 4153
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9058163372606826,
      "learning_rate": 4.729590984517498e-06,
      "loss": 0.6898,
      "step": 4154
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3294966105194963,
      "learning_rate": 4.727520930639698e-06,
      "loss": 0.608,
      "step": 4155
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3518303758688892,
      "learning_rate": 4.7254509236042105e-06,
      "loss": 0.7001,
      "step": 4156
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.497725846115204,
      "learning_rate": 4.723380963766891e-06,
      "loss": 0.7478,
      "step": 4157
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.457117039917535,
      "learning_rate": 4.72131105148359e-06,
      "loss": 0.7647,
      "step": 4158
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.192728180610705,
      "learning_rate": 4.719241187110152e-06,
      "loss": 0.7001,
      "step": 4159
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3261617242382697,
      "learning_rate": 4.717171371002407e-06,
      "loss": 0.8075,
      "step": 4160
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2276464154688296,
      "learning_rate": 4.715101603516183e-06,
      "loss": 0.8088,
      "step": 4161
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6215630590864152,
      "learning_rate": 4.713031885007296e-06,
      "loss": 0.8009,
      "step": 4162
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3667736638994845,
      "learning_rate": 4.710962215831553e-06,
      "loss": 0.7435,
      "step": 4163
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3053745263194707,
      "learning_rate": 4.708892596344758e-06,
      "loss": 0.728,
      "step": 4164
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3915460710524377,
      "learning_rate": 4.706823026902699e-06,
      "loss": 0.7445,
      "step": 4165
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.453944222824724,
      "learning_rate": 4.7047535078611626e-06,
      "loss": 0.8234,
      "step": 4166
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3112186702707418,
      "learning_rate": 4.70268403957592e-06,
      "loss": 0.7717,
      "step": 4167
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.963665385888634,
      "learning_rate": 4.700614622402737e-06,
      "loss": 0.8257,
      "step": 4168
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3844509545801815,
      "learning_rate": 4.698545256697373e-06,
      "loss": 0.7305,
      "step": 4169
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4339606002172107,
      "learning_rate": 4.696475942815573e-06,
      "loss": 0.7457,
      "step": 4170
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3784446270269786,
      "learning_rate": 4.6944066811130775e-06,
      "loss": 0.7193,
      "step": 4171
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3361916653966388,
      "learning_rate": 4.692337471945614e-06,
      "loss": 0.8001,
      "step": 4172
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.353651857876615,
      "learning_rate": 4.690268315668908e-06,
      "loss": 0.7256,
      "step": 4173
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7692574880661017,
      "learning_rate": 4.688199212638669e-06,
      "loss": 0.6437,
      "step": 4174
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.549238442635454,
      "learning_rate": 4.6861301632106005e-06,
      "loss": 0.7054,
      "step": 4175
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.038302650822431,
      "learning_rate": 4.684061167740395e-06,
      "loss": 0.8577,
      "step": 4176
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2067652353564964,
      "learning_rate": 4.6819922265837385e-06,
      "loss": 0.7313,
      "step": 4177
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5263775163936095,
      "learning_rate": 4.679923340096304e-06,
      "loss": 0.7456,
      "step": 4178
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3958159061017512,
      "learning_rate": 4.677854508633757e-06,
      "loss": 0.7131,
      "step": 4179
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9077701269375851,
      "learning_rate": 4.675785732551756e-06,
      "loss": 0.6171,
      "step": 4180
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3597581019027059,
      "learning_rate": 4.673717012205945e-06,
      "loss": 0.8152,
      "step": 4181
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4952229283179785,
      "learning_rate": 4.671648347951963e-06,
      "loss": 0.8082,
      "step": 4182
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4670915199030825,
      "learning_rate": 4.669579740145436e-06,
      "loss": 0.7361,
      "step": 4183
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3541715709135915,
      "learning_rate": 4.6675111891419835e-06,
      "loss": 0.7231,
      "step": 4184
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2495076280395137,
      "learning_rate": 4.665442695297212e-06,
      "loss": 0.8043,
      "step": 4185
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2435612337237267,
      "learning_rate": 4.663374258966718e-06,
      "loss": 0.8536,
      "step": 4186
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3458665568141341,
      "learning_rate": 4.661305880506094e-06,
      "loss": 0.8033,
      "step": 4187
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2045899409142258,
      "learning_rate": 4.659237560270914e-06,
      "loss": 0.7581,
      "step": 4188
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2971633685644814,
      "learning_rate": 4.657169298616748e-06,
      "loss": 0.7969,
      "step": 4189
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3205349058592575,
      "learning_rate": 4.655101095899154e-06,
      "loss": 0.7879,
      "step": 4190
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3758175127648717,
      "learning_rate": 4.653032952473678e-06,
      "loss": 0.7586,
      "step": 4191
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4164660465537517,
      "learning_rate": 4.650964868695863e-06,
      "loss": 0.8113,
      "step": 4192
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3179044058372218,
      "learning_rate": 4.648896844921233e-06,
      "loss": 0.7414,
      "step": 4193
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8770882310087377,
      "learning_rate": 4.6468288815053045e-06,
      "loss": 0.6429,
      "step": 4194
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.455465629317045,
      "learning_rate": 4.644760978803587e-06,
      "loss": 0.8023,
      "step": 4195
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4596500242740347,
      "learning_rate": 4.642693137171575e-06,
      "loss": 0.7049,
      "step": 4196
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2929954708119582,
      "learning_rate": 4.640625356964753e-06,
      "loss": 0.8078,
      "step": 4197
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.414391489913778,
      "learning_rate": 4.638557638538601e-06,
      "loss": 0.8461,
      "step": 4198
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.366689981516284,
      "learning_rate": 4.63648998224858e-06,
      "loss": 0.7647,
      "step": 4199
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8555484316633547,
      "learning_rate": 4.634422388450144e-06,
      "loss": 0.6752,
      "step": 4200
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2594749441579354,
      "learning_rate": 4.632354857498738e-06,
      "loss": 0.8055,
      "step": 4201
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3264462513980535,
      "learning_rate": 4.6302873897497955e-06,
      "loss": 0.8059,
      "step": 4202
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2740074178421092,
      "learning_rate": 4.628219985558737e-06,
      "loss": 0.7828,
      "step": 4203
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8203939057693425,
      "learning_rate": 4.626152645280972e-06,
      "loss": 0.7465,
      "step": 4204
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3745407885245764,
      "learning_rate": 4.624085369271902e-06,
      "loss": 0.7357,
      "step": 4205
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5303357497159928,
      "learning_rate": 4.622018157886915e-06,
      "loss": 0.836,
      "step": 4206
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.302293249756508,
      "learning_rate": 4.6199510114813875e-06,
      "loss": 0.7771,
      "step": 4207
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6357116121021475,
      "learning_rate": 4.617883930410688e-06,
      "loss": 0.7792,
      "step": 4208
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8618935375155768,
      "learning_rate": 4.615816915030169e-06,
      "loss": 0.663,
      "step": 4209
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3182839253421816,
      "learning_rate": 4.613749965695178e-06,
      "loss": 0.7138,
      "step": 4210
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.298874092501736,
      "learning_rate": 4.611683082761046e-06,
      "loss": 0.7465,
      "step": 4211
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3953683561814592,
      "learning_rate": 4.609616266583093e-06,
      "loss": 0.7521,
      "step": 4212
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8202798772216888,
      "learning_rate": 4.607549517516629e-06,
      "loss": 0.7036,
      "step": 4213
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3878429835675885,
      "learning_rate": 4.605482835916954e-06,
      "loss": 0.7435,
      "step": 4214
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4790008692919012,
      "learning_rate": 4.603416222139352e-06,
      "loss": 0.705,
      "step": 4215
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.319173669588778,
      "learning_rate": 4.6013496765391e-06,
      "loss": 0.8653,
      "step": 4216
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5827813652640879,
      "learning_rate": 4.59928319947146e-06,
      "loss": 0.7295,
      "step": 4217
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8432916635223332,
      "learning_rate": 4.597216791291681e-06,
      "loss": 0.6469,
      "step": 4218
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6276580460839805,
      "learning_rate": 4.595150452355006e-06,
      "loss": 0.7135,
      "step": 4219
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2821280261330354,
      "learning_rate": 4.593084183016664e-06,
      "loss": 0.7123,
      "step": 4220
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7134268581457566,
      "learning_rate": 4.5910179836318665e-06,
      "loss": 0.6912,
      "step": 4221
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6388343588844152,
      "learning_rate": 4.58895185455582e-06,
      "loss": 0.7775,
      "step": 4222
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8405109324753055,
      "learning_rate": 4.586885796143715e-06,
      "loss": 0.6239,
      "step": 4223
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4763108843947124,
      "learning_rate": 4.58481980875073e-06,
      "loss": 0.637,
      "step": 4224
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4886666042487149,
      "learning_rate": 4.582753892732033e-06,
      "loss": 0.75,
      "step": 4225
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3870917521229758,
      "learning_rate": 4.58068804844278e-06,
      "loss": 0.7613,
      "step": 4226
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.406105162094162,
      "learning_rate": 4.578622276238112e-06,
      "loss": 0.8498,
      "step": 4227
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.271358173566851,
      "learning_rate": 4.576556576473158e-06,
      "loss": 0.7156,
      "step": 4228
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3275917244127067,
      "learning_rate": 4.57449094950304e-06,
      "loss": 0.737,
      "step": 4229
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.517570627912461,
      "learning_rate": 4.572425395682859e-06,
      "loss": 0.7954,
      "step": 4230
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.328903217286165,
      "learning_rate": 4.570359915367711e-06,
      "loss": 0.7719,
      "step": 4231
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5742441103561922,
      "learning_rate": 4.568294508912674e-06,
      "loss": 0.6693,
      "step": 4232
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7576351178298125,
      "learning_rate": 4.566229176672815e-06,
      "loss": 0.5986,
      "step": 4233
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2709401962794395,
      "learning_rate": 4.5641639190031894e-06,
      "loss": 0.7688,
      "step": 4234
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3333276800044886,
      "learning_rate": 4.56209873625884e-06,
      "loss": 0.7745,
      "step": 4235
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4247630251599699,
      "learning_rate": 4.560033628794792e-06,
      "loss": 0.742,
      "step": 4236
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.647765515993595,
      "learning_rate": 4.557968596966063e-06,
      "loss": 0.7669,
      "step": 4237
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2370439358236978,
      "learning_rate": 4.5559036411276596e-06,
      "loss": 0.7907,
      "step": 4238
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3932233657660287,
      "learning_rate": 4.553838761634569e-06,
      "loss": 0.7158,
      "step": 4239
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2607835524466167,
      "learning_rate": 4.551773958841765e-06,
      "loss": 0.6977,
      "step": 4240
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5667584111023183,
      "learning_rate": 4.549709233104216e-06,
      "loss": 0.7976,
      "step": 4241
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3301921631194282,
      "learning_rate": 4.54764458477687e-06,
      "loss": 0.8943,
      "step": 4242
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.46650194942176,
      "learning_rate": 4.5455800142146626e-06,
      "loss": 0.7344,
      "step": 4243
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8862001129884608,
      "learning_rate": 4.54351552177252e-06,
      "loss": 0.7097,
      "step": 4244
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4324269763481932,
      "learning_rate": 4.541451107805351e-06,
      "loss": 0.7833,
      "step": 4245
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8089370224886893,
      "learning_rate": 4.53938677266805e-06,
      "loss": 0.5916,
      "step": 4246
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5267804043524635,
      "learning_rate": 4.537322516715505e-06,
      "loss": 0.841,
      "step": 4247
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3974567230950627,
      "learning_rate": 4.535258340302583e-06,
      "loss": 0.7894,
      "step": 4248
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2558922556224126,
      "learning_rate": 4.53319424378414e-06,
      "loss": 0.8113,
      "step": 4249
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.218871594658538,
      "learning_rate": 4.531130227515019e-06,
      "loss": 0.7401,
      "step": 4250
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.247656660421407,
      "learning_rate": 4.529066291850047e-06,
      "loss": 0.6573,
      "step": 4251
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8902450787518549,
      "learning_rate": 4.527002437144039e-06,
      "loss": 0.683,
      "step": 4252
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4406227145199029,
      "learning_rate": 4.524938663751796e-06,
      "loss": 0.7395,
      "step": 4253
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.282537325632134,
      "learning_rate": 4.522874972028103e-06,
      "loss": 0.7206,
      "step": 4254
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4365760379785846,
      "learning_rate": 4.520811362327732e-06,
      "loss": 0.735,
      "step": 4255
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.272062482967628,
      "learning_rate": 4.518747835005445e-06,
      "loss": 0.7232,
      "step": 4256
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7026078497368262,
      "learning_rate": 4.516684390415985e-06,
      "loss": 0.5862,
      "step": 4257
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7354059034019103,
      "learning_rate": 4.51462102891408e-06,
      "loss": 0.8204,
      "step": 4258
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2989881137343735,
      "learning_rate": 4.512557750854448e-06,
      "loss": 0.8393,
      "step": 4259
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4244935000228969,
      "learning_rate": 4.51049455659179e-06,
      "loss": 0.845,
      "step": 4260
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.513254496668033,
      "learning_rate": 4.508431446480792e-06,
      "loss": 0.7796,
      "step": 4261
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5915537019570656,
      "learning_rate": 4.506368420876127e-06,
      "loss": 0.8876,
      "step": 4262
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2044763916670447,
      "learning_rate": 4.504305480132454e-06,
      "loss": 0.7625,
      "step": 4263
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3118031519215763,
      "learning_rate": 4.502242624604413e-06,
      "loss": 0.7327,
      "step": 4264
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4060772609899252,
      "learning_rate": 4.5001798546466365e-06,
      "loss": 0.746,
      "step": 4265
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8593419329502887,
      "learning_rate": 4.498117170613738e-06,
      "loss": 0.6872,
      "step": 4266
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5334952599890304,
      "learning_rate": 4.496054572860317e-06,
      "loss": 0.7634,
      "step": 4267
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.545726425607284,
      "learning_rate": 4.493992061740956e-06,
      "loss": 0.816,
      "step": 4268
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4531914675307858,
      "learning_rate": 4.491929637610225e-06,
      "loss": 0.8268,
      "step": 4269
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7660177216991653,
      "learning_rate": 4.489867300822681e-06,
      "loss": 0.9204,
      "step": 4270
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.274416401390117,
      "learning_rate": 4.4878050517328625e-06,
      "loss": 0.7811,
      "step": 4271
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8415209572840088,
      "learning_rate": 4.485742890695292e-06,
      "loss": 0.6705,
      "step": 4272
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.472114675084001,
      "learning_rate": 4.483680818064481e-06,
      "loss": 0.7094,
      "step": 4273
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3061340645168564,
      "learning_rate": 4.481618834194921e-06,
      "loss": 0.7254,
      "step": 4274
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3183143199319731,
      "learning_rate": 4.479556939441095e-06,
      "loss": 0.6712,
      "step": 4275
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.161664319586978,
      "learning_rate": 4.477495134157464e-06,
      "loss": 0.8953,
      "step": 4276
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2566778191254766,
      "learning_rate": 4.475433418698477e-06,
      "loss": 0.8174,
      "step": 4277
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4116883319800924,
      "learning_rate": 4.473371793418567e-06,
      "loss": 0.8396,
      "step": 4278
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.461753915092708,
      "learning_rate": 4.471310258672151e-06,
      "loss": 0.771,
      "step": 4279
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8001986606688861,
      "learning_rate": 4.469248814813631e-06,
      "loss": 0.6442,
      "step": 4280
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8040896980564598,
      "learning_rate": 4.467187462197392e-06,
      "loss": 0.7044,
      "step": 4281
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3086854534539263,
      "learning_rate": 4.465126201177804e-06,
      "loss": 0.7193,
      "step": 4282
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.412164259283305,
      "learning_rate": 4.463065032109224e-06,
      "loss": 0.7915,
      "step": 4283
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.786984361306726,
      "learning_rate": 4.461003955345989e-06,
      "loss": 0.7498,
      "step": 4284
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2738596739454204,
      "learning_rate": 4.458942971242423e-06,
      "loss": 0.6158,
      "step": 4285
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.353136605273919,
      "learning_rate": 4.456882080152834e-06,
      "loss": 0.92,
      "step": 4286
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.236194268335971,
      "learning_rate": 4.454821282431511e-06,
      "loss": 0.7635,
      "step": 4287
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5192822422415266,
      "learning_rate": 4.4527605784327295e-06,
      "loss": 0.7414,
      "step": 4288
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4211685465435269,
      "learning_rate": 4.450699968510749e-06,
      "loss": 0.7661,
      "step": 4289
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4514015597268362,
      "learning_rate": 4.448639453019812e-06,
      "loss": 0.7424,
      "step": 4290
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8184947633202654,
      "learning_rate": 4.446579032314145e-06,
      "loss": 0.5971,
      "step": 4291
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5150231351054748,
      "learning_rate": 4.444518706747956e-06,
      "loss": 0.7721,
      "step": 4292
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2953106106229122,
      "learning_rate": 4.442458476675443e-06,
      "loss": 0.749,
      "step": 4293
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2143944759391194,
      "learning_rate": 4.440398342450782e-06,
      "loss": 0.7022,
      "step": 4294
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5040175379415215,
      "learning_rate": 4.438338304428132e-06,
      "loss": 0.771,
      "step": 4295
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.264305347061509,
      "learning_rate": 4.43627836296164e-06,
      "loss": 0.8224,
      "step": 4296
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.701047606778494,
      "learning_rate": 4.434218518405432e-06,
      "loss": 0.7881,
      "step": 4297
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.89109464824972,
      "learning_rate": 4.43215877111362e-06,
      "loss": 0.8286,
      "step": 4298
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4061875909519832,
      "learning_rate": 4.4300991214402986e-06,
      "loss": 0.8434,
      "step": 4299
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3493961001231163,
      "learning_rate": 4.428039569739544e-06,
      "loss": 0.6999,
      "step": 4300
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4038348225862087,
      "learning_rate": 4.425980116365419e-06,
      "loss": 0.7288,
      "step": 4301
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.507135784272756,
      "learning_rate": 4.423920761671965e-06,
      "loss": 0.7784,
      "step": 4302
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.263389999822412,
      "learning_rate": 4.421861506013213e-06,
      "loss": 0.7913,
      "step": 4303
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.804911283563948,
      "learning_rate": 4.419802349743169e-06,
      "loss": 0.7533,
      "step": 4304
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3715792699496296,
      "learning_rate": 4.417743293215827e-06,
      "loss": 0.8442,
      "step": 4305
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4832622037366012,
      "learning_rate": 4.415684336785165e-06,
      "loss": 0.7705,
      "step": 4306
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.468334754767491,
      "learning_rate": 4.413625480805138e-06,
      "loss": 0.7958,
      "step": 4307
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6816341294094526,
      "learning_rate": 4.411566725629688e-06,
      "loss": 0.8512,
      "step": 4308
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3135035413527514,
      "learning_rate": 4.409508071612741e-06,
      "loss": 0.7314,
      "step": 4309
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3972289973045284,
      "learning_rate": 4.407449519108203e-06,
      "loss": 0.7463,
      "step": 4310
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5107174365282965,
      "learning_rate": 4.405391068469958e-06,
      "loss": 0.6447,
      "step": 4311
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4606596137114418,
      "learning_rate": 4.403332720051884e-06,
      "loss": 0.8007,
      "step": 4312
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9254703055426051,
      "learning_rate": 4.401274474207833e-06,
      "loss": 0.6482,
      "step": 4313
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.458267500856942,
      "learning_rate": 4.3992163312916404e-06,
      "loss": 0.8461,
      "step": 4314
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.188797501327789,
      "learning_rate": 4.397158291657125e-06,
      "loss": 0.8451,
      "step": 4315
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2640691329865024,
      "learning_rate": 4.39510035565809e-06,
      "loss": 0.8153,
      "step": 4316
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2536276226163237,
      "learning_rate": 4.393042523648315e-06,
      "loss": 0.7136,
      "step": 4317
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3328949775106629,
      "learning_rate": 4.390984795981566e-06,
      "loss": 0.7999,
      "step": 4318
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.349487703482436,
      "learning_rate": 4.388927173011592e-06,
      "loss": 0.7671,
      "step": 4319
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.28715366829752,
      "learning_rate": 4.38686965509212e-06,
      "loss": 0.7587,
      "step": 4320
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3585268121391931,
      "learning_rate": 4.384812242576863e-06,
      "loss": 0.8778,
      "step": 4321
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8568886681835517,
      "learning_rate": 4.382754935819514e-06,
      "loss": 0.5949,
      "step": 4322
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2617531607955037,
      "learning_rate": 4.380697735173745e-06,
      "loss": 0.7045,
      "step": 4323
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8290361883030632,
      "learning_rate": 4.378640640993218e-06,
      "loss": 0.7234,
      "step": 4324
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.121627604382306,
      "learning_rate": 4.376583653631567e-06,
      "loss": 0.8234,
      "step": 4325
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3815913900755326,
      "learning_rate": 4.374526773442413e-06,
      "loss": 0.7413,
      "step": 4326
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.9157639968865805,
      "learning_rate": 4.372470000779357e-06,
      "loss": 0.6856,
      "step": 4327
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2386117364375717,
      "learning_rate": 4.370413335995985e-06,
      "loss": 0.7178,
      "step": 4328
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2266362874096686,
      "learning_rate": 4.368356779445856e-06,
      "loss": 0.7707,
      "step": 4329
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3284832113242315,
      "learning_rate": 4.36630033148252e-06,
      "loss": 0.751,
      "step": 4330
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5152656382402128,
      "learning_rate": 4.364243992459506e-06,
      "loss": 0.8243,
      "step": 4331
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6847549754261206,
      "learning_rate": 4.362187762730319e-06,
      "loss": 0.8023,
      "step": 4332
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4705770567150043,
      "learning_rate": 4.360131642648449e-06,
      "loss": 0.7286,
      "step": 4333
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7283888222228994,
      "learning_rate": 4.358075632567368e-06,
      "loss": 0.6827,
      "step": 4334
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.260284464231823,
      "learning_rate": 4.356019732840528e-06,
      "loss": 0.7356,
      "step": 4335
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.432442966208775,
      "learning_rate": 4.3539639438213606e-06,
      "loss": 0.7361,
      "step": 4336
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5692055099407614,
      "learning_rate": 4.351908265863282e-06,
      "loss": 0.772,
      "step": 4337
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2571828355164625,
      "learning_rate": 4.349852699319686e-06,
      "loss": 0.7771,
      "step": 4338
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.674258980674686,
      "learning_rate": 4.347797244543945e-06,
      "loss": 0.7477,
      "step": 4339
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3758231745590752,
      "learning_rate": 4.34574190188942e-06,
      "loss": 0.7295,
      "step": 4340
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4279968763140811,
      "learning_rate": 4.343686671709449e-06,
      "loss": 0.8118,
      "step": 4341
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4945392275430505,
      "learning_rate": 4.341631554357347e-06,
      "loss": 0.8002,
      "step": 4342
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.489833647806821,
      "learning_rate": 4.339576550186413e-06,
      "loss": 0.8028,
      "step": 4343
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.915639471463154,
      "learning_rate": 4.3375216595499254e-06,
      "loss": 0.7682,
      "step": 4344
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3244347489207047,
      "learning_rate": 4.335466882801146e-06,
      "loss": 0.7066,
      "step": 4345
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.403913401733511,
      "learning_rate": 4.333412220293313e-06,
      "loss": 0.879,
      "step": 4346
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5232043651461935,
      "learning_rate": 4.3313576723796464e-06,
      "loss": 0.7485,
      "step": 4347
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3544282116877688,
      "learning_rate": 4.329303239413346e-06,
      "loss": 0.7705,
      "step": 4348
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7521544461673013,
      "learning_rate": 4.327248921747597e-06,
      "loss": 0.6642,
      "step": 4349
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.398545200404232,
      "learning_rate": 4.325194719735557e-06,
      "loss": 0.7453,
      "step": 4350
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5649632139051273,
      "learning_rate": 4.3231406337303665e-06,
      "loss": 0.8188,
      "step": 4351
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1461798147321347,
      "learning_rate": 4.321086664085149e-06,
      "loss": 0.6885,
      "step": 4352
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3431840065895733,
      "learning_rate": 4.319032811153005e-06,
      "loss": 0.7822,
      "step": 4353
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.315191694048631,
      "learning_rate": 4.316979075287014e-06,
      "loss": 0.7506,
      "step": 4354
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3361214452291326,
      "learning_rate": 4.314925456840239e-06,
      "loss": 0.7374,
      "step": 4355
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.836417643878514,
      "learning_rate": 4.3128719561657205e-06,
      "loss": 0.684,
      "step": 4356
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.415072794252747,
      "learning_rate": 4.310818573616476e-06,
      "loss": 0.7843,
      "step": 4357
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7196102829098636,
      "learning_rate": 4.30876530954551e-06,
      "loss": 0.612,
      "step": 4358
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5547653466087663,
      "learning_rate": 4.3067121643058e-06,
      "loss": 0.8113,
      "step": 4359
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3618328656018452,
      "learning_rate": 4.304659138250309e-06,
      "loss": 0.8221,
      "step": 4360
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4828721368664637,
      "learning_rate": 4.302606231731971e-06,
      "loss": 0.8793,
      "step": 4361
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2430068405275223,
      "learning_rate": 4.300553445103707e-06,
      "loss": 0.6843,
      "step": 4362
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3615061772717274,
      "learning_rate": 4.298500778718415e-06,
      "loss": 0.8311,
      "step": 4363
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8597577932551138,
      "learning_rate": 4.296448232928971e-06,
      "loss": 0.6757,
      "step": 4364
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1299903824101407,
      "learning_rate": 4.294395808088232e-06,
      "loss": 0.7473,
      "step": 4365
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.692524587556374,
      "learning_rate": 4.292343504549032e-06,
      "loss": 0.783,
      "step": 4366
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4969069510031954,
      "learning_rate": 4.29029132266419e-06,
      "loss": 0.7027,
      "step": 4367
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.317897723206703,
      "learning_rate": 4.288239262786497e-06,
      "loss": 0.7847,
      "step": 4368
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5478420579184256,
      "learning_rate": 4.286187325268726e-06,
      "loss": 0.7047,
      "step": 4369
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.574808050939673,
      "learning_rate": 4.284135510463628e-06,
      "loss": 0.7734,
      "step": 4370
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.332248127416129,
      "learning_rate": 4.282083818723937e-06,
      "loss": 0.7636,
      "step": 4371
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5149639372625434,
      "learning_rate": 4.280032250402358e-06,
      "loss": 0.813,
      "step": 4372
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7458243821610504,
      "learning_rate": 4.2779808058515825e-06,
      "loss": 0.8586,
      "step": 4373
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5216937380495383,
      "learning_rate": 4.275929485424278e-06,
      "loss": 0.7584,
      "step": 4374
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3302477926972234,
      "learning_rate": 4.2738782894730876e-06,
      "loss": 0.739,
      "step": 4375
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2643831503359035,
      "learning_rate": 4.271827218350636e-06,
      "loss": 0.7626,
      "step": 4376
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3792469507413532,
      "learning_rate": 4.269776272409529e-06,
      "loss": 0.755,
      "step": 4377
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.460955012201751,
      "learning_rate": 4.2677254520023465e-06,
      "loss": 0.7458,
      "step": 4378
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.372338778975131,
      "learning_rate": 4.265674757481647e-06,
      "loss": 0.803,
      "step": 4379
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8894608615195487,
      "learning_rate": 4.263624189199971e-06,
      "loss": 0.6289,
      "step": 4380
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4348033041160944,
      "learning_rate": 4.261573747509833e-06,
      "loss": 0.7607,
      "step": 4381
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.263754149515486,
      "learning_rate": 4.259523432763728e-06,
      "loss": 0.6445,
      "step": 4382
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.270118397584736,
      "learning_rate": 4.25747324531413e-06,
      "loss": 0.8908,
      "step": 4383
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3675997805120435,
      "learning_rate": 4.25542318551349e-06,
      "loss": 0.7488,
      "step": 4384
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2077310262110559,
      "learning_rate": 4.2533732537142335e-06,
      "loss": 0.71,
      "step": 4385
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2716796882249812,
      "learning_rate": 4.2513234502687725e-06,
      "loss": 0.8099,
      "step": 4386
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2370255620884305,
      "learning_rate": 4.249273775529489e-06,
      "loss": 0.7115,
      "step": 4387
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0792794315920315,
      "learning_rate": 4.247224229848747e-06,
      "loss": 0.7261,
      "step": 4388
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3067067116397093,
      "learning_rate": 4.245174813578887e-06,
      "loss": 0.7533,
      "step": 4389
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8610276782082064,
      "learning_rate": 4.243125527072227e-06,
      "loss": 0.6698,
      "step": 4390
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3793201106436082,
      "learning_rate": 4.241076370681064e-06,
      "loss": 0.7472,
      "step": 4391
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7478724198838155,
      "learning_rate": 4.239027344757671e-06,
      "loss": 0.6849,
      "step": 4392
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8390852830473162,
      "learning_rate": 4.2369784496542986e-06,
      "loss": 0.6429,
      "step": 4393
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5019586950942052,
      "learning_rate": 4.234929685723175e-06,
      "loss": 0.7623,
      "step": 4394
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5545566951426977,
      "learning_rate": 4.2328810533165095e-06,
      "loss": 0.7565,
      "step": 4395
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3111024199290613,
      "learning_rate": 4.230832552786485e-06,
      "loss": 0.6695,
      "step": 4396
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.813703546405357,
      "learning_rate": 4.2287841844852595e-06,
      "loss": 0.5909,
      "step": 4397
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.264130727946485,
      "learning_rate": 4.226735948764976e-06,
      "loss": 0.6575,
      "step": 4398
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3291699654700506,
      "learning_rate": 4.2246878459777465e-06,
      "loss": 0.706,
      "step": 4399
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6324699414904227,
      "learning_rate": 4.222639876475663e-06,
      "loss": 0.789,
      "step": 4400
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2236918257255167,
      "learning_rate": 4.220592040610798e-06,
      "loss": 0.7674,
      "step": 4401
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3621137432309856,
      "learning_rate": 4.218544338735197e-06,
      "loss": 0.6997,
      "step": 4402
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4134147210710208,
      "learning_rate": 4.216496771200881e-06,
      "loss": 0.7399,
      "step": 4403
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2895540113890762,
      "learning_rate": 4.214449338359856e-06,
      "loss": 0.8099,
      "step": 4404
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.340856328487067,
      "learning_rate": 4.2124020405640955e-06,
      "loss": 0.732,
      "step": 4405
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4712244085585384,
      "learning_rate": 4.2103548781655555e-06,
      "loss": 0.8399,
      "step": 4406
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2367905420905856,
      "learning_rate": 4.2083078515161664e-06,
      "loss": 0.7245,
      "step": 4407
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.6881765457684725,
      "learning_rate": 4.206260960967836e-06,
      "loss": 0.6427,
      "step": 4408
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9124694258777122,
      "learning_rate": 4.204214206872448e-06,
      "loss": 0.7337,
      "step": 4409
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4861753367265542,
      "learning_rate": 4.202167589581863e-06,
      "loss": 0.7627,
      "step": 4410
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8610183494895414,
      "learning_rate": 4.200121109447919e-06,
      "loss": 0.6929,
      "step": 4411
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.782290076984064,
      "learning_rate": 4.198074766822429e-06,
      "loss": 0.6852,
      "step": 4412
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.454375988067462,
      "learning_rate": 4.196028562057181e-06,
      "loss": 0.7686,
      "step": 4413
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3441213971970272,
      "learning_rate": 4.193982495503946e-06,
      "loss": 0.8324,
      "step": 4414
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6339709586376918,
      "learning_rate": 4.191936567514464e-06,
      "loss": 0.8497,
      "step": 4415
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4295792610259983,
      "learning_rate": 4.189890778440454e-06,
      "loss": 0.8073,
      "step": 4416
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2669471057958492,
      "learning_rate": 4.187845128633611e-06,
      "loss": 0.7455,
      "step": 4417
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2670078961532092,
      "learning_rate": 4.185799618445605e-06,
      "loss": 0.744,
      "step": 4418
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5739716753033564,
      "learning_rate": 4.1837542482280845e-06,
      "loss": 0.6889,
      "step": 4419
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5528641713855051,
      "learning_rate": 4.181709018332672e-06,
      "loss": 0.7569,
      "step": 4420
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2651685421152326,
      "learning_rate": 4.179663929110964e-06,
      "loss": 0.8359,
      "step": 4421
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2565549778474578,
      "learning_rate": 4.177618980914536e-06,
      "loss": 0.7876,
      "step": 4422
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.479777502261943,
      "learning_rate": 4.1755741740949405e-06,
      "loss": 0.7414,
      "step": 4423
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8266808050370132,
      "learning_rate": 4.173529509003702e-06,
      "loss": 0.5876,
      "step": 4424
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.375813932040293,
      "learning_rate": 4.171484985992323e-06,
      "loss": 0.8177,
      "step": 4425
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8362551355556507,
      "learning_rate": 4.169440605412278e-06,
      "loss": 0.708,
      "step": 4426
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.063897208154976,
      "learning_rate": 4.167396367615023e-06,
      "loss": 0.6661,
      "step": 4427
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7360298125047786,
      "learning_rate": 4.165352272951985e-06,
      "loss": 0.5852,
      "step": 4428
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4267510695329788,
      "learning_rate": 4.163308321774566e-06,
      "loss": 0.7366,
      "step": 4429
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4093176571941226,
      "learning_rate": 4.161264514434148e-06,
      "loss": 0.7962,
      "step": 4430
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3921400327278175,
      "learning_rate": 4.15922085128208e-06,
      "loss": 0.7457,
      "step": 4431
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8548468998916202,
      "learning_rate": 4.157177332669698e-06,
      "loss": 0.7331,
      "step": 4432
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5287741433100683,
      "learning_rate": 4.155133958948302e-06,
      "loss": 0.7462,
      "step": 4433
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3014757508043757,
      "learning_rate": 4.153090730469174e-06,
      "loss": 0.8113,
      "step": 4434
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.293699215790574,
      "learning_rate": 4.151047647583568e-06,
      "loss": 0.7929,
      "step": 4435
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3060867182036815,
      "learning_rate": 4.149004710642712e-06,
      "loss": 0.7251,
      "step": 4436
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3208938604418763,
      "learning_rate": 4.146961919997813e-06,
      "loss": 0.7384,
      "step": 4437
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.290914131706171,
      "learning_rate": 4.144919276000048e-06,
      "loss": 0.767,
      "step": 4438
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9759336529857342,
      "learning_rate": 4.142876779000573e-06,
      "loss": 0.7166,
      "step": 4439
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4442920746856345,
      "learning_rate": 4.1408344293505154e-06,
      "loss": 0.7329,
      "step": 4440
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4200152771356853,
      "learning_rate": 4.13879222740098e-06,
      "loss": 0.8284,
      "step": 4441
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.256250928380166,
      "learning_rate": 4.136750173503046e-06,
      "loss": 0.7379,
      "step": 4442
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4208963502007652,
      "learning_rate": 4.134708268007764e-06,
      "loss": 0.6739,
      "step": 4443
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2193242308606893,
      "learning_rate": 4.132666511266162e-06,
      "loss": 0.7057,
      "step": 4444
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3181587087529165,
      "learning_rate": 4.130624903629242e-06,
      "loss": 0.7646,
      "step": 4445
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3557217266898909,
      "learning_rate": 4.12858344544798e-06,
      "loss": 0.7357,
      "step": 4446
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2431015035210984,
      "learning_rate": 4.126542137073325e-06,
      "loss": 0.6988,
      "step": 4447
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.264823450575037,
      "learning_rate": 4.124500978856204e-06,
      "loss": 0.7469,
      "step": 4448
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8480607443102528,
      "learning_rate": 4.122459971147514e-06,
      "loss": 0.6602,
      "step": 4449
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8153496212489455,
      "learning_rate": 4.120419114298127e-06,
      "loss": 0.721,
      "step": 4450
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.271042882517742,
      "learning_rate": 4.118378408658891e-06,
      "loss": 0.7988,
      "step": 4451
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2986852816513859,
      "learning_rate": 4.1163378545806286e-06,
      "loss": 0.6405,
      "step": 4452
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.410462697930429,
      "learning_rate": 4.114297452414133e-06,
      "loss": 0.8148,
      "step": 4453
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4522327776169097,
      "learning_rate": 4.112257202510173e-06,
      "loss": 0.7369,
      "step": 4454
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.252629989817148,
      "learning_rate": 4.110217105219492e-06,
      "loss": 0.7447,
      "step": 4455
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8612486989833551,
      "learning_rate": 4.108177160892807e-06,
      "loss": 0.7376,
      "step": 4456
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8908719314564313,
      "learning_rate": 4.106137369880804e-06,
      "loss": 0.6246,
      "step": 4457
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2654089762411318,
      "learning_rate": 4.104097732534153e-06,
      "loss": 0.7813,
      "step": 4458
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4234687791710539,
      "learning_rate": 4.102058249203483e-06,
      "loss": 0.7554,
      "step": 4459
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3035404851801735,
      "learning_rate": 4.1000189202394144e-06,
      "loss": 0.8182,
      "step": 4460
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.377649794528158,
      "learning_rate": 4.097979745992526e-06,
      "loss": 0.7438,
      "step": 4461
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2953828697238423,
      "learning_rate": 4.095940726813375e-06,
      "loss": 0.7373,
      "step": 4462
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.24009972884522,
      "learning_rate": 4.093901863052495e-06,
      "loss": 0.7299,
      "step": 4463
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3447935336809584,
      "learning_rate": 4.091863155060389e-06,
      "loss": 0.6605,
      "step": 4464
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.301302695029351,
      "learning_rate": 4.0898246031875346e-06,
      "loss": 0.8504,
      "step": 4465
    },
    {
      "epoch": 0.57,
      "grad_norm": 15.97932770448937,
      "learning_rate": 4.087786207784383e-06,
      "loss": 0.8016,
      "step": 4466
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2761202249469552,
      "learning_rate": 4.085747969201357e-06,
      "loss": 0.8343,
      "step": 4467
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2249154087438416,
      "learning_rate": 4.083709887788852e-06,
      "loss": 0.834,
      "step": 4468
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0450230634337974,
      "learning_rate": 4.081671963897241e-06,
      "loss": 0.6849,
      "step": 4469
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4327875162841017,
      "learning_rate": 4.079634197876867e-06,
      "loss": 0.7796,
      "step": 4470
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.278699414368355,
      "learning_rate": 4.077596590078044e-06,
      "loss": 0.7787,
      "step": 4471
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4983074433948453,
      "learning_rate": 4.075559140851061e-06,
      "loss": 0.8001,
      "step": 4472
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3928461944289199,
      "learning_rate": 4.073521850546177e-06,
      "loss": 0.7511,
      "step": 4473
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.253699347731012,
      "learning_rate": 4.07148471951363e-06,
      "loss": 0.7485,
      "step": 4474
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3867089083330328,
      "learning_rate": 4.069447748103621e-06,
      "loss": 0.8415,
      "step": 4475
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.540982873870614,
      "learning_rate": 4.067410936666335e-06,
      "loss": 0.7291,
      "step": 4476
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4562454806901761,
      "learning_rate": 4.065374285551917e-06,
      "loss": 0.7749,
      "step": 4477
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4528145327167323,
      "learning_rate": 4.063337795110497e-06,
      "loss": 0.7805,
      "step": 4478
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5746801243354143,
      "learning_rate": 4.0613014656921675e-06,
      "loss": 0.7254,
      "step": 4479
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3056321222569507,
      "learning_rate": 4.059265297646999e-06,
      "loss": 0.7163,
      "step": 4480
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5128742319129893,
      "learning_rate": 4.057229291325032e-06,
      "loss": 0.8379,
      "step": 4481
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4885385499651251,
      "learning_rate": 4.05519344707628e-06,
      "loss": 0.7715,
      "step": 4482
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.363590683030181,
      "learning_rate": 4.0531577652507256e-06,
      "loss": 0.8019,
      "step": 4483
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2455295877156476,
      "learning_rate": 4.05112224619833e-06,
      "loss": 0.7795,
      "step": 4484
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2426781373094675,
      "learning_rate": 4.049086890269021e-06,
      "loss": 0.7658,
      "step": 4485
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2304029348160033,
      "learning_rate": 4.047051697812697e-06,
      "loss": 0.6863,
      "step": 4486
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6970277115729895,
      "learning_rate": 4.0450166691792335e-06,
      "loss": 0.7362,
      "step": 4487
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.532831080418981,
      "learning_rate": 4.042981804718478e-06,
      "loss": 0.7855,
      "step": 4488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9448850183007989,
      "learning_rate": 4.040947104780244e-06,
      "loss": 0.6298,
      "step": 4489
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3889899694323917,
      "learning_rate": 4.03891256971432e-06,
      "loss": 0.7378,
      "step": 4490
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3772333519785345,
      "learning_rate": 4.036878199870469e-06,
      "loss": 0.7002,
      "step": 4491
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3670115528030515,
      "learning_rate": 4.034843995598421e-06,
      "loss": 0.707,
      "step": 4492
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.399728878738431,
      "learning_rate": 4.032809957247878e-06,
      "loss": 0.7984,
      "step": 4493
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7731934370765279,
      "learning_rate": 4.030776085168516e-06,
      "loss": 0.8027,
      "step": 4494
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3143362660256581,
      "learning_rate": 4.028742379709982e-06,
      "loss": 0.7816,
      "step": 4495
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2932469361063887,
      "learning_rate": 4.0267088412218906e-06,
      "loss": 0.7897,
      "step": 4496
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8523204147834146,
      "learning_rate": 4.024675470053836e-06,
      "loss": 0.6935,
      "step": 4497
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2688716213974902,
      "learning_rate": 4.0226422665553724e-06,
      "loss": 0.8058,
      "step": 4498
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.240971768424025,
      "learning_rate": 4.020609231076035e-06,
      "loss": 0.7682,
      "step": 4499
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2899002729547566,
      "learning_rate": 4.018576363965324e-06,
      "loss": 0.7696,
      "step": 4500
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.287331108498253,
      "learning_rate": 4.0165436655727144e-06,
      "loss": 0.8183,
      "step": 4501
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5938016280120675,
      "learning_rate": 4.01451113624765e-06,
      "loss": 0.8318,
      "step": 4502
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4470626057367435,
      "learning_rate": 4.012478776339547e-06,
      "loss": 0.8202,
      "step": 4503
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8329242650948679,
      "learning_rate": 4.010446586197788e-06,
      "loss": 0.6863,
      "step": 4504
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4100095831504302,
      "learning_rate": 4.008414566171733e-06,
      "loss": 0.7711,
      "step": 4505
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3868289012521706,
      "learning_rate": 4.006382716610711e-06,
      "loss": 0.7939,
      "step": 4506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9485664572633652,
      "learning_rate": 4.00435103786402e-06,
      "loss": 0.6888,
      "step": 4507
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2946186512804898,
      "learning_rate": 4.002319530280925e-06,
      "loss": 0.8504,
      "step": 4508
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.368711526977622,
      "learning_rate": 4.000288194210671e-06,
      "loss": 0.7777,
      "step": 4509
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3255086235986657,
      "learning_rate": 3.998257030002466e-06,
      "loss": 0.6949,
      "step": 4510
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.359628045398099,
      "learning_rate": 3.996226038005491e-06,
      "loss": 0.9335,
      "step": 4511
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6955959291338338,
      "learning_rate": 3.994195218568896e-06,
      "loss": 0.7531,
      "step": 4512
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4251672389026897,
      "learning_rate": 3.992164572041805e-06,
      "loss": 0.7366,
      "step": 4513
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3002054346213834,
      "learning_rate": 3.990134098773304e-06,
      "loss": 0.7836,
      "step": 4514
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5080349472345829,
      "learning_rate": 3.988103799112461e-06,
      "loss": 0.7878,
      "step": 4515
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2262699788060807,
      "learning_rate": 3.986073673408306e-06,
      "loss": 0.7565,
      "step": 4516
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7946919425365722,
      "learning_rate": 3.984043722009842e-06,
      "loss": 0.8679,
      "step": 4517
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.495867600309886,
      "learning_rate": 3.98201394526604e-06,
      "loss": 0.6504,
      "step": 4518
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2742976503002612,
      "learning_rate": 3.97998434352584e-06,
      "loss": 0.7191,
      "step": 4519
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.327458984440817,
      "learning_rate": 3.977954917138158e-06,
      "loss": 0.7601,
      "step": 4520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8906667759159215,
      "learning_rate": 3.975925666451875e-06,
      "loss": 0.7059,
      "step": 4521
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4912791625015818,
      "learning_rate": 3.973896591815841e-06,
      "loss": 0.7654,
      "step": 4522
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3843131043355665,
      "learning_rate": 3.971867693578879e-06,
      "loss": 0.7148,
      "step": 4523
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.378382685505619,
      "learning_rate": 3.969838972089778e-06,
      "loss": 0.7177,
      "step": 4524
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3737512494480848,
      "learning_rate": 3.967810427697301e-06,
      "loss": 0.7782,
      "step": 4525
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3235947679995759,
      "learning_rate": 3.965782060750178e-06,
      "loss": 0.7243,
      "step": 4526
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.354153777158096,
      "learning_rate": 3.96375387159711e-06,
      "loss": 0.7214,
      "step": 4527
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2158848011100971,
      "learning_rate": 3.961725860586763e-06,
      "loss": 0.7103,
      "step": 4528
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4219729417360292,
      "learning_rate": 3.9596980280677775e-06,
      "loss": 0.7291,
      "step": 4529
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3762628632886167,
      "learning_rate": 3.957670374388762e-06,
      "loss": 0.7991,
      "step": 4530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8758139784694725,
      "learning_rate": 3.955642899898293e-06,
      "loss": 0.7285,
      "step": 4531
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6622567875801364,
      "learning_rate": 3.953615604944915e-06,
      "loss": 0.7104,
      "step": 4532
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4138097939860974,
      "learning_rate": 3.9515884898771455e-06,
      "loss": 0.7623,
      "step": 4533
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5062961900721041,
      "learning_rate": 3.949561555043469e-06,
      "loss": 0.8342,
      "step": 4534
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.526877124243004,
      "learning_rate": 3.94753480079234e-06,
      "loss": 0.7315,
      "step": 4535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8006289008334465,
      "learning_rate": 3.94550822747218e-06,
      "loss": 0.6843,
      "step": 4536
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.200360826598461,
      "learning_rate": 3.943481835431379e-06,
      "loss": 0.7877,
      "step": 4537
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3391798122862109,
      "learning_rate": 3.9414556250183e-06,
      "loss": 0.7448,
      "step": 4538
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4012772539306655,
      "learning_rate": 3.939429596581271e-06,
      "loss": 0.7266,
      "step": 4539
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2219436269472574,
      "learning_rate": 3.937403750468588e-06,
      "loss": 0.7556,
      "step": 4540
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3967879452019594,
      "learning_rate": 3.935378087028521e-06,
      "loss": 0.7195,
      "step": 4541
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2966531280050426,
      "learning_rate": 3.9333526066093e-06,
      "loss": 0.6743,
      "step": 4542
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4003528097751057,
      "learning_rate": 3.931327309559133e-06,
      "loss": 0.8154,
      "step": 4543
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2301861306669337,
      "learning_rate": 3.9293021962261906e-06,
      "loss": 0.6447,
      "step": 4544
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8341070854105733,
      "learning_rate": 3.927277266958614e-06,
      "loss": 0.5994,
      "step": 4545
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.336120038353112,
      "learning_rate": 3.925252522104512e-06,
      "loss": 0.8793,
      "step": 4546
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6474340272767882,
      "learning_rate": 3.923227962011959e-06,
      "loss": 0.8066,
      "step": 4547
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6078729471545794,
      "learning_rate": 3.9212035870290035e-06,
      "loss": 0.8057,
      "step": 4548
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6277177203393405,
      "learning_rate": 3.919179397503659e-06,
      "loss": 0.7306,
      "step": 4549
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8958633186008107,
      "learning_rate": 3.917155393783903e-06,
      "loss": 0.6024,
      "step": 4550
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.309464400033865,
      "learning_rate": 3.9151315762176885e-06,
      "loss": 0.7905,
      "step": 4551
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.471615209341603,
      "learning_rate": 3.9131079451529345e-06,
      "loss": 0.6389,
      "step": 4552
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4681839617396264,
      "learning_rate": 3.9110845009375255e-06,
      "loss": 0.7845,
      "step": 4553
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2475920506264602,
      "learning_rate": 3.909061243919313e-06,
      "loss": 0.7658,
      "step": 4554
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8851000160063627,
      "learning_rate": 3.907038174446122e-06,
      "loss": 0.6678,
      "step": 4555
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8025329279974776,
      "learning_rate": 3.905015292865738e-06,
      "loss": 0.6321,
      "step": 4556
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5586089928470603,
      "learning_rate": 3.902992599525921e-06,
      "loss": 0.7357,
      "step": 4557
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3205000465570818,
      "learning_rate": 3.900970094774391e-06,
      "loss": 0.7834,
      "step": 4558
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3805165824430936,
      "learning_rate": 3.898947778958845e-06,
      "loss": 0.7396,
      "step": 4559
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.258298323877237,
      "learning_rate": 3.8969256524269395e-06,
      "loss": 0.8004,
      "step": 4560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3106752714905565,
      "learning_rate": 3.894903715526301e-06,
      "loss": 0.8815,
      "step": 4561
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.240736258270276,
      "learning_rate": 3.892881968604525e-06,
      "loss": 0.6757,
      "step": 4562
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8237792199252562,
      "learning_rate": 3.890860412009176e-06,
      "loss": 0.6272,
      "step": 4563
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3761190269383656,
      "learning_rate": 3.888839046087779e-06,
      "loss": 0.8482,
      "step": 4564
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1218180978676071,
      "learning_rate": 3.886817871187832e-06,
      "loss": 0.645,
      "step": 4565
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.272849806936162,
      "learning_rate": 3.8847968876567985e-06,
      "loss": 0.8743,
      "step": 4566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2810540075006027,
      "learning_rate": 3.882776095842108e-06,
      "loss": 0.7971,
      "step": 4567
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3639083291187362,
      "learning_rate": 3.880755496091158e-06,
      "loss": 0.8357,
      "step": 4568
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3732890652539016,
      "learning_rate": 3.878735088751315e-06,
      "loss": 0.7875,
      "step": 4569
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.269405116537701,
      "learning_rate": 3.876714874169906e-06,
      "loss": 0.7665,
      "step": 4570
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4138379633550258,
      "learning_rate": 3.874694852694236e-06,
      "loss": 0.7824,
      "step": 4571
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5053712770343108,
      "learning_rate": 3.8726750246715635e-06,
      "loss": 0.7114,
      "step": 4572
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3119756215925804,
      "learning_rate": 3.870655390449126e-06,
      "loss": 0.7825,
      "step": 4573
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.410473976982017,
      "learning_rate": 3.868635950374119e-06,
      "loss": 0.8114,
      "step": 4574
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3026150981717521,
      "learning_rate": 3.866616704793706e-06,
      "loss": 0.7316,
      "step": 4575
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8712976366931934,
      "learning_rate": 3.864597654055024e-06,
      "loss": 0.6824,
      "step": 4576
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7232470667172859,
      "learning_rate": 3.862578798505166e-06,
      "loss": 0.6242,
      "step": 4577
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3431753650166898,
      "learning_rate": 3.860560138491199e-06,
      "loss": 0.7568,
      "step": 4578
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.32124689301836,
      "learning_rate": 3.858541674360151e-06,
      "loss": 0.817,
      "step": 4579
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1528623570248495,
      "learning_rate": 3.856523406459025e-06,
      "loss": 0.7226,
      "step": 4580
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5559460170853914,
      "learning_rate": 3.854505335134781e-06,
      "loss": 0.7271,
      "step": 4581
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.329685762093376,
      "learning_rate": 3.852487460734351e-06,
      "loss": 0.7125,
      "step": 4582
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3435450728847529,
      "learning_rate": 3.850469783604628e-06,
      "loss": 0.7975,
      "step": 4583
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1753781787677582,
      "learning_rate": 3.848452304092477e-06,
      "loss": 0.6508,
      "step": 4584
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8172475397448518,
      "learning_rate": 3.846435022544724e-06,
      "loss": 0.692,
      "step": 4585
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4793257488999494,
      "learning_rate": 3.844417939308163e-06,
      "loss": 0.8006,
      "step": 4586
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3446570458543505,
      "learning_rate": 3.842401054729557e-06,
      "loss": 0.6929,
      "step": 4587
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4603605931875956,
      "learning_rate": 3.840384369155626e-06,
      "loss": 0.7928,
      "step": 4588
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3326662211977713,
      "learning_rate": 3.838367882933068e-06,
      "loss": 0.8366,
      "step": 4589
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5068980148310522,
      "learning_rate": 3.836351596408537e-06,
      "loss": 0.7691,
      "step": 4590
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3962468093200062,
      "learning_rate": 3.834335509928658e-06,
      "loss": 0.7747,
      "step": 4591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7903815257889109,
      "learning_rate": 3.832319623840018e-06,
      "loss": 0.6413,
      "step": 4592
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5011840490114945,
      "learning_rate": 3.830303938489172e-06,
      "loss": 0.7654,
      "step": 4593
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0046466184493985,
      "learning_rate": 3.8282884542226385e-06,
      "loss": 0.7297,
      "step": 4594
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5269795864346039,
      "learning_rate": 3.826273171386904e-06,
      "loss": 0.7804,
      "step": 4595
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3048939036382732,
      "learning_rate": 3.8242580903284186e-06,
      "loss": 0.8284,
      "step": 4596
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2624891549697161,
      "learning_rate": 3.822243211393596e-06,
      "loss": 0.7408,
      "step": 4597
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4646650849518603,
      "learning_rate": 3.820228534928818e-06,
      "loss": 0.8477,
      "step": 4598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8310494746657934,
      "learning_rate": 3.818214061280435e-06,
      "loss": 0.6688,
      "step": 4599
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2289098118663202,
      "learning_rate": 3.816199790794754e-06,
      "loss": 0.7484,
      "step": 4600
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2125509538828745,
      "learning_rate": 3.8141857238180537e-06,
      "loss": 0.7586,
      "step": 4601
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4906730725343442,
      "learning_rate": 3.812171860696574e-06,
      "loss": 0.8276,
      "step": 4602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7959716353618366,
      "learning_rate": 3.810158201776523e-06,
      "loss": 0.6389,
      "step": 4603
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1533087688845598,
      "learning_rate": 3.808144747404069e-06,
      "loss": 0.8258,
      "step": 4604
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5990086149428608,
      "learning_rate": 3.806131497925351e-06,
      "loss": 0.7887,
      "step": 4605
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3802576271837563,
      "learning_rate": 3.8041184536864694e-06,
      "loss": 0.8755,
      "step": 4606
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.451927559703742,
      "learning_rate": 3.802105615033488e-06,
      "loss": 0.7554,
      "step": 4607
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4927089808023313,
      "learning_rate": 3.8000929823124387e-06,
      "loss": 0.8722,
      "step": 4608
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3434499373210953,
      "learning_rate": 3.798080555869318e-06,
      "loss": 0.7742,
      "step": 4609
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4454794705361766,
      "learning_rate": 3.796068336050083e-06,
      "loss": 0.8423,
      "step": 4610
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2995624163314259,
      "learning_rate": 3.7940563232006573e-06,
      "loss": 0.7565,
      "step": 4611
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3875204435081718,
      "learning_rate": 3.792044517666931e-06,
      "loss": 0.8588,
      "step": 4612
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.36200128988264,
      "learning_rate": 3.790032919794757e-06,
      "loss": 0.7254,
      "step": 4613
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4106510010316484,
      "learning_rate": 3.788021529929949e-06,
      "loss": 0.8308,
      "step": 4614
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.367909033286175,
      "learning_rate": 3.786010348418292e-06,
      "loss": 0.8205,
      "step": 4615
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3926956362686564,
      "learning_rate": 3.783999375605527e-06,
      "loss": 0.7126,
      "step": 4616
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5238691230752945,
      "learning_rate": 3.7819886118373694e-06,
      "loss": 0.8325,
      "step": 4617
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5912667290796862,
      "learning_rate": 3.77997805745949e-06,
      "loss": 0.7203,
      "step": 4618
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2381419822530126,
      "learning_rate": 3.7779677128175247e-06,
      "loss": 0.7343,
      "step": 4619
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3651479876248964,
      "learning_rate": 3.775957578257078e-06,
      "loss": 0.847,
      "step": 4620
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1960220883370363,
      "learning_rate": 3.7739476541237137e-06,
      "loss": 0.6582,
      "step": 4621
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3651571614083926,
      "learning_rate": 3.7719379407629597e-06,
      "loss": 0.8086,
      "step": 4622
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.319188690838543,
      "learning_rate": 3.769928438520312e-06,
      "loss": 0.7034,
      "step": 4623
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3672273870587384,
      "learning_rate": 3.7679191477412263e-06,
      "loss": 0.7632,
      "step": 4624
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3255609239224582,
      "learning_rate": 3.7659100687711205e-06,
      "loss": 0.8236,
      "step": 4625
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2310244339591827,
      "learning_rate": 3.7639012019553813e-06,
      "loss": 0.6537,
      "step": 4626
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3188088078534974,
      "learning_rate": 3.761892547639357e-06,
      "loss": 0.7868,
      "step": 4627
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3853463750003665,
      "learning_rate": 3.7598841061683566e-06,
      "loss": 0.6731,
      "step": 4628
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3065451045133814,
      "learning_rate": 3.7578758778876546e-06,
      "loss": 0.8157,
      "step": 4629
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5600157463816942,
      "learning_rate": 3.75586786314249e-06,
      "loss": 0.7617,
      "step": 4630
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.362332454381591,
      "learning_rate": 3.753860062278063e-06,
      "loss": 0.7474,
      "step": 4631
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4663117955736376,
      "learning_rate": 3.7518524756395362e-06,
      "loss": 0.8116,
      "step": 4632
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3712495091094974,
      "learning_rate": 3.74984510357204e-06,
      "loss": 0.7518,
      "step": 4633
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8326383607911994,
      "learning_rate": 3.747837946420664e-06,
      "loss": 0.8546,
      "step": 4634
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4081408168573872,
      "learning_rate": 3.7458310045304586e-06,
      "loss": 0.7937,
      "step": 4635
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2196239979221766,
      "learning_rate": 3.7438242782464453e-06,
      "loss": 0.6869,
      "step": 4636
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5340878672101963,
      "learning_rate": 3.7418177679136005e-06,
      "loss": 0.7113,
      "step": 4637
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.518200733505159,
      "learning_rate": 3.7398114738768686e-06,
      "loss": 0.7086,
      "step": 4638
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4772484260552654,
      "learning_rate": 3.7378053964811534e-06,
      "loss": 0.8225,
      "step": 4639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8075368106851599,
      "learning_rate": 3.735799536071322e-06,
      "loss": 0.6175,
      "step": 4640
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3932991821763685,
      "learning_rate": 3.733793892992208e-06,
      "loss": 0.7365,
      "step": 4641
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7923320374581312,
      "learning_rate": 3.7317884675886025e-06,
      "loss": 0.5385,
      "step": 4642
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.254291999894995,
      "learning_rate": 3.7297832602052608e-06,
      "loss": 0.7434,
      "step": 4643
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4411752893657739,
      "learning_rate": 3.727778271186901e-06,
      "loss": 0.8913,
      "step": 4644
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2230955405209982,
      "learning_rate": 3.7257735008782076e-06,
      "loss": 0.7883,
      "step": 4645
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4080974956460535,
      "learning_rate": 3.7237689496238217e-06,
      "loss": 0.7986,
      "step": 4646
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.8383587988257437,
      "learning_rate": 3.721764617768347e-06,
      "loss": 0.7744,
      "step": 4647
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2777210888658925,
      "learning_rate": 3.7197605056563545e-06,
      "loss": 0.7484,
      "step": 4648
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4068503942577222,
      "learning_rate": 3.7177566136323726e-06,
      "loss": 0.7635,
      "step": 4649
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8878220093019875,
      "learning_rate": 3.7157529420408933e-06,
      "loss": 0.6735,
      "step": 4650
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.315486391800439,
      "learning_rate": 3.7137494912263723e-06,
      "loss": 0.8137,
      "step": 4651
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8898748988832961,
      "learning_rate": 3.7117462615332254e-06,
      "loss": 0.6384,
      "step": 4652
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8804575890957002,
      "learning_rate": 3.7097432533058288e-06,
      "loss": 0.6627,
      "step": 4653
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3439065352983512,
      "learning_rate": 3.7077404668885263e-06,
      "loss": 0.8532,
      "step": 4654
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6003964837157285,
      "learning_rate": 3.7057379026256185e-06,
      "loss": 0.8167,
      "step": 4655
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3527459727894604,
      "learning_rate": 3.7037355608613703e-06,
      "loss": 0.7643,
      "step": 4656
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7275708984588088,
      "learning_rate": 3.701733441940007e-06,
      "loss": 0.6239,
      "step": 4657
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.357923988022808,
      "learning_rate": 3.699731546205715e-06,
      "loss": 0.62,
      "step": 4658
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3737968590253478,
      "learning_rate": 3.6977298740026448e-06,
      "loss": 0.8724,
      "step": 4659
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3195511726493523,
      "learning_rate": 3.695728425674906e-06,
      "loss": 0.7562,
      "step": 4660
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.086881292557561,
      "learning_rate": 3.693727201566571e-06,
      "loss": 0.7562,
      "step": 4661
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2102846351436138,
      "learning_rate": 3.6917262020216727e-06,
      "loss": 0.7368,
      "step": 4662
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3061333676370794,
      "learning_rate": 3.6897254273842087e-06,
      "loss": 0.7527,
      "step": 4663
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2249161577796237,
      "learning_rate": 3.6877248779981332e-06,
      "loss": 0.8055,
      "step": 4664
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3283971640082894,
      "learning_rate": 3.685724554207365e-06,
      "loss": 0.8907,
      "step": 4665
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3323846101902828,
      "learning_rate": 3.6837244563557815e-06,
      "loss": 0.8782,
      "step": 4666
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9966916650186847,
      "learning_rate": 3.6817245847872253e-06,
      "loss": 0.6916,
      "step": 4667
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3320374090116558,
      "learning_rate": 3.679724939845494e-06,
      "loss": 0.7562,
      "step": 4668
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3962394893002905,
      "learning_rate": 3.677725521874352e-06,
      "loss": 0.6758,
      "step": 4669
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.510401709704095,
      "learning_rate": 3.6757263312175216e-06,
      "loss": 0.7652,
      "step": 4670
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.403922673130301,
      "learning_rate": 3.673727368218687e-06,
      "loss": 0.7567,
      "step": 4671
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3990047108418424,
      "learning_rate": 3.671728633221493e-06,
      "loss": 0.6421,
      "step": 4672
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3957497324522072,
      "learning_rate": 3.669730126569546e-06,
      "loss": 0.8904,
      "step": 4673
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.225159650285147,
      "learning_rate": 3.667731848606413e-06,
      "loss": 0.7481,
      "step": 4674
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3175035705014484,
      "learning_rate": 3.665733799675619e-06,
      "loss": 0.7592,
      "step": 4675
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.399999554344211,
      "learning_rate": 3.663735980120653e-06,
      "loss": 0.7787,
      "step": 4676
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2236566164179736,
      "learning_rate": 3.6617383902849645e-06,
      "loss": 0.7475,
      "step": 4677
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5827115639380378,
      "learning_rate": 3.6597410305119605e-06,
      "loss": 0.7939,
      "step": 4678
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3120690588970132,
      "learning_rate": 3.6577439011450112e-06,
      "loss": 0.7818,
      "step": 4679
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.503391331776728,
      "learning_rate": 3.6557470025274453e-06,
      "loss": 0.7813,
      "step": 4680
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0438083043996704,
      "learning_rate": 3.6537503350025525e-06,
      "loss": 0.6607,
      "step": 4681
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3114522521162641,
      "learning_rate": 3.6517538989135866e-06,
      "loss": 0.7105,
      "step": 4682
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4769005394866426,
      "learning_rate": 3.649757694603754e-06,
      "loss": 0.7632,
      "step": 4683
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8096703719173435,
      "learning_rate": 3.647761722416229e-06,
      "loss": 0.6293,
      "step": 4684
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7597983314734292,
      "learning_rate": 3.645765982694139e-06,
      "loss": 0.5646,
      "step": 4685
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3915555770213341,
      "learning_rate": 3.643770475780576e-06,
      "loss": 0.8168,
      "step": 4686
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4006254129472697,
      "learning_rate": 3.641775202018592e-06,
      "loss": 0.7775,
      "step": 4687
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8682107449774407,
      "learning_rate": 3.6397801617511965e-06,
      "loss": 0.6713,
      "step": 4688
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4028745828075548,
      "learning_rate": 3.6377853553213593e-06,
      "loss": 0.8221,
      "step": 4689
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4623374407967995,
      "learning_rate": 3.635790783072012e-06,
      "loss": 0.8121,
      "step": 4690
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.58105419789796,
      "learning_rate": 3.633796445346044e-06,
      "loss": 0.8315,
      "step": 4691
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4080486802042511,
      "learning_rate": 3.6318023424863057e-06,
      "loss": 0.7489,
      "step": 4692
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5234550984725295,
      "learning_rate": 3.6298084748356077e-06,
      "loss": 0.6896,
      "step": 4693
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6614175914654896,
      "learning_rate": 3.6278148427367154e-06,
      "loss": 0.8334,
      "step": 4694
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2897551285489715,
      "learning_rate": 3.6258214465323604e-06,
      "loss": 0.7434,
      "step": 4695
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3139514626028919,
      "learning_rate": 3.6238282865652304e-06,
      "loss": 0.8335,
      "step": 4696
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4150738589902436,
      "learning_rate": 3.62183536317797e-06,
      "loss": 0.7669,
      "step": 4697
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5294777166467468,
      "learning_rate": 3.6198426767131893e-06,
      "loss": 0.781,
      "step": 4698
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4135828643848964,
      "learning_rate": 3.6178502275134507e-06,
      "loss": 0.7607,
      "step": 4699
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.561516048891461,
      "learning_rate": 3.6158580159212833e-06,
      "loss": 0.75,
      "step": 4700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9406061130430274,
      "learning_rate": 3.6138660422791673e-06,
      "loss": 0.6397,
      "step": 4701
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3824398629778512,
      "learning_rate": 3.6118743069295503e-06,
      "loss": 0.7737,
      "step": 4702
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2254424672883903,
      "learning_rate": 3.609882810214832e-06,
      "loss": 0.6578,
      "step": 4703
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.371700507435741,
      "learning_rate": 3.6078915524773726e-06,
      "loss": 0.7166,
      "step": 4704
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.328416680656966,
      "learning_rate": 3.605900534059496e-06,
      "loss": 0.744,
      "step": 4705
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5191729433384178,
      "learning_rate": 3.603909755303479e-06,
      "loss": 0.8058,
      "step": 4706
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3237701260989279,
      "learning_rate": 3.6019192165515595e-06,
      "loss": 0.7201,
      "step": 4707
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0376539425709577,
      "learning_rate": 3.5999289181459346e-06,
      "loss": 0.6374,
      "step": 4708
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.199799666852569,
      "learning_rate": 3.597938860428758e-06,
      "loss": 0.8438,
      "step": 4709
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.320140177496371,
      "learning_rate": 3.5959490437421473e-06,
      "loss": 0.7994,
      "step": 4710
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.448952645370471,
      "learning_rate": 3.5939594684281736e-06,
      "loss": 0.685,
      "step": 4711
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3177801414489316,
      "learning_rate": 3.591970134828866e-06,
      "loss": 0.7145,
      "step": 4712
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5768873655603701,
      "learning_rate": 3.589981043286217e-06,
      "loss": 0.7838,
      "step": 4713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7792281902966326,
      "learning_rate": 3.587992194142173e-06,
      "loss": 0.5746,
      "step": 4714
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0133298636090933,
      "learning_rate": 3.586003587738639e-06,
      "loss": 0.6729,
      "step": 4715
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3133449761223845,
      "learning_rate": 3.584015224417482e-06,
      "loss": 0.7337,
      "step": 4716
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3109091853868071,
      "learning_rate": 3.5820271045205227e-06,
      "loss": 0.7052,
      "step": 4717
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.392425316700619,
      "learning_rate": 3.580039228389541e-06,
      "loss": 0.7517,
      "step": 4718
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4606228229650604,
      "learning_rate": 3.578051596366279e-06,
      "loss": 0.655,
      "step": 4719
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3788995109626767,
      "learning_rate": 3.576064208792433e-06,
      "loss": 0.7952,
      "step": 4720
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5498471052483196,
      "learning_rate": 3.574077066009657e-06,
      "loss": 0.8408,
      "step": 4721
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2941675770285612,
      "learning_rate": 3.5720901683595633e-06,
      "loss": 0.7015,
      "step": 4722
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2946376610262644,
      "learning_rate": 3.570103516183724e-06,
      "loss": 0.6655,
      "step": 4723
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4702178927731104,
      "learning_rate": 3.568117109823668e-06,
      "loss": 0.7183,
      "step": 4724
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3438745578383164,
      "learning_rate": 3.5661309496208785e-06,
      "loss": 0.8115,
      "step": 4725
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3826525672123877,
      "learning_rate": 3.564145035916803e-06,
      "loss": 0.7807,
      "step": 4726
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4321052281429894,
      "learning_rate": 3.56215936905284e-06,
      "loss": 0.8136,
      "step": 4727
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7853563497986709,
      "learning_rate": 3.5601739493703517e-06,
      "loss": 0.6105,
      "step": 4728
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.55766998693824,
      "learning_rate": 3.5581887772106536e-06,
      "loss": 0.8013,
      "step": 4729
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3278437795343874,
      "learning_rate": 3.5562038529150187e-06,
      "loss": 0.7432,
      "step": 4730
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3414677751276107,
      "learning_rate": 3.554219176824681e-06,
      "loss": 0.7544,
      "step": 4731
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.870091661985791,
      "learning_rate": 3.5522347492808272e-06,
      "loss": 0.6834,
      "step": 4732
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3284408866048572,
      "learning_rate": 3.5502505706246027e-06,
      "loss": 0.79,
      "step": 4733
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3542583487019475,
      "learning_rate": 3.5482666411971134e-06,
      "loss": 0.707,
      "step": 4734
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4438204451866203,
      "learning_rate": 3.5462829613394186e-06,
      "loss": 0.825,
      "step": 4735
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4231066984505014,
      "learning_rate": 3.544299531392533e-06,
      "loss": 0.7868,
      "step": 4736
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2918696130661476,
      "learning_rate": 3.5423163516974356e-06,
      "loss": 0.7524,
      "step": 4737
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6007281967590439,
      "learning_rate": 3.540333422595057e-06,
      "loss": 0.688,
      "step": 4738
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3546277497442991,
      "learning_rate": 3.538350744426285e-06,
      "loss": 0.6883,
      "step": 4739
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2786680293142405,
      "learning_rate": 3.5363683175319637e-06,
      "loss": 0.7296,
      "step": 4740
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4523743112583127,
      "learning_rate": 3.5343861422528973e-06,
      "loss": 0.8487,
      "step": 4741
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8081208528846716,
      "learning_rate": 3.5324042189298445e-06,
      "loss": 0.6604,
      "step": 4742
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3047637131967327,
      "learning_rate": 3.5304225479035193e-06,
      "loss": 0.7768,
      "step": 4743
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5309106232970315,
      "learning_rate": 3.5284411295145954e-06,
      "loss": 0.7795,
      "step": 4744
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2639994005037065,
      "learning_rate": 3.526459964103701e-06,
      "loss": 0.7991,
      "step": 4745
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4417899936351384,
      "learning_rate": 3.5244790520114202e-06,
      "loss": 0.798,
      "step": 4746
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3701223144466492,
      "learning_rate": 3.5224983935782975e-06,
      "loss": 0.704,
      "step": 4747
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4221908366572513,
      "learning_rate": 3.5205179891448294e-06,
      "loss": 0.9343,
      "step": 4748
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8436751184734547,
      "learning_rate": 3.518537839051471e-06,
      "loss": 0.6598,
      "step": 4749
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9191373597735866,
      "learning_rate": 3.5165579436386323e-06,
      "loss": 0.6729,
      "step": 4750
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4180092824917239,
      "learning_rate": 3.51457830324668e-06,
      "loss": 0.7883,
      "step": 4751
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.398322552932499,
      "learning_rate": 3.5125989182159393e-06,
      "loss": 0.8014,
      "step": 4752
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4058507269691753,
      "learning_rate": 3.5106197888866873e-06,
      "loss": 0.8142,
      "step": 4753
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3912166564092552,
      "learning_rate": 3.508640915599159e-06,
      "loss": 0.6557,
      "step": 4754
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.574115619404493,
      "learning_rate": 3.506662298693546e-06,
      "loss": 0.74,
      "step": 4755
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.39710644969226,
      "learning_rate": 3.5046839385099977e-06,
      "loss": 0.7666,
      "step": 4756
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4774698898659848,
      "learning_rate": 3.502705835388616e-06,
      "loss": 0.7399,
      "step": 4757
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.844277584032135,
      "learning_rate": 3.500727989669458e-06,
      "loss": 0.6905,
      "step": 4758
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.365836751099388,
      "learning_rate": 3.498750401692541e-06,
      "loss": 0.8059,
      "step": 4759
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3981950015792513,
      "learning_rate": 3.496773071797834e-06,
      "loss": 0.7832,
      "step": 4760
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3276613849600003,
      "learning_rate": 3.4947960003252614e-06,
      "loss": 0.818,
      "step": 4761
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4685534280263677,
      "learning_rate": 3.492819187614707e-06,
      "loss": 0.8086,
      "step": 4762
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3872924712411874,
      "learning_rate": 3.4908426340060075e-06,
      "loss": 0.7427,
      "step": 4763
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7786300901986133,
      "learning_rate": 3.488866339838953e-06,
      "loss": 0.6472,
      "step": 4764
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9167962695383668,
      "learning_rate": 3.4868903054532934e-06,
      "loss": 0.6605,
      "step": 4765
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9359226538476656,
      "learning_rate": 3.4849145311887335e-06,
      "loss": 0.656,
      "step": 4766
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.295282286290071,
      "learning_rate": 3.4829390173849296e-06,
      "loss": 0.789,
      "step": 4767
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.419650332909358,
      "learning_rate": 3.4809637643814965e-06,
      "loss": 0.6969,
      "step": 4768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9417031856818234,
      "learning_rate": 3.478988772518001e-06,
      "loss": 0.6285,
      "step": 4769
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7871021684689152,
      "learning_rate": 3.47701404213397e-06,
      "loss": 0.6882,
      "step": 4770
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3044763191925017,
      "learning_rate": 3.475039573568881e-06,
      "loss": 0.616,
      "step": 4771
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4768179077826584,
      "learning_rate": 3.4730653671621667e-06,
      "loss": 0.7434,
      "step": 4772
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0355866957567776,
      "learning_rate": 3.4710914232532167e-06,
      "loss": 0.6644,
      "step": 4773
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5181617365764346,
      "learning_rate": 3.4691177421813783e-06,
      "loss": 0.7892,
      "step": 4774
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3033496729733363,
      "learning_rate": 3.4671443242859465e-06,
      "loss": 0.7503,
      "step": 4775
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4502202055454099,
      "learning_rate": 3.465171169906175e-06,
      "loss": 0.8006,
      "step": 4776
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3312943420174779,
      "learning_rate": 3.4631982793812745e-06,
      "loss": 0.7872,
      "step": 4777
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2407972370275608,
      "learning_rate": 3.461225653050405e-06,
      "loss": 0.6922,
      "step": 4778
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5785971027197372,
      "learning_rate": 3.4592532912526845e-06,
      "loss": 0.7889,
      "step": 4779
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4674901905904627,
      "learning_rate": 3.4572811943271856e-06,
      "loss": 0.7011,
      "step": 4780
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2967867382909657,
      "learning_rate": 3.4553093626129343e-06,
      "loss": 0.738,
      "step": 4781
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7937171686052411,
      "learning_rate": 3.45333779644891e-06,
      "loss": 0.6679,
      "step": 4782
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4277282097874142,
      "learning_rate": 3.451366496174048e-06,
      "loss": 0.769,
      "step": 4783
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4592591314277485,
      "learning_rate": 3.4493954621272407e-06,
      "loss": 0.7287,
      "step": 4784
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2592253232200736,
      "learning_rate": 3.447424694647329e-06,
      "loss": 0.7952,
      "step": 4785
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3018090512836382,
      "learning_rate": 3.445454194073111e-06,
      "loss": 0.8227,
      "step": 4786
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4179921565894114,
      "learning_rate": 3.4434839607433396e-06,
      "loss": 0.6961,
      "step": 4787
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.558958555147389,
      "learning_rate": 3.4415139949967203e-06,
      "loss": 0.8134,
      "step": 4788
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7993316154459852,
      "learning_rate": 3.439544297171913e-06,
      "loss": 0.6624,
      "step": 4789
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.45598841735693,
      "learning_rate": 3.437574867607529e-06,
      "loss": 0.7619,
      "step": 4790
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2062535668835148,
      "learning_rate": 3.435605706642141e-06,
      "loss": 0.6699,
      "step": 4791
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1180068057417296,
      "learning_rate": 3.4336368146142653e-06,
      "loss": 0.8134,
      "step": 4792
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5676726963307224,
      "learning_rate": 3.4316681918623825e-06,
      "loss": 0.7703,
      "step": 4793
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4221276568009227,
      "learning_rate": 3.4296998387249175e-06,
      "loss": 0.763,
      "step": 4794
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2120233932747113,
      "learning_rate": 3.427731755540256e-06,
      "loss": 0.7009,
      "step": 4795
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2593585676195014,
      "learning_rate": 3.425763942646733e-06,
      "loss": 0.7492,
      "step": 4796
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.360885988630043,
      "learning_rate": 3.423796400382637e-06,
      "loss": 0.7754,
      "step": 4797
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.574878302432994,
      "learning_rate": 3.421829129086215e-06,
      "loss": 0.7203,
      "step": 4798
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5178893224326009,
      "learning_rate": 3.4198621290956603e-06,
      "loss": 0.6671,
      "step": 4799
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2728480173593413,
      "learning_rate": 3.4178954007491237e-06,
      "loss": 0.7198,
      "step": 4800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8404697216006696,
      "learning_rate": 3.4159289443847077e-06,
      "loss": 0.694,
      "step": 4801
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3411280942351356,
      "learning_rate": 3.4139627603404724e-06,
      "loss": 0.8165,
      "step": 4802
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2931651930625092,
      "learning_rate": 3.4119968489544254e-06,
      "loss": 0.6465,
      "step": 4803
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2588417399323022,
      "learning_rate": 3.4100312105645283e-06,
      "loss": 0.7447,
      "step": 4804
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4310373459895558,
      "learning_rate": 3.4080658455087e-06,
      "loss": 0.7131,
      "step": 4805
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.589483438642359,
      "learning_rate": 3.4061007541248093e-06,
      "loss": 0.7219,
      "step": 4806
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8455775730449686,
      "learning_rate": 3.404135936750674e-06,
      "loss": 0.816,
      "step": 4807
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2764144973432119,
      "learning_rate": 3.4021713937240748e-06,
      "loss": 0.7456,
      "step": 4808
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.520267991880434,
      "learning_rate": 3.4002071253827356e-06,
      "loss": 0.7643,
      "step": 4809
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.325255559615051,
      "learning_rate": 3.398243132064336e-06,
      "loss": 0.7881,
      "step": 4810
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7027057618569446,
      "learning_rate": 3.3962794141065136e-06,
      "loss": 0.7555,
      "step": 4811
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3984253493479313,
      "learning_rate": 3.3943159718468503e-06,
      "loss": 0.7501,
      "step": 4812
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3692243912109052,
      "learning_rate": 3.392352805622888e-06,
      "loss": 0.8427,
      "step": 4813
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3091486906528442,
      "learning_rate": 3.3903899157721156e-06,
      "loss": 0.7844,
      "step": 4814
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8037719084601368,
      "learning_rate": 3.388427302631975e-06,
      "loss": 0.6451,
      "step": 4815
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5364009492552344,
      "learning_rate": 3.386464966539865e-06,
      "loss": 0.8058,
      "step": 4816
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2949044902225104,
      "learning_rate": 3.3845029078331344e-06,
      "loss": 0.7841,
      "step": 4817
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3397990579084065,
      "learning_rate": 3.3825411268490803e-06,
      "loss": 0.7798,
      "step": 4818
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3760276239820197,
      "learning_rate": 3.380579623924959e-06,
      "loss": 0.7059,
      "step": 4819
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.30061106671616,
      "learning_rate": 3.378618399397972e-06,
      "loss": 0.7553,
      "step": 4820
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6008007853096207,
      "learning_rate": 3.3766574536052808e-06,
      "loss": 0.8037,
      "step": 4821
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2024583427092905,
      "learning_rate": 3.374696786883992e-06,
      "loss": 0.7067,
      "step": 4822
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2794247713280782,
      "learning_rate": 3.3727363995711695e-06,
      "loss": 0.8306,
      "step": 4823
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7983317018901692,
      "learning_rate": 3.3707762920038235e-06,
      "loss": 0.6982,
      "step": 4824
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2869921077864723,
      "learning_rate": 3.3688164645189198e-06,
      "loss": 0.7185,
      "step": 4825
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.192592375853972,
      "learning_rate": 3.366856917453377e-06,
      "loss": 0.7773,
      "step": 4826
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4050150057271253,
      "learning_rate": 3.3648976511440636e-06,
      "loss": 0.746,
      "step": 4827
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3306397936001597,
      "learning_rate": 3.3629386659277984e-06,
      "loss": 0.7419,
      "step": 4828
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4098338202463971,
      "learning_rate": 3.3609799621413554e-06,
      "loss": 0.6757,
      "step": 4829
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4117127960713416,
      "learning_rate": 3.359021540121457e-06,
      "loss": 0.7676,
      "step": 4830
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8662255647682033,
      "learning_rate": 3.3570634002047815e-06,
      "loss": 0.6938,
      "step": 4831
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5009810252074243,
      "learning_rate": 3.355105542727954e-06,
      "loss": 0.8231,
      "step": 4832
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9199855075256809,
      "learning_rate": 3.353147968027552e-06,
      "loss": 0.6119,
      "step": 4833
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.837624036041626,
      "learning_rate": 3.351190676440107e-06,
      "loss": 0.7015,
      "step": 4834
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.527578459050914,
      "learning_rate": 3.3492336683021e-06,
      "loss": 0.8028,
      "step": 4835
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7713594900500913,
      "learning_rate": 3.347276943949961e-06,
      "loss": 0.8062,
      "step": 4836
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2943965919704448,
      "learning_rate": 3.3453205037200766e-06,
      "loss": 0.8736,
      "step": 4837
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.244822191240966,
      "learning_rate": 3.3433643479487777e-06,
      "loss": 0.7665,
      "step": 4838
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.314266799723017,
      "learning_rate": 3.3414084769723554e-06,
      "loss": 0.8776,
      "step": 4839
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3461526113245355,
      "learning_rate": 3.339452891127043e-06,
      "loss": 0.7463,
      "step": 4840
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7038654907604402,
      "learning_rate": 3.33749759074903e-06,
      "loss": 0.8631,
      "step": 4841
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4471882200136108,
      "learning_rate": 3.3355425761744553e-06,
      "loss": 0.8206,
      "step": 4842
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2258648952591338,
      "learning_rate": 3.3335878477394058e-06,
      "loss": 0.8079,
      "step": 4843
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5264667267082932,
      "learning_rate": 3.331633405779926e-06,
      "loss": 0.7275,
      "step": 4844
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4268310206539476,
      "learning_rate": 3.3296792506320043e-06,
      "loss": 0.7306,
      "step": 4845
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.793939870553243,
      "learning_rate": 3.3277253826315824e-06,
      "loss": 0.6655,
      "step": 4846
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5965950367193726,
      "learning_rate": 3.325771802114555e-06,
      "loss": 0.713,
      "step": 4847
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2602597870045087,
      "learning_rate": 3.323818509416763e-06,
      "loss": 0.7632,
      "step": 4848
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3361726645315546,
      "learning_rate": 3.321865504874002e-06,
      "loss": 0.7711,
      "step": 4849
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4846849834511775,
      "learning_rate": 3.3199127888220162e-06,
      "loss": 0.6761,
      "step": 4850
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.39219896392526,
      "learning_rate": 3.317960361596498e-06,
      "loss": 0.7605,
      "step": 4851
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3915106570661475,
      "learning_rate": 3.3160082235330937e-06,
      "loss": 0.755,
      "step": 4852
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2176493603896275,
      "learning_rate": 3.3140563749673994e-06,
      "loss": 0.7552,
      "step": 4853
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3353428539441592,
      "learning_rate": 3.3121048162349577e-06,
      "loss": 0.7569,
      "step": 4854
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8027698033389573,
      "learning_rate": 3.3101535476712675e-06,
      "loss": 0.6966,
      "step": 4855
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3548656186066734,
      "learning_rate": 3.3082025696117723e-06,
      "loss": 0.7049,
      "step": 4856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3888450458312662,
      "learning_rate": 3.3062518823918664e-06,
      "loss": 0.839,
      "step": 4857
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5213048832419733,
      "learning_rate": 3.3043014863468985e-06,
      "loss": 0.7903,
      "step": 4858
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6344791048664022,
      "learning_rate": 3.3023513818121645e-06,
      "loss": 0.8814,
      "step": 4859
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.932960402440557,
      "learning_rate": 3.3004015691229086e-06,
      "loss": 0.7331,
      "step": 4860
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2435528136907625,
      "learning_rate": 3.298452048614326e-06,
      "loss": 0.757,
      "step": 4861
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3700906555391899,
      "learning_rate": 3.2965028206215627e-06,
      "loss": 0.7597,
      "step": 4862
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.268689368013913,
      "learning_rate": 3.2945538854797134e-06,
      "loss": 0.778,
      "step": 4863
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8198540229166943,
      "learning_rate": 3.292605243523821e-06,
      "loss": 0.6722,
      "step": 4864
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.908664237879591,
      "learning_rate": 3.2906568950888827e-06,
      "loss": 0.6392,
      "step": 4865
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5147322923886588,
      "learning_rate": 3.288708840509838e-06,
      "loss": 0.8065,
      "step": 4866
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4633139640729982,
      "learning_rate": 3.286761080121585e-06,
      "loss": 0.8809,
      "step": 4867
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2609190060480548,
      "learning_rate": 3.2848136142589637e-06,
      "loss": 0.8138,
      "step": 4868
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.555010007343891,
      "learning_rate": 3.282866443256765e-06,
      "loss": 0.7918,
      "step": 4869
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.30842846681626,
      "learning_rate": 3.280919567449733e-06,
      "loss": 0.7116,
      "step": 4870
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4116145994993783,
      "learning_rate": 3.278972987172556e-06,
      "loss": 0.7385,
      "step": 4871
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.86151979112464,
      "learning_rate": 3.277026702759874e-06,
      "loss": 0.6118,
      "step": 4872
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.289962352735823,
      "learning_rate": 3.275080714546277e-06,
      "loss": 0.7344,
      "step": 4873
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3022751555189245,
      "learning_rate": 3.2731350228663024e-06,
      "loss": 0.7705,
      "step": 4874
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6142895095377923,
      "learning_rate": 3.2711896280544343e-06,
      "loss": 0.816,
      "step": 4875
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.46854573680695,
      "learning_rate": 3.2692445304451128e-06,
      "loss": 0.7475,
      "step": 4876
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2733218072931345,
      "learning_rate": 3.2672997303727217e-06,
      "loss": 0.8078,
      "step": 4877
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8703520216827026,
      "learning_rate": 3.265355228171594e-06,
      "loss": 0.6802,
      "step": 4878
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3755704375066649,
      "learning_rate": 3.263411024176012e-06,
      "loss": 0.7357,
      "step": 4879
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4333142837394786,
      "learning_rate": 3.2614671187202075e-06,
      "loss": 0.8091,
      "step": 4880
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4594860606823952,
      "learning_rate": 3.2595235121383608e-06,
      "loss": 0.7563,
      "step": 4881
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6258668894486128,
      "learning_rate": 3.2575802047645977e-06,
      "loss": 0.6757,
      "step": 4882
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4913129257503954,
      "learning_rate": 3.2556371969329992e-06,
      "loss": 0.7153,
      "step": 4883
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8846170867408998,
      "learning_rate": 3.2536944889775868e-06,
      "loss": 0.6163,
      "step": 4884
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.581383543525467,
      "learning_rate": 3.2517520812323382e-06,
      "loss": 0.7263,
      "step": 4885
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.265604923398114,
      "learning_rate": 3.2498099740311752e-06,
      "loss": 0.8103,
      "step": 4886
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.336764765394436,
      "learning_rate": 3.2478681677079655e-06,
      "loss": 0.7781,
      "step": 4887
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2059086046256353,
      "learning_rate": 3.2459266625965315e-06,
      "loss": 0.8256,
      "step": 4888
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4936110718203588,
      "learning_rate": 3.24398545903064e-06,
      "loss": 0.7106,
      "step": 4889
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3310072936609538,
      "learning_rate": 3.2420445573440027e-06,
      "loss": 0.7296,
      "step": 4890
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3194083682891717,
      "learning_rate": 3.2401039578702886e-06,
      "loss": 0.735,
      "step": 4891
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5908735985065743,
      "learning_rate": 3.2381636609431065e-06,
      "loss": 0.8435,
      "step": 4892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.224272912594402,
      "learning_rate": 3.2362236668960144e-06,
      "loss": 0.8128,
      "step": 4893
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.36483864192637,
      "learning_rate": 3.234283976062521e-06,
      "loss": 0.7655,
      "step": 4894
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3863395940139456,
      "learning_rate": 3.2323445887760837e-06,
      "loss": 0.7235,
      "step": 4895
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4733772978426867,
      "learning_rate": 3.230405505370104e-06,
      "loss": 0.861,
      "step": 4896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.36832354841497,
      "learning_rate": 3.228466726177932e-06,
      "loss": 0.7825,
      "step": 4897
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3003583113096637,
      "learning_rate": 3.2265282515328676e-06,
      "loss": 0.7774,
      "step": 4898
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.367664845647285,
      "learning_rate": 3.2245900817681576e-06,
      "loss": 0.7763,
      "step": 4899
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.251075790645458,
      "learning_rate": 3.2226522172169928e-06,
      "loss": 0.8201,
      "step": 4900
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7001253034341752,
      "learning_rate": 3.220714658212518e-06,
      "loss": 0.7861,
      "step": 4901
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2482949974353956,
      "learning_rate": 3.2187774050878213e-06,
      "loss": 0.7781,
      "step": 4902
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7821446769185505,
      "learning_rate": 3.2168404581759362e-06,
      "loss": 0.6964,
      "step": 4903
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3781245181053028,
      "learning_rate": 3.21490381780985e-06,
      "loss": 0.7335,
      "step": 4904
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2279991675440136,
      "learning_rate": 3.212967484322491e-06,
      "loss": 0.9167,
      "step": 4905
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2766250050369214,
      "learning_rate": 3.2110314580467404e-06,
      "loss": 0.7077,
      "step": 4906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.300552898921065,
      "learning_rate": 3.209095739315421e-06,
      "loss": 0.753,
      "step": 4907
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9570621929936852,
      "learning_rate": 3.2071603284613052e-06,
      "loss": 0.6667,
      "step": 4908
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4263412537385254,
      "learning_rate": 3.2052252258171142e-06,
      "loss": 0.6661,
      "step": 4909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4945971681532564,
      "learning_rate": 3.2032904317155133e-06,
      "loss": 0.6569,
      "step": 4910
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3626182210714066,
      "learning_rate": 3.2013559464891162e-06,
      "loss": 0.8798,
      "step": 4911
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4736922001202304,
      "learning_rate": 3.199421770470482e-06,
      "loss": 0.7872,
      "step": 4912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2641269043381014,
      "learning_rate": 3.1974879039921214e-06,
      "loss": 0.7157,
      "step": 4913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4005171666322065,
      "learning_rate": 3.1955543473864868e-06,
      "loss": 0.8252,
      "step": 4914
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4201830642730495,
      "learning_rate": 3.1936211009859786e-06,
      "loss": 0.8053,
      "step": 4915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9361679191372612,
      "learning_rate": 3.1916881651229447e-06,
      "loss": 0.7811,
      "step": 4916
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4612384073352487,
      "learning_rate": 3.189755540129679e-06,
      "loss": 0.8636,
      "step": 4917
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7828872465579902,
      "learning_rate": 3.1878232263384214e-06,
      "loss": 0.6149,
      "step": 4918
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8927635069076937,
      "learning_rate": 3.1858912240813607e-06,
      "loss": 0.7043,
      "step": 4919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4684650506268686,
      "learning_rate": 3.183959533690629e-06,
      "loss": 0.7145,
      "step": 4920
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3923712206163592,
      "learning_rate": 3.1820281554983056e-06,
      "loss": 0.711,
      "step": 4921
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2838270517269865,
      "learning_rate": 3.1800970898364194e-06,
      "loss": 0.6862,
      "step": 4922
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2063975086025103,
      "learning_rate": 3.178166337036941e-06,
      "loss": 0.7415,
      "step": 4923
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.327730756362626,
      "learning_rate": 3.176235897431791e-06,
      "loss": 0.8129,
      "step": 4924
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8662559181573487,
      "learning_rate": 3.1743057713528325e-06,
      "loss": 0.6538,
      "step": 4925
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3669000660842574,
      "learning_rate": 3.172375959131876e-06,
      "loss": 0.7183,
      "step": 4926
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4751912760471488,
      "learning_rate": 3.1704464611006813e-06,
      "loss": 0.7797,
      "step": 4927
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9450788241945975,
      "learning_rate": 3.16851727759095e-06,
      "loss": 0.771,
      "step": 4928
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2466915909930538,
      "learning_rate": 3.1665884089343296e-06,
      "loss": 0.7332,
      "step": 4929
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9157870422785112,
      "learning_rate": 3.1646598554624174e-06,
      "loss": 0.7024,
      "step": 4930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5958541399706834,
      "learning_rate": 3.1627316175067515e-06,
      "loss": 0.778,
      "step": 4931
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3374881509271321,
      "learning_rate": 3.160803695398822e-06,
      "loss": 0.825,
      "step": 4932
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6894937834537302,
      "learning_rate": 3.1588760894700575e-06,
      "loss": 0.7831,
      "step": 4933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5435339944961275,
      "learning_rate": 3.156948800051839e-06,
      "loss": 0.666,
      "step": 4934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.294696732005118,
      "learning_rate": 3.1550218274754887e-06,
      "loss": 0.8486,
      "step": 4935
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3144450488892252,
      "learning_rate": 3.153095172072273e-06,
      "loss": 0.7139,
      "step": 4936
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4309630045165829,
      "learning_rate": 3.15116883417341e-06,
      "loss": 0.8092,
      "step": 4937
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4180288373709282,
      "learning_rate": 3.1492428141100575e-06,
      "loss": 0.8609,
      "step": 4938
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5303072677566052,
      "learning_rate": 3.1473171122133207e-06,
      "loss": 0.7621,
      "step": 4939
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9735158323998732,
      "learning_rate": 3.14539172881425e-06,
      "loss": 0.8539,
      "step": 4940
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8808176043228757,
      "learning_rate": 3.1434666642438416e-06,
      "loss": 0.637,
      "step": 4941
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.210075337818038,
      "learning_rate": 3.141541918833037e-06,
      "loss": 0.7025,
      "step": 4942
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.1601450968906994,
      "learning_rate": 3.139617492912722e-06,
      "loss": 0.7101,
      "step": 4943
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4492500527821226,
      "learning_rate": 3.1376933868137267e-06,
      "loss": 0.7113,
      "step": 4944
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5361883474345939,
      "learning_rate": 3.1357696008668285e-06,
      "loss": 0.79,
      "step": 4945
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9926070686420738,
      "learning_rate": 3.133846135402748e-06,
      "loss": 0.7269,
      "step": 4946
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5490324432241225,
      "learning_rate": 3.1319229907521502e-06,
      "loss": 0.7383,
      "step": 4947
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3183388862753593,
      "learning_rate": 3.130000167245648e-06,
      "loss": 0.7808,
      "step": 4948
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2871495163594708,
      "learning_rate": 3.128077665213794e-06,
      "loss": 0.8096,
      "step": 4949
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3847718902776287,
      "learning_rate": 3.1261554849870925e-06,
      "loss": 0.8335,
      "step": 4950
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4539464179260717,
      "learning_rate": 3.124233626895985e-06,
      "loss": 0.8398,
      "step": 4951
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3029814381052276,
      "learning_rate": 3.1223120912708647e-06,
      "loss": 0.8664,
      "step": 4952
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3150908499121314,
      "learning_rate": 3.1203908784420635e-06,
      "loss": 0.8818,
      "step": 4953
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0218947563929892,
      "learning_rate": 3.1184699887398594e-06,
      "loss": 0.6347,
      "step": 4954
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4779338440937941,
      "learning_rate": 3.116549422494478e-06,
      "loss": 0.7069,
      "step": 4955
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.33041609624482,
      "learning_rate": 3.1146291800360863e-06,
      "loss": 0.7463,
      "step": 4956
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4765959034317697,
      "learning_rate": 3.1127092616947935e-06,
      "loss": 0.8185,
      "step": 4957
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8552237181908856,
      "learning_rate": 3.1107896678006577e-06,
      "loss": 0.6695,
      "step": 4958
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5018121226941399,
      "learning_rate": 3.108870398683681e-06,
      "loss": 0.7057,
      "step": 4959
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3666939553297368,
      "learning_rate": 3.106951454673808e-06,
      "loss": 0.802,
      "step": 4960
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.575653657627692,
      "learning_rate": 3.105032836100925e-06,
      "loss": 0.7813,
      "step": 4961
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.393116586091643,
      "learning_rate": 3.1031145432948644e-06,
      "loss": 0.8457,
      "step": 4962
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1990279726770716,
      "learning_rate": 3.1011965765854056e-06,
      "loss": 0.7085,
      "step": 4963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6579386672618108,
      "learning_rate": 3.099278936302269e-06,
      "loss": 0.7668,
      "step": 4964
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4054731724585787,
      "learning_rate": 3.097361622775117e-06,
      "loss": 0.7269,
      "step": 4965
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6065682700756592,
      "learning_rate": 3.095444636333559e-06,
      "loss": 0.8147,
      "step": 4966
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.327254678763877,
      "learning_rate": 3.0935279773071492e-06,
      "loss": 0.7554,
      "step": 4967
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3232323392962784,
      "learning_rate": 3.0916116460253787e-06,
      "loss": 0.8117,
      "step": 4968
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1747737989550628,
      "learning_rate": 3.0896956428176916e-06,
      "loss": 0.7191,
      "step": 4969
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2129706822234725,
      "learning_rate": 3.0877799680134713e-06,
      "loss": 0.7818,
      "step": 4970
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4687400310412786,
      "learning_rate": 3.085864621942042e-06,
      "loss": 0.793,
      "step": 4971
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.34565006903012,
      "learning_rate": 3.0839496049326745e-06,
      "loss": 0.7961,
      "step": 4972
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.376987880150734,
      "learning_rate": 3.0820349173145823e-06,
      "loss": 0.9435,
      "step": 4973
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3696163510516641,
      "learning_rate": 3.080120559416924e-06,
      "loss": 0.6572,
      "step": 4974
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3205520970508848,
      "learning_rate": 3.0782065315687975e-06,
      "loss": 0.7959,
      "step": 4975
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3128281192272655,
      "learning_rate": 3.0762928340992483e-06,
      "loss": 0.7362,
      "step": 4976
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3527676273191533,
      "learning_rate": 3.0743794673372605e-06,
      "loss": 0.6777,
      "step": 4977
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4306171177988192,
      "learning_rate": 3.072466431611768e-06,
      "loss": 0.6944,
      "step": 4978
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4981572106357328,
      "learning_rate": 3.0705537272516406e-06,
      "loss": 0.8158,
      "step": 4979
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4609971455749986,
      "learning_rate": 3.0686413545856975e-06,
      "loss": 0.8447,
      "step": 4980
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6309854217092445,
      "learning_rate": 3.0667293139426944e-06,
      "loss": 0.7261,
      "step": 4981
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3193826542203237,
      "learning_rate": 3.064817605651336e-06,
      "loss": 0.7151,
      "step": 4982
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3510987592362103,
      "learning_rate": 3.0629062300402636e-06,
      "loss": 0.8144,
      "step": 4983
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3128118406864335,
      "learning_rate": 3.0609951874380685e-06,
      "loss": 0.7763,
      "step": 4984
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.75480558412316,
      "learning_rate": 3.05908447817328e-06,
      "loss": 0.7487,
      "step": 4985
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4029038448506026,
      "learning_rate": 3.057174102574367e-06,
      "loss": 0.7754,
      "step": 4986
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.543473121047097,
      "learning_rate": 3.0552640609697514e-06,
      "loss": 0.7791,
      "step": 4987
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.544617601318806,
      "learning_rate": 3.0533543536877897e-06,
      "loss": 0.8004,
      "step": 4988
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3692540244812366,
      "learning_rate": 3.0514449810567827e-06,
      "loss": 0.6985,
      "step": 4989
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.543386524670057,
      "learning_rate": 3.049535943404971e-06,
      "loss": 0.7563,
      "step": 4990
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3227881309563132,
      "learning_rate": 3.047627241060544e-06,
      "loss": 0.8139,
      "step": 4991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8583933946353851,
      "learning_rate": 3.0457188743516275e-06,
      "loss": 0.7534,
      "step": 4992
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2629274329196238,
      "learning_rate": 3.043810843606292e-06,
      "loss": 0.8634,
      "step": 4993
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8366056005401544,
      "learning_rate": 3.041903149152552e-06,
      "loss": 0.6129,
      "step": 4994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4671333813327918,
      "learning_rate": 3.039995791318358e-06,
      "loss": 0.8555,
      "step": 4995
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6702397383481513,
      "learning_rate": 3.038088770431612e-06,
      "loss": 0.7347,
      "step": 4996
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3321820243523161,
      "learning_rate": 3.0361820868201497e-06,
      "loss": 0.7715,
      "step": 4997
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5932485066755766,
      "learning_rate": 3.0342757408117544e-06,
      "loss": 0.7554,
      "step": 4998
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.333136399635792,
      "learning_rate": 3.0323697327341483e-06,
      "loss": 0.8096,
      "step": 4999
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8116679357029372,
      "learning_rate": 3.030464062914995e-06,
      "loss": 0.6545,
      "step": 5000
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.266027514670323,
      "learning_rate": 3.028558731681902e-06,
      "loss": 0.79,
      "step": 5001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2821009723429835,
      "learning_rate": 3.0266537393624185e-06,
      "loss": 0.7409,
      "step": 5002
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3203393674686346,
      "learning_rate": 3.024749086284034e-06,
      "loss": 0.679,
      "step": 5003
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1953271438874113,
      "learning_rate": 3.02284477277418e-06,
      "loss": 0.7222,
      "step": 5004
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2757988715727382,
      "learning_rate": 3.020940799160229e-06,
      "loss": 0.7233,
      "step": 5005
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3622627123411155,
      "learning_rate": 3.0190371657695005e-06,
      "loss": 0.7375,
      "step": 5006
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.578305719887606,
      "learning_rate": 3.017133872929249e-06,
      "loss": 0.8093,
      "step": 5007
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7650940815722593,
      "learning_rate": 3.0152309209666703e-06,
      "loss": 0.7043,
      "step": 5008
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3686927157833801,
      "learning_rate": 3.0133283102089067e-06,
      "loss": 0.7613,
      "step": 5009
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3492410340305439,
      "learning_rate": 3.0114260409830386e-06,
      "loss": 0.7625,
      "step": 5010
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8591024134523766,
      "learning_rate": 3.009524113616086e-06,
      "loss": 0.6279,
      "step": 5011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2998994529005452,
      "learning_rate": 3.0076225284350147e-06,
      "loss": 0.7971,
      "step": 5012
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.560363464311795,
      "learning_rate": 3.005721285766729e-06,
      "loss": 0.7456,
      "step": 5013
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2760506748611766,
      "learning_rate": 3.0038203859380717e-06,
      "loss": 0.7235,
      "step": 5014
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.385693394491089,
      "learning_rate": 3.0019198292758327e-06,
      "loss": 0.6973,
      "step": 5015
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8057683705000906,
      "learning_rate": 3.0000196161067398e-06,
      "loss": 0.652,
      "step": 5016
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2626483152685295,
      "learning_rate": 2.998119746757461e-06,
      "loss": 0.7494,
      "step": 5017
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3227165121523932,
      "learning_rate": 2.9962202215546043e-06,
      "loss": 0.7426,
      "step": 5018
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2867305168416665,
      "learning_rate": 2.994321040824722e-06,
      "loss": 0.8045,
      "step": 5019
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2692164956294334,
      "learning_rate": 2.9924222048943046e-06,
      "loss": 0.8171,
      "step": 5020
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4108345648695166,
      "learning_rate": 2.990523714089785e-06,
      "loss": 0.6537,
      "step": 5021
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4488378782378484,
      "learning_rate": 2.988625568737532e-06,
      "loss": 0.6984,
      "step": 5022
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4239976041259264,
      "learning_rate": 2.986727769163862e-06,
      "loss": 0.679,
      "step": 5023
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3954353605463095,
      "learning_rate": 2.9848303156950297e-06,
      "loss": 0.7103,
      "step": 5024
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4355785022193013,
      "learning_rate": 2.982933208657228e-06,
      "loss": 0.7894,
      "step": 5025
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1973311600820624,
      "learning_rate": 2.9810364483765913e-06,
      "loss": 0.7312,
      "step": 5026
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3306336619546575,
      "learning_rate": 2.9791400351791954e-06,
      "loss": 0.7343,
      "step": 5027
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3837970250207203,
      "learning_rate": 2.9772439693910554e-06,
      "loss": 0.7756,
      "step": 5028
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.716648021950409,
      "learning_rate": 2.975348251338126e-06,
      "loss": 0.7813,
      "step": 5029
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3731187770743527,
      "learning_rate": 2.973452881346305e-06,
      "loss": 0.8268,
      "step": 5030
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2385008798493642,
      "learning_rate": 2.9715578597414268e-06,
      "loss": 0.687,
      "step": 5031
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.214149449206696,
      "learning_rate": 2.9696631868492664e-06,
      "loss": 0.6324,
      "step": 5032
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.312012903613179,
      "learning_rate": 2.9677688629955438e-06,
      "loss": 0.7331,
      "step": 5033
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9148711360264706,
      "learning_rate": 2.965874888505913e-06,
      "loss": 0.6618,
      "step": 5034
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3383945277530513,
      "learning_rate": 2.9639812637059717e-06,
      "loss": 0.7833,
      "step": 5035
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8077538820744984,
      "learning_rate": 2.962087988921253e-06,
      "loss": 0.6158,
      "step": 5036
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4634856200740423,
      "learning_rate": 2.9601950644772352e-06,
      "loss": 0.7981,
      "step": 5037
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2172524528364543,
      "learning_rate": 2.958302490699334e-06,
      "loss": 0.7268,
      "step": 5038
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5513408908007351,
      "learning_rate": 2.9564102679129027e-06,
      "loss": 0.7685,
      "step": 5039
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.267610340865318,
      "learning_rate": 2.954518396443239e-06,
      "loss": 0.7418,
      "step": 5040
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3864541269710602,
      "learning_rate": 2.9526268766155753e-06,
      "loss": 0.7806,
      "step": 5041
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2279109692787253,
      "learning_rate": 2.9507357087550857e-06,
      "loss": 0.7305,
      "step": 5042
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2696211381702827,
      "learning_rate": 2.948844893186886e-06,
      "loss": 0.6992,
      "step": 5043
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.269823335285335,
      "learning_rate": 2.9469544302360283e-06,
      "loss": 0.7098,
      "step": 5044
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1978655958942934,
      "learning_rate": 2.945064320227505e-06,
      "loss": 0.7177,
      "step": 5045
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7861453264568413,
      "learning_rate": 2.9431745634862484e-06,
      "loss": 0.605,
      "step": 5046
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7294717263982475,
      "learning_rate": 2.9412851603371277e-06,
      "loss": 0.7359,
      "step": 5047
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2463334284289938,
      "learning_rate": 2.9393961111049564e-06,
      "loss": 0.7249,
      "step": 5048
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7597563456852254,
      "learning_rate": 2.937507416114481e-06,
      "loss": 0.7855,
      "step": 5049
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3969446693101928,
      "learning_rate": 2.9356190756903913e-06,
      "loss": 0.8192,
      "step": 5050
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5098880250040405,
      "learning_rate": 2.9337310901573134e-06,
      "loss": 0.6863,
      "step": 5051
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.885570285013463,
      "learning_rate": 2.9318434598398173e-06,
      "loss": 0.6306,
      "step": 5052
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.599506148409178,
      "learning_rate": 2.929956185062407e-06,
      "loss": 0.7247,
      "step": 5053
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8756672254719365,
      "learning_rate": 2.9280692661495247e-06,
      "loss": 0.6081,
      "step": 5054
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.288418633759493,
      "learning_rate": 2.9261827034255575e-06,
      "loss": 0.7533,
      "step": 5055
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3250460052286983,
      "learning_rate": 2.9242964972148245e-06,
      "loss": 0.6415,
      "step": 5056
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3518384995217154,
      "learning_rate": 2.922410647841586e-06,
      "loss": 0.7474,
      "step": 5057
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8708111354146257,
      "learning_rate": 2.920525155630044e-06,
      "loss": 0.7084,
      "step": 5058
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5275219837995677,
      "learning_rate": 2.918640020904334e-06,
      "loss": 0.8154,
      "step": 5059
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.314847894641115,
      "learning_rate": 2.9167552439885333e-06,
      "loss": 0.7355,
      "step": 5060
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6190691686445273,
      "learning_rate": 2.914870825206659e-06,
      "loss": 0.7145,
      "step": 5061
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3876865836556498,
      "learning_rate": 2.9129867648826623e-06,
      "loss": 0.7611,
      "step": 5062
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.267076964769963,
      "learning_rate": 2.9111030633404354e-06,
      "loss": 0.789,
      "step": 5063
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.353642582150885,
      "learning_rate": 2.9092197209038087e-06,
      "loss": 0.7931,
      "step": 5064
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8703413380523196,
      "learning_rate": 2.90733673789655e-06,
      "loss": 0.6959,
      "step": 5065
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3745433518841381,
      "learning_rate": 2.905454114642365e-06,
      "loss": 0.8028,
      "step": 5066
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.314432749442512,
      "learning_rate": 2.9035718514649013e-06,
      "loss": 0.8569,
      "step": 5067
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.524091177297494,
      "learning_rate": 2.9016899486877404e-06,
      "loss": 0.6796,
      "step": 5068
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7862573422604997,
      "learning_rate": 2.8998084066344005e-06,
      "loss": 0.6706,
      "step": 5069
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4529231685052928,
      "learning_rate": 2.8979272256283453e-06,
      "loss": 0.7906,
      "step": 5070
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6354748245150126,
      "learning_rate": 2.89604640599297e-06,
      "loss": 0.7325,
      "step": 5071
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4474598146528113,
      "learning_rate": 2.8941659480516083e-06,
      "loss": 0.7838,
      "step": 5072
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2707169957415583,
      "learning_rate": 2.892285852127532e-06,
      "loss": 0.8138,
      "step": 5073
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3855806091281735,
      "learning_rate": 2.8904061185439513e-06,
      "loss": 0.7579,
      "step": 5074
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3691690619573764,
      "learning_rate": 2.8885267476240173e-06,
      "loss": 0.8022,
      "step": 5075
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5629730650690299,
      "learning_rate": 2.8866477396908143e-06,
      "loss": 0.8205,
      "step": 5076
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2014756220585627,
      "learning_rate": 2.8847690950673642e-06,
      "loss": 0.658,
      "step": 5077
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8406073482980997,
      "learning_rate": 2.882890814076629e-06,
      "loss": 0.6899,
      "step": 5078
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1831390057830793,
      "learning_rate": 2.8810128970415035e-06,
      "loss": 0.7421,
      "step": 5079
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4462761449565507,
      "learning_rate": 2.879135344284829e-06,
      "loss": 0.7314,
      "step": 5080
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.271957818484182,
      "learning_rate": 2.8772581561293743e-06,
      "loss": 0.8268,
      "step": 5081
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2658051176932275,
      "learning_rate": 2.875381332897853e-06,
      "loss": 0.8607,
      "step": 5082
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6785949679968024,
      "learning_rate": 2.873504874912911e-06,
      "loss": 0.8425,
      "step": 5083
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.675945329678281,
      "learning_rate": 2.8716287824971344e-06,
      "loss": 0.7875,
      "step": 5084
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.285745798109743,
      "learning_rate": 2.869753055973044e-06,
      "loss": 0.7539,
      "step": 5085
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.319105478584304,
      "learning_rate": 2.8678776956630995e-06,
      "loss": 0.7243,
      "step": 5086
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4236442832088072,
      "learning_rate": 2.866002701889694e-06,
      "loss": 0.8138,
      "step": 5087
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8167521169738956,
      "learning_rate": 2.8641280749751655e-06,
      "loss": 0.6625,
      "step": 5088
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.388406766628848,
      "learning_rate": 2.8622538152417804e-06,
      "loss": 0.8235,
      "step": 5089
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3521933152942518,
      "learning_rate": 2.8603799230117495e-06,
      "loss": 0.7452,
      "step": 5090
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.379238435996149,
      "learning_rate": 2.858506398607214e-06,
      "loss": 0.7562,
      "step": 5091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4603325115202082,
      "learning_rate": 2.8566332423502553e-06,
      "loss": 0.7995,
      "step": 5092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1232371476468854,
      "learning_rate": 2.8547604545628903e-06,
      "loss": 0.7479,
      "step": 5093
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2015330966896198,
      "learning_rate": 2.852888035567073e-06,
      "loss": 0.7728,
      "step": 5094
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3384716874551557,
      "learning_rate": 2.851015985684691e-06,
      "loss": 0.8236,
      "step": 5095
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2123437967894088,
      "learning_rate": 2.8491443052375767e-06,
      "loss": 0.724,
      "step": 5096
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5060260798942133,
      "learning_rate": 2.84727299454749e-06,
      "loss": 0.7426,
      "step": 5097
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4115488399901377,
      "learning_rate": 2.845402053936133e-06,
      "loss": 0.7998,
      "step": 5098
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9307920615905705,
      "learning_rate": 2.8435314837251414e-06,
      "loss": 0.8026,
      "step": 5099
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2920511536376669,
      "learning_rate": 2.841661284236088e-06,
      "loss": 0.8462,
      "step": 5100
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2912830678392357,
      "learning_rate": 2.8397914557904816e-06,
      "loss": 0.7579,
      "step": 5101
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3664990890388713,
      "learning_rate": 2.837921998709765e-06,
      "loss": 0.8212,
      "step": 5102
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4512151090798946,
      "learning_rate": 2.8360529133153237e-06,
      "loss": 0.6955,
      "step": 5103
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4501470145879938,
      "learning_rate": 2.8341841999284725e-06,
      "loss": 0.7146,
      "step": 5104
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3722204796786457,
      "learning_rate": 2.832315858870467e-06,
      "loss": 0.7776,
      "step": 5105
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9486486608418191,
      "learning_rate": 2.830447890462492e-06,
      "loss": 0.676,
      "step": 5106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3727340357288254,
      "learning_rate": 2.828580295025679e-06,
      "loss": 0.7309,
      "step": 5107
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4761668604689104,
      "learning_rate": 2.8267130728810866e-06,
      "loss": 0.7896,
      "step": 5108
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4018123192887428,
      "learning_rate": 2.824846224349712e-06,
      "loss": 0.8919,
      "step": 5109
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.297940002770349,
      "learning_rate": 2.822979749752486e-06,
      "loss": 0.7703,
      "step": 5110
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3135808850512436,
      "learning_rate": 2.821113649410281e-06,
      "loss": 0.7,
      "step": 5111
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3229140294785093,
      "learning_rate": 2.8192479236438987e-06,
      "loss": 0.7967,
      "step": 5112
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.250267481470256,
      "learning_rate": 2.8173825727740808e-06,
      "loss": 0.685,
      "step": 5113
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3395735773827178,
      "learning_rate": 2.815517597121501e-06,
      "loss": 0.7653,
      "step": 5114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3711616871978858,
      "learning_rate": 2.8136529970067715e-06,
      "loss": 0.7499,
      "step": 5115
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.506700801729369,
      "learning_rate": 2.8117887727504355e-06,
      "loss": 0.6967,
      "step": 5116
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3495009817146488,
      "learning_rate": 2.809924924672978e-06,
      "loss": 0.813,
      "step": 5117
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4215554090636935,
      "learning_rate": 2.8080614530948174e-06,
      "loss": 0.8372,
      "step": 5118
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1761961198922297,
      "learning_rate": 2.806198358336304e-06,
      "loss": 0.679,
      "step": 5119
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3596072427486723,
      "learning_rate": 2.804335640717726e-06,
      "loss": 0.778,
      "step": 5120
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8485638022074072,
      "learning_rate": 2.802473300559305e-06,
      "loss": 0.6527,
      "step": 5121
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.35641706845247,
      "learning_rate": 2.8006113381812e-06,
      "loss": 0.7762,
      "step": 5122
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.527947455351264,
      "learning_rate": 2.798749753903501e-06,
      "loss": 0.7424,
      "step": 5123
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4596848150424873,
      "learning_rate": 2.7968885480462406e-06,
      "loss": 0.766,
      "step": 5124
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.032764906802939,
      "learning_rate": 2.7950277209293774e-06,
      "loss": 0.739,
      "step": 5125
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9414235067404799,
      "learning_rate": 2.7931672728728137e-06,
      "loss": 0.5965,
      "step": 5126
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3760809034098203,
      "learning_rate": 2.7913072041963797e-06,
      "loss": 0.682,
      "step": 5127
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8305066894871062,
      "learning_rate": 2.789447515219842e-06,
      "loss": 0.6496,
      "step": 5128
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3139340865622477,
      "learning_rate": 2.7875882062629033e-06,
      "loss": 0.8027,
      "step": 5129
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.15571078868022,
      "learning_rate": 2.7857292776452003e-06,
      "loss": 0.6886,
      "step": 5130
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5964913879500795,
      "learning_rate": 2.783870729686303e-06,
      "loss": 0.7374,
      "step": 5131
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.515182166060998,
      "learning_rate": 2.782012562705721e-06,
      "loss": 0.7892,
      "step": 5132
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1618805362658806,
      "learning_rate": 2.780154777022892e-06,
      "loss": 0.7701,
      "step": 5133
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.300696334115789,
      "learning_rate": 2.77829737295719e-06,
      "loss": 0.6928,
      "step": 5134
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5239523799045958,
      "learning_rate": 2.7764403508279275e-06,
      "loss": 0.737,
      "step": 5135
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7839888973171174,
      "learning_rate": 2.774583710954346e-06,
      "loss": 0.572,
      "step": 5136
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3676716502820567,
      "learning_rate": 2.7727274536556248e-06,
      "loss": 0.7394,
      "step": 5137
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5038460392866513,
      "learning_rate": 2.770871579250872e-06,
      "loss": 0.7871,
      "step": 5138
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.35328097533708,
      "learning_rate": 2.76901608805914e-06,
      "loss": 0.7783,
      "step": 5139
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3540533373644794,
      "learning_rate": 2.767160980399406e-06,
      "loss": 0.707,
      "step": 5140
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5571380242023445,
      "learning_rate": 2.7653062565905842e-06,
      "loss": 0.7872,
      "step": 5141
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.87064331204223,
      "learning_rate": 2.7634519169515237e-06,
      "loss": 0.635,
      "step": 5142
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3098850422889068,
      "learning_rate": 2.7615979618010057e-06,
      "loss": 0.8067,
      "step": 5143
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4274121371531514,
      "learning_rate": 2.759744391457749e-06,
      "loss": 0.7678,
      "step": 5144
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3477905711117653,
      "learning_rate": 2.757891206240404e-06,
      "loss": 0.7476,
      "step": 5145
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4539216688312762,
      "learning_rate": 2.7560384064675505e-06,
      "loss": 0.7044,
      "step": 5146
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2678029502555288,
      "learning_rate": 2.7541859924577106e-06,
      "loss": 0.7832,
      "step": 5147
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4250109823302972,
      "learning_rate": 2.7523339645293357e-06,
      "loss": 0.6545,
      "step": 5148
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8725240960703048,
      "learning_rate": 2.750482323000808e-06,
      "loss": 0.6604,
      "step": 5149
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4868162710164192,
      "learning_rate": 2.748631068190449e-06,
      "loss": 0.8319,
      "step": 5150
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.576829543382453,
      "learning_rate": 2.746780200416509e-06,
      "loss": 0.7544,
      "step": 5151
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3447608478656228,
      "learning_rate": 2.7449297199971713e-06,
      "loss": 0.7288,
      "step": 5152
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.570926797123869,
      "learning_rate": 2.743079627250559e-06,
      "loss": 0.6707,
      "step": 5153
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.29065689020619,
      "learning_rate": 2.741229922494724e-06,
      "loss": 0.7513,
      "step": 5154
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5366055213705538,
      "learning_rate": 2.7393806060476525e-06,
      "loss": 0.7667,
      "step": 5155
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3683517167699175,
      "learning_rate": 2.737531678227261e-06,
      "loss": 0.8459,
      "step": 5156
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9461462749924361,
      "learning_rate": 2.7356831393514033e-06,
      "loss": 0.7257,
      "step": 5157
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1904245639932303,
      "learning_rate": 2.733834989737864e-06,
      "loss": 0.6836,
      "step": 5158
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8212427419009817,
      "learning_rate": 2.7319872297043604e-06,
      "loss": 0.679,
      "step": 5159
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.818997016154652,
      "learning_rate": 2.7301398595685456e-06,
      "loss": 0.7893,
      "step": 5160
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4650361027936736,
      "learning_rate": 2.7282928796480047e-06,
      "loss": 0.7939,
      "step": 5161
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2526908200156068,
      "learning_rate": 2.726446290260251e-06,
      "loss": 0.7014,
      "step": 5162
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4578354918774297,
      "learning_rate": 2.72460009172274e-06,
      "loss": 0.769,
      "step": 5163
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.525497340154369,
      "learning_rate": 2.722754284352851e-06,
      "loss": 0.7343,
      "step": 5164
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.664029586371584,
      "learning_rate": 2.7209088684679015e-06,
      "loss": 0.7426,
      "step": 5165
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4146731653622044,
      "learning_rate": 2.719063844385138e-06,
      "loss": 0.8526,
      "step": 5166
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.334823659864901,
      "learning_rate": 2.7172192124217418e-06,
      "loss": 0.7591,
      "step": 5167
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5458461034114548,
      "learning_rate": 2.7153749728948286e-06,
      "loss": 0.7936,
      "step": 5168
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3813653421196093,
      "learning_rate": 2.7135311261214425e-06,
      "loss": 0.7864,
      "step": 5169
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3675217520075083,
      "learning_rate": 2.711687672418564e-06,
      "loss": 0.7117,
      "step": 5170
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8272882417925642,
      "learning_rate": 2.7098446121031008e-06,
      "loss": 0.6354,
      "step": 5171
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4811799633818867,
      "learning_rate": 2.7080019454919e-06,
      "loss": 0.7733,
      "step": 5172
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8300193896027374,
      "learning_rate": 2.706159672901736e-06,
      "loss": 0.7164,
      "step": 5173
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5612370983853785,
      "learning_rate": 2.7043177946493148e-06,
      "loss": 0.7121,
      "step": 5174
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3528153464618566,
      "learning_rate": 2.7024763110512808e-06,
      "loss": 0.7976,
      "step": 5175
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.399500189205405,
      "learning_rate": 2.700635222424204e-06,
      "loss": 0.7729,
      "step": 5176
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2845353048694035,
      "learning_rate": 2.6987945290845884e-06,
      "loss": 0.8007,
      "step": 5177
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3606057310529254,
      "learning_rate": 2.6969542313488716e-06,
      "loss": 0.793,
      "step": 5178
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2712888270356977,
      "learning_rate": 2.695114329533422e-06,
      "loss": 0.766,
      "step": 5179
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8795272029610002,
      "learning_rate": 2.693274823954537e-06,
      "loss": 0.6469,
      "step": 5180
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.91262603116069,
      "learning_rate": 2.6914357149284522e-06,
      "loss": 0.6484,
      "step": 5181
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.3107877995026764,
      "learning_rate": 2.689597002771333e-06,
      "loss": 0.7891,
      "step": 5182
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5629008475934936,
      "learning_rate": 2.6877586877992735e-06,
      "loss": 0.7648,
      "step": 5183
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3517355374687743,
      "learning_rate": 2.6859207703283012e-06,
      "loss": 0.7901,
      "step": 5184
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8315350691298811,
      "learning_rate": 2.6840832506743757e-06,
      "loss": 0.6454,
      "step": 5185
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2176311074502866,
      "learning_rate": 2.6822461291533874e-06,
      "loss": 0.7675,
      "step": 5186
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8352597648047673,
      "learning_rate": 2.6804094060811604e-06,
      "loss": 0.7075,
      "step": 5187
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8203961943786504,
      "learning_rate": 2.6785730817734446e-06,
      "loss": 0.7809,
      "step": 5188
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.694099527297083,
      "learning_rate": 2.67673715654593e-06,
      "loss": 0.8033,
      "step": 5189
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4356659550633497,
      "learning_rate": 2.6749016307142307e-06,
      "loss": 0.7852,
      "step": 5190
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7185789548065826,
      "learning_rate": 2.6730665045938975e-06,
      "loss": 0.8226,
      "step": 5191
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.486558608222929,
      "learning_rate": 2.671231778500408e-06,
      "loss": 0.7017,
      "step": 5192
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7128109957233821,
      "learning_rate": 2.669397452749173e-06,
      "loss": 0.8521,
      "step": 5193
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3853828783700264,
      "learning_rate": 2.667563527655535e-06,
      "loss": 0.7695,
      "step": 5194
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5988690261136058,
      "learning_rate": 2.6657300035347643e-06,
      "loss": 0.796,
      "step": 5195
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.354666855791736,
      "learning_rate": 2.6638968807020693e-06,
      "loss": 0.8238,
      "step": 5196
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3125236522926826,
      "learning_rate": 2.6620641594725834e-06,
      "loss": 0.7743,
      "step": 5197
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.859276583301977,
      "learning_rate": 2.6602318401613715e-06,
      "loss": 0.6866,
      "step": 5198
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4980231427351247,
      "learning_rate": 2.6583999230834294e-06,
      "loss": 0.7059,
      "step": 5199
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8342262904047454,
      "learning_rate": 2.656568408553689e-06,
      "loss": 0.665,
      "step": 5200
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7256601403799114,
      "learning_rate": 2.654737296887006e-06,
      "loss": 0.7736,
      "step": 5201
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4434011175702095,
      "learning_rate": 2.652906588398171e-06,
      "loss": 0.7547,
      "step": 5202
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5153738723781407,
      "learning_rate": 2.6510762834019006e-06,
      "loss": 0.7379,
      "step": 5203
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3108143076667338,
      "learning_rate": 2.64924638221285e-06,
      "loss": 0.7644,
      "step": 5204
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8806549734253943,
      "learning_rate": 2.6474168851455995e-06,
      "loss": 0.6345,
      "step": 5205
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2376250229303556,
      "learning_rate": 2.645587792514659e-06,
      "loss": 0.7263,
      "step": 5206
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4642506493476264,
      "learning_rate": 2.6437591046344724e-06,
      "loss": 0.7618,
      "step": 5207
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5287567609991481,
      "learning_rate": 2.641930821819409e-06,
      "loss": 0.7803,
      "step": 5208
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.428889758156576,
      "learning_rate": 2.6401029443837767e-06,
      "loss": 0.7201,
      "step": 5209
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2951065258295429,
      "learning_rate": 2.6382754726418047e-06,
      "loss": 0.8651,
      "step": 5210
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.24305621837569,
      "learning_rate": 2.63644840690766e-06,
      "loss": 0.7863,
      "step": 5211
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2555062542296407,
      "learning_rate": 2.634621747495435e-06,
      "loss": 0.7885,
      "step": 5212
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4000935618318708,
      "learning_rate": 2.632795494719154e-06,
      "loss": 0.7616,
      "step": 5213
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9735839504320942,
      "learning_rate": 2.6309696488927692e-06,
      "loss": 0.6814,
      "step": 5214
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5099751131680368,
      "learning_rate": 2.6291442103301667e-06,
      "loss": 0.754,
      "step": 5215
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4602325080765308,
      "learning_rate": 2.6273191793451575e-06,
      "loss": 0.7993,
      "step": 5216
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3794785470020134,
      "learning_rate": 2.6254945562514877e-06,
      "loss": 0.7818,
      "step": 5217
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3325688044876842,
      "learning_rate": 2.623670341362834e-06,
      "loss": 0.8292,
      "step": 5218
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4785704975948548,
      "learning_rate": 2.621846534992797e-06,
      "loss": 0.764,
      "step": 5219
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4302854668440044,
      "learning_rate": 2.62002313745491e-06,
      "loss": 0.8498,
      "step": 5220
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.316688599768095,
      "learning_rate": 2.6182001490626374e-06,
      "loss": 0.7321,
      "step": 5221
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4368807379724164,
      "learning_rate": 2.6163775701293714e-06,
      "loss": 0.8734,
      "step": 5222
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.295702792425549,
      "learning_rate": 2.6145554009684343e-06,
      "loss": 0.7459,
      "step": 5223
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2957859856525415,
      "learning_rate": 2.612733641893076e-06,
      "loss": 0.8237,
      "step": 5224
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3536754796902266,
      "learning_rate": 2.610912293216483e-06,
      "loss": 0.7333,
      "step": 5225
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2407254105673837,
      "learning_rate": 2.6090913552517626e-06,
      "loss": 0.8407,
      "step": 5226
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.424438349814412,
      "learning_rate": 2.6072708283119543e-06,
      "loss": 0.7651,
      "step": 5227
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2298984233280745,
      "learning_rate": 2.6054507127100315e-06,
      "loss": 0.694,
      "step": 5228
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7987101607450438,
      "learning_rate": 2.6036310087588913e-06,
      "loss": 0.6518,
      "step": 5229
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3687221604606874,
      "learning_rate": 2.601811716771362e-06,
      "loss": 0.7358,
      "step": 5230
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8131694428160581,
      "learning_rate": 2.599992837060198e-06,
      "loss": 0.7224,
      "step": 5231
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2473262877356668,
      "learning_rate": 2.598174369938091e-06,
      "loss": 0.8121,
      "step": 5232
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.43564417849554,
      "learning_rate": 2.596356315717654e-06,
      "loss": 0.716,
      "step": 5233
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3456653506608858,
      "learning_rate": 2.594538674711432e-06,
      "loss": 0.7522,
      "step": 5234
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2350378834570852,
      "learning_rate": 2.5927214472318983e-06,
      "loss": 0.7343,
      "step": 5235
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2343173716320168,
      "learning_rate": 2.5909046335914533e-06,
      "loss": 0.7571,
      "step": 5236
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6459698719727636,
      "learning_rate": 2.5890882341024328e-06,
      "loss": 0.8859,
      "step": 5237
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4142217882342718,
      "learning_rate": 2.5872722490770917e-06,
      "loss": 0.8171,
      "step": 5238
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.551847100634189,
      "learning_rate": 2.5854566788276247e-06,
      "loss": 0.7319,
      "step": 5239
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2850862403848338,
      "learning_rate": 2.583641523666146e-06,
      "loss": 0.7136,
      "step": 5240
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.252232430471832,
      "learning_rate": 2.5818267839047018e-06,
      "loss": 0.7806,
      "step": 5241
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3239212283783655,
      "learning_rate": 2.5800124598552676e-06,
      "loss": 0.7347,
      "step": 5242
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2552901678014206,
      "learning_rate": 2.5781985518297468e-06,
      "loss": 0.7347,
      "step": 5243
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4682423231693644,
      "learning_rate": 2.576385060139971e-06,
      "loss": 0.7639,
      "step": 5244
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2420052385727631,
      "learning_rate": 2.574571985097698e-06,
      "loss": 0.7844,
      "step": 5245
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5774158046524405,
      "learning_rate": 2.572759327014618e-06,
      "loss": 0.6553,
      "step": 5246
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5683729675382214,
      "learning_rate": 2.5709470862023517e-06,
      "loss": 0.686,
      "step": 5247
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8645765072031575,
      "learning_rate": 2.5691352629724407e-06,
      "loss": 0.6511,
      "step": 5248
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3418567946209479,
      "learning_rate": 2.567323857636359e-06,
      "loss": 0.6911,
      "step": 5249
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.385024526554139,
      "learning_rate": 2.565512870505509e-06,
      "loss": 0.9188,
      "step": 5250
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5480665451452706,
      "learning_rate": 2.5637023018912187e-06,
      "loss": 0.7625,
      "step": 5251
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6490258349289422,
      "learning_rate": 2.561892152104745e-06,
      "loss": 0.7617,
      "step": 5252
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7910826158461067,
      "learning_rate": 2.5600824214572774e-06,
      "loss": 0.6721,
      "step": 5253
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4330412432497939,
      "learning_rate": 2.558273110259925e-06,
      "loss": 0.7687,
      "step": 5254
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.359293268391157,
      "learning_rate": 2.5564642188237345e-06,
      "loss": 0.7383,
      "step": 5255
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.377324753989795,
      "learning_rate": 2.5546557474596725e-06,
      "loss": 0.7429,
      "step": 5256
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3614563110837294,
      "learning_rate": 2.5528476964786365e-06,
      "loss": 0.6997,
      "step": 5257
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8007406527518054,
      "learning_rate": 2.5510400661914503e-06,
      "loss": 0.6557,
      "step": 5258
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.470133734465315,
      "learning_rate": 2.5492328569088666e-06,
      "loss": 0.7906,
      "step": 5259
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.298661966928032,
      "learning_rate": 2.5474260689415687e-06,
      "loss": 0.8834,
      "step": 5260
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3705298468329066,
      "learning_rate": 2.5456197026001613e-06,
      "loss": 0.7622,
      "step": 5261
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2652590020783223,
      "learning_rate": 2.543813758195181e-06,
      "loss": 0.6494,
      "step": 5262
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3236422667777714,
      "learning_rate": 2.54200823603709e-06,
      "loss": 0.7236,
      "step": 5263
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4730929562267892,
      "learning_rate": 2.5402031364362772e-06,
      "loss": 0.8161,
      "step": 5264
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3117972665774673,
      "learning_rate": 2.5383984597030638e-06,
      "loss": 0.7914,
      "step": 5265
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.689869817623836,
      "learning_rate": 2.5365942061476924e-06,
      "loss": 0.7877,
      "step": 5266
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6402804921216811,
      "learning_rate": 2.5347903760803327e-06,
      "loss": 0.7908,
      "step": 5267
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6516828549094436,
      "learning_rate": 2.532986969811089e-06,
      "loss": 0.754,
      "step": 5268
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7264725311518074,
      "learning_rate": 2.5311839876499843e-06,
      "loss": 0.7378,
      "step": 5269
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3219585364709663,
      "learning_rate": 2.5293814299069742e-06,
      "loss": 0.7912,
      "step": 5270
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2606488446645223,
      "learning_rate": 2.5275792968919376e-06,
      "loss": 0.7133,
      "step": 5271
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3215139051803788,
      "learning_rate": 2.525777588914683e-06,
      "loss": 0.6744,
      "step": 5272
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4235131695576995,
      "learning_rate": 2.5239763062849424e-06,
      "loss": 0.7244,
      "step": 5273
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2859553997148159,
      "learning_rate": 2.5221754493123798e-06,
      "loss": 0.8433,
      "step": 5274
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3486625963260908,
      "learning_rate": 2.520375018306584e-06,
      "loss": 0.7732,
      "step": 5275
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6883762206785367,
      "learning_rate": 2.5185750135770694e-06,
      "loss": 0.7432,
      "step": 5276
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4016225005575425,
      "learning_rate": 2.516775435433276e-06,
      "loss": 0.7004,
      "step": 5277
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4268854768179413,
      "learning_rate": 2.514976284184575e-06,
      "loss": 0.7731,
      "step": 5278
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5047018596327852,
      "learning_rate": 2.5131775601402588e-06,
      "loss": 0.7682,
      "step": 5279
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4755809194079375,
      "learning_rate": 2.511379263609547e-06,
      "loss": 0.7978,
      "step": 5280
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2645536258166936,
      "learning_rate": 2.5095813949015934e-06,
      "loss": 0.8574,
      "step": 5281
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2904252845633852,
      "learning_rate": 2.507783954325467e-06,
      "loss": 0.7672,
      "step": 5282
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7464795098097924,
      "learning_rate": 2.505986942190173e-06,
      "loss": 0.603,
      "step": 5283
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7559899265982822,
      "learning_rate": 2.504190358804637e-06,
      "loss": 0.6331,
      "step": 5284
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3953913681902872,
      "learning_rate": 2.5023942044777114e-06,
      "loss": 0.8214,
      "step": 5285
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5544402514213471,
      "learning_rate": 2.5005984795181783e-06,
      "loss": 0.7867,
      "step": 5286
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4840762046135036,
      "learning_rate": 2.498803184234741e-06,
      "loss": 0.759,
      "step": 5287
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4816288122155048,
      "learning_rate": 2.497008318936031e-06,
      "loss": 0.7402,
      "step": 5288
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2490922363008565,
      "learning_rate": 2.49521388393061e-06,
      "loss": 0.7287,
      "step": 5289
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.443788180485834,
      "learning_rate": 2.4934198795269603e-06,
      "loss": 0.8007,
      "step": 5290
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.306525164921852,
      "learning_rate": 2.4916263060334903e-06,
      "loss": 0.6725,
      "step": 5291
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3484091147702313,
      "learning_rate": 2.48983316375854e-06,
      "loss": 0.7248,
      "step": 5292
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4575479362987784,
      "learning_rate": 2.488040453010369e-06,
      "loss": 0.8225,
      "step": 5293
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.262507910392504,
      "learning_rate": 2.486248174097165e-06,
      "loss": 0.7475,
      "step": 5294
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3362882040456465,
      "learning_rate": 2.4844563273270404e-06,
      "loss": 0.6744,
      "step": 5295
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8835311387110436,
      "learning_rate": 2.482664913008037e-06,
      "loss": 0.6121,
      "step": 5296
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3908145494450352,
      "learning_rate": 2.4808739314481196e-06,
      "loss": 0.7209,
      "step": 5297
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6696723206892634,
      "learning_rate": 2.4790833829551764e-06,
      "loss": 0.6988,
      "step": 5298
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2740755894938052,
      "learning_rate": 2.4772932678370255e-06,
      "loss": 0.7227,
      "step": 5299
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2588764793815361,
      "learning_rate": 2.475503586401407e-06,
      "loss": 0.802,
      "step": 5300
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.903271593253689,
      "learning_rate": 2.473714338955987e-06,
      "loss": 0.6477,
      "step": 5301
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4639820424552048,
      "learning_rate": 2.4719255258083614e-06,
      "loss": 0.8177,
      "step": 5302
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.957452369661961,
      "learning_rate": 2.4701371472660433e-06,
      "loss": 0.8447,
      "step": 5303
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3524657374979934,
      "learning_rate": 2.4683492036364805e-06,
      "loss": 0.7567,
      "step": 5304
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2805095321130848,
      "learning_rate": 2.466561695227039e-06,
      "loss": 0.7692,
      "step": 5305
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.520883712620318,
      "learning_rate": 2.4647746223450123e-06,
      "loss": 0.8004,
      "step": 5306
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8557015903395269,
      "learning_rate": 2.462987985297619e-06,
      "loss": 0.7101,
      "step": 5307
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3980432017562143,
      "learning_rate": 2.461201784392002e-06,
      "loss": 0.8178,
      "step": 5308
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4597955031649075,
      "learning_rate": 2.459416019935229e-06,
      "loss": 0.8132,
      "step": 5309
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.365146042848515,
      "learning_rate": 2.4576306922342956e-06,
      "loss": 0.7297,
      "step": 5310
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.275998995094752,
      "learning_rate": 2.4558458015961213e-06,
      "loss": 0.788,
      "step": 5311
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9835934528928842,
      "learning_rate": 2.4540613483275484e-06,
      "loss": 0.7903,
      "step": 5312
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6090075256804253,
      "learning_rate": 2.4522773327353446e-06,
      "loss": 0.8043,
      "step": 5313
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4049109025745286,
      "learning_rate": 2.4504937551262024e-06,
      "loss": 0.8302,
      "step": 5314
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.875855012964347,
      "learning_rate": 2.448710615806741e-06,
      "loss": 0.7163,
      "step": 5315
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1969988308185642,
      "learning_rate": 2.4469279150834996e-06,
      "loss": 0.7652,
      "step": 5316
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3580453330401092,
      "learning_rate": 2.445145653262949e-06,
      "loss": 0.714,
      "step": 5317
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2647019337390415,
      "learning_rate": 2.4433638306514794e-06,
      "loss": 0.7449,
      "step": 5318
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.547394639134293,
      "learning_rate": 2.4415824475554046e-06,
      "loss": 0.804,
      "step": 5319
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3027247247802423,
      "learning_rate": 2.439801504280968e-06,
      "loss": 0.7722,
      "step": 5320
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3236110418546911,
      "learning_rate": 2.438021001134334e-06,
      "loss": 0.6962,
      "step": 5321
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2277634800086685,
      "learning_rate": 2.436240938421591e-06,
      "loss": 0.6826,
      "step": 5322
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8347651820535736,
      "learning_rate": 2.4344613164487513e-06,
      "loss": 0.7029,
      "step": 5323
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.288934731241606,
      "learning_rate": 2.432682135521752e-06,
      "loss": 0.7174,
      "step": 5324
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.395082082448222,
      "learning_rate": 2.430903395946459e-06,
      "loss": 0.745,
      "step": 5325
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4742267084353382,
      "learning_rate": 2.429125098028655e-06,
      "loss": 0.7294,
      "step": 5326
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2111970804138086,
      "learning_rate": 2.427347242074051e-06,
      "loss": 0.7834,
      "step": 5327
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.227317842160665,
      "learning_rate": 2.4255698283882794e-06,
      "loss": 0.7464,
      "step": 5328
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3141438078002077,
      "learning_rate": 2.4237928572769016e-06,
      "loss": 0.6915,
      "step": 5329
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.092845193444028,
      "learning_rate": 2.4220163290453973e-06,
      "loss": 0.7184,
      "step": 5330
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.302375047416975,
      "learning_rate": 2.4202402439991708e-06,
      "loss": 0.8275,
      "step": 5331
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2552261323144216,
      "learning_rate": 2.4184646024435547e-06,
      "loss": 0.7427,
      "step": 5332
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3583724871495535,
      "learning_rate": 2.416689404683802e-06,
      "loss": 0.78,
      "step": 5333
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.37374549518612,
      "learning_rate": 2.4149146510250885e-06,
      "loss": 0.7205,
      "step": 5334
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5418106770499467,
      "learning_rate": 2.4131403417725156e-06,
      "loss": 0.7035,
      "step": 5335
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4381864615309967,
      "learning_rate": 2.4113664772311076e-06,
      "loss": 0.8312,
      "step": 5336
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.375249172474336,
      "learning_rate": 2.40959305770581e-06,
      "loss": 0.7221,
      "step": 5337
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2170790228029096,
      "learning_rate": 2.4078200835014983e-06,
      "loss": 0.7697,
      "step": 5338
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3599496292364028,
      "learning_rate": 2.4060475549229634e-06,
      "loss": 0.7389,
      "step": 5339
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4712513827266647,
      "learning_rate": 2.4042754722749274e-06,
      "loss": 0.7388,
      "step": 5340
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4237934360181919,
      "learning_rate": 2.40250383586203e-06,
      "loss": 0.7896,
      "step": 5341
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.305944317929744,
      "learning_rate": 2.400732645988836e-06,
      "loss": 0.8292,
      "step": 5342
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2501076487659994,
      "learning_rate": 2.398961902959833e-06,
      "loss": 0.6746,
      "step": 5343
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7167261321254235,
      "learning_rate": 2.3971916070794337e-06,
      "loss": 0.7023,
      "step": 5344
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4414017365376444,
      "learning_rate": 2.3954217586519692e-06,
      "loss": 0.768,
      "step": 5345
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.882094274840785,
      "learning_rate": 2.3936523579817005e-06,
      "loss": 0.6613,
      "step": 5346
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3383299277145801,
      "learning_rate": 2.391883405372805e-06,
      "loss": 0.8015,
      "step": 5347
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3110800779062401,
      "learning_rate": 2.3901149011293906e-06,
      "loss": 0.8521,
      "step": 5348
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5538051302120288,
      "learning_rate": 2.388346845555481e-06,
      "loss": 0.7263,
      "step": 5349
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4578539973430562,
      "learning_rate": 2.386579238955025e-06,
      "loss": 0.7021,
      "step": 5350
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9253320915824798,
      "learning_rate": 2.384812081631895e-06,
      "loss": 0.6969,
      "step": 5351
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6639568265865616,
      "learning_rate": 2.3830453738898852e-06,
      "loss": 0.7738,
      "step": 5352
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2500864177646223,
      "learning_rate": 2.3812791160327154e-06,
      "loss": 0.7781,
      "step": 5353
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3525351057111854,
      "learning_rate": 2.379513308364024e-06,
      "loss": 0.8558,
      "step": 5354
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3798101331444466,
      "learning_rate": 2.377747951187375e-06,
      "loss": 0.7414,
      "step": 5355
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8200958453474702,
      "learning_rate": 2.3759830448062497e-06,
      "loss": 0.7083,
      "step": 5356
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9600843001920936,
      "learning_rate": 2.3742185895240623e-06,
      "loss": 0.7171,
      "step": 5357
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7413946934453755,
      "learning_rate": 2.3724545856441394e-06,
      "loss": 0.7109,
      "step": 5358
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3592944388181143,
      "learning_rate": 2.370691033469735e-06,
      "loss": 0.7553,
      "step": 5359
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4493990078059926,
      "learning_rate": 2.368927933304021e-06,
      "loss": 0.8065,
      "step": 5360
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.32158160887908,
      "learning_rate": 2.3671652854500995e-06,
      "loss": 0.7948,
      "step": 5361
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8533226217936031,
      "learning_rate": 2.3654030902109877e-06,
      "loss": 0.581,
      "step": 5362
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5703357433707728,
      "learning_rate": 2.3636413478896276e-06,
      "loss": 0.8418,
      "step": 5363
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.904438814033383,
      "learning_rate": 2.361880058788883e-06,
      "loss": 0.6795,
      "step": 5364
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4695722790040766,
      "learning_rate": 2.3601192232115376e-06,
      "loss": 0.8537,
      "step": 5365
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.273762866830312,
      "learning_rate": 2.3583588414603046e-06,
      "loss": 0.7548,
      "step": 5366
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.320961579809818,
      "learning_rate": 2.3565989138378094e-06,
      "loss": 0.8151,
      "step": 5367
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4060555185809454,
      "learning_rate": 2.3548394406466074e-06,
      "loss": 0.7441,
      "step": 5368
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9828560747903382,
      "learning_rate": 2.3530804221891707e-06,
      "loss": 0.7396,
      "step": 5369
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2692831311561286,
      "learning_rate": 2.3513218587678952e-06,
      "loss": 0.8195,
      "step": 5370
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8085779539519241,
      "learning_rate": 2.349563750685099e-06,
      "loss": 0.6755,
      "step": 5371
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.296135182464106,
      "learning_rate": 2.3478060982430193e-06,
      "loss": 0.7807,
      "step": 5372
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3047482564493003,
      "learning_rate": 2.3460489017438176e-06,
      "loss": 0.6711,
      "step": 5373
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.288352579387273,
      "learning_rate": 2.3442921614895783e-06,
      "loss": 0.678,
      "step": 5374
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2816887646751798,
      "learning_rate": 2.342535877782302e-06,
      "loss": 0.7255,
      "step": 5375
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9039621933157176,
      "learning_rate": 2.340780050923918e-06,
      "loss": 0.6912,
      "step": 5376
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.391732368534112,
      "learning_rate": 2.339024681216272e-06,
      "loss": 0.7596,
      "step": 5377
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3700840262454757,
      "learning_rate": 2.3372697689611317e-06,
      "loss": 0.7758,
      "step": 5378
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3768686712756995,
      "learning_rate": 2.3355153144601873e-06,
      "loss": 0.912,
      "step": 5379
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8772612040492723,
      "learning_rate": 2.3337613180150497e-06,
      "loss": 0.7043,
      "step": 5380
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3977092637846094,
      "learning_rate": 2.3320077799272493e-06,
      "loss": 0.7163,
      "step": 5381
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1751010021273534,
      "learning_rate": 2.3302547004982434e-06,
      "loss": 0.7104,
      "step": 5382
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.363165410116041,
      "learning_rate": 2.328502080029405e-06,
      "loss": 0.6798,
      "step": 5383
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5067491801210233,
      "learning_rate": 2.326749918822028e-06,
      "loss": 0.7019,
      "step": 5384
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7776724115660583,
      "learning_rate": 2.3249982171773322e-06,
      "loss": 0.7152,
      "step": 5385
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3366112198212199,
      "learning_rate": 2.323246975396454e-06,
      "loss": 0.8284,
      "step": 5386
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4606406549013482,
      "learning_rate": 2.321496193780453e-06,
      "loss": 0.7083,
      "step": 5387
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8661193980060463,
      "learning_rate": 2.3197458726303064e-06,
      "loss": 0.6261,
      "step": 5388
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2857715073711382,
      "learning_rate": 2.3179960122469173e-06,
      "loss": 0.8217,
      "step": 5389
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1757257051342296,
      "learning_rate": 2.3162466129311074e-06,
      "loss": 0.8638,
      "step": 5390
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1821235685925744,
      "learning_rate": 2.314497674983617e-06,
      "loss": 0.7557,
      "step": 5391
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.410241793898247,
      "learning_rate": 2.3127491987051086e-06,
      "loss": 0.822,
      "step": 5392
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8297133512366451,
      "learning_rate": 2.311001184396165e-06,
      "loss": 0.6399,
      "step": 5393
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.37393494983216,
      "learning_rate": 2.3092536323572933e-06,
      "loss": 0.6579,
      "step": 5394
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4799575201316415,
      "learning_rate": 2.307506542888916e-06,
      "loss": 0.8166,
      "step": 5395
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3931819302844493,
      "learning_rate": 2.305759916291376e-06,
      "loss": 0.7399,
      "step": 5396
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8090437146468374,
      "learning_rate": 2.3040137528649424e-06,
      "loss": 0.6651,
      "step": 5397
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2971267536082354,
      "learning_rate": 2.3022680529097995e-06,
      "loss": 0.82,
      "step": 5398
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.265624597870133,
      "learning_rate": 2.300522816726053e-06,
      "loss": 0.7526,
      "step": 5399
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3336976889108576,
      "learning_rate": 2.298778044613729e-06,
      "loss": 0.7096,
      "step": 5400
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3093644681715217,
      "learning_rate": 2.2970337368727753e-06,
      "loss": 0.7141,
      "step": 5401
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3955137151160983,
      "learning_rate": 2.2952898938030554e-06,
      "loss": 0.7399,
      "step": 5402
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4769145023158137,
      "learning_rate": 2.293546515704358e-06,
      "loss": 0.725,
      "step": 5403
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6821669991300368,
      "learning_rate": 2.2918036028763928e-06,
      "loss": 0.6143,
      "step": 5404
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3627218523440425,
      "learning_rate": 2.290061155618784e-06,
      "loss": 0.7236,
      "step": 5405
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4974110875460345,
      "learning_rate": 2.2883191742310795e-06,
      "loss": 0.7772,
      "step": 5406
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5808942026442139,
      "learning_rate": 2.2865776590127447e-06,
      "loss": 0.7035,
      "step": 5407
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7926511789034693,
      "learning_rate": 2.2848366102631674e-06,
      "loss": 0.6794,
      "step": 5408
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4333165630947957,
      "learning_rate": 2.283096028281652e-06,
      "loss": 0.7707,
      "step": 5409
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8416155528409693,
      "learning_rate": 2.281355913367428e-06,
      "loss": 0.6021,
      "step": 5410
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5372705393286354,
      "learning_rate": 2.2796162658196397e-06,
      "loss": 0.7689,
      "step": 5411
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.366333519633303,
      "learning_rate": 2.2778770859373504e-06,
      "loss": 0.758,
      "step": 5412
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.318579368900042,
      "learning_rate": 2.27613837401955e-06,
      "loss": 0.8041,
      "step": 5413
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3205694459136283,
      "learning_rate": 2.2744001303651407e-06,
      "loss": 0.7171,
      "step": 5414
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6603633354755114,
      "learning_rate": 2.2726623552729473e-06,
      "loss": 0.7991,
      "step": 5415
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7703027320075537,
      "learning_rate": 2.2709250490417124e-06,
      "loss": 0.6348,
      "step": 5416
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7544129257182866,
      "learning_rate": 2.2691882119700983e-06,
      "loss": 0.7607,
      "step": 5417
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2373618625815719,
      "learning_rate": 2.2674518443566908e-06,
      "loss": 0.8505,
      "step": 5418
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.247872376610169,
      "learning_rate": 2.265715946499989e-06,
      "loss": 0.7154,
      "step": 5419
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2591799021018746,
      "learning_rate": 2.2639805186984148e-06,
      "loss": 0.7848,
      "step": 5420
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8559963409479053,
      "learning_rate": 2.2622455612503064e-06,
      "loss": 0.6985,
      "step": 5421
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8806514678616033,
      "learning_rate": 2.2605110744539266e-06,
      "loss": 0.6435,
      "step": 5422
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.444807643447216,
      "learning_rate": 2.258777058607452e-06,
      "loss": 0.6839,
      "step": 5423
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.867735470291927,
      "learning_rate": 2.257043514008978e-06,
      "loss": 0.7537,
      "step": 5424
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9084569878524518,
      "learning_rate": 2.255310440956525e-06,
      "loss": 0.6902,
      "step": 5425
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3748769667419518,
      "learning_rate": 2.2535778397480263e-06,
      "loss": 0.7574,
      "step": 5426
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2344941011792554,
      "learning_rate": 2.251845710681336e-06,
      "loss": 0.7502,
      "step": 5427
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3470707314731443,
      "learning_rate": 2.250114054054228e-06,
      "loss": 0.8092,
      "step": 5428
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.762863380908543,
      "learning_rate": 2.2483828701643933e-06,
      "loss": 0.7575,
      "step": 5429
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5683509524867973,
      "learning_rate": 2.2466521593094404e-06,
      "loss": 0.7495,
      "step": 5430
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8650004200178515,
      "learning_rate": 2.2449219217869013e-06,
      "loss": 0.6089,
      "step": 5431
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.44130468420237,
      "learning_rate": 2.243192157894225e-06,
      "loss": 0.7059,
      "step": 5432
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3583068594918044,
      "learning_rate": 2.241462867928777e-06,
      "loss": 0.7726,
      "step": 5433
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6472638482997242,
      "learning_rate": 2.239734052187841e-06,
      "loss": 0.8841,
      "step": 5434
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3602903535822757,
      "learning_rate": 2.2380057109686213e-06,
      "loss": 0.7743,
      "step": 5435
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8400766504762304,
      "learning_rate": 2.2362778445682403e-06,
      "loss": 0.712,
      "step": 5436
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3206544137329082,
      "learning_rate": 2.234550453283737e-06,
      "loss": 0.7397,
      "step": 5437
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6150784518865495,
      "learning_rate": 2.2328235374120694e-06,
      "loss": 0.7647,
      "step": 5438
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.314796790461794,
      "learning_rate": 2.231097097250115e-06,
      "loss": 0.8639,
      "step": 5439
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.296771839878465,
      "learning_rate": 2.229371133094671e-06,
      "loss": 0.6957,
      "step": 5440
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4616668318409958,
      "learning_rate": 2.22764564524245e-06,
      "loss": 0.8471,
      "step": 5441
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7591904604924602,
      "learning_rate": 2.225920633990082e-06,
      "loss": 0.6835,
      "step": 5442
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2608185208609386,
      "learning_rate": 2.2241960996341166e-06,
      "loss": 0.7575,
      "step": 5443
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3777489427791854,
      "learning_rate": 2.2224720424710222e-06,
      "loss": 0.7613,
      "step": 5444
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.329471717543302,
      "learning_rate": 2.2207484627971817e-06,
      "loss": 0.7263,
      "step": 5445
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2276839630543492,
      "learning_rate": 2.2190253609089014e-06,
      "loss": 0.7289,
      "step": 5446
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.016014901592212,
      "learning_rate": 2.217302737102402e-06,
      "loss": 0.7546,
      "step": 5447
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.60911204182805,
      "learning_rate": 2.2155805916738215e-06,
      "loss": 0.6237,
      "step": 5448
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2518083927714634,
      "learning_rate": 2.2138589249192156e-06,
      "loss": 0.752,
      "step": 5449
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2828777030555414,
      "learning_rate": 2.212137737134562e-06,
      "loss": 0.6808,
      "step": 5450
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.255302117896302,
      "learning_rate": 2.2104170286157506e-06,
      "loss": 0.8045,
      "step": 5451
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3225610770856802,
      "learning_rate": 2.20869679965859e-06,
      "loss": 0.6498,
      "step": 5452
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4197596465275057,
      "learning_rate": 2.206977050558811e-06,
      "loss": 0.8004,
      "step": 5453
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9020737800544577,
      "learning_rate": 2.205257781612057e-06,
      "loss": 0.678,
      "step": 5454
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3742784161489092,
      "learning_rate": 2.2035389931138896e-06,
      "loss": 0.806,
      "step": 5455
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.833682943343842,
      "learning_rate": 2.201820685359788e-06,
      "loss": 0.686,
      "step": 5456
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4819698274181665,
      "learning_rate": 2.20010285864515e-06,
      "loss": 0.7234,
      "step": 5457
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3796190098688856,
      "learning_rate": 2.198385513265289e-06,
      "loss": 0.7409,
      "step": 5458
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4006182394486344,
      "learning_rate": 2.1966686495154375e-06,
      "loss": 0.8117,
      "step": 5459
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.295266746679111,
      "learning_rate": 2.194952267690744e-06,
      "loss": 0.7547,
      "step": 5460
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.878109019248761,
      "learning_rate": 2.193236368086275e-06,
      "loss": 0.7536,
      "step": 5461
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4616715777618257,
      "learning_rate": 2.191520950997014e-06,
      "loss": 0.8141,
      "step": 5462
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4950890342732053,
      "learning_rate": 2.1898060167178604e-06,
      "loss": 0.7324,
      "step": 5463
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4126805568984555,
      "learning_rate": 2.18809156554363e-06,
      "loss": 0.7167,
      "step": 5464
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5126243595067521,
      "learning_rate": 2.1863775977690588e-06,
      "loss": 0.8219,
      "step": 5465
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7600043841492393,
      "learning_rate": 2.1846641136887947e-06,
      "loss": 0.5947,
      "step": 5466
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3981047799428905,
      "learning_rate": 2.182951113597408e-06,
      "loss": 0.7624,
      "step": 5467
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5268464196630083,
      "learning_rate": 2.1812385977893844e-06,
      "loss": 0.7937,
      "step": 5468
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3805357559994804,
      "learning_rate": 2.1795265665591236e-06,
      "loss": 0.8498,
      "step": 5469
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3529082079571642,
      "learning_rate": 2.177815020200944e-06,
      "loss": 0.7067,
      "step": 5470
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8885057471553971,
      "learning_rate": 2.17610395900908e-06,
      "loss": 0.6594,
      "step": 5471
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.506150607487622,
      "learning_rate": 2.174393383277683e-06,
      "loss": 0.8065,
      "step": 5472
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2886670496876205,
      "learning_rate": 2.1726832933008185e-06,
      "loss": 0.6926,
      "step": 5473
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3100817404156484,
      "learning_rate": 2.170973689372475e-06,
      "loss": 0.8413,
      "step": 5474
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7714833016416528,
      "learning_rate": 2.1692645717865515e-06,
      "loss": 0.6814,
      "step": 5475
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3210552701363798,
      "learning_rate": 2.1675559408368623e-06,
      "loss": 0.7052,
      "step": 5476
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3384850896916494,
      "learning_rate": 2.165847796817145e-06,
      "loss": 0.7227,
      "step": 5477
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2077615217035436,
      "learning_rate": 2.164140140021049e-06,
      "loss": 0.7175,
      "step": 5478
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4588259377580743,
      "learning_rate": 2.1624329707421374e-06,
      "loss": 0.7122,
      "step": 5479
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4635471968108549,
      "learning_rate": 2.1607262892738956e-06,
      "loss": 0.8185,
      "step": 5480
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.662938876059917,
      "learning_rate": 2.1590200959097173e-06,
      "loss": 0.736,
      "step": 5481
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.531152200370719,
      "learning_rate": 2.157314390942923e-06,
      "loss": 0.7515,
      "step": 5482
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4517260794213886,
      "learning_rate": 2.1556091746667392e-06,
      "loss": 0.7802,
      "step": 5483
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6463616335482005,
      "learning_rate": 2.1539044473743136e-06,
      "loss": 0.8219,
      "step": 5484
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2861138129593093,
      "learning_rate": 2.152200209358708e-06,
      "loss": 0.7064,
      "step": 5485
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.837371884014507,
      "learning_rate": 2.1504964609128994e-06,
      "loss": 0.6257,
      "step": 5486
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6503503115732534,
      "learning_rate": 2.1487932023297853e-06,
      "loss": 0.7405,
      "step": 5487
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6118176622432903,
      "learning_rate": 2.147090433902172e-06,
      "loss": 0.7459,
      "step": 5488
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.89001917020629,
      "learning_rate": 2.1453881559227883e-06,
      "loss": 0.6367,
      "step": 5489
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3122803305333641,
      "learning_rate": 2.143686368684274e-06,
      "loss": 0.8369,
      "step": 5490
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.33263785999752,
      "learning_rate": 2.141985072479187e-06,
      "loss": 0.8324,
      "step": 5491
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3697883372759434,
      "learning_rate": 2.1402842675999978e-06,
      "loss": 0.7698,
      "step": 5492
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.836079406425686,
      "learning_rate": 2.1385839543390967e-06,
      "loss": 0.631,
      "step": 5493
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5594932674949529,
      "learning_rate": 2.136884132988784e-06,
      "loss": 0.7416,
      "step": 5494
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6895823930691032,
      "learning_rate": 2.1351848038412832e-06,
      "loss": 0.6715,
      "step": 5495
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3339574072878548,
      "learning_rate": 2.1334859671887236e-06,
      "loss": 0.7948,
      "step": 5496
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4171451059371507,
      "learning_rate": 2.1317876233231606e-06,
      "loss": 0.7152,
      "step": 5497
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5122194886420532,
      "learning_rate": 2.1300897725365555e-06,
      "loss": 0.702,
      "step": 5498
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3657905802648083,
      "learning_rate": 2.12839241512079e-06,
      "loss": 0.7252,
      "step": 5499
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4384604457390875,
      "learning_rate": 2.1266955513676584e-06,
      "loss": 0.8027,
      "step": 5500
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8204742296870837,
      "learning_rate": 2.124999181568872e-06,
      "loss": 0.7467,
      "step": 5501
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.248230085381593,
      "learning_rate": 2.1233033060160542e-06,
      "loss": 0.8134,
      "step": 5502
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3125588306932043,
      "learning_rate": 2.121607925000749e-06,
      "loss": 0.7687,
      "step": 5503
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5298490678057515,
      "learning_rate": 2.1199130388144098e-06,
      "loss": 0.7645,
      "step": 5504
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3056076206236225,
      "learning_rate": 2.1182186477484094e-06,
      "loss": 0.7067,
      "step": 5505
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3806734455575136,
      "learning_rate": 2.1165247520940317e-06,
      "loss": 0.6992,
      "step": 5506
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3472919057645731,
      "learning_rate": 2.114831352142478e-06,
      "loss": 0.8217,
      "step": 5507
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2592308019227023,
      "learning_rate": 2.1131384481848614e-06,
      "loss": 0.8511,
      "step": 5508
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3041813435023613,
      "learning_rate": 2.111446040512212e-06,
      "loss": 0.7562,
      "step": 5509
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2817643023725103,
      "learning_rate": 2.1097541294154773e-06,
      "loss": 0.7695,
      "step": 5510
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.34537769186271,
      "learning_rate": 2.108062715185514e-06,
      "loss": 0.8087,
      "step": 5511
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4322089889304699,
      "learning_rate": 2.1063717981130952e-06,
      "loss": 0.7101,
      "step": 5512
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.29687369811538,
      "learning_rate": 2.10468137848891e-06,
      "loss": 0.7058,
      "step": 5513
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3321518586543557,
      "learning_rate": 2.102991456603562e-06,
      "loss": 0.7401,
      "step": 5514
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.013618645542914,
      "learning_rate": 2.1013020327475683e-06,
      "loss": 0.7289,
      "step": 5515
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2703897771852657,
      "learning_rate": 2.099613107211359e-06,
      "loss": 0.73,
      "step": 5516
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4993356138455851,
      "learning_rate": 2.0979246802852794e-06,
      "loss": 0.8191,
      "step": 5517
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.595715238855854,
      "learning_rate": 2.096236752259592e-06,
      "loss": 0.7256,
      "step": 5518
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3137535841604682,
      "learning_rate": 2.0945493234244714e-06,
      "loss": 0.6951,
      "step": 5519
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3245280276715539,
      "learning_rate": 2.0928623940700044e-06,
      "loss": 0.7697,
      "step": 5520
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3973948161849747,
      "learning_rate": 2.091175964486194e-06,
      "loss": 0.7466,
      "step": 5521
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3657844571679179,
      "learning_rate": 2.0894900349629576e-06,
      "loss": 0.7434,
      "step": 5522
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4050927520185703,
      "learning_rate": 2.087804605790124e-06,
      "loss": 0.8085,
      "step": 5523
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.643123749258105,
      "learning_rate": 2.08611967725744e-06,
      "loss": 0.8717,
      "step": 5524
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3039577146178085,
      "learning_rate": 2.0844352496545652e-06,
      "loss": 0.7555,
      "step": 5525
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.44423343237518,
      "learning_rate": 2.0827513232710716e-06,
      "loss": 0.8622,
      "step": 5526
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.321169499269285,
      "learning_rate": 2.081067898396445e-06,
      "loss": 0.7297,
      "step": 5527
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3831163930208228,
      "learning_rate": 2.0793849753200855e-06,
      "loss": 0.7877,
      "step": 5528
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4305389885986322,
      "learning_rate": 2.077702554331308e-06,
      "loss": 0.7462,
      "step": 5529
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.273623772733337,
      "learning_rate": 2.0760206357193373e-06,
      "loss": 0.7953,
      "step": 5530
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.421855106128933,
      "learning_rate": 2.0743392197733193e-06,
      "loss": 0.8069,
      "step": 5531
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4160265727173504,
      "learning_rate": 2.0726583067823046e-06,
      "loss": 0.809,
      "step": 5532
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4782846844032023,
      "learning_rate": 2.0709778970352657e-06,
      "loss": 0.7514,
      "step": 5533
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4912763472160468,
      "learning_rate": 2.069297990821082e-06,
      "loss": 0.7088,
      "step": 5534
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.418635807301236,
      "learning_rate": 2.0676185884285495e-06,
      "loss": 0.6961,
      "step": 5535
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.377394143112526,
      "learning_rate": 2.0659396901463764e-06,
      "loss": 0.8153,
      "step": 5536
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.467435740702294,
      "learning_rate": 2.064261296263185e-06,
      "loss": 0.785,
      "step": 5537
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3119152884894136,
      "learning_rate": 2.0625834070675094e-06,
      "loss": 0.7652,
      "step": 5538
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.554541308522712,
      "learning_rate": 2.0609060228478017e-06,
      "loss": 0.7004,
      "step": 5539
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.42221747413887,
      "learning_rate": 2.0592291438924213e-06,
      "loss": 0.6819,
      "step": 5540
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3327347719133475,
      "learning_rate": 2.0575527704896414e-06,
      "loss": 0.806,
      "step": 5541
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2503782426950492,
      "learning_rate": 2.055876902927654e-06,
      "loss": 0.6876,
      "step": 5542
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3678740109994096,
      "learning_rate": 2.0542015414945577e-06,
      "loss": 0.759,
      "step": 5543
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4144573035263293,
      "learning_rate": 2.0525266864783676e-06,
      "loss": 0.7218,
      "step": 5544
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8592065963968647,
      "learning_rate": 2.050852338167008e-06,
      "loss": 0.6261,
      "step": 5545
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3851699956205827,
      "learning_rate": 2.049178496848323e-06,
      "loss": 0.8261,
      "step": 5546
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3127693983222652,
      "learning_rate": 2.0475051628100635e-06,
      "loss": 0.7561,
      "step": 5547
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8937293458886028,
      "learning_rate": 2.045832336339894e-06,
      "loss": 0.6416,
      "step": 5548
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.419668521264381,
      "learning_rate": 2.044160017725394e-06,
      "loss": 0.8134,
      "step": 5549
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3523395852823086,
      "learning_rate": 2.042488207254054e-06,
      "loss": 0.7605,
      "step": 5550
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8525483203385024,
      "learning_rate": 2.040816905213276e-06,
      "loss": 0.6608,
      "step": 5551
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7970751996752882,
      "learning_rate": 2.0391461118903788e-06,
      "loss": 0.6883,
      "step": 5552
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.664209340351805,
      "learning_rate": 2.0374758275725893e-06,
      "loss": 0.7583,
      "step": 5553
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.352918066254143,
      "learning_rate": 2.0358060525470507e-06,
      "loss": 0.82,
      "step": 5554
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.48626205769909,
      "learning_rate": 2.0341367871008154e-06,
      "loss": 0.7372,
      "step": 5555
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5806984299148685,
      "learning_rate": 2.0324680315208505e-06,
      "loss": 0.6783,
      "step": 5556
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2269119150295469,
      "learning_rate": 2.0307997860940333e-06,
      "loss": 0.7614,
      "step": 5557
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8724541651894815,
      "learning_rate": 2.0291320511071544e-06,
      "loss": 0.6186,
      "step": 5558
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2381350674768337,
      "learning_rate": 2.0274648268469154e-06,
      "loss": 0.743,
      "step": 5559
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.77222904603695,
      "learning_rate": 2.025798113599933e-06,
      "loss": 0.6694,
      "step": 5560
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5322838750451135,
      "learning_rate": 2.0241319116527376e-06,
      "loss": 0.6864,
      "step": 5561
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.316110434931232,
      "learning_rate": 2.022466221291765e-06,
      "loss": 0.6966,
      "step": 5562
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3927854962452662,
      "learning_rate": 2.0208010428033675e-06,
      "loss": 0.691,
      "step": 5563
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5345684724758943,
      "learning_rate": 2.0191363764738087e-06,
      "loss": 0.8072,
      "step": 5564
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3253261127695204,
      "learning_rate": 2.017472222589264e-06,
      "loss": 0.7762,
      "step": 5565
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.93304046965223,
      "learning_rate": 2.0158085814358187e-06,
      "loss": 0.7863,
      "step": 5566
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2487291489446457,
      "learning_rate": 2.014145453299476e-06,
      "loss": 0.7861,
      "step": 5567
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.7953322655214117,
      "learning_rate": 2.012482838466145e-06,
      "loss": 0.7369,
      "step": 5568
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2373231361537826,
      "learning_rate": 2.010820737221646e-06,
      "loss": 0.6622,
      "step": 5569
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.249179132406768,
      "learning_rate": 2.0091591498517184e-06,
      "loss": 0.8515,
      "step": 5570
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3870810581762576,
      "learning_rate": 2.007498076642005e-06,
      "loss": 0.8041,
      "step": 5571
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2954314631521988,
      "learning_rate": 2.0058375178780644e-06,
      "loss": 0.801,
      "step": 5572
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.600884430146354,
      "learning_rate": 2.004177473845366e-06,
      "loss": 0.6973,
      "step": 5573
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2755547284432893,
      "learning_rate": 2.0025179448292886e-06,
      "loss": 0.8187,
      "step": 5574
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8667210352609789,
      "learning_rate": 2.000858931115128e-06,
      "loss": 0.713,
      "step": 5575
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3303834986159688,
      "learning_rate": 1.999200432988086e-06,
      "loss": 0.7698,
      "step": 5576
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4462461037569154,
      "learning_rate": 1.997542450733278e-06,
      "loss": 0.7332,
      "step": 5577
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3768553678845221,
      "learning_rate": 1.9958849846357287e-06,
      "loss": 0.7419,
      "step": 5578
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.736022714625748,
      "learning_rate": 1.994228034980378e-06,
      "loss": 0.8652,
      "step": 5579
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3009243387947282,
      "learning_rate": 1.992571602052075e-06,
      "loss": 0.829,
      "step": 5580
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2783165022150031,
      "learning_rate": 1.9909156861355767e-06,
      "loss": 0.7198,
      "step": 5581
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5593960607915462,
      "learning_rate": 1.9892602875155582e-06,
      "loss": 0.7215,
      "step": 5582
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4594541303323274,
      "learning_rate": 1.9876054064765993e-06,
      "loss": 0.6914,
      "step": 5583
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4398237533191502,
      "learning_rate": 1.9859510433031943e-06,
      "loss": 0.6939,
      "step": 5584
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.387355099628399,
      "learning_rate": 1.984297198279746e-06,
      "loss": 0.7244,
      "step": 5585
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6388464413334967,
      "learning_rate": 1.982643871690571e-06,
      "loss": 0.8004,
      "step": 5586
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1987144700517693,
      "learning_rate": 1.980991063819893e-06,
      "loss": 0.7213,
      "step": 5587
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.612157110467907,
      "learning_rate": 1.9793387749518517e-06,
      "loss": 0.7123,
      "step": 5588
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2845655591412428,
      "learning_rate": 1.9776870053704917e-06,
      "loss": 0.7703,
      "step": 5589
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.274414823989568,
      "learning_rate": 1.976035755359775e-06,
      "loss": 0.7123,
      "step": 5590
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3106755030555663,
      "learning_rate": 1.974385025203569e-06,
      "loss": 0.6794,
      "step": 5591
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3404620284208284,
      "learning_rate": 1.9727348151856535e-06,
      "loss": 0.78,
      "step": 5592
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.391511443556897,
      "learning_rate": 1.9710851255897173e-06,
      "loss": 0.7568,
      "step": 5593
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9073542021949946,
      "learning_rate": 1.969435956699363e-06,
      "loss": 0.6593,
      "step": 5594
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3599634815995594,
      "learning_rate": 1.967787308798099e-06,
      "loss": 0.7724,
      "step": 5595
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7703576897693354,
      "learning_rate": 1.966139182169351e-06,
      "loss": 0.6215,
      "step": 5596
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8259598004541163,
      "learning_rate": 1.9644915770964472e-06,
      "loss": 0.6769,
      "step": 5597
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5337385409782769,
      "learning_rate": 1.9628444938626336e-06,
      "loss": 0.7805,
      "step": 5598
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3923532349387098,
      "learning_rate": 1.9611979327510617e-06,
      "loss": 0.7841,
      "step": 5599
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3053400575118845,
      "learning_rate": 1.9595518940447933e-06,
      "loss": 0.6548,
      "step": 5600
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3438217756680335,
      "learning_rate": 1.9579063780268026e-06,
      "loss": 0.7402,
      "step": 5601
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.312384738425409,
      "learning_rate": 1.9562613849799704e-06,
      "loss": 0.7156,
      "step": 5602
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3205603935394288,
      "learning_rate": 1.9546169151870943e-06,
      "loss": 0.7942,
      "step": 5603
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4107077970524569,
      "learning_rate": 1.9529729689308756e-06,
      "loss": 0.816,
      "step": 5604
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8440319578376458,
      "learning_rate": 1.9513295464939274e-06,
      "loss": 0.6748,
      "step": 5605
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3433074501461169,
      "learning_rate": 1.9496866481587717e-06,
      "loss": 0.7823,
      "step": 5606
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3655393508587472,
      "learning_rate": 1.9480442742078455e-06,
      "loss": 0.8636,
      "step": 5607
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4074225365563249,
      "learning_rate": 1.9464024249234895e-06,
      "loss": 0.8129,
      "step": 5608
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8230864782822553,
      "learning_rate": 1.9447611005879573e-06,
      "loss": 0.7068,
      "step": 5609
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4132149016313,
      "learning_rate": 1.9431203014834093e-06,
      "loss": 0.6995,
      "step": 5610
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4464488140330793,
      "learning_rate": 1.9414800278919223e-06,
      "loss": 0.7377,
      "step": 5611
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3017814229198745,
      "learning_rate": 1.9398402800954746e-06,
      "loss": 0.7815,
      "step": 5612
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1819102270203266,
      "learning_rate": 1.9382010583759604e-06,
      "loss": 0.8067,
      "step": 5613
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0153449715688,
      "learning_rate": 1.936562363015179e-06,
      "loss": 0.8237,
      "step": 5614
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3632202671737497,
      "learning_rate": 1.9349241942948405e-06,
      "loss": 0.8248,
      "step": 5615
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3944935328637544,
      "learning_rate": 1.9332865524965677e-06,
      "loss": 0.8054,
      "step": 5616
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6114881763405282,
      "learning_rate": 1.9316494379018876e-06,
      "loss": 0.7804,
      "step": 5617
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4086807707786004,
      "learning_rate": 1.9300128507922417e-06,
      "loss": 0.8117,
      "step": 5618
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2664776190205247,
      "learning_rate": 1.9283767914489777e-06,
      "loss": 0.761,
      "step": 5619
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.248365289490369,
      "learning_rate": 1.926741260153352e-06,
      "loss": 0.7217,
      "step": 5620
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2976249608388404,
      "learning_rate": 1.925106257186532e-06,
      "loss": 0.7683,
      "step": 5621
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2982213093941077,
      "learning_rate": 1.923471782829594e-06,
      "loss": 0.6986,
      "step": 5622
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2539179643375808,
      "learning_rate": 1.921837837363521e-06,
      "loss": 0.6778,
      "step": 5623
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8715693905008348,
      "learning_rate": 1.92020442106921e-06,
      "loss": 0.6363,
      "step": 5624
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2280078092959918,
      "learning_rate": 1.918571534227462e-06,
      "loss": 0.7278,
      "step": 5625
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4853955069856146,
      "learning_rate": 1.9169391771189915e-06,
      "loss": 0.7953,
      "step": 5626
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.560399059471543,
      "learning_rate": 1.915307350024419e-06,
      "loss": 0.7434,
      "step": 5627
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2441462333678026,
      "learning_rate": 1.913676053224273e-06,
      "loss": 0.7307,
      "step": 5628
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4555268830166876,
      "learning_rate": 1.9120452869989943e-06,
      "loss": 0.7348,
      "step": 5629
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.065718385097586,
      "learning_rate": 1.9104150516289283e-06,
      "loss": 0.7858,
      "step": 5630
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3058903649276064,
      "learning_rate": 1.9087853473943313e-06,
      "loss": 0.7389,
      "step": 5631
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2410324210028738,
      "learning_rate": 1.9071561745753715e-06,
      "loss": 0.6634,
      "step": 5632
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9041708550596113,
      "learning_rate": 1.90552753345212e-06,
      "loss": 0.8025,
      "step": 5633
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4250060690795994,
      "learning_rate": 1.9038994243045582e-06,
      "loss": 0.7013,
      "step": 5634
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8653985873296466,
      "learning_rate": 1.90227184741258e-06,
      "loss": 0.7036,
      "step": 5635
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.43721005847259,
      "learning_rate": 1.9006448030559832e-06,
      "loss": 0.7983,
      "step": 5636
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2776639445145266,
      "learning_rate": 1.899018291514476e-06,
      "loss": 0.8286,
      "step": 5637
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3908414670540532,
      "learning_rate": 1.897392313067672e-06,
      "loss": 0.7538,
      "step": 5638
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3258397718271915,
      "learning_rate": 1.8957668679950997e-06,
      "loss": 0.7458,
      "step": 5639
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7584176762849943,
      "learning_rate": 1.8941419565761903e-06,
      "loss": 0.8399,
      "step": 5640
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.518421948534142,
      "learning_rate": 1.892517579090285e-06,
      "loss": 0.789,
      "step": 5641
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3088686624921848,
      "learning_rate": 1.8908937358166323e-06,
      "loss": 0.7517,
      "step": 5642
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4530520523714756,
      "learning_rate": 1.8892704270343887e-06,
      "loss": 0.7994,
      "step": 5643
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.31248162040613,
      "learning_rate": 1.8876476530226235e-06,
      "loss": 0.7209,
      "step": 5644
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5124486898801335,
      "learning_rate": 1.8860254140603063e-06,
      "loss": 0.7207,
      "step": 5645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7570771624094238,
      "learning_rate": 1.8844037104263225e-06,
      "loss": 0.6497,
      "step": 5646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8899924172676822,
      "learning_rate": 1.8827825423994595e-06,
      "loss": 0.7133,
      "step": 5647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9557373384819562,
      "learning_rate": 1.8811619102584155e-06,
      "loss": 0.6871,
      "step": 5648
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3509748386986962,
      "learning_rate": 1.8795418142817962e-06,
      "loss": 0.7309,
      "step": 5649
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.4188116239959103,
      "learning_rate": 1.877922254748114e-06,
      "loss": 0.7105,
      "step": 5650
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5084376289793506,
      "learning_rate": 1.876303231935791e-06,
      "loss": 0.7357,
      "step": 5651
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3767130749938643,
      "learning_rate": 1.8746847461231533e-06,
      "loss": 0.793,
      "step": 5652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8986502079264147,
      "learning_rate": 1.8730667975884398e-06,
      "loss": 0.6906,
      "step": 5653
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8247183326055831,
      "learning_rate": 1.8714493866097955e-06,
      "loss": 0.667,
      "step": 5654
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.286623874803463,
      "learning_rate": 1.8698325134652711e-06,
      "loss": 0.7694,
      "step": 5655
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.467818271230568,
      "learning_rate": 1.8682161784328262e-06,
      "loss": 0.6898,
      "step": 5656
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5529584123816802,
      "learning_rate": 1.8666003817903267e-06,
      "loss": 0.8101,
      "step": 5657
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8459913915269548,
      "learning_rate": 1.8649851238155465e-06,
      "loss": 0.6801,
      "step": 5658
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.077249882241598,
      "learning_rate": 1.8633704047861667e-06,
      "loss": 0.8513,
      "step": 5659
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1526439651409934,
      "learning_rate": 1.8617562249797788e-06,
      "loss": 0.8558,
      "step": 5660
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.98449508859404,
      "learning_rate": 1.8601425846738775e-06,
      "loss": 0.8512,
      "step": 5661
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8688232359645467,
      "learning_rate": 1.858529484145864e-06,
      "loss": 0.6976,
      "step": 5662
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3585669216931726,
      "learning_rate": 1.8569169236730533e-06,
      "loss": 0.7058,
      "step": 5663
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2611459816015729,
      "learning_rate": 1.8553049035326615e-06,
      "loss": 0.7628,
      "step": 5664
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.236210243378992,
      "learning_rate": 1.8536934240018129e-06,
      "loss": 0.7632,
      "step": 5665
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.751870164982865,
      "learning_rate": 1.852082485357538e-06,
      "loss": 0.5962,
      "step": 5666
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3687436618448574,
      "learning_rate": 1.8504720878767797e-06,
      "loss": 0.7746,
      "step": 5667
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.239681360579283,
      "learning_rate": 1.8488622318363814e-06,
      "loss": 0.6973,
      "step": 5668
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9385070119038771,
      "learning_rate": 1.847252917513097e-06,
      "loss": 0.5933,
      "step": 5669
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2990156837205107,
      "learning_rate": 1.845644145183586e-06,
      "loss": 0.7749,
      "step": 5670
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7317314895284792,
      "learning_rate": 1.844035915124413e-06,
      "loss": 0.8329,
      "step": 5671
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3880809675140793,
      "learning_rate": 1.8424282276120547e-06,
      "loss": 0.7111,
      "step": 5672
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3137167374066385,
      "learning_rate": 1.84082108292289e-06,
      "loss": 0.7668,
      "step": 5673
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3702034137415346,
      "learning_rate": 1.8392144813332041e-06,
      "loss": 0.7579,
      "step": 5674
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2223514912251607,
      "learning_rate": 1.8376084231191932e-06,
      "loss": 0.6445,
      "step": 5675
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2989479618810427,
      "learning_rate": 1.8360029085569558e-06,
      "loss": 0.7486,
      "step": 5676
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3626719923022603,
      "learning_rate": 1.8343979379224991e-06,
      "loss": 0.8055,
      "step": 5677
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6369796413407511,
      "learning_rate": 1.832793511491735e-06,
      "loss": 0.7918,
      "step": 5678
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8384548768546164,
      "learning_rate": 1.831189629540484e-06,
      "loss": 0.5961,
      "step": 5679
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.76505650113712,
      "learning_rate": 1.8295862923444702e-06,
      "loss": 0.5656,
      "step": 5680
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1897775168567688,
      "learning_rate": 1.8279835001793272e-06,
      "loss": 0.7283,
      "step": 5681
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4187901732402028,
      "learning_rate": 1.8263812533205955e-06,
      "loss": 0.7058,
      "step": 5682
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1588180358452793,
      "learning_rate": 1.8247795520437177e-06,
      "loss": 0.7942,
      "step": 5683
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3377084466146316,
      "learning_rate": 1.8231783966240458e-06,
      "loss": 0.7494,
      "step": 5684
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2497264402149146,
      "learning_rate": 1.8215777873368363e-06,
      "loss": 0.7843,
      "step": 5685
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2148786841592205,
      "learning_rate": 1.8199777244572525e-06,
      "loss": 0.7577,
      "step": 5686
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.424549089922762,
      "learning_rate": 1.8183782082603618e-06,
      "loss": 0.6341,
      "step": 5687
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1876155733700708,
      "learning_rate": 1.8167792390211435e-06,
      "loss": 0.7513,
      "step": 5688
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4773277374236655,
      "learning_rate": 1.8151808170144751e-06,
      "loss": 0.7053,
      "step": 5689
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1843674252331515,
      "learning_rate": 1.813582942515148e-06,
      "loss": 0.7256,
      "step": 5690
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3397528844865303,
      "learning_rate": 1.8119856157978534e-06,
      "loss": 0.5813,
      "step": 5691
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.37742585099615,
      "learning_rate": 1.8103888371371898e-06,
      "loss": 0.8475,
      "step": 5692
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8626717926544304,
      "learning_rate": 1.8087926068076622e-06,
      "loss": 0.7467,
      "step": 5693
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4173940719603413,
      "learning_rate": 1.8071969250836813e-06,
      "loss": 0.8297,
      "step": 5694
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3384887204078915,
      "learning_rate": 1.8056017922395607e-06,
      "loss": 0.7887,
      "step": 5695
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4426296694876826,
      "learning_rate": 1.8040072085495276e-06,
      "loss": 0.8361,
      "step": 5696
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5211388157070653,
      "learning_rate": 1.8024131742877054e-06,
      "loss": 0.7952,
      "step": 5697
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3078321257558079,
      "learning_rate": 1.8008196897281287e-06,
      "loss": 0.7879,
      "step": 5698
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4496480254780384,
      "learning_rate": 1.7992267551447334e-06,
      "loss": 0.7958,
      "step": 5699
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.550132669892018,
      "learning_rate": 1.7976343708113675e-06,
      "loss": 0.7935,
      "step": 5700
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8034118539884391,
      "learning_rate": 1.7960425370017782e-06,
      "loss": 0.7025,
      "step": 5701
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3192351835793548,
      "learning_rate": 1.794451253989618e-06,
      "loss": 0.8116,
      "step": 5702
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2940053919489802,
      "learning_rate": 1.7928605220484513e-06,
      "loss": 0.7302,
      "step": 5703
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.303272811678168,
      "learning_rate": 1.7912703414517413e-06,
      "loss": 0.6876,
      "step": 5704
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2799571043014444,
      "learning_rate": 1.7896807124728582e-06,
      "loss": 0.7207,
      "step": 5705
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4005114744507983,
      "learning_rate": 1.788091635385078e-06,
      "loss": 0.7429,
      "step": 5706
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4768898366230012,
      "learning_rate": 1.7865031104615809e-06,
      "loss": 0.7369,
      "step": 5707
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8364333744052899,
      "learning_rate": 1.784915137975452e-06,
      "loss": 0.685,
      "step": 5708
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4052920480011957,
      "learning_rate": 1.783327718199685e-06,
      "loss": 0.7457,
      "step": 5709
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6443148931097233,
      "learning_rate": 1.7817408514071722e-06,
      "loss": 0.6382,
      "step": 5710
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8183823028977274,
      "learning_rate": 1.780154537870718e-06,
      "loss": 0.6666,
      "step": 5711
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7086594268889386,
      "learning_rate": 1.7785687778630268e-06,
      "loss": 0.7635,
      "step": 5712
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5069046364726275,
      "learning_rate": 1.776983571656708e-06,
      "loss": 0.8235,
      "step": 5713
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2829038869215843,
      "learning_rate": 1.7753989195242782e-06,
      "loss": 0.7042,
      "step": 5714
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4725077583469968,
      "learning_rate": 1.7738148217381568e-06,
      "loss": 0.818,
      "step": 5715
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5937969110796455,
      "learning_rate": 1.772231278570667e-06,
      "loss": 0.7398,
      "step": 5716
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3775096851520352,
      "learning_rate": 1.7706482902940397e-06,
      "loss": 0.8198,
      "step": 5717
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3596092667392994,
      "learning_rate": 1.7690658571804109e-06,
      "loss": 0.7647,
      "step": 5718
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8095265205419651,
      "learning_rate": 1.7674839795018173e-06,
      "loss": 0.7034,
      "step": 5719
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3462191771372516,
      "learning_rate": 1.7659026575302025e-06,
      "loss": 0.6983,
      "step": 5720
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3654512682893958,
      "learning_rate": 1.764321891537414e-06,
      "loss": 0.793,
      "step": 5721
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3060044039500311,
      "learning_rate": 1.7627416817952032e-06,
      "loss": 0.7669,
      "step": 5722
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7840234608500902,
      "learning_rate": 1.7611620285752246e-06,
      "loss": 0.6901,
      "step": 5723
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.953080862776942,
      "learning_rate": 1.7595829321490437e-06,
      "loss": 0.7466,
      "step": 5724
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.289719392129933,
      "learning_rate": 1.7580043927881224e-06,
      "loss": 0.8195,
      "step": 5725
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3892097721090917,
      "learning_rate": 1.756426410763829e-06,
      "loss": 0.7224,
      "step": 5726
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5008617084532598,
      "learning_rate": 1.7548489863474393e-06,
      "loss": 0.6582,
      "step": 5727
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.349281400956698,
      "learning_rate": 1.753272119810131e-06,
      "loss": 0.8519,
      "step": 5728
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6427903465834066,
      "learning_rate": 1.7516958114229837e-06,
      "loss": 0.7256,
      "step": 5729
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.506265266723014,
      "learning_rate": 1.7501200614569847e-06,
      "loss": 0.7709,
      "step": 5730
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4159865907149194,
      "learning_rate": 1.7485448701830205e-06,
      "loss": 0.6979,
      "step": 5731
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4078147154087726,
      "learning_rate": 1.7469702378718894e-06,
      "loss": 0.713,
      "step": 5732
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3003879917315895,
      "learning_rate": 1.7453961647942868e-06,
      "loss": 0.812,
      "step": 5733
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3118891333975362,
      "learning_rate": 1.743822651220814e-06,
      "loss": 0.7161,
      "step": 5734
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3159673076345153,
      "learning_rate": 1.7422496974219761e-06,
      "loss": 0.7165,
      "step": 5735
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4072111223089132,
      "learning_rate": 1.7406773036681807e-06,
      "loss": 0.7098,
      "step": 5736
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.240908392011209,
      "learning_rate": 1.7391054702297439e-06,
      "loss": 0.7123,
      "step": 5737
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2312975346695303,
      "learning_rate": 1.737534197376879e-06,
      "loss": 0.8074,
      "step": 5738
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.887349669942999,
      "learning_rate": 1.7359634853797081e-06,
      "loss": 0.6287,
      "step": 5739
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7460283851419521,
      "learning_rate": 1.7343933345082547e-06,
      "loss": 0.6953,
      "step": 5740
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.270766867340872,
      "learning_rate": 1.7328237450324454e-06,
      "loss": 0.7551,
      "step": 5741
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.356417919288171,
      "learning_rate": 1.73125471722211e-06,
      "loss": 0.7367,
      "step": 5742
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2843563705828198,
      "learning_rate": 1.7296862513469836e-06,
      "loss": 0.825,
      "step": 5743
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2963318657902463,
      "learning_rate": 1.7281183476767016e-06,
      "loss": 0.69,
      "step": 5744
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3454948237073878,
      "learning_rate": 1.7265510064808084e-06,
      "loss": 0.7229,
      "step": 5745
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3518525354410371,
      "learning_rate": 1.7249842280287442e-06,
      "loss": 0.7974,
      "step": 5746
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2744937022288882,
      "learning_rate": 1.7234180125898608e-06,
      "loss": 0.8545,
      "step": 5747
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3091644014886998,
      "learning_rate": 1.721852360433406e-06,
      "loss": 0.7639,
      "step": 5748
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2820036498711445,
      "learning_rate": 1.7202872718285341e-06,
      "loss": 0.7526,
      "step": 5749
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.376077790777326,
      "learning_rate": 1.7187227470443013e-06,
      "loss": 0.8593,
      "step": 5750
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8174844422883358,
      "learning_rate": 1.7171587863496686e-06,
      "loss": 0.6584,
      "step": 5751
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3087899171959858,
      "learning_rate": 1.715595390013497e-06,
      "loss": 0.8073,
      "step": 5752
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7144134795015005,
      "learning_rate": 1.7140325583045553e-06,
      "loss": 0.7623,
      "step": 5753
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3685570434024754,
      "learning_rate": 1.7124702914915097e-06,
      "loss": 0.6838,
      "step": 5754
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4268811349029071,
      "learning_rate": 1.7109085898429345e-06,
      "loss": 0.7338,
      "step": 5755
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7979087344321483,
      "learning_rate": 1.7093474536273037e-06,
      "loss": 0.63,
      "step": 5756
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6046586382131196,
      "learning_rate": 1.7077868831129935e-06,
      "loss": 0.7884,
      "step": 5757
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.480997239234184,
      "learning_rate": 1.7062268785682852e-06,
      "loss": 0.7538,
      "step": 5758
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.875001835687192,
      "learning_rate": 1.7046674402613594e-06,
      "loss": 0.8694,
      "step": 5759
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2604110254992196,
      "learning_rate": 1.703108568460305e-06,
      "loss": 0.7329,
      "step": 5760
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.525899870775405,
      "learning_rate": 1.7015502634331083e-06,
      "loss": 0.8609,
      "step": 5761
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9291857886142361,
      "learning_rate": 1.6999925254476606e-06,
      "loss": 0.7094,
      "step": 5762
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6118571258596675,
      "learning_rate": 1.698435354771753e-06,
      "loss": 0.837,
      "step": 5763
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4915615098606896,
      "learning_rate": 1.6968787516730845e-06,
      "loss": 0.6957,
      "step": 5764
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3632668411925253,
      "learning_rate": 1.6953227164192516e-06,
      "loss": 0.7225,
      "step": 5765
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3529437456675477,
      "learning_rate": 1.6937672492777547e-06,
      "loss": 0.8104,
      "step": 5766
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8578172590404405,
      "learning_rate": 1.6922123505159955e-06,
      "loss": 0.7276,
      "step": 5767
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.486696972390848,
      "learning_rate": 1.6906580204012818e-06,
      "loss": 0.8059,
      "step": 5768
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.707677426295243,
      "learning_rate": 1.68910425920082e-06,
      "loss": 0.7564,
      "step": 5769
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2446657046002831,
      "learning_rate": 1.6875510671817186e-06,
      "loss": 0.7142,
      "step": 5770
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3271540352986873,
      "learning_rate": 1.6859984446109906e-06,
      "loss": 0.752,
      "step": 5771
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4636802590931732,
      "learning_rate": 1.6844463917555487e-06,
      "loss": 0.8018,
      "step": 5772
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.380346884983304,
      "learning_rate": 1.6828949088822077e-06,
      "loss": 0.6914,
      "step": 5773
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5929686619701977,
      "learning_rate": 1.6813439962576872e-06,
      "loss": 0.6703,
      "step": 5774
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3943705618274167,
      "learning_rate": 1.6797936541486082e-06,
      "loss": 0.8441,
      "step": 5775
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3916332702564962,
      "learning_rate": 1.6782438828214913e-06,
      "loss": 0.8148,
      "step": 5776
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8643824606373232,
      "learning_rate": 1.6766946825427605e-06,
      "loss": 0.7512,
      "step": 5777
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3192011402142692,
      "learning_rate": 1.6751460535787407e-06,
      "loss": 0.731,
      "step": 5778
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.864684560122918,
      "learning_rate": 1.6735979961956588e-06,
      "loss": 0.6795,
      "step": 5779
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4604563566014936,
      "learning_rate": 1.6720505106596429e-06,
      "loss": 0.7823,
      "step": 5780
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5427704718141508,
      "learning_rate": 1.6705035972367272e-06,
      "loss": 0.8104,
      "step": 5781
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3183192436051991,
      "learning_rate": 1.6689572561928397e-06,
      "loss": 0.7609,
      "step": 5782
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3562792600904936,
      "learning_rate": 1.6674114877938185e-06,
      "loss": 0.7498,
      "step": 5783
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3455316909874695,
      "learning_rate": 1.6658662923053974e-06,
      "loss": 0.6412,
      "step": 5784
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5162653074969024,
      "learning_rate": 1.664321669993213e-06,
      "loss": 0.6874,
      "step": 5785
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.261556671649307,
      "learning_rate": 1.6627776211228041e-06,
      "loss": 0.686,
      "step": 5786
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2770724876200752,
      "learning_rate": 1.661234145959611e-06,
      "loss": 0.6831,
      "step": 5787
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4712163854012972,
      "learning_rate": 1.6596912447689723e-06,
      "loss": 0.8075,
      "step": 5788
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4669509384913262,
      "learning_rate": 1.6581489178161348e-06,
      "loss": 0.7742,
      "step": 5789
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.34033149048171,
      "learning_rate": 1.6566071653662403e-06,
      "loss": 0.7926,
      "step": 5790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8113676121886806,
      "learning_rate": 1.6550659876843317e-06,
      "loss": 0.662,
      "step": 5791
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3399081938900117,
      "learning_rate": 1.65352538503536e-06,
      "loss": 0.7421,
      "step": 5792
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2524504723005798,
      "learning_rate": 1.6519853576841698e-06,
      "loss": 0.7786,
      "step": 5793
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.495425944256309,
      "learning_rate": 1.6504459058955108e-06,
      "loss": 0.7321,
      "step": 5794
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.865513591758512,
      "learning_rate": 1.6489070299340298e-06,
      "loss": 0.773,
      "step": 5795
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3385566735551961,
      "learning_rate": 1.6473687300642815e-06,
      "loss": 0.7877,
      "step": 5796
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6151367772824865,
      "learning_rate": 1.6458310065507154e-06,
      "loss": 0.6641,
      "step": 5797
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2327351862314344,
      "learning_rate": 1.6442938596576842e-06,
      "loss": 0.6895,
      "step": 5798
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.248252507024352,
      "learning_rate": 1.6427572896494408e-06,
      "loss": 0.7158,
      "step": 5799
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6892163554107436,
      "learning_rate": 1.6412212967901386e-06,
      "loss": 0.8272,
      "step": 5800
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3847146137399686,
      "learning_rate": 1.6396858813438355e-06,
      "loss": 0.7157,
      "step": 5801
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4121464780804902,
      "learning_rate": 1.638151043574485e-06,
      "loss": 0.728,
      "step": 5802
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8665926526270457,
      "learning_rate": 1.6366167837459429e-06,
      "loss": 0.6484,
      "step": 5803
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3065659979100321,
      "learning_rate": 1.6350831021219686e-06,
      "loss": 0.7736,
      "step": 5804
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3471058273811447,
      "learning_rate": 1.6335499989662185e-06,
      "loss": 0.7155,
      "step": 5805
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3939396378541968,
      "learning_rate": 1.6320174745422513e-06,
      "loss": 0.6869,
      "step": 5806
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3829421973187446,
      "learning_rate": 1.6304855291135247e-06,
      "loss": 0.6974,
      "step": 5807
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8326328836919101,
      "learning_rate": 1.6289541629433986e-06,
      "loss": 0.6939,
      "step": 5808
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3620033745155968,
      "learning_rate": 1.6274233762951308e-06,
      "loss": 0.7808,
      "step": 5809
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3231260295623755,
      "learning_rate": 1.6258931694318831e-06,
      "loss": 0.6846,
      "step": 5810
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3285610342396765,
      "learning_rate": 1.624363542616717e-06,
      "loss": 0.6557,
      "step": 5811
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3004693544458075,
      "learning_rate": 1.6228344961125914e-06,
      "loss": 0.8037,
      "step": 5812
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7276018167304528,
      "learning_rate": 1.6213060301823674e-06,
      "loss": 0.6888,
      "step": 5813
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3032272995166245,
      "learning_rate": 1.6197781450888067e-06,
      "loss": 0.8053,
      "step": 5814
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7573296425312237,
      "learning_rate": 1.618250841094569e-06,
      "loss": 0.554,
      "step": 5815
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4048062999899864,
      "learning_rate": 1.616724118462215e-06,
      "loss": 0.7435,
      "step": 5816
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5436570751994103,
      "learning_rate": 1.6151979774542087e-06,
      "loss": 0.8183,
      "step": 5817
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3890367149810088,
      "learning_rate": 1.6136724183329106e-06,
      "loss": 0.8381,
      "step": 5818
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4545580003135643,
      "learning_rate": 1.6121474413605792e-06,
      "loss": 0.7567,
      "step": 5819
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2981467155490838,
      "learning_rate": 1.61062304679938e-06,
      "loss": 0.7894,
      "step": 5820
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3517799080632515,
      "learning_rate": 1.609099234911372e-06,
      "loss": 0.7639,
      "step": 5821
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5266497554896032,
      "learning_rate": 1.6075760059585166e-06,
      "loss": 0.7224,
      "step": 5822
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3048605905634538,
      "learning_rate": 1.6060533602026734e-06,
      "loss": 0.6746,
      "step": 5823
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5169179782360134,
      "learning_rate": 1.6045312979056027e-06,
      "loss": 0.7534,
      "step": 5824
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.460665251429382,
      "learning_rate": 1.6030098193289667e-06,
      "loss": 0.7579,
      "step": 5825
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.473199003204683,
      "learning_rate": 1.601488924734324e-06,
      "loss": 0.8017,
      "step": 5826
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8377247271854122,
      "learning_rate": 1.5999686143831344e-06,
      "loss": 0.6852,
      "step": 5827
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6436179048563018,
      "learning_rate": 1.5984488885367543e-06,
      "loss": 0.8749,
      "step": 5828
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3700817660249605,
      "learning_rate": 1.5969297474564465e-06,
      "loss": 0.6815,
      "step": 5829
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.394017766866857,
      "learning_rate": 1.595411191403367e-06,
      "loss": 0.7307,
      "step": 5830
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3660001565848272,
      "learning_rate": 1.5938932206385716e-06,
      "loss": 0.7966,
      "step": 5831
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0247416595104824,
      "learning_rate": 1.5923758354230196e-06,
      "loss": 0.723,
      "step": 5832
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2923055557846375,
      "learning_rate": 1.5908590360175663e-06,
      "loss": 0.6989,
      "step": 5833
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4910042271790727,
      "learning_rate": 1.5893428226829672e-06,
      "loss": 0.7577,
      "step": 5834
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3037303155137205,
      "learning_rate": 1.5878271956798762e-06,
      "loss": 0.7558,
      "step": 5835
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8490044054069534,
      "learning_rate": 1.5863121552688477e-06,
      "loss": 0.6803,
      "step": 5836
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6066895718247158,
      "learning_rate": 1.5847977017103327e-06,
      "loss": 0.7519,
      "step": 5837
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4962955624963583,
      "learning_rate": 1.5832838352646856e-06,
      "loss": 0.7851,
      "step": 5838
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5594923807452385,
      "learning_rate": 1.5817705561921587e-06,
      "loss": 0.7309,
      "step": 5839
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0412627348806462,
      "learning_rate": 1.5802578647529005e-06,
      "loss": 0.8192,
      "step": 5840
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3571178705040898,
      "learning_rate": 1.5787457612069607e-06,
      "loss": 0.727,
      "step": 5841
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7826087572633142,
      "learning_rate": 1.577234245814287e-06,
      "loss": 0.8984,
      "step": 5842
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.392693806175986,
      "learning_rate": 1.5757233188347265e-06,
      "loss": 0.7326,
      "step": 5843
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.370807174845917,
      "learning_rate": 1.5742129805280249e-06,
      "loss": 0.754,
      "step": 5844
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3726390142844833,
      "learning_rate": 1.572703231153826e-06,
      "loss": 0.7026,
      "step": 5845
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9046471413214449,
      "learning_rate": 1.5711940709716755e-06,
      "loss": 0.6731,
      "step": 5846
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3830034769431,
      "learning_rate": 1.5696855002410127e-06,
      "loss": 0.7264,
      "step": 5847
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2465687267506513,
      "learning_rate": 1.5681775192211819e-06,
      "loss": 0.7273,
      "step": 5848
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2159380300616038,
      "learning_rate": 1.5666701281714202e-06,
      "loss": 0.7439,
      "step": 5849
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3880921149981178,
      "learning_rate": 1.5651633273508666e-06,
      "loss": 0.7909,
      "step": 5850
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3196424245824014,
      "learning_rate": 1.5636571170185565e-06,
      "loss": 0.7547,
      "step": 5851
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.146552036124323,
      "learning_rate": 1.5621514974334246e-06,
      "loss": 0.8333,
      "step": 5852
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4042174432767154,
      "learning_rate": 1.5606464688543066e-06,
      "loss": 0.6617,
      "step": 5853
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.43656505205816,
      "learning_rate": 1.5591420315399324e-06,
      "loss": 0.8983,
      "step": 5854
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3348714292251873,
      "learning_rate": 1.5576381857489337e-06,
      "loss": 0.7147,
      "step": 5855
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.20120079098583,
      "learning_rate": 1.556134931739836e-06,
      "loss": 0.8014,
      "step": 5856
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3023351398155891,
      "learning_rate": 1.5546322697710697e-06,
      "loss": 0.6557,
      "step": 5857
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8864764554577573,
      "learning_rate": 1.553130200100959e-06,
      "loss": 0.696,
      "step": 5858
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2410441018845966,
      "learning_rate": 1.5516287229877242e-06,
      "loss": 0.7483,
      "step": 5859
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4763573311816767,
      "learning_rate": 1.5501278386894907e-06,
      "loss": 0.7671,
      "step": 5860
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.342455719948755,
      "learning_rate": 1.5486275474642765e-06,
      "loss": 0.7476,
      "step": 5861
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6860885831936419,
      "learning_rate": 1.5471278495699982e-06,
      "loss": 0.7137,
      "step": 5862
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3429339122524755,
      "learning_rate": 1.545628745264472e-06,
      "loss": 0.7184,
      "step": 5863
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8325850078438258,
      "learning_rate": 1.5441302348054105e-06,
      "loss": 0.6197,
      "step": 5864
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4064735298893494,
      "learning_rate": 1.5426323184504244e-06,
      "loss": 0.8085,
      "step": 5865
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3091353619621318,
      "learning_rate": 1.5411349964570254e-06,
      "loss": 0.685,
      "step": 5866
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8316222317409866,
      "learning_rate": 1.5396382690826173e-06,
      "loss": 0.6597,
      "step": 5867
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8769076755822864,
      "learning_rate": 1.5381421365845083e-06,
      "loss": 0.6835,
      "step": 5868
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.472955587351609,
      "learning_rate": 1.5366465992198997e-06,
      "loss": 0.8031,
      "step": 5869
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3580663002785671,
      "learning_rate": 1.5351516572458913e-06,
      "loss": 0.6958,
      "step": 5870
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2549714042514541,
      "learning_rate": 1.5336573109194807e-06,
      "loss": 0.7155,
      "step": 5871
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.951551619527363,
      "learning_rate": 1.5321635604975637e-06,
      "loss": 0.703,
      "step": 5872
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4680564948054289,
      "learning_rate": 1.530670406236932e-06,
      "loss": 0.6847,
      "step": 5873
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3489429406666174,
      "learning_rate": 1.5291778483942771e-06,
      "loss": 0.7525,
      "step": 5874
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.300299366662436,
      "learning_rate": 1.527685887226189e-06,
      "loss": 0.6763,
      "step": 5875
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5194154568940026,
      "learning_rate": 1.5261945229891512e-06,
      "loss": 0.8078,
      "step": 5876
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.210932245545554,
      "learning_rate": 1.5247037559395467e-06,
      "loss": 0.6846,
      "step": 5877
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.320062831459752,
      "learning_rate": 1.5232135863336556e-06,
      "loss": 0.7385,
      "step": 5878
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9094198145161939,
      "learning_rate": 1.5217240144276558e-06,
      "loss": 0.6505,
      "step": 5879
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.19018138212865,
      "learning_rate": 1.5202350404776196e-06,
      "loss": 0.6822,
      "step": 5880
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2404117417012006,
      "learning_rate": 1.5187466647395227e-06,
      "loss": 0.6899,
      "step": 5881
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5339024475554834,
      "learning_rate": 1.5172588874692318e-06,
      "loss": 0.671,
      "step": 5882
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4834066623199353,
      "learning_rate": 1.5157717089225144e-06,
      "loss": 0.8059,
      "step": 5883
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7905352649357696,
      "learning_rate": 1.5142851293550303e-06,
      "loss": 0.6651,
      "step": 5884
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.363071761489037,
      "learning_rate": 1.5127991490223449e-06,
      "loss": 0.8,
      "step": 5885
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.415763059385711,
      "learning_rate": 1.5113137681799123e-06,
      "loss": 0.6509,
      "step": 5886
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3341120711933725,
      "learning_rate": 1.5098289870830869e-06,
      "loss": 0.7508,
      "step": 5887
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8255884500294752,
      "learning_rate": 1.508344805987119e-06,
      "loss": 0.6846,
      "step": 5888
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9145222625235202,
      "learning_rate": 1.5068612251471592e-06,
      "loss": 0.6654,
      "step": 5889
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3011659722517426,
      "learning_rate": 1.5053782448182509e-06,
      "loss": 0.654,
      "step": 5890
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.424418245535245,
      "learning_rate": 1.5038958652553354e-06,
      "loss": 0.8721,
      "step": 5891
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3091890954144314,
      "learning_rate": 1.502414086713251e-06,
      "loss": 0.7569,
      "step": 5892
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4184533649184499,
      "learning_rate": 1.5009329094467313e-06,
      "loss": 0.7473,
      "step": 5893
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4369520174981545,
      "learning_rate": 1.49945233371041e-06,
      "loss": 0.6995,
      "step": 5894
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.368027885714629,
      "learning_rate": 1.497972359758813e-06,
      "loss": 0.7077,
      "step": 5895
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3473049524608143,
      "learning_rate": 1.4964929878463685e-06,
      "loss": 0.7441,
      "step": 5896
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4884435563394385,
      "learning_rate": 1.4950142182273947e-06,
      "loss": 0.7979,
      "step": 5897
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.330590894560788,
      "learning_rate": 1.493536051156111e-06,
      "loss": 0.8255,
      "step": 5898
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1788525587217005,
      "learning_rate": 1.4920584868866295e-06,
      "loss": 0.816,
      "step": 5899
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6137655021936583,
      "learning_rate": 1.4905815256729621e-06,
      "loss": 0.7045,
      "step": 5900
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2228051241502984,
      "learning_rate": 1.4891051677690156e-06,
      "loss": 0.6188,
      "step": 5901
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4066301920997857,
      "learning_rate": 1.4876294134285902e-06,
      "loss": 0.779,
      "step": 5902
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.194177892195602,
      "learning_rate": 1.4861542629053882e-06,
      "loss": 0.7788,
      "step": 5903
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3794190938118052,
      "learning_rate": 1.4846797164530051e-06,
      "loss": 0.6979,
      "step": 5904
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3613498831648079,
      "learning_rate": 1.4832057743249329e-06,
      "loss": 0.7542,
      "step": 5905
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3334767216856465,
      "learning_rate": 1.4817324367745573e-06,
      "loss": 0.8509,
      "step": 5906
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8871458825529216,
      "learning_rate": 1.4802597040551636e-06,
      "loss": 0.6906,
      "step": 5907
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3715431986131539,
      "learning_rate": 1.4787875764199312e-06,
      "loss": 0.6745,
      "step": 5908
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4803192398039355,
      "learning_rate": 1.4773160541219338e-06,
      "loss": 0.842,
      "step": 5909
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2696228083318448,
      "learning_rate": 1.4758451374141469e-06,
      "loss": 0.7403,
      "step": 5910
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.345303290083377,
      "learning_rate": 1.474374826549435e-06,
      "loss": 0.7949,
      "step": 5911
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4793377343648424,
      "learning_rate": 1.4729051217805645e-06,
      "loss": 0.7564,
      "step": 5912
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.289553435514134,
      "learning_rate": 1.4714360233601933e-06,
      "loss": 0.7447,
      "step": 5913
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2433574670153495,
      "learning_rate": 1.4699675315408756e-06,
      "loss": 0.7673,
      "step": 5914
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3217910186314865,
      "learning_rate": 1.468499646575064e-06,
      "loss": 0.8055,
      "step": 5915
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3298731983209149,
      "learning_rate": 1.4670323687151012e-06,
      "loss": 0.8488,
      "step": 5916
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3333963500662065,
      "learning_rate": 1.4655656982132338e-06,
      "loss": 0.8129,
      "step": 5917
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4165341792351889,
      "learning_rate": 1.464099635321598e-06,
      "loss": 0.7354,
      "step": 5918
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.45974929920277,
      "learning_rate": 1.4626341802922262e-06,
      "loss": 0.7885,
      "step": 5919
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5697015095436193,
      "learning_rate": 1.461169333377047e-06,
      "loss": 0.7551,
      "step": 5920
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5868448162132074,
      "learning_rate": 1.459705094827884e-06,
      "loss": 0.7285,
      "step": 5921
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.184072630006091,
      "learning_rate": 1.4582414648964594e-06,
      "loss": 0.7677,
      "step": 5922
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3694900865149169,
      "learning_rate": 1.4567784438343868e-06,
      "loss": 0.7789,
      "step": 5923
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.443209909828809,
      "learning_rate": 1.455316031893174e-06,
      "loss": 0.7615,
      "step": 5924
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8823036442386076,
      "learning_rate": 1.4538542293242307e-06,
      "loss": 0.6951,
      "step": 5925
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3054658085280164,
      "learning_rate": 1.4523930363788562e-06,
      "loss": 0.7348,
      "step": 5926
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.285475617375825,
      "learning_rate": 1.450932453308246e-06,
      "loss": 0.9382,
      "step": 5927
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4947659812961416,
      "learning_rate": 1.4494724803634912e-06,
      "loss": 0.8213,
      "step": 5928
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.262085922452415,
      "learning_rate": 1.448013117795578e-06,
      "loss": 0.755,
      "step": 5929
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7378382395632949,
      "learning_rate": 1.446554365855387e-06,
      "loss": 0.6064,
      "step": 5930
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.41531292573159,
      "learning_rate": 1.445096224793695e-06,
      "loss": 0.7593,
      "step": 5931
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3480729656558728,
      "learning_rate": 1.4436386948611763e-06,
      "loss": 0.6709,
      "step": 5932
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8453205498937308,
      "learning_rate": 1.442181776308394e-06,
      "loss": 0.5858,
      "step": 5933
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.829431835582106,
      "learning_rate": 1.4407254693858108e-06,
      "loss": 0.6584,
      "step": 5934
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3429665128898158,
      "learning_rate": 1.4392697743437816e-06,
      "loss": 0.7368,
      "step": 5935
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2623326523398153,
      "learning_rate": 1.437814691432558e-06,
      "loss": 0.7361,
      "step": 5936
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8654219782790915,
      "learning_rate": 1.4363602209022837e-06,
      "loss": 0.6404,
      "step": 5937
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.337646753277978,
      "learning_rate": 1.4349063630030018e-06,
      "loss": 0.8494,
      "step": 5938
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5077239468576455,
      "learning_rate": 1.4334531179846455e-06,
      "loss": 0.7959,
      "step": 5939
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5849191254734143,
      "learning_rate": 1.432000486097046e-06,
      "loss": 0.6933,
      "step": 5940
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4098682456315639,
      "learning_rate": 1.4305484675899272e-06,
      "loss": 0.8018,
      "step": 5941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8509124150728742,
      "learning_rate": 1.4290970627129075e-06,
      "loss": 0.6517,
      "step": 5942
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6640732045387474,
      "learning_rate": 1.4276462717154999e-06,
      "loss": 0.7256,
      "step": 5943
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.4027311825061486,
      "learning_rate": 1.4261960948471122e-06,
      "loss": 0.8286,
      "step": 5944
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4417864869037806,
      "learning_rate": 1.424746532357046e-06,
      "loss": 0.7505,
      "step": 5945
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4075154587555603,
      "learning_rate": 1.4232975844944997e-06,
      "loss": 0.773,
      "step": 5946
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8698648413367877,
      "learning_rate": 1.4218492515085636e-06,
      "loss": 0.6768,
      "step": 5947
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.827600529808237,
      "learning_rate": 1.4204015336482213e-06,
      "loss": 0.6424,
      "step": 5948
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7130145326303733,
      "learning_rate": 1.418954431162355e-06,
      "loss": 0.6038,
      "step": 5949
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.479640288885274,
      "learning_rate": 1.417507944299737e-06,
      "loss": 0.7667,
      "step": 5950
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3997306140925918,
      "learning_rate": 1.4160620733090351e-06,
      "loss": 0.7877,
      "step": 5951
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6484409340160757,
      "learning_rate": 1.4146168184388099e-06,
      "loss": 0.8071,
      "step": 5952
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3933183450016973,
      "learning_rate": 1.413172179937521e-06,
      "loss": 0.7779,
      "step": 5953
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7665209234129379,
      "learning_rate": 1.4117281580535158e-06,
      "loss": 0.6149,
      "step": 5954
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3346498340378772,
      "learning_rate": 1.410284753035039e-06,
      "loss": 0.7084,
      "step": 5955
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2817430484107895,
      "learning_rate": 1.4088419651302288e-06,
      "loss": 0.8271,
      "step": 5956
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3785001797029677,
      "learning_rate": 1.407399794587117e-06,
      "loss": 0.7793,
      "step": 5957
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3295062852696484,
      "learning_rate": 1.4059582416536282e-06,
      "loss": 0.7141,
      "step": 5958
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.545873978428979,
      "learning_rate": 1.4045173065775852e-06,
      "loss": 0.7395,
      "step": 5959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.829448700098905,
      "learning_rate": 1.4030769896066975e-06,
      "loss": 0.6472,
      "step": 5960
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.851964796436071,
      "learning_rate": 1.4016372909885762e-06,
      "loss": 0.5867,
      "step": 5961
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7591773128592674,
      "learning_rate": 1.4001982109707201e-06,
      "loss": 0.7245,
      "step": 5962
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4273676181440098,
      "learning_rate": 1.3987597498005245e-06,
      "loss": 0.8144,
      "step": 5963
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1461033278833104,
      "learning_rate": 1.397321907725277e-06,
      "loss": 0.7947,
      "step": 5964
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3327649019348808,
      "learning_rate": 1.3958846849921593e-06,
      "loss": 0.7288,
      "step": 5965
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2955206589691797,
      "learning_rate": 1.3944480818482448e-06,
      "loss": 0.899,
      "step": 5966
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3421073208132528,
      "learning_rate": 1.393012098540505e-06,
      "loss": 0.688,
      "step": 5967
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1640517983282375,
      "learning_rate": 1.3915767353158022e-06,
      "loss": 0.6858,
      "step": 5968
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.151616052791559,
      "learning_rate": 1.3901419924208908e-06,
      "loss": 0.7538,
      "step": 5969
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3385974310167712,
      "learning_rate": 1.3887078701024204e-06,
      "loss": 0.8045,
      "step": 5970
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5456403557069969,
      "learning_rate": 1.3872743686069328e-06,
      "loss": 0.8402,
      "step": 5971
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9061787746113646,
      "learning_rate": 1.3858414881808634e-06,
      "loss": 0.7145,
      "step": 5972
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2895142655601888,
      "learning_rate": 1.3844092290705396e-06,
      "loss": 0.8492,
      "step": 5973
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3844911247056548,
      "learning_rate": 1.382977591522186e-06,
      "loss": 0.7144,
      "step": 5974
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9177222439791317,
      "learning_rate": 1.3815465757819174e-06,
      "loss": 0.7115,
      "step": 5975
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4094492262851923,
      "learning_rate": 1.3801161820957386e-06,
      "loss": 0.8309,
      "step": 5976
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5593825298084703,
      "learning_rate": 1.378686410709556e-06,
      "loss": 0.7868,
      "step": 5977
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.333555938786691,
      "learning_rate": 1.3772572618691604e-06,
      "loss": 0.7752,
      "step": 5978
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7757196789150979,
      "learning_rate": 1.3758287358202404e-06,
      "loss": 0.6633,
      "step": 5979
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.30252219943976,
      "learning_rate": 1.3744008328083758e-06,
      "loss": 0.7919,
      "step": 5980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9154968687788405,
      "learning_rate": 1.3729735530790378e-06,
      "loss": 0.6789,
      "step": 5981
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2982915211987731,
      "learning_rate": 1.3715468968775952e-06,
      "loss": 0.6974,
      "step": 5982
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3624679625544112,
      "learning_rate": 1.3701208644493064e-06,
      "loss": 0.7493,
      "step": 5983
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.296807120225464,
      "learning_rate": 1.3686954560393218e-06,
      "loss": 0.7555,
      "step": 5984
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3431167698153281,
      "learning_rate": 1.3672706718926849e-06,
      "loss": 0.8422,
      "step": 5985
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8396260862443846,
      "learning_rate": 1.3658465122543346e-06,
      "loss": 0.672,
      "step": 5986
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5624457282087365,
      "learning_rate": 1.3644229773690997e-06,
      "loss": 0.6787,
      "step": 5987
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2936843298087213,
      "learning_rate": 1.3630000674817011e-06,
      "loss": 0.6121,
      "step": 5988
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2618367893096643,
      "learning_rate": 1.361577782836756e-06,
      "loss": 0.6981,
      "step": 5989
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.64140747736905,
      "learning_rate": 1.3601561236787702e-06,
      "loss": 0.7687,
      "step": 5990
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4999275750919596,
      "learning_rate": 1.3587350902521435e-06,
      "loss": 0.736,
      "step": 5991
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3976478524873097,
      "learning_rate": 1.357314682801168e-06,
      "loss": 0.7674,
      "step": 5992
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.354444878092562,
      "learning_rate": 1.3558949015700278e-06,
      "loss": 0.7926,
      "step": 5993
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6473276276856719,
      "learning_rate": 1.3544757468027986e-06,
      "loss": 0.8109,
      "step": 5994
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4144794871385025,
      "learning_rate": 1.3530572187434531e-06,
      "loss": 0.7431,
      "step": 5995
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3008014561626562,
      "learning_rate": 1.351639317635849e-06,
      "loss": 0.6446,
      "step": 5996
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3388353612195456,
      "learning_rate": 1.3502220437237429e-06,
      "loss": 0.7435,
      "step": 5997
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.252600968428779,
      "learning_rate": 1.3488053972507792e-06,
      "loss": 0.7901,
      "step": 5998
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2520267635568194,
      "learning_rate": 1.3473893784604963e-06,
      "loss": 0.7766,
      "step": 5999
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3124099675482659,
      "learning_rate": 1.345973987596324e-06,
      "loss": 0.7815,
      "step": 6000
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8350894999097039,
      "learning_rate": 1.3445592249015843e-06,
      "loss": 0.5865,
      "step": 6001
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4406031242495339,
      "learning_rate": 1.3431450906194892e-06,
      "loss": 0.7917,
      "step": 6002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9537416423108472,
      "learning_rate": 1.3417315849931495e-06,
      "loss": 0.6485,
      "step": 6003
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.124372404008063,
      "learning_rate": 1.3403187082655584e-06,
      "loss": 0.7068,
      "step": 6004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.264652823471934,
      "learning_rate": 1.3389064606796098e-06,
      "loss": 0.6578,
      "step": 6005
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.250274991738625,
      "learning_rate": 1.3374948424780836e-06,
      "loss": 0.756,
      "step": 6006
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1264910894244322,
      "learning_rate": 1.336083853903653e-06,
      "loss": 0.7049,
      "step": 6007
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3046850294189898,
      "learning_rate": 1.3346734951988844e-06,
      "loss": 0.8121,
      "step": 6008
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5200201002570983,
      "learning_rate": 1.333263766606232e-06,
      "loss": 0.7399,
      "step": 6009
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5923839849961574,
      "learning_rate": 1.3318546683680483e-06,
      "loss": 0.7175,
      "step": 6010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8290304535134018,
      "learning_rate": 1.3304462007265716e-06,
      "loss": 0.6129,
      "step": 6011
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5120870856182582,
      "learning_rate": 1.3290383639239347e-06,
      "loss": 0.7584,
      "step": 6012
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3816232611709571,
      "learning_rate": 1.3276311582021583e-06,
      "loss": 0.6927,
      "step": 6013
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4339802230415324,
      "learning_rate": 1.3262245838031618e-06,
      "loss": 0.7365,
      "step": 6014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8253301732468628,
      "learning_rate": 1.3248186409687491e-06,
      "loss": 0.6057,
      "step": 6015
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2758690260467878,
      "learning_rate": 1.3234133299406183e-06,
      "loss": 0.6956,
      "step": 6016
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3892663078768752,
      "learning_rate": 1.3220086509603569e-06,
      "loss": 0.786,
      "step": 6017
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.284350738947654,
      "learning_rate": 1.3206046042694493e-06,
      "loss": 0.7377,
      "step": 6018
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.523744466211635,
      "learning_rate": 1.3192011901092654e-06,
      "loss": 0.7363,
      "step": 6019
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5862853028009107,
      "learning_rate": 1.3177984087210682e-06,
      "loss": 0.7578,
      "step": 6020
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.258712569426316,
      "learning_rate": 1.3163962603460123e-06,
      "loss": 0.8046,
      "step": 6021
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2521135366422191,
      "learning_rate": 1.3149947452251422e-06,
      "loss": 0.7352,
      "step": 6022
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3673949683815876,
      "learning_rate": 1.3135938635993966e-06,
      "loss": 0.7714,
      "step": 6023
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.508346954037812,
      "learning_rate": 1.312193615709601e-06,
      "loss": 0.8026,
      "step": 6024
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3090822456988713,
      "learning_rate": 1.310794001796477e-06,
      "loss": 0.7842,
      "step": 6025
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3829922424698857,
      "learning_rate": 1.3093950221006329e-06,
      "loss": 0.8008,
      "step": 6026
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3729879328833499,
      "learning_rate": 1.30799667686257e-06,
      "loss": 0.7096,
      "step": 6027
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5025228064794214,
      "learning_rate": 1.3065989663226797e-06,
      "loss": 0.8319,
      "step": 6028
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5766906155358054,
      "learning_rate": 1.3052018907212448e-06,
      "loss": 0.7658,
      "step": 6029
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3207104978410402,
      "learning_rate": 1.303805450298437e-06,
      "loss": 0.7483,
      "step": 6030
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2771440077555345,
      "learning_rate": 1.3024096452943236e-06,
      "loss": 0.6929,
      "step": 6031
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4755585502327764,
      "learning_rate": 1.301014475948857e-06,
      "loss": 0.7443,
      "step": 6032
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.752104553948452,
      "learning_rate": 1.2996199425018858e-06,
      "loss": 0.7331,
      "step": 6033
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4216031038840862,
      "learning_rate": 1.2982260451931445e-06,
      "loss": 0.8586,
      "step": 6034
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8231038732564049,
      "learning_rate": 1.2968327842622612e-06,
      "loss": 0.6359,
      "step": 6035
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.281460030074009,
      "learning_rate": 1.2954401599487531e-06,
      "loss": 0.7428,
      "step": 6036
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5345646439050211,
      "learning_rate": 1.2940481724920284e-06,
      "loss": 0.7953,
      "step": 6037
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9499682692318145,
      "learning_rate": 1.292656822131384e-06,
      "loss": 0.6907,
      "step": 6038
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7552371611945086,
      "learning_rate": 1.291266109106013e-06,
      "loss": 0.6815,
      "step": 6039
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3565472758754573,
      "learning_rate": 1.2898760336549931e-06,
      "loss": 0.799,
      "step": 6040
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.309648698631096,
      "learning_rate": 1.2884865960172931e-06,
      "loss": 0.7802,
      "step": 6041
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4094077885567704,
      "learning_rate": 1.287097796431776e-06,
      "loss": 0.7518,
      "step": 6042
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3179307879597533,
      "learning_rate": 1.2857096351371917e-06,
      "loss": 0.8507,
      "step": 6043
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7789153055471323,
      "learning_rate": 1.2843221123721804e-06,
      "loss": 0.7821,
      "step": 6044
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5741740737770304,
      "learning_rate": 1.2829352283752728e-06,
      "loss": 0.8341,
      "step": 6045
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1836654819205392,
      "learning_rate": 1.2815489833848927e-06,
      "loss": 0.833,
      "step": 6046
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7571614400848973,
      "learning_rate": 1.28016337763935e-06,
      "loss": 0.6392,
      "step": 6047
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2491351263371275,
      "learning_rate": 1.278778411376847e-06,
      "loss": 0.842,
      "step": 6048
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.889728927153808,
      "learning_rate": 1.2773940848354754e-06,
      "loss": 0.7803,
      "step": 6049
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2889638694446885,
      "learning_rate": 1.2760103982532152e-06,
      "loss": 0.8062,
      "step": 6050
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4176378151526998,
      "learning_rate": 1.274627351867941e-06,
      "loss": 0.691,
      "step": 6051
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.370315822339242,
      "learning_rate": 1.2732449459174134e-06,
      "loss": 0.6917,
      "step": 6052
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.385396226929748,
      "learning_rate": 1.2718631806392823e-06,
      "loss": 0.8114,
      "step": 6053
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4894389244544401,
      "learning_rate": 1.2704820562710923e-06,
      "loss": 0.7731,
      "step": 6054
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3540272248497798,
      "learning_rate": 1.2691015730502732e-06,
      "loss": 0.8301,
      "step": 6055
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8168746968849019,
      "learning_rate": 1.2677217312141455e-06,
      "loss": 0.6713,
      "step": 6056
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3248534690647376,
      "learning_rate": 1.2663425309999205e-06,
      "loss": 0.7912,
      "step": 6057
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3199590796201626,
      "learning_rate": 1.2649639726446994e-06,
      "loss": 0.7619,
      "step": 6058
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3216205272250923,
      "learning_rate": 1.2635860563854695e-06,
      "loss": 0.8573,
      "step": 6059
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4479004645878317,
      "learning_rate": 1.2622087824591129e-06,
      "loss": 0.7243,
      "step": 6060
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3024948044779323,
      "learning_rate": 1.2608321511024007e-06,
      "loss": 0.7067,
      "step": 6061
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4055509002801807,
      "learning_rate": 1.2594561625519891e-06,
      "loss": 0.8817,
      "step": 6062
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5764260185919696,
      "learning_rate": 1.258080817044428e-06,
      "loss": 0.8526,
      "step": 6063
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8383146296076587,
      "learning_rate": 1.256706114816154e-06,
      "loss": 0.7892,
      "step": 6064
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8690021791777875,
      "learning_rate": 1.2553320561034955e-06,
      "loss": 0.6894,
      "step": 6065
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.394739596133423,
      "learning_rate": 1.2539586411426664e-06,
      "loss": 0.7335,
      "step": 6066
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3820800735217922,
      "learning_rate": 1.2525858701697762e-06,
      "loss": 0.709,
      "step": 6067
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4777093015841156,
      "learning_rate": 1.2512137434208188e-06,
      "loss": 0.7867,
      "step": 6068
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3892792112048915,
      "learning_rate": 1.2498422611316767e-06,
      "loss": 0.6703,
      "step": 6069
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4684456935569818,
      "learning_rate": 1.2484714235381278e-06,
      "loss": 0.7345,
      "step": 6070
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3897389884858053,
      "learning_rate": 1.2471012308758324e-06,
      "loss": 0.8763,
      "step": 6071
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3182287307728455,
      "learning_rate": 1.2457316833803424e-06,
      "loss": 0.756,
      "step": 6072
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.422739254236175,
      "learning_rate": 1.244362781287099e-06,
      "loss": 0.7468,
      "step": 6073
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3295988279809816,
      "learning_rate": 1.2429945248314317e-06,
      "loss": 0.7157,
      "step": 6074
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3449939840844654,
      "learning_rate": 1.2416269142485615e-06,
      "loss": 0.7075,
      "step": 6075
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6863039041925425,
      "learning_rate": 1.2402599497735961e-06,
      "loss": 0.7927,
      "step": 6076
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5127795160898583,
      "learning_rate": 1.2388936316415317e-06,
      "loss": 0.6841,
      "step": 6077
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3764114733466832,
      "learning_rate": 1.237527960087253e-06,
      "loss": 0.6986,
      "step": 6078
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8064861600174835,
      "learning_rate": 1.2361629353455378e-06,
      "loss": 0.7271,
      "step": 6079
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.254493313276773,
      "learning_rate": 1.234798557651048e-06,
      "loss": 0.7586,
      "step": 6080
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8730378254505564,
      "learning_rate": 1.233434827238335e-06,
      "loss": 0.6315,
      "step": 6081
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3865314807646867,
      "learning_rate": 1.2320717443418422e-06,
      "loss": 0.8217,
      "step": 6082
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.234359542386287,
      "learning_rate": 1.2307093091958983e-06,
      "loss": 0.6848,
      "step": 6083
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8368349924840262,
      "learning_rate": 1.2293475220347212e-06,
      "loss": 0.6819,
      "step": 6084
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2404842994074121,
      "learning_rate": 1.2279863830924183e-06,
      "loss": 0.7239,
      "step": 6085
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4330409634121046,
      "learning_rate": 1.2266258926029851e-06,
      "loss": 0.7084,
      "step": 6086
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3922402937469665,
      "learning_rate": 1.2252660508003045e-06,
      "loss": 0.7684,
      "step": 6087
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3034835790476915,
      "learning_rate": 1.2239068579181497e-06,
      "loss": 0.8512,
      "step": 6088
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2486637806190808,
      "learning_rate": 1.222548314190184e-06,
      "loss": 0.7391,
      "step": 6089
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5852654871595353,
      "learning_rate": 1.2211904198499551e-06,
      "loss": 0.764,
      "step": 6090
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.392363580412767,
      "learning_rate": 1.2198331751309006e-06,
      "loss": 0.7136,
      "step": 6091
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3721301135842325,
      "learning_rate": 1.2184765802663468e-06,
      "loss": 0.6937,
      "step": 6092
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3665869971693965,
      "learning_rate": 1.2171206354895081e-06,
      "loss": 0.7456,
      "step": 6093
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.29599783947087,
      "learning_rate": 1.2157653410334875e-06,
      "loss": 0.6865,
      "step": 6094
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.642462662116725,
      "learning_rate": 1.214410697131273e-06,
      "loss": 0.6919,
      "step": 6095
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4316104610508509,
      "learning_rate": 1.2130567040157465e-06,
      "loss": 0.7576,
      "step": 6096
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3184903157635572,
      "learning_rate": 1.2117033619196762e-06,
      "loss": 0.8179,
      "step": 6097
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8636927282727044,
      "learning_rate": 1.2103506710757156e-06,
      "loss": 0.5991,
      "step": 6098
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4402953970992478,
      "learning_rate": 1.2089986317164075e-06,
      "loss": 0.7806,
      "step": 6099
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5506914008390698,
      "learning_rate": 1.2076472440741844e-06,
      "loss": 0.8335,
      "step": 6100
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.821020484100913,
      "learning_rate": 1.2062965083813643e-06,
      "loss": 0.7313,
      "step": 6101
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.421050397391485,
      "learning_rate": 1.2049464248701537e-06,
      "loss": 0.73,
      "step": 6102
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4831509235462368,
      "learning_rate": 1.20359699377265e-06,
      "loss": 0.693,
      "step": 6103
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4121250173185882,
      "learning_rate": 1.202248215320835e-06,
      "loss": 0.7696,
      "step": 6104
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.785558036029735,
      "learning_rate": 1.2009000897465782e-06,
      "loss": 0.6844,
      "step": 6105
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2172968517750093,
      "learning_rate": 1.1995526172816385e-06,
      "loss": 0.7975,
      "step": 6106
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3640273012937303,
      "learning_rate": 1.1982057981576634e-06,
      "loss": 0.7777,
      "step": 6107
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4935640859806287,
      "learning_rate": 1.196859632606186e-06,
      "loss": 0.843,
      "step": 6108
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.435882693963624,
      "learning_rate": 1.1955141208586257e-06,
      "loss": 0.7104,
      "step": 6109
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4285041817200743,
      "learning_rate": 1.1941692631462954e-06,
      "loss": 0.9088,
      "step": 6110
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.569343799293258,
      "learning_rate": 1.1928250597003893e-06,
      "loss": 0.6634,
      "step": 6111
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2561847496054295,
      "learning_rate": 1.1914815107519922e-06,
      "loss": 0.696,
      "step": 6112
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3492797039230187,
      "learning_rate": 1.1901386165320755e-06,
      "loss": 0.7875,
      "step": 6113
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4665977913382626,
      "learning_rate": 1.1887963772714982e-06,
      "loss": 0.7666,
      "step": 6114
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4873028270727195,
      "learning_rate": 1.1874547932010054e-06,
      "loss": 0.8116,
      "step": 6115
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.437783592361763,
      "learning_rate": 1.1861138645512343e-06,
      "loss": 0.8451,
      "step": 6116
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.545902573497661,
      "learning_rate": 1.1847735915527026e-06,
      "loss": 0.7194,
      "step": 6117
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.246364945266745,
      "learning_rate": 1.183433974435822e-06,
      "loss": 0.7378,
      "step": 6118
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2892512771225308,
      "learning_rate": 1.1820950134308862e-06,
      "loss": 0.8129,
      "step": 6119
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7256635996149163,
      "learning_rate": 1.1807567087680787e-06,
      "loss": 0.8469,
      "step": 6120
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2139005462776147,
      "learning_rate": 1.1794190606774696e-06,
      "loss": 0.755,
      "step": 6121
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3091108289084055,
      "learning_rate": 1.178082069389016e-06,
      "loss": 0.8008,
      "step": 6122
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2619547520854542,
      "learning_rate": 1.1767457351325605e-06,
      "loss": 0.7043,
      "step": 6123
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4758272194233364,
      "learning_rate": 1.1754100581378365e-06,
      "loss": 0.8499,
      "step": 6124
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.429437494014957,
      "learning_rate": 1.174075038634463e-06,
      "loss": 0.7518,
      "step": 6125
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2510325765683967,
      "learning_rate": 1.1727406768519444e-06,
      "loss": 0.7462,
      "step": 6126
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3939962732858338,
      "learning_rate": 1.171406973019673e-06,
      "loss": 0.7272,
      "step": 6127
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4508770660475023,
      "learning_rate": 1.1700739273669277e-06,
      "loss": 0.806,
      "step": 6128
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3834111564181768,
      "learning_rate": 1.168741540122874e-06,
      "loss": 0.775,
      "step": 6129
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3656331334433611,
      "learning_rate": 1.1674098115165645e-06,
      "loss": 0.7852,
      "step": 6130
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3414728603421684,
      "learning_rate": 1.166078741776941e-06,
      "loss": 0.8111,
      "step": 6131
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6253262875729453,
      "learning_rate": 1.1647483311328285e-06,
      "loss": 0.8376,
      "step": 6132
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5035843564222857,
      "learning_rate": 1.1634185798129383e-06,
      "loss": 0.6728,
      "step": 6133
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8861616021758028,
      "learning_rate": 1.1620894880458732e-06,
      "loss": 0.7565,
      "step": 6134
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4521911079193948,
      "learning_rate": 1.1607610560601179e-06,
      "loss": 0.7405,
      "step": 6135
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.280829192318533,
      "learning_rate": 1.1594332840840455e-06,
      "loss": 0.8466,
      "step": 6136
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2402645927994635,
      "learning_rate": 1.1581061723459153e-06,
      "loss": 0.6603,
      "step": 6137
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2537474715188466,
      "learning_rate": 1.1567797210738713e-06,
      "loss": 0.7342,
      "step": 6138
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4805662517695095,
      "learning_rate": 1.1554539304959494e-06,
      "loss": 0.8207,
      "step": 6139
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6739943924219114,
      "learning_rate": 1.1541288008400665e-06,
      "loss": 0.8051,
      "step": 6140
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.382837873969134,
      "learning_rate": 1.1528043323340281e-06,
      "loss": 0.8287,
      "step": 6141
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8716513869536437,
      "learning_rate": 1.151480525205525e-06,
      "loss": 0.6104,
      "step": 6142
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2492153663127963,
      "learning_rate": 1.1501573796821348e-06,
      "loss": 0.6803,
      "step": 6143
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.391365368857445,
      "learning_rate": 1.148834895991323e-06,
      "loss": 0.7948,
      "step": 6144
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2326044181307498,
      "learning_rate": 1.147513074360438e-06,
      "loss": 0.7093,
      "step": 6145
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3326172669636236,
      "learning_rate": 1.1461919150167189e-06,
      "loss": 0.7756,
      "step": 6146
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3974736634566018,
      "learning_rate": 1.1448714181872867e-06,
      "loss": 0.7378,
      "step": 6147
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5496415518636095,
      "learning_rate": 1.1435515840991502e-06,
      "loss": 0.7392,
      "step": 6148
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3926375102232453,
      "learning_rate": 1.1422324129792039e-06,
      "loss": 0.7509,
      "step": 6149
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2678752954861015,
      "learning_rate": 1.1409139050542295e-06,
      "loss": 0.8046,
      "step": 6150
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5571501073360456,
      "learning_rate": 1.1395960605508916e-06,
      "loss": 0.7539,
      "step": 6151
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.213203367455713,
      "learning_rate": 1.1382788796957456e-06,
      "loss": 0.7869,
      "step": 6152
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.321807772481058,
      "learning_rate": 1.1369623627152276e-06,
      "loss": 0.722,
      "step": 6153
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.476613020511071,
      "learning_rate": 1.1356465098356656e-06,
      "loss": 0.777,
      "step": 6154
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2196831287031724,
      "learning_rate": 1.1343313212832674e-06,
      "loss": 0.7263,
      "step": 6155
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3256636011113176,
      "learning_rate": 1.13301679728413e-06,
      "loss": 0.6628,
      "step": 6156
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.396142464123598,
      "learning_rate": 1.1317029380642353e-06,
      "loss": 0.8521,
      "step": 6157
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4268209464955963,
      "learning_rate": 1.1303897438494503e-06,
      "loss": 0.7727,
      "step": 6158
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8400329953065457,
      "learning_rate": 1.1290772148655278e-06,
      "loss": 0.6215,
      "step": 6159
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3676864415612382,
      "learning_rate": 1.1277653513381083e-06,
      "loss": 0.6918,
      "step": 6160
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8071510249160705,
      "learning_rate": 1.1264541534927148e-06,
      "loss": 0.6742,
      "step": 6161
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5837112009473073,
      "learning_rate": 1.1251436215547596e-06,
      "loss": 0.7564,
      "step": 6162
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.348255283973617,
      "learning_rate": 1.1238337557495372e-06,
      "loss": 0.8868,
      "step": 6163
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3659835588151816,
      "learning_rate": 1.1225245563022285e-06,
      "loss": 0.8669,
      "step": 6164
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1568568724729713,
      "learning_rate": 1.1212160234378999e-06,
      "loss": 0.7263,
      "step": 6165
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8221576702834175,
      "learning_rate": 1.1199081573815023e-06,
      "loss": 0.7005,
      "step": 6166
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6165492603055645,
      "learning_rate": 1.1186009583578761e-06,
      "loss": 0.8109,
      "step": 6167
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4896529580424094,
      "learning_rate": 1.1172944265917419e-06,
      "loss": 0.7792,
      "step": 6168
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6458948013659873,
      "learning_rate": 1.1159885623077076e-06,
      "loss": 0.7595,
      "step": 6169
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5098823823424563,
      "learning_rate": 1.1146833657302659e-06,
      "loss": 0.7335,
      "step": 6170
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5506246565807142,
      "learning_rate": 1.1133788370837972e-06,
      "loss": 0.651,
      "step": 6171
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3308493056400765,
      "learning_rate": 1.1120749765925643e-06,
      "loss": 0.7956,
      "step": 6172
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2241476264928686,
      "learning_rate": 1.1107717844807153e-06,
      "loss": 0.7616,
      "step": 6173
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8853474738103346,
      "learning_rate": 1.1094692609722829e-06,
      "loss": 0.6307,
      "step": 6174
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5301525600441404,
      "learning_rate": 1.108167406291189e-06,
      "loss": 0.7309,
      "step": 6175
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4809237463319826,
      "learning_rate": 1.1068662206612363e-06,
      "loss": 0.8063,
      "step": 6176
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5894514205756154,
      "learning_rate": 1.1055657043061124e-06,
      "loss": 0.8116,
      "step": 6177
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3736539856976366,
      "learning_rate": 1.104265857449393e-06,
      "loss": 0.7968,
      "step": 6178
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2724997299750431,
      "learning_rate": 1.1029666803145356e-06,
      "loss": 0.7414,
      "step": 6179
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8151084175474145,
      "learning_rate": 1.101668173124883e-06,
      "loss": 0.6445,
      "step": 6180
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3858427270420037,
      "learning_rate": 1.1003703361036644e-06,
      "loss": 0.7518,
      "step": 6181
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2842205733181993,
      "learning_rate": 1.0990731694739947e-06,
      "loss": 0.7496,
      "step": 6182
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.471750253698032,
      "learning_rate": 1.0977766734588707e-06,
      "loss": 0.7676,
      "step": 6183
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6478498405785362,
      "learning_rate": 1.0964808482811751e-06,
      "loss": 0.7189,
      "step": 6184
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.3334801049602407,
      "learning_rate": 1.0951856941636752e-06,
      "loss": 0.8264,
      "step": 6185
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3403846445434722,
      "learning_rate": 1.093891211329023e-06,
      "loss": 0.7429,
      "step": 6186
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2919868796441274,
      "learning_rate": 1.0925973999997535e-06,
      "loss": 0.6564,
      "step": 6187
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5233493769827613,
      "learning_rate": 1.091304260398291e-06,
      "loss": 0.7822,
      "step": 6188
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4070301062850359,
      "learning_rate": 1.0900117927469384e-06,
      "loss": 0.7651,
      "step": 6189
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.352506241405715,
      "learning_rate": 1.088719997267888e-06,
      "loss": 0.7259,
      "step": 6190
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8341041384178243,
      "learning_rate": 1.087428874183214e-06,
      "loss": 0.6852,
      "step": 6191
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7437648504678275,
      "learning_rate": 1.0861384237148749e-06,
      "loss": 0.6303,
      "step": 6192
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3931889479703137,
      "learning_rate": 1.0848486460847135e-06,
      "loss": 0.6753,
      "step": 6193
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1317596322462986,
      "learning_rate": 1.083559541514459e-06,
      "loss": 0.662,
      "step": 6194
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2067463067942312,
      "learning_rate": 1.0822711102257205e-06,
      "loss": 0.8008,
      "step": 6195
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3256336467845513,
      "learning_rate": 1.0809833524399981e-06,
      "loss": 0.8024,
      "step": 6196
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9540083051981237,
      "learning_rate": 1.079696268378671e-06,
      "loss": 0.7225,
      "step": 6197
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.49738106069295,
      "learning_rate": 1.0784098582630015e-06,
      "loss": 0.7432,
      "step": 6198
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3179710588669378,
      "learning_rate": 1.0771241223141415e-06,
      "loss": 0.7362,
      "step": 6199
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.233306407978452,
      "learning_rate": 1.0758390607531232e-06,
      "loss": 0.7537,
      "step": 6200
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.437581058808887,
      "learning_rate": 1.0745546738008627e-06,
      "loss": 0.7941,
      "step": 6201
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8169675684359025,
      "learning_rate": 1.0732709616781606e-06,
      "loss": 0.6232,
      "step": 6202
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3243083996870642,
      "learning_rate": 1.071987924605704e-06,
      "loss": 0.7533,
      "step": 6203
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4168151503060304,
      "learning_rate": 1.0707055628040602e-06,
      "loss": 0.7595,
      "step": 6204
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3269998561452199,
      "learning_rate": 1.0694238764936827e-06,
      "loss": 0.7451,
      "step": 6205
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2979147735134233,
      "learning_rate": 1.0681428658949083e-06,
      "loss": 0.8126,
      "step": 6206
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5881665064978114,
      "learning_rate": 1.0668625312279552e-06,
      "loss": 0.756,
      "step": 6207
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6238543056931432,
      "learning_rate": 1.0655828727129319e-06,
      "loss": 0.7699,
      "step": 6208
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6780151833787254,
      "learning_rate": 1.0643038905698239e-06,
      "loss": 0.7111,
      "step": 6209
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.493722988619987,
      "learning_rate": 1.0630255850185024e-06,
      "loss": 0.7593,
      "step": 6210
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3271426516797145,
      "learning_rate": 1.0617479562787258e-06,
      "loss": 0.7189,
      "step": 6211
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.304407252635322,
      "learning_rate": 1.060471004570131e-06,
      "loss": 0.6758,
      "step": 6212
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4356891303852675,
      "learning_rate": 1.0591947301122424e-06,
      "loss": 0.7614,
      "step": 6213
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3420669009060733,
      "learning_rate": 1.0579191331244648e-06,
      "loss": 0.7031,
      "step": 6214
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5141269905446317,
      "learning_rate": 1.0566442138260885e-06,
      "loss": 0.6768,
      "step": 6215
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4228671069424246,
      "learning_rate": 1.0553699724362866e-06,
      "loss": 0.7735,
      "step": 6216
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3544974297440322,
      "learning_rate": 1.0540964091741157e-06,
      "loss": 0.909,
      "step": 6217
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8593172943907064,
      "learning_rate": 1.0528235242585188e-06,
      "loss": 0.6965,
      "step": 6218
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8193341654696165,
      "learning_rate": 1.0515513179083176e-06,
      "loss": 0.7648,
      "step": 6219
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.397606150937335,
      "learning_rate": 1.050279790342219e-06,
      "loss": 0.8346,
      "step": 6220
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8059286499320995,
      "learning_rate": 1.0490089417788135e-06,
      "loss": 0.6431,
      "step": 6221
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3941704777326442,
      "learning_rate": 1.0477387724365752e-06,
      "loss": 0.6901,
      "step": 6222
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.45777416401969,
      "learning_rate": 1.0464692825338584e-06,
      "loss": 0.7382,
      "step": 6223
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2794751317500548,
      "learning_rate": 1.045200472288907e-06,
      "loss": 0.6986,
      "step": 6224
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6466812352540765,
      "learning_rate": 1.0439323419198422e-06,
      "loss": 0.8249,
      "step": 6225
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6560193751653447,
      "learning_rate": 1.0426648916446692e-06,
      "loss": 0.6824,
      "step": 6226
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8816418208549006,
      "learning_rate": 1.0413981216812802e-06,
      "loss": 0.6552,
      "step": 6227
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4064675456282656,
      "learning_rate": 1.0401320322474456e-06,
      "loss": 0.655,
      "step": 6228
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.819011250761754,
      "learning_rate": 1.0388666235608219e-06,
      "loss": 0.7592,
      "step": 6229
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.338630427152045,
      "learning_rate": 1.0376018958389472e-06,
      "loss": 0.7113,
      "step": 6230
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7821062208524142,
      "learning_rate": 1.036337849299241e-06,
      "loss": 0.6521,
      "step": 6231
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2532805745535696,
      "learning_rate": 1.0350744841590106e-06,
      "loss": 0.7492,
      "step": 6232
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2807233450874755,
      "learning_rate": 1.0338118006354413e-06,
      "loss": 0.7165,
      "step": 6233
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2730165015256196,
      "learning_rate": 1.032549798945604e-06,
      "loss": 0.7102,
      "step": 6234
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4616170020048442,
      "learning_rate": 1.0312884793064493e-06,
      "loss": 0.7359,
      "step": 6235
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3824230026306314,
      "learning_rate": 1.0300278419348158e-06,
      "loss": 0.6837,
      "step": 6236
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.957869833475758,
      "learning_rate": 1.02876788704742e-06,
      "loss": 0.7775,
      "step": 6237
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2864456041221983,
      "learning_rate": 1.0275086148608614e-06,
      "loss": 0.7485,
      "step": 6238
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2901000003614576,
      "learning_rate": 1.026250025591627e-06,
      "loss": 0.6917,
      "step": 6239
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7994745167332581,
      "learning_rate": 1.0249921194560803e-06,
      "loss": 0.6667,
      "step": 6240
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3697789045864552,
      "learning_rate": 1.0237348966704708e-06,
      "loss": 0.7388,
      "step": 6241
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1476790193979807,
      "learning_rate": 1.022478357450929e-06,
      "loss": 0.7904,
      "step": 6242
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7006969803769763,
      "learning_rate": 1.0212225020134693e-06,
      "loss": 0.8182,
      "step": 6243
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4932668284245543,
      "learning_rate": 1.0199673305739854e-06,
      "loss": 0.7809,
      "step": 6244
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.100691776748293,
      "learning_rate": 1.0187128433482601e-06,
      "loss": 0.7235,
      "step": 6245
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.325450264906094,
      "learning_rate": 1.0174590405519502e-06,
      "loss": 0.7559,
      "step": 6246
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5565249113071402,
      "learning_rate": 1.0162059224006027e-06,
      "loss": 0.7212,
      "step": 6247
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.316908803052905,
      "learning_rate": 1.0149534891096408e-06,
      "loss": 0.7089,
      "step": 6248
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4466144854101795,
      "learning_rate": 1.0137017408943729e-06,
      "loss": 0.7477,
      "step": 6249
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.260727113657528,
      "learning_rate": 1.0124506779699882e-06,
      "loss": 0.7,
      "step": 6250
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4486267253767606,
      "learning_rate": 1.0112003005515603e-06,
      "loss": 0.7833,
      "step": 6251
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2941475402893865,
      "learning_rate": 1.0099506088540418e-06,
      "loss": 0.7415,
      "step": 6252
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8426428023699226,
      "learning_rate": 1.0087016030922709e-06,
      "loss": 0.8217,
      "step": 6253
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2361318881496017,
      "learning_rate": 1.0074532834809647e-06,
      "loss": 0.7209,
      "step": 6254
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2496168502805884,
      "learning_rate": 1.0062056502347257e-06,
      "loss": 0.7247,
      "step": 6255
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3810589463494662,
      "learning_rate": 1.0049587035680353e-06,
      "loss": 0.708,
      "step": 6256
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.254940111924028,
      "learning_rate": 1.003712443695259e-06,
      "loss": 0.7933,
      "step": 6257
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3033569497145072,
      "learning_rate": 1.0024668708306418e-06,
      "loss": 0.6643,
      "step": 6258
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5097100457140828,
      "learning_rate": 1.001221985188312e-06,
      "loss": 0.6815,
      "step": 6259
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8719618352414873,
      "learning_rate": 9.99977786982282e-07,
      "loss": 0.6278,
      "step": 6260
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3155041001785428,
      "learning_rate": 9.987342764264424e-07,
      "loss": 0.6788,
      "step": 6261
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7560330310563603,
      "learning_rate": 9.974914537345675e-07,
      "loss": 0.6148,
      "step": 6262
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3289259261329471,
      "learning_rate": 9.96249319120311e-07,
      "loss": 0.8178,
      "step": 6263
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3476609601701006,
      "learning_rate": 9.95007872797214e-07,
      "loss": 0.6845,
      "step": 6264
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3616410262682304,
      "learning_rate": 9.937671149786933e-07,
      "loss": 0.7843,
      "step": 6265
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4455958589227387,
      "learning_rate": 9.925270458780496e-07,
      "loss": 0.7561,
      "step": 6266
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5111811721997728,
      "learning_rate": 9.91287665708464e-07,
      "loss": 0.7683,
      "step": 6267
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5471834377586196,
      "learning_rate": 9.900489746830034e-07,
      "loss": 0.783,
      "step": 6268
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4707094770999525,
      "learning_rate": 9.888109730146112e-07,
      "loss": 0.7459,
      "step": 6269
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.408365860161609,
      "learning_rate": 9.87573660916114e-07,
      "loss": 0.7531,
      "step": 6270
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.26969149446807,
      "learning_rate": 9.863370386002214e-07,
      "loss": 0.7499,
      "step": 6271
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3051993372205153,
      "learning_rate": 9.851011062795201e-07,
      "loss": 0.6993,
      "step": 6272
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3261236279302502,
      "learning_rate": 9.83865864166485e-07,
      "loss": 0.6952,
      "step": 6273
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9317252730926264,
      "learning_rate": 9.826313124734654e-07,
      "loss": 0.6368,
      "step": 6274
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.360992390720312,
      "learning_rate": 9.813974514126977e-07,
      "loss": 0.8149,
      "step": 6275
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3561346257287006,
      "learning_rate": 9.801642811962964e-07,
      "loss": 0.7379,
      "step": 6276
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3641554691538427,
      "learning_rate": 9.789318020362564e-07,
      "loss": 0.7942,
      "step": 6277
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3868995067890963,
      "learning_rate": 9.777000141444564e-07,
      "loss": 0.8022,
      "step": 6278
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.551256069801074,
      "learning_rate": 9.764689177326542e-07,
      "loss": 0.7249,
      "step": 6279
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4566177052650666,
      "learning_rate": 9.75238513012488e-07,
      "loss": 0.6847,
      "step": 6280
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3065447467616393,
      "learning_rate": 9.74008800195481e-07,
      "loss": 0.6954,
      "step": 6281
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6601879575072402,
      "learning_rate": 9.727797794930361e-07,
      "loss": 0.7119,
      "step": 6282
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3148447532135983,
      "learning_rate": 9.715514511164343e-07,
      "loss": 0.7521,
      "step": 6283
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9325044709299295,
      "learning_rate": 9.703238152768402e-07,
      "loss": 0.736,
      "step": 6284
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3863549840243734,
      "learning_rate": 9.690968721852978e-07,
      "loss": 0.7389,
      "step": 6285
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.360066581921867,
      "learning_rate": 9.678706220527333e-07,
      "loss": 0.8247,
      "step": 6286
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9238064864001538,
      "learning_rate": 9.666450650899533e-07,
      "loss": 0.7112,
      "step": 6287
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7985537107586317,
      "learning_rate": 9.654202015076442e-07,
      "loss": 0.6439,
      "step": 6288
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3118744212987814,
      "learning_rate": 9.641960315163768e-07,
      "loss": 0.7842,
      "step": 6289
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2328885798289761,
      "learning_rate": 9.629725553265983e-07,
      "loss": 0.795,
      "step": 6290
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3332546594803723,
      "learning_rate": 9.617497731486374e-07,
      "loss": 0.7318,
      "step": 6291
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3567330196931262,
      "learning_rate": 9.605276851927075e-07,
      "loss": 0.7987,
      "step": 6292
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.341412865718263,
      "learning_rate": 9.593062916688982e-07,
      "loss": 0.7308,
      "step": 6293
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4093680715304069,
      "learning_rate": 9.580855927871808e-07,
      "loss": 0.7321,
      "step": 6294
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.260615202965336,
      "learning_rate": 9.56865588757407e-07,
      "loss": 0.7108,
      "step": 6295
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4002974987511305,
      "learning_rate": 9.556462797893113e-07,
      "loss": 0.7804,
      "step": 6296
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2948997648129492,
      "learning_rate": 9.544276660925067e-07,
      "loss": 0.7218,
      "step": 6297
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1295611340001868,
      "learning_rate": 9.532097478764862e-07,
      "loss": 0.7496,
      "step": 6298
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3480744484271254,
      "learning_rate": 9.519925253506246e-07,
      "loss": 0.7548,
      "step": 6299
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.358047917110724,
      "learning_rate": 9.507759987241755e-07,
      "loss": 0.7582,
      "step": 6300
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4733120296092557,
      "learning_rate": 9.495601682062755e-07,
      "loss": 0.7018,
      "step": 6301
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2211402721059954,
      "learning_rate": 9.483450340059386e-07,
      "loss": 0.7846,
      "step": 6302
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.124263464900056,
      "learning_rate": 9.47130596332062e-07,
      "loss": 0.7016,
      "step": 6303
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5475233308912264,
      "learning_rate": 9.459168553934211e-07,
      "loss": 0.7067,
      "step": 6304
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5825134120921664,
      "learning_rate": 9.447038113986717e-07,
      "loss": 0.7452,
      "step": 6305
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3173174203352642,
      "learning_rate": 9.434914645563498e-07,
      "loss": 0.7353,
      "step": 6306
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2909064640737677,
      "learning_rate": 9.422798150748724e-07,
      "loss": 0.7886,
      "step": 6307
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.353014715033268,
      "learning_rate": 9.410688631625364e-07,
      "loss": 0.7492,
      "step": 6308
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2607262271362245,
      "learning_rate": 9.398586090275164e-07,
      "loss": 0.7466,
      "step": 6309
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.827481939927637,
      "learning_rate": 9.386490528778702e-07,
      "loss": 0.6292,
      "step": 6310
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2823502469625234,
      "learning_rate": 9.374401949215367e-07,
      "loss": 0.7694,
      "step": 6311
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.375254540698704,
      "learning_rate": 9.362320353663313e-07,
      "loss": 0.8287,
      "step": 6312
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8569269831897288,
      "learning_rate": 9.350245744199499e-07,
      "loss": 0.8865,
      "step": 6313
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5213208309490183,
      "learning_rate": 9.338178122899693e-07,
      "loss": 0.707,
      "step": 6314
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1635419131486138,
      "learning_rate": 9.32611749183846e-07,
      "loss": 0.7224,
      "step": 6315
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1649571500814386,
      "learning_rate": 9.31406385308915e-07,
      "loss": 0.7147,
      "step": 6316
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2712900054801053,
      "learning_rate": 9.302017208723951e-07,
      "loss": 0.7806,
      "step": 6317
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4282949151335471,
      "learning_rate": 9.289977560813789e-07,
      "loss": 0.7472,
      "step": 6318
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8371500718327528,
      "learning_rate": 9.27794491142845e-07,
      "loss": 0.6662,
      "step": 6319
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4711793315736994,
      "learning_rate": 9.265919262636469e-07,
      "loss": 0.6711,
      "step": 6320
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3779898093058438,
      "learning_rate": 9.253900616505202e-07,
      "loss": 0.6571,
      "step": 6321
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4616640769381812,
      "learning_rate": 9.241888975100782e-07,
      "loss": 0.7952,
      "step": 6322
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.324359658130686,
      "learning_rate": 9.229884340488149e-07,
      "loss": 0.7752,
      "step": 6323
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.348285947202773,
      "learning_rate": 9.217886714731056e-07,
      "loss": 0.7784,
      "step": 6324
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3894296373097215,
      "learning_rate": 9.205896099892019e-07,
      "loss": 0.7522,
      "step": 6325
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8405584018225507,
      "learning_rate": 9.193912498032376e-07,
      "loss": 0.7181,
      "step": 6326
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.299667898093061,
      "learning_rate": 9.181935911212231e-07,
      "loss": 0.6259,
      "step": 6327
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2958161486211488,
      "learning_rate": 9.169966341490499e-07,
      "loss": 0.7146,
      "step": 6328
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4196643056948008,
      "learning_rate": 9.158003790924908e-07,
      "loss": 0.7289,
      "step": 6329
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.384681825509665,
      "learning_rate": 9.146048261571944e-07,
      "loss": 0.7808,
      "step": 6330
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3704487346897962,
      "learning_rate": 9.134099755486892e-07,
      "loss": 0.8219,
      "step": 6331
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3595931854932957,
      "learning_rate": 9.122158274723863e-07,
      "loss": 0.6901,
      "step": 6332
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.395398787714282,
      "learning_rate": 9.110223821335723e-07,
      "loss": 0.8123,
      "step": 6333
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5399486128536295,
      "learning_rate": 9.098296397374146e-07,
      "loss": 0.7004,
      "step": 6334
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2453577351554583,
      "learning_rate": 9.086376004889591e-07,
      "loss": 0.7274,
      "step": 6335
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2146112018775819,
      "learning_rate": 9.074462645931309e-07,
      "loss": 0.7594,
      "step": 6336
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.384387033073176,
      "learning_rate": 9.062556322547333e-07,
      "loss": 0.7014,
      "step": 6337
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8727732705535173,
      "learning_rate": 9.050657036784516e-07,
      "loss": 0.6888,
      "step": 6338
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5204854451835705,
      "learning_rate": 9.038764790688492e-07,
      "loss": 0.7704,
      "step": 6339
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4292489745763712,
      "learning_rate": 9.026879586303666e-07,
      "loss": 0.7038,
      "step": 6340
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3777729895743014,
      "learning_rate": 9.015001425673242e-07,
      "loss": 0.7528,
      "step": 6341
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4023988682129618,
      "learning_rate": 9.003130310839203e-07,
      "loss": 0.6692,
      "step": 6342
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3998950086858406,
      "learning_rate": 8.99126624384235e-07,
      "loss": 0.7007,
      "step": 6343
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6249022420753672,
      "learning_rate": 8.979409226722224e-07,
      "loss": 0.8089,
      "step": 6344
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2243861221693144,
      "learning_rate": 8.967559261517217e-07,
      "loss": 0.7819,
      "step": 6345
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3996799938141558,
      "learning_rate": 8.955716350264454e-07,
      "loss": 0.7565,
      "step": 6346
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4943947835537865,
      "learning_rate": 8.943880494999884e-07,
      "loss": 0.7287,
      "step": 6347
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.286676763511708,
      "learning_rate": 8.932051697758227e-07,
      "loss": 0.7561,
      "step": 6348
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.628523704162936,
      "learning_rate": 8.920229960572973e-07,
      "loss": 0.815,
      "step": 6349
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.262054443580081,
      "learning_rate": 8.908415285476435e-07,
      "loss": 0.7776,
      "step": 6350
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1902381852377226,
      "learning_rate": 8.89660767449968e-07,
      "loss": 0.764,
      "step": 6351
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4027594491620563,
      "learning_rate": 8.884807129672568e-07,
      "loss": 0.7,
      "step": 6352
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.420966064575447,
      "learning_rate": 8.873013653023765e-07,
      "loss": 0.8045,
      "step": 6353
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4199611346892136,
      "learning_rate": 8.861227246580706e-07,
      "loss": 0.687,
      "step": 6354
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5860190692181495,
      "learning_rate": 8.849447912369591e-07,
      "loss": 0.7961,
      "step": 6355
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7207392378332553,
      "learning_rate": 8.837675652415451e-07,
      "loss": 0.6587,
      "step": 6356
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3882699915649777,
      "learning_rate": 8.82591046874206e-07,
      "loss": 0.7867,
      "step": 6357
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.346152583820256,
      "learning_rate": 8.814152363371992e-07,
      "loss": 0.8566,
      "step": 6358
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5381617346387406,
      "learning_rate": 8.802401338326582e-07,
      "loss": 0.7383,
      "step": 6359
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.315126697158174,
      "learning_rate": 8.790657395626001e-07,
      "loss": 0.823,
      "step": 6360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8394635061422139,
      "learning_rate": 8.778920537289154e-07,
      "loss": 0.7177,
      "step": 6361
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9922063496332953,
      "learning_rate": 8.767190765333744e-07,
      "loss": 0.7053,
      "step": 6362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7477752970879756,
      "learning_rate": 8.755468081776252e-07,
      "loss": 0.6502,
      "step": 6363
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2956981996614796,
      "learning_rate": 8.743752488631946e-07,
      "loss": 0.6749,
      "step": 6364
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8071526353808115,
      "learning_rate": 8.732043987914856e-07,
      "loss": 0.6096,
      "step": 6365
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5437834394605943,
      "learning_rate": 8.720342581637836e-07,
      "loss": 0.7478,
      "step": 6366
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4371015967708103,
      "learning_rate": 8.708648271812469e-07,
      "loss": 0.8138,
      "step": 6367
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.33350901478372,
      "learning_rate": 8.696961060449166e-07,
      "loss": 0.8094,
      "step": 6368
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.284963295874719,
      "learning_rate": 8.685280949557084e-07,
      "loss": 0.8042,
      "step": 6369
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2497047456190573,
      "learning_rate": 8.673607941144169e-07,
      "loss": 0.6912,
      "step": 6370
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2885832734504448,
      "learning_rate": 8.661942037217141e-07,
      "loss": 0.8484,
      "step": 6371
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4014823943670587,
      "learning_rate": 8.650283239781515e-07,
      "loss": 0.7848,
      "step": 6372
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4157488317859916,
      "learning_rate": 8.638631550841553e-07,
      "loss": 0.7218,
      "step": 6373
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4785542125725257,
      "learning_rate": 8.626986972400326e-07,
      "loss": 0.7808,
      "step": 6374
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.901918219645304,
      "learning_rate": 8.615349506459691e-07,
      "loss": 0.8055,
      "step": 6375
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2127671544236427,
      "learning_rate": 8.603719155020246e-07,
      "loss": 0.6717,
      "step": 6376
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2103276016885671,
      "learning_rate": 8.592095920081383e-07,
      "loss": 0.7043,
      "step": 6377
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.898847819255937,
      "learning_rate": 8.580479803641279e-07,
      "loss": 0.8229,
      "step": 6378
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6081884486821563,
      "learning_rate": 8.568870807696872e-07,
      "loss": 0.7812,
      "step": 6379
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8831094167453524,
      "learning_rate": 8.557268934243868e-07,
      "loss": 0.5628,
      "step": 6380
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9113611425394479,
      "learning_rate": 8.545674185276792e-07,
      "loss": 0.6113,
      "step": 6381
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.474206397564138,
      "learning_rate": 8.534086562788907e-07,
      "loss": 0.7309,
      "step": 6382
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3004805770268024,
      "learning_rate": 8.52250606877224e-07,
      "loss": 0.714,
      "step": 6383
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5375334362255675,
      "learning_rate": 8.510932705217645e-07,
      "loss": 0.7544,
      "step": 6384
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3275930649237286,
      "learning_rate": 8.499366474114695e-07,
      "loss": 0.7134,
      "step": 6385
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2318225290332916,
      "learning_rate": 8.487807377451767e-07,
      "loss": 0.7972,
      "step": 6386
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8288692941351955,
      "learning_rate": 8.476255417216007e-07,
      "loss": 0.6932,
      "step": 6387
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5667701303607746,
      "learning_rate": 8.464710595393306e-07,
      "loss": 0.6991,
      "step": 6388
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2913885139202876,
      "learning_rate": 8.453172913968382e-07,
      "loss": 0.7892,
      "step": 6389
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4448115774503805,
      "learning_rate": 8.441642374924692e-07,
      "loss": 0.8157,
      "step": 6390
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2824099605791404,
      "learning_rate": 8.430118980244462e-07,
      "loss": 0.7229,
      "step": 6391
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1965968947457881,
      "learning_rate": 8.418602731908687e-07,
      "loss": 0.7021,
      "step": 6392
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.339683345945202,
      "learning_rate": 8.407093631897168e-07,
      "loss": 0.8187,
      "step": 6393
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2650396556558314,
      "learning_rate": 8.395591682188442e-07,
      "loss": 0.7041,
      "step": 6394
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7988475610718332,
      "learning_rate": 8.384096884759807e-07,
      "loss": 0.6526,
      "step": 6395
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3508213519486918,
      "learning_rate": 8.372609241587387e-07,
      "loss": 0.7355,
      "step": 6396
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6547974785696113,
      "learning_rate": 8.361128754646025e-07,
      "loss": 0.7424,
      "step": 6397
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9327752406786054,
      "learning_rate": 8.349655425909348e-07,
      "loss": 0.6814,
      "step": 6398
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1575672287180092,
      "learning_rate": 8.338189257349755e-07,
      "loss": 0.7795,
      "step": 6399
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4711567067089648,
      "learning_rate": 8.326730250938414e-07,
      "loss": 0.7286,
      "step": 6400
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.554355241750073,
      "learning_rate": 8.31527840864525e-07,
      "loss": 0.7887,
      "step": 6401
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.444099811633456,
      "learning_rate": 8.303833732438988e-07,
      "loss": 0.7403,
      "step": 6402
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4198117259742322,
      "learning_rate": 8.29239622428708e-07,
      "loss": 0.7911,
      "step": 6403
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.288490122318964,
      "learning_rate": 8.280965886155789e-07,
      "loss": 0.7076,
      "step": 6404
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7635269612942861,
      "learning_rate": 8.26954272001011e-07,
      "loss": 0.6193,
      "step": 6405
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3556108095572847,
      "learning_rate": 8.25812672781382e-07,
      "loss": 0.6979,
      "step": 6406
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3176459276409926,
      "learning_rate": 8.246717911529456e-07,
      "loss": 0.734,
      "step": 6407
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8043880483096452,
      "learning_rate": 8.235316273118333e-07,
      "loss": 0.6633,
      "step": 6408
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.428702342191749,
      "learning_rate": 8.223921814540503e-07,
      "loss": 0.7552,
      "step": 6409
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4186566229996485,
      "learning_rate": 8.212534537754841e-07,
      "loss": 0.8049,
      "step": 6410
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8868408831809587,
      "learning_rate": 8.201154444718917e-07,
      "loss": 0.7988,
      "step": 6411
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3756037026638606,
      "learning_rate": 8.189781537389135e-07,
      "loss": 0.777,
      "step": 6412
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3741347164613766,
      "learning_rate": 8.178415817720609e-07,
      "loss": 0.7289,
      "step": 6413
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5401406419770645,
      "learning_rate": 8.16705728766724e-07,
      "loss": 0.788,
      "step": 6414
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4113615945614828,
      "learning_rate": 8.155705949181691e-07,
      "loss": 0.7169,
      "step": 6415
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2416599359098774,
      "learning_rate": 8.144361804215384e-07,
      "loss": 0.7709,
      "step": 6416
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6363987860210885,
      "learning_rate": 8.133024854718524e-07,
      "loss": 0.761,
      "step": 6417
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2210930313297756,
      "learning_rate": 8.121695102640053e-07,
      "loss": 0.6198,
      "step": 6418
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.317620378295458,
      "learning_rate": 8.110372549927692e-07,
      "loss": 0.7773,
      "step": 6419
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.292812295281077,
      "learning_rate": 8.099057198527899e-07,
      "loss": 0.8264,
      "step": 6420
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2274102324898963,
      "learning_rate": 8.087749050385952e-07,
      "loss": 0.6781,
      "step": 6421
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2830642214573995,
      "learning_rate": 8.076448107445822e-07,
      "loss": 0.7305,
      "step": 6422
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2571825962891174,
      "learning_rate": 8.065154371650286e-07,
      "loss": 0.8167,
      "step": 6423
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2954552946959612,
      "learning_rate": 8.053867844940855e-07,
      "loss": 0.7207,
      "step": 6424
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3540860550752083,
      "learning_rate": 8.042588529257828e-07,
      "loss": 0.7441,
      "step": 6425
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.30584350786939,
      "learning_rate": 8.031316426540254e-07,
      "loss": 0.6863,
      "step": 6426
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3505019550236415,
      "learning_rate": 8.02005153872592e-07,
      "loss": 0.8604,
      "step": 6427
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5099527837554294,
      "learning_rate": 8.008793867751402e-07,
      "loss": 0.7381,
      "step": 6428
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5314766159586228,
      "learning_rate": 7.997543415552011e-07,
      "loss": 0.8587,
      "step": 6429
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8126650197737797,
      "learning_rate": 7.986300184061857e-07,
      "loss": 0.7311,
      "step": 6430
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4233976113014188,
      "learning_rate": 7.975064175213748e-07,
      "loss": 0.8263,
      "step": 6431
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3400936833246047,
      "learning_rate": 7.963835390939317e-07,
      "loss": 0.8699,
      "step": 6432
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3836024982547732,
      "learning_rate": 7.952613833168909e-07,
      "loss": 0.7307,
      "step": 6433
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4745383490492896,
      "learning_rate": 7.941399503831637e-07,
      "loss": 0.8077,
      "step": 6434
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3088062573604413,
      "learning_rate": 7.930192404855375e-07,
      "loss": 0.7659,
      "step": 6435
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3524368617305065,
      "learning_rate": 7.918992538166753e-07,
      "loss": 0.7521,
      "step": 6436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8191020454854524,
      "learning_rate": 7.907799905691144e-07,
      "loss": 0.6612,
      "step": 6437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8558805597806265,
      "learning_rate": 7.896614509352724e-07,
      "loss": 0.6661,
      "step": 6438
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0162260578676063,
      "learning_rate": 7.885436351074355e-07,
      "loss": 0.7455,
      "step": 6439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8435494309033367,
      "learning_rate": 7.874265432777728e-07,
      "loss": 0.7006,
      "step": 6440
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1590871899870239,
      "learning_rate": 7.863101756383235e-07,
      "loss": 0.772,
      "step": 6441
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4186040598676724,
      "learning_rate": 7.851945323810045e-07,
      "loss": 0.8783,
      "step": 6442
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1866553060392269,
      "learning_rate": 7.840796136976075e-07,
      "loss": 0.7528,
      "step": 6443
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4138146807907281,
      "learning_rate": 7.829654197797998e-07,
      "loss": 0.812,
      "step": 6444
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4747788468607201,
      "learning_rate": 7.818519508191236e-07,
      "loss": 0.7518,
      "step": 6445
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3784090583801243,
      "learning_rate": 7.807392070069992e-07,
      "loss": 0.7579,
      "step": 6446
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.205520041277843,
      "learning_rate": 7.796271885347189e-07,
      "loss": 0.7134,
      "step": 6447
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5380705389987548,
      "learning_rate": 7.785158955934508e-07,
      "loss": 0.681,
      "step": 6448
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3103116154234409,
      "learning_rate": 7.774053283742406e-07,
      "loss": 0.7025,
      "step": 6449
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3804884898519116,
      "learning_rate": 7.762954870680067e-07,
      "loss": 0.7156,
      "step": 6450
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4880342887776299,
      "learning_rate": 7.751863718655444e-07,
      "loss": 0.8722,
      "step": 6451
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2791508741214295,
      "learning_rate": 7.740779829575218e-07,
      "loss": 0.7379,
      "step": 6452
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7036840716065345,
      "learning_rate": 7.729703205344863e-07,
      "loss": 0.7581,
      "step": 6453
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2242598926271286,
      "learning_rate": 7.718633847868568e-07,
      "loss": 0.767,
      "step": 6454
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7899087210035195,
      "learning_rate": 7.707571759049281e-07,
      "loss": 0.6686,
      "step": 6455
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2650247902600946,
      "learning_rate": 7.696516940788701e-07,
      "loss": 0.8402,
      "step": 6456
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7125489319476614,
      "learning_rate": 7.685469394987271e-07,
      "loss": 0.7397,
      "step": 6457
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2299315354941858,
      "learning_rate": 7.67442912354422e-07,
      "loss": 0.7567,
      "step": 6458
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3772636590792546,
      "learning_rate": 7.663396128357481e-07,
      "loss": 0.8231,
      "step": 6459
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4746138130488518,
      "learning_rate": 7.652370411323745e-07,
      "loss": 0.7873,
      "step": 6460
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2111962436921204,
      "learning_rate": 7.641351974338478e-07,
      "loss": 0.705,
      "step": 6461
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8534446601088784,
      "learning_rate": 7.630340819295879e-07,
      "loss": 0.6015,
      "step": 6462
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4821546775859904,
      "learning_rate": 7.619336948088879e-07,
      "loss": 0.658,
      "step": 6463
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5315840252714754,
      "learning_rate": 7.608340362609174e-07,
      "loss": 0.8196,
      "step": 6464
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2969205000635506,
      "learning_rate": 7.597351064747211e-07,
      "loss": 0.6675,
      "step": 6465
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4492733765353445,
      "learning_rate": 7.586369056392162e-07,
      "loss": 0.8276,
      "step": 6466
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.577623104219131,
      "learning_rate": 7.575394339431969e-07,
      "loss": 0.7129,
      "step": 6467
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2741884054166441,
      "learning_rate": 7.564426915753331e-07,
      "loss": 0.8086,
      "step": 6468
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.227165693063055,
      "learning_rate": 7.553466787241665e-07,
      "loss": 0.7697,
      "step": 6469
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.382562703255245,
      "learning_rate": 7.542513955781139e-07,
      "loss": 0.7772,
      "step": 6470
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.788626812823415,
      "learning_rate": 7.53156842325467e-07,
      "loss": 0.6752,
      "step": 6471
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3100465231044307,
      "learning_rate": 7.520630191543932e-07,
      "loss": 0.7119,
      "step": 6472
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3646528142101952,
      "learning_rate": 7.509699262529308e-07,
      "loss": 0.8056,
      "step": 6473
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1579700241396222,
      "learning_rate": 7.498775638089989e-07,
      "loss": 0.7149,
      "step": 6474
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3590101759556759,
      "learning_rate": 7.487859320103847e-07,
      "loss": 0.6834,
      "step": 6475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8107259812656938,
      "learning_rate": 7.476950310447523e-07,
      "loss": 0.6153,
      "step": 6476
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6240406087633836,
      "learning_rate": 7.466048610996423e-07,
      "loss": 0.7461,
      "step": 6477
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2053056316248731,
      "learning_rate": 7.455154223624661e-07,
      "loss": 0.7287,
      "step": 6478
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5043525490372565,
      "learning_rate": 7.444267150205108e-07,
      "loss": 0.7805,
      "step": 6479
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5841737953043733,
      "learning_rate": 7.433387392609387e-07,
      "loss": 0.8651,
      "step": 6480
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2481207526351747,
      "learning_rate": 7.422514952707832e-07,
      "loss": 0.7195,
      "step": 6481
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.451339419252236,
      "learning_rate": 7.411649832369566e-07,
      "loss": 0.7232,
      "step": 6482
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6475761008988288,
      "learning_rate": 7.400792033462428e-07,
      "loss": 0.8524,
      "step": 6483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7620346643994805,
      "learning_rate": 7.389941557852987e-07,
      "loss": 0.6469,
      "step": 6484
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6328772539467273,
      "learning_rate": 7.379098407406554e-07,
      "loss": 0.8228,
      "step": 6485
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.287296121566819,
      "learning_rate": 7.368262583987229e-07,
      "loss": 0.7383,
      "step": 6486
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2466933503886197,
      "learning_rate": 7.357434089457788e-07,
      "loss": 0.7331,
      "step": 6487
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5392023996979056,
      "learning_rate": 7.346612925679774e-07,
      "loss": 0.7293,
      "step": 6488
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.323890053982999,
      "learning_rate": 7.33579909451349e-07,
      "loss": 0.831,
      "step": 6489
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.088015779715576,
      "learning_rate": 7.324992597817948e-07,
      "loss": 0.7326,
      "step": 6490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8272867964520567,
      "learning_rate": 7.314193437450911e-07,
      "loss": 0.6301,
      "step": 6491
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4572531129615407,
      "learning_rate": 7.30340161526888e-07,
      "loss": 0.7143,
      "step": 6492
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4747383043367166,
      "learning_rate": 7.292617133127083e-07,
      "loss": 0.7552,
      "step": 6493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7855917861765429,
      "learning_rate": 7.281839992879503e-07,
      "loss": 0.6911,
      "step": 6494
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3198922746777384,
      "learning_rate": 7.271070196378859e-07,
      "loss": 0.7377,
      "step": 6495
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3208459399877002,
      "learning_rate": 7.260307745476619e-07,
      "loss": 0.7091,
      "step": 6496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7612119240072056,
      "learning_rate": 7.249552642022956e-07,
      "loss": 0.6385,
      "step": 6497
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.370889369344029,
      "learning_rate": 7.238804887866796e-07,
      "loss": 0.7835,
      "step": 6498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7963692564698166,
      "learning_rate": 7.228064484855807e-07,
      "loss": 0.6766,
      "step": 6499
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5258157844152909,
      "learning_rate": 7.217331434836395e-07,
      "loss": 0.7327,
      "step": 6500
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5030300519204691,
      "learning_rate": 7.206605739653683e-07,
      "loss": 0.8213,
      "step": 6501
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3560613526788738,
      "learning_rate": 7.195887401151536e-07,
      "loss": 0.701,
      "step": 6502
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3946368560544715,
      "learning_rate": 7.185176421172573e-07,
      "loss": 0.6789,
      "step": 6503
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4052400601976205,
      "learning_rate": 7.174472801558147e-07,
      "loss": 0.7866,
      "step": 6504
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.272952304909098,
      "learning_rate": 7.163776544148321e-07,
      "loss": 0.7168,
      "step": 6505
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.734614721874703,
      "learning_rate": 7.153087650781909e-07,
      "loss": 0.7661,
      "step": 6506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.757918597035831,
      "learning_rate": 7.142406123296452e-07,
      "loss": 0.6624,
      "step": 6507
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3580289925434286,
      "learning_rate": 7.131731963528232e-07,
      "loss": 0.8524,
      "step": 6508
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4111271104894814,
      "learning_rate": 7.121065173312253e-07,
      "loss": 0.7685,
      "step": 6509
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2657760184664182,
      "learning_rate": 7.110405754482269e-07,
      "loss": 0.7898,
      "step": 6510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9262031225494032,
      "learning_rate": 7.09975370887076e-07,
      "loss": 0.6571,
      "step": 6511
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4525140961207241,
      "learning_rate": 7.089109038308928e-07,
      "loss": 0.9082,
      "step": 6512
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3232321706358738,
      "learning_rate": 7.078471744626708e-07,
      "loss": 0.7254,
      "step": 6513
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3560639538470487,
      "learning_rate": 7.067841829652794e-07,
      "loss": 0.8187,
      "step": 6514
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4389441109822507,
      "learning_rate": 7.057219295214579e-07,
      "loss": 0.7618,
      "step": 6515
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3769097968918824,
      "learning_rate": 7.046604143138198e-07,
      "loss": 0.7325,
      "step": 6516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8313589634935863,
      "learning_rate": 7.035996375248527e-07,
      "loss": 0.6037,
      "step": 6517
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3382065423984084,
      "learning_rate": 7.025395993369166e-07,
      "loss": 0.7476,
      "step": 6518
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3550528551078327,
      "learning_rate": 7.01480299932244e-07,
      "loss": 0.8375,
      "step": 6519
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4098575691987423,
      "learning_rate": 7.004217394929402e-07,
      "loss": 0.7588,
      "step": 6520
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.3766946801650315,
      "learning_rate": 6.993639182009843e-07,
      "loss": 0.7268,
      "step": 6521
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3565377544382546,
      "learning_rate": 6.983068362382272e-07,
      "loss": 0.7364,
      "step": 6522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7822038666635137,
      "learning_rate": 6.972504937863955e-07,
      "loss": 0.6436,
      "step": 6523
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4385722955544127,
      "learning_rate": 6.961948910270844e-07,
      "loss": 0.7227,
      "step": 6524
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6033688898754046,
      "learning_rate": 6.951400281417669e-07,
      "loss": 0.7556,
      "step": 6525
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.30777827603168,
      "learning_rate": 6.940859053117843e-07,
      "loss": 0.6766,
      "step": 6526
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4825466952651147,
      "learning_rate": 6.930325227183537e-07,
      "loss": 0.7086,
      "step": 6527
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1756121415772105,
      "learning_rate": 6.919798805425626e-07,
      "loss": 0.6526,
      "step": 6528
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3219537122385638,
      "learning_rate": 6.909279789653734e-07,
      "loss": 0.6844,
      "step": 6529
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3750605195942083,
      "learning_rate": 6.89876818167618e-07,
      "loss": 0.7636,
      "step": 6530
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.488340907104074,
      "learning_rate": 6.888263983300048e-07,
      "loss": 0.6992,
      "step": 6531
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8552032849236635,
      "learning_rate": 6.877767196331147e-07,
      "loss": 0.7171,
      "step": 6532
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3848529757256443,
      "learning_rate": 6.867277822573975e-07,
      "loss": 0.8028,
      "step": 6533
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2152785434210895,
      "learning_rate": 6.856795863831789e-07,
      "loss": 0.781,
      "step": 6534
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3088423809484102,
      "learning_rate": 6.846321321906551e-07,
      "loss": 0.7255,
      "step": 6535
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.401331922957025,
      "learning_rate": 6.835854198598957e-07,
      "loss": 0.7966,
      "step": 6536
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4769172553045646,
      "learning_rate": 6.825394495708415e-07,
      "loss": 0.7091,
      "step": 6537
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2672012015518344,
      "learning_rate": 6.814942215033099e-07,
      "loss": 0.67,
      "step": 6538
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3355489328123769,
      "learning_rate": 6.804497358369855e-07,
      "loss": 0.7986,
      "step": 6539
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8905897595428591,
      "learning_rate": 6.794059927514268e-07,
      "loss": 0.7284,
      "step": 6540
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5033004208327192,
      "learning_rate": 6.783629924260682e-07,
      "loss": 0.6779,
      "step": 6541
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.328134484379832,
      "learning_rate": 6.773207350402117e-07,
      "loss": 0.7544,
      "step": 6542
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3176901606411031,
      "learning_rate": 6.762792207730334e-07,
      "loss": 0.7971,
      "step": 6543
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4287031356565438,
      "learning_rate": 6.752384498035824e-07,
      "loss": 0.8085,
      "step": 6544
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4089075430187255,
      "learning_rate": 6.741984223107773e-07,
      "loss": 0.695,
      "step": 6545
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3516657423936533,
      "learning_rate": 6.731591384734138e-07,
      "loss": 0.7482,
      "step": 6546
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.858012981333028,
      "learning_rate": 6.721205984701551e-07,
      "loss": 0.7092,
      "step": 6547
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.326197819595887,
      "learning_rate": 6.71082802479539e-07,
      "loss": 0.7832,
      "step": 6548
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7733597067286032,
      "learning_rate": 6.70045750679974e-07,
      "loss": 0.6267,
      "step": 6549
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.7040832140800903,
      "learning_rate": 6.690094432497407e-07,
      "loss": 0.7597,
      "step": 6550
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2519582962436504,
      "learning_rate": 6.679738803669944e-07,
      "loss": 0.7266,
      "step": 6551
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7599230182059242,
      "learning_rate": 6.669390622097577e-07,
      "loss": 0.6804,
      "step": 6552
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4500895075828941,
      "learning_rate": 6.65904988955931e-07,
      "loss": 0.8062,
      "step": 6553
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2186149487612974,
      "learning_rate": 6.648716607832811e-07,
      "loss": 0.7733,
      "step": 6554
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4190979812699775,
      "learning_rate": 6.638390778694504e-07,
      "loss": 0.7534,
      "step": 6555
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5595223996215364,
      "learning_rate": 6.628072403919511e-07,
      "loss": 0.7632,
      "step": 6556
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.288381512082193,
      "learning_rate": 6.617761485281687e-07,
      "loss": 0.7241,
      "step": 6557
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2347694033059602,
      "learning_rate": 6.607458024553576e-07,
      "loss": 0.79,
      "step": 6558
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4360100501139108,
      "learning_rate": 6.597162023506492e-07,
      "loss": 0.7347,
      "step": 6559
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.784659463508592,
      "learning_rate": 6.586873483910416e-07,
      "loss": 0.6519,
      "step": 6560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.819231825210462,
      "learning_rate": 6.576592407534088e-07,
      "loss": 0.6835,
      "step": 6561
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.487159216795858,
      "learning_rate": 6.566318796144933e-07,
      "loss": 0.7842,
      "step": 6562
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2654326924215145,
      "learning_rate": 6.556052651509104e-07,
      "loss": 0.7637,
      "step": 6563
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5059275162332628,
      "learning_rate": 6.545793975391468e-07,
      "loss": 0.8089,
      "step": 6564
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.674239172384998,
      "learning_rate": 6.535542769555609e-07,
      "loss": 0.7788,
      "step": 6565
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4458096564791416,
      "learning_rate": 6.525299035763827e-07,
      "loss": 0.7788,
      "step": 6566
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4353362088529604,
      "learning_rate": 6.515062775777148e-07,
      "loss": 0.7681,
      "step": 6567
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4588915905746143,
      "learning_rate": 6.504833991355292e-07,
      "loss": 0.7481,
      "step": 6568
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.400489021497777,
      "learning_rate": 6.494612684256718e-07,
      "loss": 0.7567,
      "step": 6569
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5011486634380784,
      "learning_rate": 6.484398856238582e-07,
      "loss": 0.8225,
      "step": 6570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7466216404219554,
      "learning_rate": 6.474192509056759e-07,
      "loss": 0.6029,
      "step": 6571
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4010465138475205,
      "learning_rate": 6.463993644465843e-07,
      "loss": 0.7115,
      "step": 6572
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3349218328292425,
      "learning_rate": 6.453802264219117e-07,
      "loss": 0.7346,
      "step": 6573
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.360280873916413,
      "learning_rate": 6.443618370068622e-07,
      "loss": 0.7051,
      "step": 6574
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3916243023743127,
      "learning_rate": 6.43344196376508e-07,
      "loss": 0.7241,
      "step": 6575
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3648786916942248,
      "learning_rate": 6.423273047057932e-07,
      "loss": 0.8091,
      "step": 6576
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2726721539796015,
      "learning_rate": 6.413111621695322e-07,
      "loss": 0.7062,
      "step": 6577
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3794362326751897,
      "learning_rate": 6.402957689424139e-07,
      "loss": 0.7715,
      "step": 6578
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5274477846547772,
      "learning_rate": 6.392811251989944e-07,
      "loss": 0.728,
      "step": 6579
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1976275280287476,
      "learning_rate": 6.382672311137039e-07,
      "loss": 0.9521,
      "step": 6580
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3898466025489091,
      "learning_rate": 6.372540868608401e-07,
      "loss": 0.7854,
      "step": 6581
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.377637605529255,
      "learning_rate": 6.362416926145775e-07,
      "loss": 0.7331,
      "step": 6582
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.443139445180012,
      "learning_rate": 6.352300485489571e-07,
      "loss": 0.7772,
      "step": 6583
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8142645602227154,
      "learning_rate": 6.342191548378923e-07,
      "loss": 0.7332,
      "step": 6584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7742661187356349,
      "learning_rate": 6.33209011655167e-07,
      "loss": 0.6651,
      "step": 6585
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5846482030465914,
      "learning_rate": 6.321996191744368e-07,
      "loss": 0.789,
      "step": 6586
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.452663988383989,
      "learning_rate": 6.311909775692265e-07,
      "loss": 0.6765,
      "step": 6587
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3993971873305724,
      "learning_rate": 6.301830870129349e-07,
      "loss": 0.7511,
      "step": 6588
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4806236719095331,
      "learning_rate": 6.291759476788312e-07,
      "loss": 0.8245,
      "step": 6589
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2484577633725635,
      "learning_rate": 6.281695597400533e-07,
      "loss": 0.7364,
      "step": 6590
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.675193048157943,
      "learning_rate": 6.271639233696103e-07,
      "loss": 0.7455,
      "step": 6591
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.509966030853173,
      "learning_rate": 6.261590387403832e-07,
      "loss": 0.7671,
      "step": 6592
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3368388130957825,
      "learning_rate": 6.251549060251233e-07,
      "loss": 0.7951,
      "step": 6593
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6619292257808014,
      "learning_rate": 6.241515253964515e-07,
      "loss": 0.8055,
      "step": 6594
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4735143066754857,
      "learning_rate": 6.231488970268628e-07,
      "loss": 0.7643,
      "step": 6595
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2709341019380314,
      "learning_rate": 6.221470210887182e-07,
      "loss": 0.6758,
      "step": 6596
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.298811284778196,
      "learning_rate": 6.211458977542545e-07,
      "loss": 0.7059,
      "step": 6597
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.608582540194366,
      "learning_rate": 6.201455271955747e-07,
      "loss": 0.722,
      "step": 6598
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4632798111728251,
      "learning_rate": 6.191459095846547e-07,
      "loss": 0.7342,
      "step": 6599
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3079951253277835,
      "learning_rate": 6.181470450933397e-07,
      "loss": 0.7777,
      "step": 6600
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3528032697238304,
      "learning_rate": 6.171489338933467e-07,
      "loss": 0.8363,
      "step": 6601
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0049258273167085,
      "learning_rate": 6.161515761562614e-07,
      "loss": 0.7601,
      "step": 6602
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5443790431290567,
      "learning_rate": 6.151549720535433e-07,
      "loss": 0.7792,
      "step": 6603
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5009242338660818,
      "learning_rate": 6.141591217565185e-07,
      "loss": 0.6747,
      "step": 6604
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3006302418396776,
      "learning_rate": 6.131640254363847e-07,
      "loss": 0.7454,
      "step": 6605
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1997871232663275,
      "learning_rate": 6.121696832642126e-07,
      "loss": 0.7505,
      "step": 6606
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2588812124098891,
      "learning_rate": 6.111760954109402e-07,
      "loss": 0.7222,
      "step": 6607
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4816579015520481,
      "learning_rate": 6.101832620473763e-07,
      "loss": 0.8529,
      "step": 6608
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3775106223481082,
      "learning_rate": 6.091911833441999e-07,
      "loss": 0.6847,
      "step": 6609
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4075541502834994,
      "learning_rate": 6.081998594719629e-07,
      "loss": 0.8414,
      "step": 6610
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9286523085651344,
      "learning_rate": 6.07209290601084e-07,
      "loss": 0.6369,
      "step": 6611
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4938046823209952,
      "learning_rate": 6.062194769018542e-07,
      "loss": 0.8197,
      "step": 6612
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5217684031972314,
      "learning_rate": 6.05230418544433e-07,
      "loss": 0.783,
      "step": 6613
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4096688432264821,
      "learning_rate": 6.042421156988498e-07,
      "loss": 0.7448,
      "step": 6614
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4886346719241625,
      "learning_rate": 6.032545685350088e-07,
      "loss": 0.7779,
      "step": 6615
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.352426389912077,
      "learning_rate": 6.022677772226781e-07,
      "loss": 0.7003,
      "step": 6616
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3808178737678636,
      "learning_rate": 6.012817419314992e-07,
      "loss": 0.8087,
      "step": 6617
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.321682233488096,
      "learning_rate": 6.002964628309838e-07,
      "loss": 0.7192,
      "step": 6618
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3589576902564746,
      "learning_rate": 5.993119400905123e-07,
      "loss": 0.7615,
      "step": 6619
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.360058316955191,
      "learning_rate": 5.983281738793351e-07,
      "loss": 0.8217,
      "step": 6620
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2721949951431635,
      "learning_rate": 5.973451643665734e-07,
      "loss": 0.7654,
      "step": 6621
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3720047435380947,
      "learning_rate": 5.963629117212183e-07,
      "loss": 0.6964,
      "step": 6622
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.352612095387425,
      "learning_rate": 5.95381416112129e-07,
      "loss": 0.7511,
      "step": 6623
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5709553749622989,
      "learning_rate": 5.944006777080363e-07,
      "loss": 0.8078,
      "step": 6624
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3460983120800718,
      "learning_rate": 5.934206966775429e-07,
      "loss": 0.7104,
      "step": 6625
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4234219889398982,
      "learning_rate": 5.924414731891171e-07,
      "loss": 0.8102,
      "step": 6626
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8151429157272675,
      "learning_rate": 5.914630074110989e-07,
      "loss": 0.6143,
      "step": 6627
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5386359707358295,
      "learning_rate": 5.904852995116977e-07,
      "loss": 0.7464,
      "step": 6628
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2376964261053331,
      "learning_rate": 5.895083496589932e-07,
      "loss": 0.7035,
      "step": 6629
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3494186397519279,
      "learning_rate": 5.885321580209324e-07,
      "loss": 0.7997,
      "step": 6630
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.420158139190958,
      "learning_rate": 5.875567247653374e-07,
      "loss": 0.6428,
      "step": 6631
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3769236928291733,
      "learning_rate": 5.865820500598951e-07,
      "loss": 0.7897,
      "step": 6632
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9233754630367048,
      "learning_rate": 5.85608134072162e-07,
      "loss": 0.6789,
      "step": 6633
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5771715726889373,
      "learning_rate": 5.846349769695675e-07,
      "loss": 0.836,
      "step": 6634
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3022193395137682,
      "learning_rate": 5.836625789194078e-07,
      "loss": 0.7056,
      "step": 6635
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4655544864596972,
      "learning_rate": 5.826909400888492e-07,
      "loss": 0.7705,
      "step": 6636
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3805823513476745,
      "learning_rate": 5.81720060644928e-07,
      "loss": 0.7673,
      "step": 6637
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2959432610006407,
      "learning_rate": 5.807499407545486e-07,
      "loss": 0.7839,
      "step": 6638
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4015052772824828,
      "learning_rate": 5.797805805844875e-07,
      "loss": 0.8508,
      "step": 6639
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.607594737701152,
      "learning_rate": 5.788119803013881e-07,
      "loss": 0.6986,
      "step": 6640
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3465044110579905,
      "learning_rate": 5.778441400717644e-07,
      "loss": 0.7226,
      "step": 6641
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.42598975217019,
      "learning_rate": 5.768770600619978e-07,
      "loss": 0.8572,
      "step": 6642
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7770293038301694,
      "learning_rate": 5.759107404383429e-07,
      "loss": 0.6258,
      "step": 6643
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.293388155825038,
      "learning_rate": 5.749451813669205e-07,
      "loss": 0.795,
      "step": 6644
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.44013514517514,
      "learning_rate": 5.739803830137192e-07,
      "loss": 0.7672,
      "step": 6645
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.560791772432291,
      "learning_rate": 5.730163455446025e-07,
      "loss": 0.7711,
      "step": 6646
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2121796183552342,
      "learning_rate": 5.720530691252979e-07,
      "loss": 0.8023,
      "step": 6647
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7532651600239574,
      "learning_rate": 5.71090553921404e-07,
      "loss": 0.6559,
      "step": 6648
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3172489343849332,
      "learning_rate": 5.701288000983884e-07,
      "loss": 0.7946,
      "step": 6649
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2603575087609749,
      "learning_rate": 5.691678078215873e-07,
      "loss": 0.7738,
      "step": 6650
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3652289932167085,
      "learning_rate": 5.682075772562051e-07,
      "loss": 0.7249,
      "step": 6651
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3088477679854864,
      "learning_rate": 5.672481085673199e-07,
      "loss": 0.6915,
      "step": 6652
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.325050869890945,
      "learning_rate": 5.662894019198722e-07,
      "loss": 0.8371,
      "step": 6653
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3113385457144036,
      "learning_rate": 5.653314574786778e-07,
      "loss": 0.7404,
      "step": 6654
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3381922148644696,
      "learning_rate": 5.643742754084164e-07,
      "loss": 0.7249,
      "step": 6655
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5441204169948117,
      "learning_rate": 5.634178558736397e-07,
      "loss": 0.7962,
      "step": 6656
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8247114924279126,
      "learning_rate": 5.624621990387669e-07,
      "loss": 0.6426,
      "step": 6657
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3640007628376989,
      "learning_rate": 5.615073050680859e-07,
      "loss": 0.7909,
      "step": 6658
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5334940481248576,
      "learning_rate": 5.605531741257536e-07,
      "loss": 0.8238,
      "step": 6659
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3034138728237028,
      "learning_rate": 5.595998063757985e-07,
      "loss": 0.7025,
      "step": 6660
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3459837751104435,
      "learning_rate": 5.586472019821132e-07,
      "loss": 0.7816,
      "step": 6661
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7750513958336677,
      "learning_rate": 5.576953611084635e-07,
      "loss": 0.6179,
      "step": 6662
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.349183905685506,
      "learning_rate": 5.567442839184805e-07,
      "loss": 0.7004,
      "step": 6663
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4504492841829846,
      "learning_rate": 5.557939705756665e-07,
      "loss": 0.7508,
      "step": 6664
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.292651527543143,
      "learning_rate": 5.548444212433901e-07,
      "loss": 0.7701,
      "step": 6665
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1859410417112433,
      "learning_rate": 5.538956360848891e-07,
      "loss": 0.7313,
      "step": 6666
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5229492909270919,
      "learning_rate": 5.529476152632735e-07,
      "loss": 0.6971,
      "step": 6667
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8043098068554799,
      "learning_rate": 5.520003589415168e-07,
      "loss": 0.6458,
      "step": 6668
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1983435747896936,
      "learning_rate": 5.510538672824645e-07,
      "loss": 0.7156,
      "step": 6669
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4144513962812713,
      "learning_rate": 5.501081404488279e-07,
      "loss": 0.9089,
      "step": 6670
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7848356309616653,
      "learning_rate": 5.491631786031904e-07,
      "loss": 0.5819,
      "step": 6671
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1359525530536791,
      "learning_rate": 5.482189819080014e-07,
      "loss": 0.6586,
      "step": 6672
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3375888121420023,
      "learning_rate": 5.472755505255783e-07,
      "loss": 0.673,
      "step": 6673
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.381406430577585,
      "learning_rate": 5.463328846181081e-07,
      "loss": 0.7573,
      "step": 6674
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5843352080220916,
      "learning_rate": 5.453909843476468e-07,
      "loss": 0.7884,
      "step": 6675
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8953811822299179,
      "learning_rate": 5.444498498761181e-07,
      "loss": 0.63,
      "step": 6676
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2182067733415303,
      "learning_rate": 5.43509481365313e-07,
      "loss": 0.778,
      "step": 6677
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4293349499141554,
      "learning_rate": 5.425698789768924e-07,
      "loss": 0.7345,
      "step": 6678
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4517011514317653,
      "learning_rate": 5.416310428723837e-07,
      "loss": 0.7458,
      "step": 6679
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4230343798312275,
      "learning_rate": 5.40692973213185e-07,
      "loss": 0.703,
      "step": 6680
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3457578054803985,
      "learning_rate": 5.397556701605605e-07,
      "loss": 0.6775,
      "step": 6681
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4300471031155282,
      "learning_rate": 5.388191338756443e-07,
      "loss": 0.6362,
      "step": 6682
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.51675647350434,
      "learning_rate": 5.378833645194375e-07,
      "loss": 0.7528,
      "step": 6683
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3604826069971068,
      "learning_rate": 5.369483622528104e-07,
      "loss": 0.7275,
      "step": 6684
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5446676334084024,
      "learning_rate": 5.360141272364994e-07,
      "loss": 0.7695,
      "step": 6685
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2391860188542339,
      "learning_rate": 5.350806596311109e-07,
      "loss": 0.7011,
      "step": 6686
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4369857137777386,
      "learning_rate": 5.341479595971177e-07,
      "loss": 0.7813,
      "step": 6687
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2829145591042967,
      "learning_rate": 5.332160272948628e-07,
      "loss": 0.7205,
      "step": 6688
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3943819033142248,
      "learning_rate": 5.322848628845578e-07,
      "loss": 0.748,
      "step": 6689
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2740874964141342,
      "learning_rate": 5.313544665262782e-07,
      "loss": 0.7954,
      "step": 6690
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.364762694933837,
      "learning_rate": 5.304248383799715e-07,
      "loss": 0.7165,
      "step": 6691
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6400445453752972,
      "learning_rate": 5.294959786054505e-07,
      "loss": 0.7672,
      "step": 6692
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3684468929489177,
      "learning_rate": 5.285678873623973e-07,
      "loss": 0.7831,
      "step": 6693
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4622845230090311,
      "learning_rate": 5.276405648103616e-07,
      "loss": 0.7462,
      "step": 6694
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3966290424710663,
      "learning_rate": 5.267140111087604e-07,
      "loss": 0.747,
      "step": 6695
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.27192180339987,
      "learning_rate": 5.257882264168795e-07,
      "loss": 0.7837,
      "step": 6696
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4672198008018167,
      "learning_rate": 5.248632108938728e-07,
      "loss": 0.7582,
      "step": 6697
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.384420481501508,
      "learning_rate": 5.239389646987592e-07,
      "loss": 0.7584,
      "step": 6698
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4269393989776302,
      "learning_rate": 5.230154879904292e-07,
      "loss": 0.7903,
      "step": 6699
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2640212002535431,
      "learning_rate": 5.220927809276383e-07,
      "loss": 0.726,
      "step": 6700
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.080907974271279,
      "learning_rate": 5.211708436690111e-07,
      "loss": 0.818,
      "step": 6701
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3973454533379508,
      "learning_rate": 5.20249676373038e-07,
      "loss": 0.8093,
      "step": 6702
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8084272128870913,
      "learning_rate": 5.193292791980803e-07,
      "loss": 0.6471,
      "step": 6703
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3226558082248978,
      "learning_rate": 5.184096523023635e-07,
      "loss": 0.7355,
      "step": 6704
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3889792297840085,
      "learning_rate": 5.174907958439829e-07,
      "loss": 0.7782,
      "step": 6705
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9402435357677437,
      "learning_rate": 5.165727099808998e-07,
      "loss": 0.6951,
      "step": 6706
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.50975342812221,
      "learning_rate": 5.156553948709436e-07,
      "loss": 0.8693,
      "step": 6707
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.338168380608131,
      "learning_rate": 5.147388506718127e-07,
      "loss": 0.6972,
      "step": 6708
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4207042872545166,
      "learning_rate": 5.138230775410713e-07,
      "loss": 0.7557,
      "step": 6709
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6217546176883226,
      "learning_rate": 5.129080756361499e-07,
      "loss": 0.7521,
      "step": 6710
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5499429197184127,
      "learning_rate": 5.119938451143502e-07,
      "loss": 0.7413,
      "step": 6711
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5100589702968017,
      "learning_rate": 5.110803861328378e-07,
      "loss": 0.6875,
      "step": 6712
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3891758674915224,
      "learning_rate": 5.10167698848647e-07,
      "loss": 0.7765,
      "step": 6713
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2904396079852645,
      "learning_rate": 5.092557834186795e-07,
      "loss": 0.7425,
      "step": 6714
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.316505346391187,
      "learning_rate": 5.083446399997044e-07,
      "loss": 0.7618,
      "step": 6715
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3834839085062434,
      "learning_rate": 5.074342687483557e-07,
      "loss": 0.7662,
      "step": 6716
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4621815549771564,
      "learning_rate": 5.06524669821139e-07,
      "loss": 0.7465,
      "step": 6717
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3724713963263757,
      "learning_rate": 5.056158433744251e-07,
      "loss": 0.7384,
      "step": 6718
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.434757719831997,
      "learning_rate": 5.047077895644514e-07,
      "loss": 0.7734,
      "step": 6719
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5263604792890315,
      "learning_rate": 5.038005085473224e-07,
      "loss": 0.7449,
      "step": 6720
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2645592598580557,
      "learning_rate": 5.028940004790106e-07,
      "loss": 0.7717,
      "step": 6721
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3123275346088668,
      "learning_rate": 5.019882655153547e-07,
      "loss": 0.7672,
      "step": 6722
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8625471295551985,
      "learning_rate": 5.010833038120605e-07,
      "loss": 0.6885,
      "step": 6723
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3555753565996453,
      "learning_rate": 5.001791155247032e-07,
      "loss": 0.7028,
      "step": 6724
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8540927696556507,
      "learning_rate": 4.992757008087218e-07,
      "loss": 0.6693,
      "step": 6725
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4332598712133218,
      "learning_rate": 4.983730598194247e-07,
      "loss": 0.8504,
      "step": 6726
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4251005239234649,
      "learning_rate": 4.974711927119863e-07,
      "loss": 0.6811,
      "step": 6727
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4801552500864301,
      "learning_rate": 4.96570099641448e-07,
      "loss": 0.7223,
      "step": 6728
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3699932381356916,
      "learning_rate": 4.95669780762717e-07,
      "loss": 0.8071,
      "step": 6729
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.56516456513793,
      "learning_rate": 4.9477023623057e-07,
      "loss": 0.8085,
      "step": 6730
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.482796247663638,
      "learning_rate": 4.938714661996469e-07,
      "loss": 0.7928,
      "step": 6731
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2919440044253454,
      "learning_rate": 4.929734708244599e-07,
      "loss": 0.7116,
      "step": 6732
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4555439967461736,
      "learning_rate": 4.920762502593829e-07,
      "loss": 0.7978,
      "step": 6733
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2691145155951657,
      "learning_rate": 4.911798046586591e-07,
      "loss": 0.7015,
      "step": 6734
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5175440838504055,
      "learning_rate": 4.902841341763964e-07,
      "loss": 0.798,
      "step": 6735
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4279817706833047,
      "learning_rate": 4.893892389665728e-07,
      "loss": 0.7563,
      "step": 6736
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.9114095735379095,
      "learning_rate": 4.884951191830312e-07,
      "loss": 0.6875,
      "step": 6737
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4060833461846332,
      "learning_rate": 4.876017749794787e-07,
      "loss": 0.8142,
      "step": 6738
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3462447669789774,
      "learning_rate": 4.867092065094947e-07,
      "loss": 0.7671,
      "step": 6739
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6341145319788215,
      "learning_rate": 4.858174139265209e-07,
      "loss": 0.7369,
      "step": 6740
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3815440148763904,
      "learning_rate": 4.849263973838664e-07,
      "loss": 0.7974,
      "step": 6741
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3878149887105613,
      "learning_rate": 4.840361570347069e-07,
      "loss": 0.7115,
      "step": 6742
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.27489181845753,
      "learning_rate": 4.831466930320861e-07,
      "loss": 0.8022,
      "step": 6743
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4594785327224797,
      "learning_rate": 4.822580055289117e-07,
      "loss": 0.847,
      "step": 6744
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8416592941339244,
      "learning_rate": 4.813700946779598e-07,
      "loss": 0.6812,
      "step": 6745
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4445570318917487,
      "learning_rate": 4.804829606318745e-07,
      "loss": 0.7251,
      "step": 6746
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5140336233316791,
      "learning_rate": 4.795966035431626e-07,
      "loss": 0.8204,
      "step": 6747
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3325867503257838,
      "learning_rate": 4.787110235642001e-07,
      "loss": 0.7332,
      "step": 6748
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4527614764103967,
      "learning_rate": 4.778262208472273e-07,
      "loss": 0.8314,
      "step": 6749
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2360673785979992,
      "learning_rate": 4.769421955443531e-07,
      "loss": 0.79,
      "step": 6750
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3886113509098255,
      "learning_rate": 4.7605894780755113e-07,
      "loss": 0.6631,
      "step": 6751
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4391397813501254,
      "learning_rate": 4.751764777886614e-07,
      "loss": 0.7198,
      "step": 6752
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4980677678520522,
      "learning_rate": 4.742947856393909e-07,
      "loss": 0.7256,
      "step": 6753
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3647655716764557,
      "learning_rate": 4.734138715113146e-07,
      "loss": 0.7345,
      "step": 6754
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7114350793398487,
      "learning_rate": 4.7253373555587014e-07,
      "loss": 0.7456,
      "step": 6755
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.412449281978685,
      "learning_rate": 4.7165437792436373e-07,
      "loss": 0.7624,
      "step": 6756
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8083013450945286,
      "learning_rate": 4.7077579876796675e-07,
      "loss": 0.659,
      "step": 6757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4956834594034223,
      "learning_rate": 4.698979982377172e-07,
      "loss": 0.8131,
      "step": 6758
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8496158715337923,
      "learning_rate": 4.690209764845183e-07,
      "loss": 0.8267,
      "step": 6759
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2308613540557305,
      "learning_rate": 4.6814473365914217e-07,
      "loss": 0.7188,
      "step": 6760
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7750223404813886,
      "learning_rate": 4.6726926991222386e-07,
      "loss": 0.6193,
      "step": 6761
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8213239156795404,
      "learning_rate": 4.663945853942653e-07,
      "loss": 0.6616,
      "step": 6762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3198597993708232,
      "learning_rate": 4.655206802556361e-07,
      "loss": 0.7277,
      "step": 6763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1616875539057288,
      "learning_rate": 4.646475546465706e-07,
      "loss": 0.681,
      "step": 6764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.220770100271119,
      "learning_rate": 4.6377520871716874e-07,
      "loss": 0.7302,
      "step": 6765
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5958725491579566,
      "learning_rate": 4.629036426173955e-07,
      "loss": 0.7709,
      "step": 6766
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4565641138614573,
      "learning_rate": 4.6203285649708605e-07,
      "loss": 0.7988,
      "step": 6767
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5065956804232739,
      "learning_rate": 4.6116285050593733e-07,
      "loss": 0.7419,
      "step": 6768
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.214403204513957,
      "learning_rate": 4.6029362479351303e-07,
      "loss": 0.6903,
      "step": 6769
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8506328384007382,
      "learning_rate": 4.5942517950924305e-07,
      "loss": 0.853,
      "step": 6770
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3003560310123392,
      "learning_rate": 4.5855751480242404e-07,
      "loss": 0.807,
      "step": 6771
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2848351372012101,
      "learning_rate": 4.5769063082221623e-07,
      "loss": 0.7641,
      "step": 6772
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4087560203406382,
      "learning_rate": 4.568245277176486e-07,
      "loss": 0.7724,
      "step": 6773
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2158940419946949,
      "learning_rate": 4.559592056376133e-07,
      "loss": 0.7478,
      "step": 6774
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2301959501303477,
      "learning_rate": 4.5509466473087017e-07,
      "loss": 0.8061,
      "step": 6775
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1903638974954007,
      "learning_rate": 4.5423090514604305e-07,
      "loss": 0.8182,
      "step": 6776
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2846230154524556,
      "learning_rate": 4.533679270316227e-07,
      "loss": 0.7408,
      "step": 6777
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9363633512229559,
      "learning_rate": 4.525057305359648e-07,
      "loss": 0.6747,
      "step": 6778
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4554175246626542,
      "learning_rate": 4.5164431580729087e-07,
      "loss": 0.6415,
      "step": 6779
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.409724039133721,
      "learning_rate": 4.507836829936868e-07,
      "loss": 0.7656,
      "step": 6780
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4584077888672784,
      "learning_rate": 4.499238322431071e-07,
      "loss": 0.7278,
      "step": 6781
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4664130972970162,
      "learning_rate": 4.4906476370337017e-07,
      "loss": 0.7728,
      "step": 6782
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.625070817592833,
      "learning_rate": 4.482064775221595e-07,
      "loss": 0.7174,
      "step": 6783
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8595504515011854,
      "learning_rate": 4.4734897384702434e-07,
      "loss": 0.6583,
      "step": 6784
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3528605770278672,
      "learning_rate": 4.4649225282537954e-07,
      "loss": 0.7216,
      "step": 6785
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8239655538088743,
      "learning_rate": 4.4563631460450506e-07,
      "loss": 0.6516,
      "step": 6786
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8538276885951754,
      "learning_rate": 4.4478115933154597e-07,
      "loss": 0.6933,
      "step": 6787
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4187144371575438,
      "learning_rate": 4.439267871535152e-07,
      "loss": 0.8492,
      "step": 6788
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3844608634613016,
      "learning_rate": 4.430731982172887e-07,
      "loss": 0.7163,
      "step": 6789
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4483186411062376,
      "learning_rate": 4.4222039266960616e-07,
      "loss": 0.7188,
      "step": 6790
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7373594127763625,
      "learning_rate": 4.413683706570776e-07,
      "loss": 0.7208,
      "step": 6791
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4110775101007513,
      "learning_rate": 4.4051713232617423e-07,
      "loss": 0.6801,
      "step": 6792
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4224276261248314,
      "learning_rate": 4.396666778232345e-07,
      "loss": 0.6791,
      "step": 6793
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3449688212136452,
      "learning_rate": 4.388170072944603e-07,
      "loss": 0.7772,
      "step": 6794
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8552605819782367,
      "learning_rate": 4.3796812088591934e-07,
      "loss": 0.6236,
      "step": 6795
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3106163931034245,
      "learning_rate": 4.37120018743547e-07,
      "loss": 0.7231,
      "step": 6796
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.419404826031721,
      "learning_rate": 4.362727010131407e-07,
      "loss": 0.7299,
      "step": 6797
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6694787568682288,
      "learning_rate": 4.354261678403643e-07,
      "loss": 0.7774,
      "step": 6798
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4396888105386219,
      "learning_rate": 4.345804193707459e-07,
      "loss": 0.7859,
      "step": 6799
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9581167341867851,
      "learning_rate": 4.3373545574968143e-07,
      "loss": 0.7935,
      "step": 6800
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3307660084854378,
      "learning_rate": 4.3289127712242864e-07,
      "loss": 0.8005,
      "step": 6801
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3334437917582342,
      "learning_rate": 4.320478836341113e-07,
      "loss": 0.769,
      "step": 6802
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8785423126015932,
      "learning_rate": 4.312052754297197e-07,
      "loss": 0.6689,
      "step": 6803
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3645474024716393,
      "learning_rate": 4.303634526541067e-07,
      "loss": 0.817,
      "step": 6804
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.331744483618294,
      "learning_rate": 4.2952241545199226e-07,
      "loss": 0.7953,
      "step": 6805
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3508363259915455,
      "learning_rate": 4.2868216396796057e-07,
      "loss": 0.7692,
      "step": 6806
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2360297536817877,
      "learning_rate": 4.2784269834645955e-07,
      "loss": 0.6994,
      "step": 6807
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2064897778413652,
      "learning_rate": 4.270040187318031e-07,
      "loss": 0.6451,
      "step": 6808
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4745373452615969,
      "learning_rate": 4.2616612526817146e-07,
      "loss": 0.8031,
      "step": 6809
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6929551857158969,
      "learning_rate": 4.253290180996067e-07,
      "loss": 0.8148,
      "step": 6810
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.330477833793548,
      "learning_rate": 4.244926973700181e-07,
      "loss": 0.7804,
      "step": 6811
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.439869972538188,
      "learning_rate": 4.2365716322317894e-07,
      "loss": 0.7505,
      "step": 6812
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3467766201020546,
      "learning_rate": 4.228224158027272e-07,
      "loss": 0.7331,
      "step": 6813
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7794758936152315,
      "learning_rate": 4.2198845525216524e-07,
      "loss": 0.6914,
      "step": 6814
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3153307046661888,
      "learning_rate": 4.2115528171486063e-07,
      "loss": 0.7269,
      "step": 6815
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3267568305932127,
      "learning_rate": 4.2032289533404434e-07,
      "loss": 0.755,
      "step": 6816
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8786352423586374,
      "learning_rate": 4.194912962528158e-07,
      "loss": 0.6475,
      "step": 6817
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4771556870401459,
      "learning_rate": 4.1866048461413454e-07,
      "loss": 0.7176,
      "step": 6818
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.586665274611312,
      "learning_rate": 4.17830460560828e-07,
      "loss": 0.7661,
      "step": 6819
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2657361492498738,
      "learning_rate": 4.170012242355864e-07,
      "loss": 0.7297,
      "step": 6820
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4117961259570462,
      "learning_rate": 4.161727757809653e-07,
      "loss": 0.7312,
      "step": 6821
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3911596191617932,
      "learning_rate": 4.153451153393839e-07,
      "loss": 0.7069,
      "step": 6822
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2496412983408365,
      "learning_rate": 4.145182430531264e-07,
      "loss": 0.7142,
      "step": 6823
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2423655992514966,
      "learning_rate": 4.1369215906434337e-07,
      "loss": 0.7343,
      "step": 6824
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2153312755295924,
      "learning_rate": 4.1286686351504735e-07,
      "loss": 0.74,
      "step": 6825
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7935743885538289,
      "learning_rate": 4.120423565471154e-07,
      "loss": 0.6241,
      "step": 6826
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3536977923678621,
      "learning_rate": 4.1121863830229037e-07,
      "loss": 0.7152,
      "step": 6827
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5651225377946962,
      "learning_rate": 4.1039570892217993e-07,
      "loss": 0.7344,
      "step": 6828
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.435595902265918,
      "learning_rate": 4.0957356854825436e-07,
      "loss": 0.7111,
      "step": 6829
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3850819159535042,
      "learning_rate": 4.087522173218489e-07,
      "loss": 0.7496,
      "step": 6830
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2979896587157698,
      "learning_rate": 4.079316553841628e-07,
      "loss": 0.7472,
      "step": 6831
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9085601064719785,
      "learning_rate": 4.0711188287626156e-07,
      "loss": 0.7005,
      "step": 6832
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.310472694210475,
      "learning_rate": 4.062928999390736e-07,
      "loss": 0.7946,
      "step": 6833
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.818726863272997,
      "learning_rate": 4.0547470671339075e-07,
      "loss": 0.6582,
      "step": 6834
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2603480576146648,
      "learning_rate": 4.0465730333986983e-07,
      "loss": 0.7016,
      "step": 6835
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4438328507589,
      "learning_rate": 4.0384068995903136e-07,
      "loss": 0.6434,
      "step": 6836
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.718917569642871,
      "learning_rate": 4.03024866711263e-07,
      "loss": 0.7585,
      "step": 6837
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2317334691544888,
      "learning_rate": 4.0220983373681133e-07,
      "loss": 0.7705,
      "step": 6838
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.526043241504706,
      "learning_rate": 4.013955911757922e-07,
      "loss": 0.7311,
      "step": 6839
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.720513675661349,
      "learning_rate": 4.00582139168183e-07,
      "loss": 0.8048,
      "step": 6840
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3871747551475546,
      "learning_rate": 3.997694778538247e-07,
      "loss": 0.8158,
      "step": 6841
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.480170325407711,
      "learning_rate": 3.9895760737242384e-07,
      "loss": 0.7333,
      "step": 6842
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6444740281670065,
      "learning_rate": 3.9814652786355046e-07,
      "loss": 0.7311,
      "step": 6843
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4523751940176837,
      "learning_rate": 3.973362394666369e-07,
      "loss": 0.738,
      "step": 6844
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.566181270289874,
      "learning_rate": 3.965267423209834e-07,
      "loss": 0.7845,
      "step": 6845
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2591335739388934,
      "learning_rate": 3.957180365657498e-07,
      "loss": 0.7879,
      "step": 6846
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.431693107195416,
      "learning_rate": 3.949101223399643e-07,
      "loss": 0.7121,
      "step": 6847
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9467303691990406,
      "learning_rate": 3.9410299978251523e-07,
      "loss": 0.7198,
      "step": 6848
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.347359995065006,
      "learning_rate": 3.932966690321571e-07,
      "loss": 0.6979,
      "step": 6849
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.222332950400392,
      "learning_rate": 3.92491130227507e-07,
      "loss": 0.7477,
      "step": 6850
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3330173462425567,
      "learning_rate": 3.9168638350704614e-07,
      "loss": 0.7901,
      "step": 6851
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.319193925639298,
      "learning_rate": 3.9088242900911957e-07,
      "loss": 0.7658,
      "step": 6852
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3186247106151645,
      "learning_rate": 3.900792668719372e-07,
      "loss": 0.7899,
      "step": 6853
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3346517955899866,
      "learning_rate": 3.8927689723357253e-07,
      "loss": 0.7866,
      "step": 6854
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.542148349212004,
      "learning_rate": 3.8847532023195967e-07,
      "loss": 0.7356,
      "step": 6855
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.31953937587631,
      "learning_rate": 3.876745360049017e-07,
      "loss": 0.7586,
      "step": 6856
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.426869565241233,
      "learning_rate": 3.868745446900618e-07,
      "loss": 0.7321,
      "step": 6857
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8276150387027572,
      "learning_rate": 3.860753464249678e-07,
      "loss": 0.6632,
      "step": 6858
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2746574346716857,
      "learning_rate": 3.8527694134700975e-07,
      "loss": 0.7315,
      "step": 6859
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5874442136896498,
      "learning_rate": 3.844793295934451e-07,
      "loss": 0.7828,
      "step": 6860
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7115584761134903,
      "learning_rate": 3.836825113013909e-07,
      "loss": 0.7274,
      "step": 6861
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.350766285083245,
      "learning_rate": 3.8288648660783023e-07,
      "loss": 0.7278,
      "step": 6862
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7972141850432748,
      "learning_rate": 3.820912556496087e-07,
      "loss": 0.6032,
      "step": 6863
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4239733131306405,
      "learning_rate": 3.8129681856343526e-07,
      "loss": 0.7206,
      "step": 6864
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2481966645621212,
      "learning_rate": 3.8050317548588343e-07,
      "loss": 0.7181,
      "step": 6865
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5523530416170022,
      "learning_rate": 3.7971032655339026e-07,
      "loss": 0.7737,
      "step": 6866
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5204246449567924,
      "learning_rate": 3.789182719022538e-07,
      "loss": 0.8217,
      "step": 6867
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3457346457352624,
      "learning_rate": 3.781270116686392e-07,
      "loss": 0.732,
      "step": 6868
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.380968395356263,
      "learning_rate": 3.7733654598857303e-07,
      "loss": 0.8535,
      "step": 6869
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.356149186026385,
      "learning_rate": 3.7654687499794453e-07,
      "loss": 0.7885,
      "step": 6870
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.370700626889758,
      "learning_rate": 3.757579988325083e-07,
      "loss": 0.7011,
      "step": 6871
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.333739228148393,
      "learning_rate": 3.7496991762788095e-07,
      "loss": 0.8395,
      "step": 6872
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3271537075068391,
      "learning_rate": 3.7418263151954184e-07,
      "loss": 0.7913,
      "step": 6873
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.329786308189006,
      "learning_rate": 3.7339614064283545e-07,
      "loss": 0.7764,
      "step": 6874
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4123848481859147,
      "learning_rate": 3.7261044513296927e-07,
      "loss": 0.8041,
      "step": 6875
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3329982793856303,
      "learning_rate": 3.71825545125013e-07,
      "loss": 0.7386,
      "step": 6876
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2704690374936174,
      "learning_rate": 3.710414407538998e-07,
      "loss": 0.6825,
      "step": 6877
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2878153465837054,
      "learning_rate": 3.7025813215442685e-07,
      "loss": 0.6813,
      "step": 6878
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7921654192635328,
      "learning_rate": 3.694756194612531e-07,
      "loss": 0.6474,
      "step": 6879
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4036898318564814,
      "learning_rate": 3.686939028089015e-07,
      "loss": 0.8181,
      "step": 6880
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4932689174304536,
      "learning_rate": 3.6791298233175955e-07,
      "loss": 0.6795,
      "step": 6881
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.277672951986295,
      "learning_rate": 3.671328581640754e-07,
      "loss": 0.877,
      "step": 6882
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4078668115201298,
      "learning_rate": 3.663535304399607e-07,
      "loss": 0.6985,
      "step": 6883
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2915911813159742,
      "learning_rate": 3.6557499929339325e-07,
      "loss": 0.6003,
      "step": 6884
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4540442209433149,
      "learning_rate": 3.647972648582104e-07,
      "loss": 0.7765,
      "step": 6885
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4475129844775012,
      "learning_rate": 3.640203272681131e-07,
      "loss": 0.8298,
      "step": 6886
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4079423600120835,
      "learning_rate": 3.6324418665666717e-07,
      "loss": 0.7782,
      "step": 6887
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.211604563648341,
      "learning_rate": 3.624688431572981e-07,
      "loss": 0.6985,
      "step": 6888
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7904339466106568,
      "learning_rate": 3.6169429690329825e-07,
      "loss": 0.6398,
      "step": 6889
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.389261598072675,
      "learning_rate": 3.6092054802782107e-07,
      "loss": 0.7606,
      "step": 6890
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.773005115660842,
      "learning_rate": 3.601475966638829e-07,
      "loss": 0.7906,
      "step": 6891
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7090084535787513,
      "learning_rate": 3.5937544294436134e-07,
      "loss": 0.6405,
      "step": 6892
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.205520052444165,
      "learning_rate": 3.586040870020008e-07,
      "loss": 0.6831,
      "step": 6893
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.821533134554622,
      "learning_rate": 3.5783352896940513e-07,
      "loss": 0.6353,
      "step": 6894
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3483906229313067,
      "learning_rate": 3.570637689790418e-07,
      "loss": 0.721,
      "step": 6895
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5989378945018886,
      "learning_rate": 3.562948071632427e-07,
      "loss": 0.7038,
      "step": 6896
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4153603811740325,
      "learning_rate": 3.55526643654201e-07,
      "loss": 0.8026,
      "step": 6897
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.208779151694664,
      "learning_rate": 3.5475927858397263e-07,
      "loss": 0.7398,
      "step": 6898
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.428998764968605,
      "learning_rate": 3.539927120844766e-07,
      "loss": 0.7224,
      "step": 6899
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3233809516832922,
      "learning_rate": 3.532269442874942e-07,
      "loss": 0.8589,
      "step": 6900
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2812603018332283,
      "learning_rate": 3.524619753246694e-07,
      "loss": 0.8053,
      "step": 6901
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.825095244711166,
      "learning_rate": 3.516978053275111e-07,
      "loss": 0.679,
      "step": 6902
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7969497024400959,
      "learning_rate": 3.5093443442738627e-07,
      "loss": 0.6261,
      "step": 6903
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.974701542910457,
      "learning_rate": 3.5017186275553004e-07,
      "loss": 0.6428,
      "step": 6904
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2568206543958926,
      "learning_rate": 3.494100904430359e-07,
      "loss": 0.6736,
      "step": 6905
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4926600597504986,
      "learning_rate": 3.486491176208617e-07,
      "loss": 0.775,
      "step": 6906
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2438519362771399,
      "learning_rate": 3.478889444198269e-07,
      "loss": 0.6675,
      "step": 6907
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3566326837051375,
      "learning_rate": 3.471295709706146e-07,
      "loss": 0.7416,
      "step": 6908
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4094700981391433,
      "learning_rate": 3.4637099740376934e-07,
      "loss": 0.8196,
      "step": 6909
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.399265870682317,
      "learning_rate": 3.4561322384969843e-07,
      "loss": 0.7314,
      "step": 6910
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1961805497914746,
      "learning_rate": 3.448562504386738e-07,
      "loss": 0.6633,
      "step": 6911
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5306565419926033,
      "learning_rate": 3.4410007730082685e-07,
      "loss": 0.7344,
      "step": 6912
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3669355542112456,
      "learning_rate": 3.4334470456615255e-07,
      "loss": 0.7225,
      "step": 6913
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.366451549572782,
      "learning_rate": 3.4259013236450755e-07,
      "loss": 0.7865,
      "step": 6914
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3909646634090775,
      "learning_rate": 3.4183636082561257e-07,
      "loss": 0.7937,
      "step": 6915
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5105407403452051,
      "learning_rate": 3.4108339007904834e-07,
      "loss": 0.7801,
      "step": 6916
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7803881411981001,
      "learning_rate": 3.4033122025426077e-07,
      "loss": 0.6718,
      "step": 6917
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2546600853252785,
      "learning_rate": 3.395798514805565e-07,
      "loss": 0.7275,
      "step": 6918
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9255095204200736,
      "learning_rate": 3.3882928388710376e-07,
      "loss": 0.6599,
      "step": 6919
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5782415509131082,
      "learning_rate": 3.380795176029328e-07,
      "loss": 0.7616,
      "step": 6920
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2790975930869157,
      "learning_rate": 3.3733055275693983e-07,
      "loss": 0.7615,
      "step": 6921
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4622513137730857,
      "learning_rate": 3.3658238947787857e-07,
      "loss": 0.77,
      "step": 6922
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5474968171088261,
      "learning_rate": 3.3583502789436783e-07,
      "loss": 0.7199,
      "step": 6923
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.828115258532469,
      "learning_rate": 3.3508846813488647e-07,
      "loss": 0.6083,
      "step": 6924
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3917576025766407,
      "learning_rate": 3.3434271032777856e-07,
      "loss": 0.7454,
      "step": 6925
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4100956303517667,
      "learning_rate": 3.335977546012481e-07,
      "loss": 0.7916,
      "step": 6926
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4037891240518021,
      "learning_rate": 3.328536010833605e-07,
      "loss": 0.68,
      "step": 6927
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2975368390943762,
      "learning_rate": 3.3211024990204565e-07,
      "loss": 0.7365,
      "step": 6928
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2412268644101967,
      "learning_rate": 3.3136770118509243e-07,
      "loss": 0.7074,
      "step": 6929
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3891493792569447,
      "learning_rate": 3.3062595506015594e-07,
      "loss": 0.7918,
      "step": 6930
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4245827562181048,
      "learning_rate": 3.2988501165474864e-07,
      "loss": 0.8247,
      "step": 6931
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3574626197919497,
      "learning_rate": 3.291448710962497e-07,
      "loss": 0.6916,
      "step": 6932
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3165941129645442,
      "learning_rate": 3.284055335118963e-07,
      "loss": 0.7176,
      "step": 6933
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4021015370989713,
      "learning_rate": 3.276669990287895e-07,
      "loss": 0.6991,
      "step": 6934
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.261388490105171,
      "learning_rate": 3.269292677738922e-07,
      "loss": 0.7398,
      "step": 6935
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4146506920967707,
      "learning_rate": 3.261923398740285e-07,
      "loss": 0.7449,
      "step": 6936
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2673664520554222,
      "learning_rate": 3.2545621545588434e-07,
      "loss": 0.7532,
      "step": 6937
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4512213930934788,
      "learning_rate": 3.2472089464600844e-07,
      "loss": 0.7534,
      "step": 6938
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7804552246467398,
      "learning_rate": 3.2398637757081187e-07,
      "loss": 0.7088,
      "step": 6939
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1575364203120084,
      "learning_rate": 3.232526643565664e-07,
      "loss": 0.6592,
      "step": 6940
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4670715522804574,
      "learning_rate": 3.2251975512940516e-07,
      "loss": 0.776,
      "step": 6941
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5520774636212034,
      "learning_rate": 3.2178765001532386e-07,
      "loss": 0.8596,
      "step": 6942
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.523814876388246,
      "learning_rate": 3.210563491401808e-07,
      "loss": 0.7464,
      "step": 6943
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2948783348407202,
      "learning_rate": 3.203258526296937e-07,
      "loss": 0.7784,
      "step": 6944
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3046660061190425,
      "learning_rate": 3.195961606094433e-07,
      "loss": 0.757,
      "step": 6945
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3300953940169806,
      "learning_rate": 3.188672732048731e-07,
      "loss": 0.7229,
      "step": 6946
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2486625382977465,
      "learning_rate": 3.181391905412867e-07,
      "loss": 0.8385,
      "step": 6947
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3400413996965017,
      "learning_rate": 3.1741191274385076e-07,
      "loss": 0.7576,
      "step": 6948
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4909302315102408,
      "learning_rate": 3.1668543993759293e-07,
      "loss": 0.7797,
      "step": 6949
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7583288196136828,
      "learning_rate": 3.159597722474006e-07,
      "loss": 0.6037,
      "step": 6950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8786831558040961,
      "learning_rate": 3.152349097980262e-07,
      "loss": 0.6473,
      "step": 6951
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2322534827629532,
      "learning_rate": 3.1451085271408053e-07,
      "loss": 0.763,
      "step": 6952
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4153037562527693,
      "learning_rate": 3.137876011200386e-07,
      "loss": 0.676,
      "step": 6953
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.444006742308799,
      "learning_rate": 3.130651551402353e-07,
      "loss": 0.7473,
      "step": 6954
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3533010262914242,
      "learning_rate": 3.123435148988674e-07,
      "loss": 0.766,
      "step": 6955
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1571890119472525,
      "learning_rate": 3.116226805199929e-07,
      "loss": 0.708,
      "step": 6956
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3117048703187084,
      "learning_rate": 3.1090265212753214e-07,
      "loss": 0.8283,
      "step": 6957
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.333545072397867,
      "learning_rate": 3.101834298452661e-07,
      "loss": 0.7898,
      "step": 6958
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8204481563676043,
      "learning_rate": 3.094650137968369e-07,
      "loss": 0.6267,
      "step": 6959
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4258279354257994,
      "learning_rate": 3.0874740410575033e-07,
      "loss": 0.7996,
      "step": 6960
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.384159846473715,
      "learning_rate": 3.080306008953704e-07,
      "loss": 0.7609,
      "step": 6961
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2705241783002545,
      "learning_rate": 3.0731460428892414e-07,
      "loss": 0.667,
      "step": 6962
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7750658218373713,
      "learning_rate": 3.065994144094997e-07,
      "loss": 0.646,
      "step": 6963
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.339765253277992,
      "learning_rate": 3.0588503138004597e-07,
      "loss": 0.7392,
      "step": 6964
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4671846575083862,
      "learning_rate": 3.051714553233748e-07,
      "loss": 0.8067,
      "step": 6965
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5564346314918804,
      "learning_rate": 3.044586863621568e-07,
      "loss": 0.7535,
      "step": 6966
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8003287526887876,
      "learning_rate": 3.0374672461892574e-07,
      "loss": 0.5976,
      "step": 6967
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3631219009689908,
      "learning_rate": 3.0303557021607754e-07,
      "loss": 0.8252,
      "step": 6968
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7421808436218347,
      "learning_rate": 3.023252232758667e-07,
      "loss": 0.7733,
      "step": 6969
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7770704659520775,
      "learning_rate": 3.0161568392040986e-07,
      "loss": 0.7424,
      "step": 6970
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3926871663166325,
      "learning_rate": 3.009069522716851e-07,
      "loss": 0.8335,
      "step": 6971
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4127611724751261,
      "learning_rate": 3.0019902845153216e-07,
      "loss": 0.6635,
      "step": 6972
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2822507374381429,
      "learning_rate": 2.994919125816498e-07,
      "loss": 0.802,
      "step": 6973
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.318116316292185,
      "learning_rate": 2.9878560478360186e-07,
      "loss": 0.7973,
      "step": 6974
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4436539851755843,
      "learning_rate": 2.980801051788085e-07,
      "loss": 0.8093,
      "step": 6975
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.234715024164726,
      "learning_rate": 2.973754138885554e-07,
      "loss": 0.693,
      "step": 6976
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9387715089582827,
      "learning_rate": 2.9667153103398573e-07,
      "loss": 0.7882,
      "step": 6977
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1770377496410704,
      "learning_rate": 2.9596845673610597e-07,
      "loss": 0.6628,
      "step": 6978
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3543029585132726,
      "learning_rate": 2.9526619111578223e-07,
      "loss": 0.845,
      "step": 6979
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4115816752019028,
      "learning_rate": 2.945647342937413e-07,
      "loss": 0.7085,
      "step": 6980
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.186028921926996,
      "learning_rate": 2.9386408639057394e-07,
      "loss": 0.7809,
      "step": 6981
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2947359403780705,
      "learning_rate": 2.9316424752672766e-07,
      "loss": 0.7148,
      "step": 6982
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3413422996708912,
      "learning_rate": 2.9246521782251403e-07,
      "loss": 0.6604,
      "step": 6983
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3512962837179365,
      "learning_rate": 2.91766997398103e-07,
      "loss": 0.7759,
      "step": 6984
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.439681011010638,
      "learning_rate": 2.91069586373528e-07,
      "loss": 0.7576,
      "step": 6985
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7640687522216076,
      "learning_rate": 2.9037298486868205e-07,
      "loss": 0.7401,
      "step": 6986
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3378921439623603,
      "learning_rate": 2.8967719300331875e-07,
      "loss": 0.7349,
      "step": 6987
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5706609082854859,
      "learning_rate": 2.8898221089705194e-07,
      "loss": 0.727,
      "step": 6988
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6235358473409038,
      "learning_rate": 2.882880386693582e-07,
      "loss": 0.7248,
      "step": 6989
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3575322488190713,
      "learning_rate": 2.8759467643957375e-07,
      "loss": 0.8461,
      "step": 6990
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.369008813730514,
      "learning_rate": 2.8690212432689546e-07,
      "loss": 0.6858,
      "step": 6991
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2874921386649025,
      "learning_rate": 2.862103824503809e-07,
      "loss": 0.7267,
      "step": 6992
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4356789405126213,
      "learning_rate": 2.855194509289483e-07,
      "loss": 0.8681,
      "step": 6993
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2674730494858628,
      "learning_rate": 2.8482932988137647e-07,
      "loss": 0.7015,
      "step": 6994
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3723814632074949,
      "learning_rate": 2.8414001942630556e-07,
      "loss": 0.7536,
      "step": 6995
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2012509154244568,
      "learning_rate": 2.834515196822374e-07,
      "loss": 0.7493,
      "step": 6996
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3073575762691834,
      "learning_rate": 2.8276383076753175e-07,
      "loss": 0.7391,
      "step": 6997
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4141956420594681,
      "learning_rate": 2.8207695280041025e-07,
      "loss": 0.904,
      "step": 6998
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.548829388425757,
      "learning_rate": 2.813908858989556e-07,
      "loss": 0.7698,
      "step": 6999
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.443484871474467,
      "learning_rate": 2.8070563018111063e-07,
      "loss": 0.7045,
      "step": 7000
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2845680744710557,
      "learning_rate": 2.8002118576467784e-07,
      "loss": 0.8832,
      "step": 7001
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4542640938258735,
      "learning_rate": 2.7933755276732257e-07,
      "loss": 0.6989,
      "step": 7002
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.332493632208149,
      "learning_rate": 2.7865473130656794e-07,
      "loss": 0.7738,
      "step": 7003
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8195993829597136,
      "learning_rate": 2.779727214997996e-07,
      "loss": 0.6495,
      "step": 7004
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2387529339078767,
      "learning_rate": 2.7729152346426366e-07,
      "loss": 0.7505,
      "step": 7005
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3593145685718866,
      "learning_rate": 2.766111373170649e-07,
      "loss": 0.8338,
      "step": 7006
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3003499208140978,
      "learning_rate": 2.7593156317516966e-07,
      "loss": 0.8309,
      "step": 7007
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3934885567788906,
      "learning_rate": 2.752528011554051e-07,
      "loss": 0.7603,
      "step": 7008
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.685978188212703,
      "learning_rate": 2.7457485137445725e-07,
      "loss": 0.7766,
      "step": 7009
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8004980060176349,
      "learning_rate": 2.738977139488752e-07,
      "loss": 0.6687,
      "step": 7010
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2453815885827106,
      "learning_rate": 2.732213889950652e-07,
      "loss": 0.7459,
      "step": 7011
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3929238413206173,
      "learning_rate": 2.725458766292954e-07,
      "loss": 0.7731,
      "step": 7012
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.78981674100926,
      "learning_rate": 2.718711769676957e-07,
      "loss": 0.6288,
      "step": 7013
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.387949558736922,
      "learning_rate": 2.711972901262538e-07,
      "loss": 0.878,
      "step": 7014
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4476181859473911,
      "learning_rate": 2.705242162208188e-07,
      "loss": 0.7205,
      "step": 7015
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4220123765272852,
      "learning_rate": 2.698519553670992e-07,
      "loss": 0.6595,
      "step": 7016
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3649197296277313,
      "learning_rate": 2.6918050768066527e-07,
      "loss": 0.6863,
      "step": 7017
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3892349525102092,
      "learning_rate": 2.68509873276947e-07,
      "loss": 0.6506,
      "step": 7018
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3515773196795702,
      "learning_rate": 2.678400522712332e-07,
      "loss": 0.6764,
      "step": 7019
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8377293705304095,
      "learning_rate": 2.6717104477867464e-07,
      "loss": 0.6328,
      "step": 7020
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9906805550400357,
      "learning_rate": 2.665028509142803e-07,
      "loss": 0.6742,
      "step": 7021
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2684500377992145,
      "learning_rate": 2.6583547079292224e-07,
      "loss": 0.7378,
      "step": 7022
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6050152570127398,
      "learning_rate": 2.651689045293293e-07,
      "loss": 0.731,
      "step": 7023
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2758442575482396,
      "learning_rate": 2.645031522380925e-07,
      "loss": 0.7287,
      "step": 7024
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2588342265310495,
      "learning_rate": 2.638382140336626e-07,
      "loss": 0.6928,
      "step": 7025
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6102222630841627,
      "learning_rate": 2.631740900303503e-07,
      "loss": 0.7524,
      "step": 7026
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.360344497985407,
      "learning_rate": 2.6251078034232605e-07,
      "loss": 0.6601,
      "step": 7027
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2887969803143142,
      "learning_rate": 2.6184828508362016e-07,
      "loss": 0.7221,
      "step": 7028
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.278986992770721,
      "learning_rate": 2.6118660436812326e-07,
      "loss": 0.7663,
      "step": 7029
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.326842973879593,
      "learning_rate": 2.605257383095855e-07,
      "loss": 0.6862,
      "step": 7030
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.470286238011905,
      "learning_rate": 2.5986568702161817e-07,
      "loss": 0.7761,
      "step": 7031
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5871277572754494,
      "learning_rate": 2.5920645061769225e-07,
      "loss": 0.8415,
      "step": 7032
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.367393862317749,
      "learning_rate": 2.585480292111375e-07,
      "loss": 0.8001,
      "step": 7033
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3542799330053668,
      "learning_rate": 2.578904229151441e-07,
      "loss": 0.7392,
      "step": 7034
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.393202350612744,
      "learning_rate": 2.5723363184276207e-07,
      "loss": 0.7451,
      "step": 7035
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2853919473094364,
      "learning_rate": 2.5657765610690223e-07,
      "loss": 0.7123,
      "step": 7036
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.561738143094931,
      "learning_rate": 2.559224958203321e-07,
      "loss": 0.7028,
      "step": 7037
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4328034049497398,
      "learning_rate": 2.552681510956845e-07,
      "loss": 0.8051,
      "step": 7038
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9885808520545493,
      "learning_rate": 2.546146220454471e-07,
      "loss": 0.6381,
      "step": 7039
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2442900411721105,
      "learning_rate": 2.539619087819689e-07,
      "loss": 0.7149,
      "step": 7040
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4440526877638271,
      "learning_rate": 2.533100114174597e-07,
      "loss": 0.8268,
      "step": 7041
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3308230616910413,
      "learning_rate": 2.526589300639881e-07,
      "loss": 0.7441,
      "step": 7042
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3238436510101481,
      "learning_rate": 2.520086648334824e-07,
      "loss": 0.6801,
      "step": 7043
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2533148050630034,
      "learning_rate": 2.5135921583773036e-07,
      "loss": 0.8215,
      "step": 7044
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3745349068590758,
      "learning_rate": 2.507105831883794e-07,
      "loss": 0.7287,
      "step": 7045
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3982443417266528,
      "learning_rate": 2.5006276699693854e-07,
      "loss": 0.7137,
      "step": 7046
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2782732542966517,
      "learning_rate": 2.4941576737477435e-07,
      "loss": 0.7546,
      "step": 7047
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3507857420069356,
      "learning_rate": 2.487695844331128e-07,
      "loss": 0.8179,
      "step": 7048
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5576180346362263,
      "learning_rate": 2.481242182830401e-07,
      "loss": 0.7586,
      "step": 7049
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5689518293548546,
      "learning_rate": 2.4747966903550355e-07,
      "loss": 0.794,
      "step": 7050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8559530638045303,
      "learning_rate": 2.4683593680130734e-07,
      "loss": 0.724,
      "step": 7051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7710128041737156,
      "learning_rate": 2.461930216911168e-07,
      "loss": 0.6086,
      "step": 7052
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3656160892894766,
      "learning_rate": 2.455509238154574e-07,
      "loss": 0.7608,
      "step": 7053
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6563824040247228,
      "learning_rate": 2.4490964328471257e-07,
      "loss": 0.7863,
      "step": 7054
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6195511370887083,
      "learning_rate": 2.442691802091257e-07,
      "loss": 0.7491,
      "step": 7055
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5856604201742819,
      "learning_rate": 2.4362953469879934e-07,
      "loss": 0.7985,
      "step": 7056
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3652505839326008,
      "learning_rate": 2.429907068636972e-07,
      "loss": 0.7996,
      "step": 7057
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.355368289510794,
      "learning_rate": 2.423526968136397e-07,
      "loss": 0.6849,
      "step": 7058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8478848688906734,
      "learning_rate": 2.4171550465830974e-07,
      "loss": 0.6819,
      "step": 7059
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2687744899970903,
      "learning_rate": 2.4107913050724627e-07,
      "loss": 0.7336,
      "step": 7060
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2730921752099331,
      "learning_rate": 2.4044357446985134e-07,
      "loss": 0.7624,
      "step": 7061
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.483894290042859,
      "learning_rate": 2.398088366553836e-07,
      "loss": 0.8306,
      "step": 7062
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3486568391378215,
      "learning_rate": 2.3917491717296184e-07,
      "loss": 0.742,
      "step": 7063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9437657871540698,
      "learning_rate": 2.385418161315639e-07,
      "loss": 0.6433,
      "step": 7064
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3030748821994773,
      "learning_rate": 2.3790953364002722e-07,
      "loss": 0.7615,
      "step": 7065
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3997212337836125,
      "learning_rate": 2.372780698070476e-07,
      "loss": 0.676,
      "step": 7066
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3210103321335074,
      "learning_rate": 2.3664742474118317e-07,
      "loss": 0.7799,
      "step": 7067
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5761220824769733,
      "learning_rate": 2.3601759855084672e-07,
      "loss": 0.7577,
      "step": 7068
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4378881515987394,
      "learning_rate": 2.3538859134431547e-07,
      "loss": 0.7316,
      "step": 7069
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.817096341308286,
      "learning_rate": 2.347604032297207e-07,
      "loss": 0.6525,
      "step": 7070
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2949613619397666,
      "learning_rate": 2.3413303431505606e-07,
      "loss": 0.8568,
      "step": 7071
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3171536703429245,
      "learning_rate": 2.3350648470817416e-07,
      "loss": 0.732,
      "step": 7072
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4587598307139331,
      "learning_rate": 2.3288075451678381e-07,
      "loss": 0.7214,
      "step": 7073
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.444005153256164,
      "learning_rate": 2.3225584384845845e-07,
      "loss": 0.8585,
      "step": 7074
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.055043452958912,
      "learning_rate": 2.3163175281062545e-07,
      "loss": 0.7275,
      "step": 7075
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.435074536260803,
      "learning_rate": 2.31008481510574e-07,
      "loss": 0.6984,
      "step": 7076
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2149387983019218,
      "learning_rate": 2.3038603005545113e-07,
      "loss": 0.7514,
      "step": 7077
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.298442264163376,
      "learning_rate": 2.2976439855226406e-07,
      "loss": 0.7975,
      "step": 7078
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.243177505410053,
      "learning_rate": 2.2914358710787842e-07,
      "loss": 0.7715,
      "step": 7079
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.240525062711261,
      "learning_rate": 2.2852359582901828e-07,
      "loss": 0.679,
      "step": 7080
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6263450160187864,
      "learning_rate": 2.2790442482226727e-07,
      "loss": 0.6558,
      "step": 7081
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3439034579362399,
      "learning_rate": 2.2728607419406967e-07,
      "loss": 0.7574,
      "step": 7082
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3140204068713182,
      "learning_rate": 2.2666854405072546e-07,
      "loss": 0.7718,
      "step": 7083
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.32007208234889,
      "learning_rate": 2.2605183449839585e-07,
      "loss": 0.8149,
      "step": 7084
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4226660861086986,
      "learning_rate": 2.2543594564309989e-07,
      "loss": 0.7636,
      "step": 7085
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3570692195787741,
      "learning_rate": 2.2482087759071625e-07,
      "loss": 0.7016,
      "step": 7086
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3885486294286487,
      "learning_rate": 2.2420663044698254e-07,
      "loss": 0.6841,
      "step": 7087
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3444001138137451,
      "learning_rate": 2.2359320431749432e-07,
      "loss": 0.8258,
      "step": 7088
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.272382583064905,
      "learning_rate": 2.2298059930770833e-07,
      "loss": 0.7193,
      "step": 7089
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5003445662062698,
      "learning_rate": 2.2236881552293642e-07,
      "loss": 0.7461,
      "step": 7090
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3354868458636147,
      "learning_rate": 2.2175785306835285e-07,
      "loss": 0.7151,
      "step": 7091
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2011748694017061,
      "learning_rate": 2.21147712048988e-07,
      "loss": 0.784,
      "step": 7092
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8655310607976513,
      "learning_rate": 2.2053839256973297e-07,
      "loss": 0.6592,
      "step": 7093
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3557004804280655,
      "learning_rate": 2.1992989473533566e-07,
      "loss": 0.7839,
      "step": 7094
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5510745920709659,
      "learning_rate": 2.1932221865040572e-07,
      "loss": 0.8118,
      "step": 7095
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2879993353711114,
      "learning_rate": 2.187153644194079e-07,
      "loss": 0.7422,
      "step": 7096
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8292796954126816,
      "learning_rate": 2.1810933214666928e-07,
      "loss": 0.6992,
      "step": 7097
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3290400679787266,
      "learning_rate": 2.1750412193637216e-07,
      "loss": 0.6736,
      "step": 7098
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8702445210648265,
      "learning_rate": 2.1689973389256047e-07,
      "loss": 0.6051,
      "step": 7099
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.273044486087193,
      "learning_rate": 2.1629616811913502e-07,
      "loss": 0.7947,
      "step": 7100
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3481648788705225,
      "learning_rate": 2.1569342471985556e-07,
      "loss": 0.6533,
      "step": 7101
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3189693075265316,
      "learning_rate": 2.150915037983403e-07,
      "loss": 0.7593,
      "step": 7102
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2454601576442497,
      "learning_rate": 2.1449040545806766e-07,
      "loss": 0.659,
      "step": 7103
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4409997676718571,
      "learning_rate": 2.1389012980237267e-07,
      "loss": 0.8099,
      "step": 7104
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5191535013207842,
      "learning_rate": 2.1329067693444893e-07,
      "loss": 0.7429,
      "step": 7105
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1815117685976675,
      "learning_rate": 2.126920469573507e-07,
      "loss": 0.7063,
      "step": 7106
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3957307530038718,
      "learning_rate": 2.1209423997398893e-07,
      "loss": 0.741,
      "step": 7107
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5062814027657534,
      "learning_rate": 2.1149725608713368e-07,
      "loss": 0.8041,
      "step": 7108
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7933816132007303,
      "learning_rate": 2.109010953994123e-07,
      "loss": 0.617,
      "step": 7109
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5122499136085834,
      "learning_rate": 2.1030575801331332e-07,
      "loss": 0.76,
      "step": 7110
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3315480224393912,
      "learning_rate": 2.097112440311816e-07,
      "loss": 0.7421,
      "step": 7111
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3203303072733152,
      "learning_rate": 2.0911755355522089e-07,
      "loss": 0.6809,
      "step": 7112
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2687083301458966,
      "learning_rate": 2.0852468668749294e-07,
      "loss": 0.7562,
      "step": 7113
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.69537535457251,
      "learning_rate": 2.0793264352991894e-07,
      "loss": 0.632,
      "step": 7114
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3049649689329132,
      "learning_rate": 2.0734142418427806e-07,
      "loss": 0.7096,
      "step": 7115
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7240162797194815,
      "learning_rate": 2.067510287522073e-07,
      "loss": 0.6369,
      "step": 7116
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8087460150224073,
      "learning_rate": 2.0616145733520276e-07,
      "loss": 0.6028,
      "step": 7117
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8670744042792302,
      "learning_rate": 2.0557271003461942e-07,
      "loss": 0.727,
      "step": 7118
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3968259599085442,
      "learning_rate": 2.0498478695166857e-07,
      "loss": 0.7168,
      "step": 7119
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2954351679174876,
      "learning_rate": 2.0439768818742156e-07,
      "loss": 0.8295,
      "step": 7120
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4551764536505247,
      "learning_rate": 2.0381141384280711e-07,
      "loss": 0.9189,
      "step": 7121
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3616381779972409,
      "learning_rate": 2.0322596401861294e-07,
      "loss": 0.7014,
      "step": 7122
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3306026833860611,
      "learning_rate": 2.026413388154841e-07,
      "loss": 0.7164,
      "step": 7123
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4252276979305365,
      "learning_rate": 2.020575383339246e-07,
      "loss": 0.716,
      "step": 7124
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3011586904285508,
      "learning_rate": 2.0147456267429754e-07,
      "loss": 0.6902,
      "step": 7125
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7738962794240908,
      "learning_rate": 2.0089241193682273e-07,
      "loss": 0.5753,
      "step": 7126
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3373151444456697,
      "learning_rate": 2.003110862215779e-07,
      "loss": 0.8075,
      "step": 7127
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1776736222109667,
      "learning_rate": 1.9973058562850033e-07,
      "loss": 0.6101,
      "step": 7128
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7593234567023552,
      "learning_rate": 1.9915091025738464e-07,
      "loss": 0.6387,
      "step": 7129
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6449526388277442,
      "learning_rate": 1.985720602078828e-07,
      "loss": 0.7562,
      "step": 7130
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5598411597802404,
      "learning_rate": 1.9799403557950793e-07,
      "loss": 0.6698,
      "step": 7131
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7855627737048932,
      "learning_rate": 1.9741683647162724e-07,
      "loss": 0.5821,
      "step": 7132
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3243878932477215,
      "learning_rate": 1.9684046298346858e-07,
      "loss": 0.7246,
      "step": 7133
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3683179911960353,
      "learning_rate": 1.9626491521411773e-07,
      "loss": 0.7872,
      "step": 7134
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3257154908128332,
      "learning_rate": 1.956901932625177e-07,
      "loss": 0.7915,
      "step": 7135
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3583978838772643,
      "learning_rate": 1.9511629722747004e-07,
      "loss": 0.6653,
      "step": 7136
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3322260742266352,
      "learning_rate": 1.9454322720763364e-07,
      "loss": 0.7667,
      "step": 7137
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5962886974577348,
      "learning_rate": 1.939709833015252e-07,
      "loss": 0.8311,
      "step": 7138
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3478073535209427,
      "learning_rate": 1.9339956560752216e-07,
      "loss": 0.763,
      "step": 7139
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.22272956400226,
      "learning_rate": 1.9282897422385593e-07,
      "loss": 0.8067,
      "step": 7140
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7973265736360133,
      "learning_rate": 1.9225920924861917e-07,
      "loss": 0.6145,
      "step": 7141
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5474630948143457,
      "learning_rate": 1.9169027077975965e-07,
      "loss": 0.7281,
      "step": 7142
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2828398764383948,
      "learning_rate": 1.9112215891508635e-07,
      "loss": 0.6968,
      "step": 7143
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4794855901848,
      "learning_rate": 1.905548737522628e-07,
      "loss": 0.6932,
      "step": 7144
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3099628785148378,
      "learning_rate": 1.899884153888115e-07,
      "loss": 0.7725,
      "step": 7145
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4476314963131203,
      "learning_rate": 1.8942278392211466e-07,
      "loss": 0.7438,
      "step": 7146
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2871410692101897,
      "learning_rate": 1.8885797944941052e-07,
      "loss": 0.7411,
      "step": 7147
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3852085836320254,
      "learning_rate": 1.8829400206779536e-07,
      "loss": 0.7313,
      "step": 7148
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3668989786663137,
      "learning_rate": 1.8773085187422325e-07,
      "loss": 0.7646,
      "step": 7149
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5719416301451454,
      "learning_rate": 1.8716852896550618e-07,
      "loss": 0.7977,
      "step": 7150
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3103134547792548,
      "learning_rate": 1.8660703343831354e-07,
      "loss": 0.6501,
      "step": 7151
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3450874746665507,
      "learning_rate": 1.8604636538917365e-07,
      "loss": 0.7438,
      "step": 7152
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.269751245916588,
      "learning_rate": 1.8548652491447217e-07,
      "loss": 0.8047,
      "step": 7153
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4401562345366292,
      "learning_rate": 1.8492751211045156e-07,
      "loss": 0.7041,
      "step": 7154
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3392799029344877,
      "learning_rate": 1.8436932707321276e-07,
      "loss": 0.7495,
      "step": 7155
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7271270725070618,
      "learning_rate": 1.8381196989871453e-07,
      "loss": 0.7302,
      "step": 7156
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3563794447349442,
      "learning_rate": 1.8325544068277244e-07,
      "loss": 0.7603,
      "step": 7157
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3532115993340554,
      "learning_rate": 1.8269973952106057e-07,
      "loss": 0.7049,
      "step": 7158
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3014612990352517,
      "learning_rate": 1.8214486650911022e-07,
      "loss": 0.6952,
      "step": 7159
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2685857975958326,
      "learning_rate": 1.8159082174231012e-07,
      "loss": 0.7186,
      "step": 7160
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3643066529953534,
      "learning_rate": 1.8103760531590851e-07,
      "loss": 0.7264,
      "step": 7161
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3162443511980995,
      "learning_rate": 1.8048521732500878e-07,
      "loss": 0.72,
      "step": 7162
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4948597467165177,
      "learning_rate": 1.7993365786457217e-07,
      "loss": 0.7515,
      "step": 7163
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2923323133795084,
      "learning_rate": 1.793829270294195e-07,
      "loss": 0.7483,
      "step": 7164
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7188583091212315,
      "learning_rate": 1.7883302491422673e-07,
      "loss": 0.8228,
      "step": 7165
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.281878071502012,
      "learning_rate": 1.782839516135282e-07,
      "loss": 0.7107,
      "step": 7166
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6642382466038426,
      "learning_rate": 1.777357072217173e-07,
      "loss": 0.8618,
      "step": 7167
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3562091171559605,
      "learning_rate": 1.7718829183304254e-07,
      "loss": 0.597,
      "step": 7168
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3296300126407952,
      "learning_rate": 1.766417055416114e-07,
      "loss": 0.7586,
      "step": 7169
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5087542898596449,
      "learning_rate": 1.7609594844138767e-07,
      "loss": 0.7747,
      "step": 7170
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.554265349435973,
      "learning_rate": 1.7555102062619454e-07,
      "loss": 0.8624,
      "step": 7171
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4850189432813186,
      "learning_rate": 1.7500692218971048e-07,
      "loss": 0.7222,
      "step": 7172
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4788549666197863,
      "learning_rate": 1.7446365322547231e-07,
      "loss": 0.6889,
      "step": 7173
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7956103542993447,
      "learning_rate": 1.7392121382687533e-07,
      "loss": 0.631,
      "step": 7174
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2492885891327563,
      "learning_rate": 1.733796040871699e-07,
      "loss": 0.7307,
      "step": 7175
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6600940425021524,
      "learning_rate": 1.7283882409946552e-07,
      "loss": 0.829,
      "step": 7176
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7377896666474855,
      "learning_rate": 1.7229887395672884e-07,
      "loss": 0.7049,
      "step": 7177
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0090097425963374,
      "learning_rate": 1.7175975375178343e-07,
      "loss": 0.6319,
      "step": 7178
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4626896252637973,
      "learning_rate": 1.7122146357730908e-07,
      "loss": 0.7108,
      "step": 7179
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4580039051164417,
      "learning_rate": 1.7068400352584613e-07,
      "loss": 0.7271,
      "step": 7180
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5183022537288156,
      "learning_rate": 1.7014737368978795e-07,
      "loss": 0.8373,
      "step": 7181
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0871653525469327,
      "learning_rate": 1.6961157416139018e-07,
      "loss": 0.7296,
      "step": 7182
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5264380154585977,
      "learning_rate": 1.690766050327608e-07,
      "loss": 0.8325,
      "step": 7183
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.593451742591584,
      "learning_rate": 1.685424663958679e-07,
      "loss": 0.7452,
      "step": 7184
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.335795157631955,
      "learning_rate": 1.680091583425364e-07,
      "loss": 0.837,
      "step": 7185
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.432491066212867,
      "learning_rate": 1.674766809644479e-07,
      "loss": 0.7818,
      "step": 7186
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2343479386565157,
      "learning_rate": 1.6694503435314035e-07,
      "loss": 0.6748,
      "step": 7187
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4825983365330655,
      "learning_rate": 1.6641421860001172e-07,
      "loss": 0.7177,
      "step": 7188
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3845202441215634,
      "learning_rate": 1.6588423379631458e-07,
      "loss": 0.8497,
      "step": 7189
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3718222583012802,
      "learning_rate": 1.6535508003315937e-07,
      "loss": 0.7368,
      "step": 7190
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2408627590151142,
      "learning_rate": 1.6482675740151444e-07,
      "loss": 0.7107,
      "step": 7191
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5625315610035826,
      "learning_rate": 1.642992659922038e-07,
      "loss": 0.6454,
      "step": 7192
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7463630327919009,
      "learning_rate": 1.6377260589590939e-07,
      "loss": 0.6477,
      "step": 7193
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2613813922269799,
      "learning_rate": 1.632467772031704e-07,
      "loss": 0.7771,
      "step": 7194
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7751690758830457,
      "learning_rate": 1.6272178000438288e-07,
      "loss": 0.6748,
      "step": 7195
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.378319961548833,
      "learning_rate": 1.621976143898002e-07,
      "loss": 0.8231,
      "step": 7196
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4171339875238296,
      "learning_rate": 1.6167428044953138e-07,
      "loss": 0.7063,
      "step": 7197
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8107044517122938,
      "learning_rate": 1.6115177827354556e-07,
      "loss": 0.6539,
      "step": 7198
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7255723682987805,
      "learning_rate": 1.6063010795166533e-07,
      "loss": 0.7642,
      "step": 7199
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4779197353991957,
      "learning_rate": 1.6010926957357232e-07,
      "loss": 0.8606,
      "step": 7200
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3122540491814643,
      "learning_rate": 1.5958926322880487e-07,
      "loss": 0.7199,
      "step": 7201
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3372526831768372,
      "learning_rate": 1.590700890067576e-07,
      "loss": 0.8088,
      "step": 7202
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.33207947808989,
      "learning_rate": 1.5855174699668298e-07,
      "loss": 0.7323,
      "step": 7203
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6433716507286835,
      "learning_rate": 1.580342372876903e-07,
      "loss": 0.7326,
      "step": 7204
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2711771361472313,
      "learning_rate": 1.5751755996874452e-07,
      "loss": 0.8169,
      "step": 7205
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3565107459135795,
      "learning_rate": 1.5700171512866956e-07,
      "loss": 0.7859,
      "step": 7206
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3111121152194536,
      "learning_rate": 1.5648670285614397e-07,
      "loss": 0.701,
      "step": 7207
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3524642280907553,
      "learning_rate": 1.559725232397058e-07,
      "loss": 0.6609,
      "step": 7208
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.553193689591855,
      "learning_rate": 1.5545917636774655e-07,
      "loss": 0.8935,
      "step": 7209
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2966985823572765,
      "learning_rate": 1.5494666232851896e-07,
      "loss": 0.7521,
      "step": 7210
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.474883165953931,
      "learning_rate": 1.5443498121012813e-07,
      "loss": 0.6572,
      "step": 7211
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3650454991243246,
      "learning_rate": 1.5392413310053866e-07,
      "loss": 0.7376,
      "step": 7212
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4912419681198095,
      "learning_rate": 1.5341411808757146e-07,
      "loss": 0.7901,
      "step": 7213
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5469047598435306,
      "learning_rate": 1.5290493625890413e-07,
      "loss": 0.8125,
      "step": 7214
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2621190758091276,
      "learning_rate": 1.5239658770206945e-07,
      "loss": 0.7957,
      "step": 7215
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2854728568558564,
      "learning_rate": 1.5188907250446028e-07,
      "loss": 0.7354,
      "step": 7216
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3660585982410474,
      "learning_rate": 1.513823907533235e-07,
      "loss": 0.7833,
      "step": 7217
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5597274514522765,
      "learning_rate": 1.508765425357639e-07,
      "loss": 0.7748,
      "step": 7218
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3793543531651333,
      "learning_rate": 1.5037152793874244e-07,
      "loss": 0.7121,
      "step": 7219
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9084058433549572,
      "learning_rate": 1.4986734704907745e-07,
      "loss": 0.6887,
      "step": 7220
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6579438477733224,
      "learning_rate": 1.4936399995344298e-07,
      "loss": 0.7084,
      "step": 7221
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.268074674009846,
      "learning_rate": 1.4886148673836975e-07,
      "loss": 0.6696,
      "step": 7222
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2534994141406453,
      "learning_rate": 1.4835980749024592e-07,
      "loss": 0.7212,
      "step": 7223
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4766084635531236,
      "learning_rate": 1.4785896229531692e-07,
      "loss": 0.6946,
      "step": 7224
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2507622818203739,
      "learning_rate": 1.473589512396828e-07,
      "loss": 0.6811,
      "step": 7225
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6396192333079647,
      "learning_rate": 1.46859774409302e-07,
      "loss": 0.7859,
      "step": 7226
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.488877276364441,
      "learning_rate": 1.4636143188998808e-07,
      "loss": 0.7969,
      "step": 7227
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.818649609573922,
      "learning_rate": 1.4586392376741254e-07,
      "loss": 0.8223,
      "step": 7228
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3765641985359736,
      "learning_rate": 1.4536725012710252e-07,
      "loss": 0.7775,
      "step": 7229
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.611605915658955,
      "learning_rate": 1.4487141105444136e-07,
      "loss": 0.7755,
      "step": 7230
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8447342032374391,
      "learning_rate": 1.4437640663467034e-07,
      "loss": 0.6662,
      "step": 7231
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3386702014886627,
      "learning_rate": 1.4388223695288695e-07,
      "loss": 0.8447,
      "step": 7232
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.408221846071062,
      "learning_rate": 1.433889020940432e-07,
      "loss": 0.832,
      "step": 7233
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.626352656564721,
      "learning_rate": 1.4289640214294963e-07,
      "loss": 0.8062,
      "step": 7234
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4394788398483576,
      "learning_rate": 1.424047371842735e-07,
      "loss": 0.8233,
      "step": 7235
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8443383398051414,
      "learning_rate": 1.4191390730253718e-07,
      "loss": 0.6592,
      "step": 7236
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.249390085153082,
      "learning_rate": 1.4142391258211985e-07,
      "loss": 0.7074,
      "step": 7237
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3852104877129396,
      "learning_rate": 1.409347531072569e-07,
      "loss": 0.6931,
      "step": 7238
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2510574070793996,
      "learning_rate": 1.4044642896204107e-07,
      "loss": 0.6934,
      "step": 7239
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.309056322310479,
      "learning_rate": 1.3995894023042135e-07,
      "loss": 0.7032,
      "step": 7240
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.366105662959338,
      "learning_rate": 1.394722869962023e-07,
      "loss": 0.8243,
      "step": 7241
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.41127116808141,
      "learning_rate": 1.3898646934304538e-07,
      "loss": 0.7295,
      "step": 7242
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5479965053431008,
      "learning_rate": 1.3850148735446767e-07,
      "loss": 0.7724,
      "step": 7243
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.611716478203782,
      "learning_rate": 1.38017341113843e-07,
      "loss": 0.7524,
      "step": 7244
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.842498671176308,
      "learning_rate": 1.3753403070440263e-07,
      "loss": 0.6293,
      "step": 7245
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5572380209628633,
      "learning_rate": 1.3705155620923337e-07,
      "loss": 0.7444,
      "step": 7246
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3750832111619373,
      "learning_rate": 1.3656991771127781e-07,
      "loss": 0.7509,
      "step": 7247
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2660204428466582,
      "learning_rate": 1.3608911529333467e-07,
      "loss": 0.7505,
      "step": 7248
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3944427253766274,
      "learning_rate": 1.3560914903806065e-07,
      "loss": 0.7431,
      "step": 7249
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4500506913324813,
      "learning_rate": 1.3513001902796642e-07,
      "loss": 0.7021,
      "step": 7250
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2470471534881675,
      "learning_rate": 1.3465172534541936e-07,
      "loss": 0.7726,
      "step": 7251
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.337178351396768,
      "learning_rate": 1.34174268072646e-07,
      "loss": 0.8253,
      "step": 7252
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3241682617140227,
      "learning_rate": 1.3369764729172453e-07,
      "loss": 0.7401,
      "step": 7253
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5130839788980093,
      "learning_rate": 1.3322186308459274e-07,
      "loss": 0.7179,
      "step": 7254
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.466874918552609,
      "learning_rate": 1.3274691553304352e-07,
      "loss": 0.7668,
      "step": 7255
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4328137781730048,
      "learning_rate": 1.322728047187255e-07,
      "loss": 0.7747,
      "step": 7256
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3357758627965588,
      "learning_rate": 1.31799530723144e-07,
      "loss": 0.7678,
      "step": 7257
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2711283900316968,
      "learning_rate": 1.3132709362766006e-07,
      "loss": 0.7797,
      "step": 7258
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3093266947462059,
      "learning_rate": 1.308554935134909e-07,
      "loss": 0.7688,
      "step": 7259
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6706246016211088,
      "learning_rate": 1.3038473046171063e-07,
      "loss": 0.7859,
      "step": 7260
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1746223107950888,
      "learning_rate": 1.299148045532489e-07,
      "loss": 0.6975,
      "step": 7261
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4385470568527037,
      "learning_rate": 1.2944571586888998e-07,
      "loss": 0.7655,
      "step": 7262
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2959182173903336,
      "learning_rate": 1.2897746448927828e-07,
      "loss": 0.7077,
      "step": 7263
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.451249351103663,
      "learning_rate": 1.2851005049490939e-07,
      "loss": 0.7306,
      "step": 7264
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4317610422965932,
      "learning_rate": 1.2804347396613848e-07,
      "loss": 0.767,
      "step": 7265
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5965042909408216,
      "learning_rate": 1.2757773498317416e-07,
      "loss": 0.7816,
      "step": 7266
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5458141918840034,
      "learning_rate": 1.2711283362608351e-07,
      "loss": 0.8322,
      "step": 7267
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3240888534321638,
      "learning_rate": 1.266487699747887e-07,
      "loss": 0.8089,
      "step": 7268
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3230290284884056,
      "learning_rate": 1.2618554410906648e-07,
      "loss": 0.7715,
      "step": 7269
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4413041729470155,
      "learning_rate": 1.2572315610855201e-07,
      "loss": 0.8067,
      "step": 7270
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6434286123101491,
      "learning_rate": 1.252616060527334e-07,
      "loss": 0.7066,
      "step": 7271
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3384854586331938,
      "learning_rate": 1.248008940209583e-07,
      "loss": 0.7695,
      "step": 7272
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.256726721864605,
      "learning_rate": 1.243410200924272e-07,
      "loss": 0.7414,
      "step": 7273
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3991835108314592,
      "learning_rate": 1.2388198434619803e-07,
      "loss": 0.8415,
      "step": 7274
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4161123728007725,
      "learning_rate": 1.2342378686118538e-07,
      "loss": 0.7411,
      "step": 7275
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8653035608564554,
      "learning_rate": 1.2296642771615741e-07,
      "loss": 0.6536,
      "step": 7276
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3687052367482995,
      "learning_rate": 1.2250990698974009e-07,
      "loss": 0.7856,
      "step": 7277
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3371329657213573,
      "learning_rate": 1.2205422476041452e-07,
      "loss": 0.786,
      "step": 7278
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.554290556215938,
      "learning_rate": 1.2159938110651803e-07,
      "loss": 0.7108,
      "step": 7279
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.321079767419032,
      "learning_rate": 1.2114537610624255e-07,
      "loss": 0.7517,
      "step": 7280
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3972196910718153,
      "learning_rate": 1.206922098376373e-07,
      "loss": 0.7669,
      "step": 7281
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4324624051148918,
      "learning_rate": 1.2023988237860718e-07,
      "loss": 0.848,
      "step": 7282
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8648470698292594,
      "learning_rate": 1.1978839380691277e-07,
      "loss": 0.6502,
      "step": 7283
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3742936627303952,
      "learning_rate": 1.1933774420016974e-07,
      "loss": 0.7236,
      "step": 7284
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.63385329795789,
      "learning_rate": 1.188879336358495e-07,
      "loss": 0.765,
      "step": 7285
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8246475412323737,
      "learning_rate": 1.184389621912807e-07,
      "loss": 0.722,
      "step": 7286
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2943841833138852,
      "learning_rate": 1.1799082994364553e-07,
      "loss": 0.6637,
      "step": 7287
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.591798661355766,
      "learning_rate": 1.17543536969984e-07,
      "loss": 0.812,
      "step": 7288
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.336906894182228,
      "learning_rate": 1.1709708334719128e-07,
      "loss": 0.8502,
      "step": 7289
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7577452970664041,
      "learning_rate": 1.1665146915201652e-07,
      "loss": 0.7663,
      "step": 7290
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.854910303684077,
      "learning_rate": 1.1620669446106735e-07,
      "loss": 0.6236,
      "step": 7291
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3450285060327394,
      "learning_rate": 1.157627593508054e-07,
      "loss": 0.803,
      "step": 7292
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.59019688601346,
      "learning_rate": 1.153196638975479e-07,
      "loss": 0.8082,
      "step": 7293
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9147217935623254,
      "learning_rate": 1.148774081774684e-07,
      "loss": 0.6244,
      "step": 7294
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.830302009515638,
      "learning_rate": 1.1443599226659497e-07,
      "loss": 0.6687,
      "step": 7295
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5530011725629722,
      "learning_rate": 1.1399541624081357e-07,
      "loss": 0.75,
      "step": 7296
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3864819494335505,
      "learning_rate": 1.1355568017586305e-07,
      "loss": 0.7109,
      "step": 7297
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.298255009580745,
      "learning_rate": 1.1311678414734018e-07,
      "loss": 0.7655,
      "step": 7298
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8536725065061338,
      "learning_rate": 1.1267872823069459e-07,
      "loss": 0.6319,
      "step": 7299
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4860335261601816,
      "learning_rate": 1.1224151250123549e-07,
      "loss": 0.7785,
      "step": 7300
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.330047484830679,
      "learning_rate": 1.1180513703412388e-07,
      "loss": 0.7036,
      "step": 7301
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.185408391035694,
      "learning_rate": 1.1136960190437751e-07,
      "loss": 0.7587,
      "step": 7302
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5325981661803565,
      "learning_rate": 1.1093490718687094e-07,
      "loss": 0.7577,
      "step": 7303
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3901457807213176,
      "learning_rate": 1.1050105295633274e-07,
      "loss": 0.731,
      "step": 7304
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3041663396808447,
      "learning_rate": 1.1006803928734711e-07,
      "loss": 0.6887,
      "step": 7305
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3116630026364728,
      "learning_rate": 1.0963586625435507e-07,
      "loss": 0.8168,
      "step": 7306
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2770646147255187,
      "learning_rate": 1.0920453393165109e-07,
      "loss": 0.7544,
      "step": 7307
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3654143784393262,
      "learning_rate": 1.087740423933864e-07,
      "loss": 0.7936,
      "step": 7308
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.419989540695564,
      "learning_rate": 1.0834439171356848e-07,
      "loss": 0.7824,
      "step": 7309
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8964374587949022,
      "learning_rate": 1.0791558196605823e-07,
      "loss": 0.7235,
      "step": 7310
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.317390956178209,
      "learning_rate": 1.0748761322457334e-07,
      "loss": 0.7339,
      "step": 7311
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4243136088312545,
      "learning_rate": 1.0706048556268667e-07,
      "loss": 0.7869,
      "step": 7312
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.251639157119592,
      "learning_rate": 1.0663419905382666e-07,
      "loss": 0.8055,
      "step": 7313
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3058169750803073,
      "learning_rate": 1.0620875377127637e-07,
      "loss": 0.6946,
      "step": 7314
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3919387303203976,
      "learning_rate": 1.0578414978817508e-07,
      "loss": 0.6086,
      "step": 7315
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2093131341246712,
      "learning_rate": 1.0536038717751607e-07,
      "loss": 0.8236,
      "step": 7316
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.247102706193398,
      "learning_rate": 1.0493746601215105e-07,
      "loss": 0.7688,
      "step": 7317
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2675910779859587,
      "learning_rate": 1.0451538636478353e-07,
      "loss": 0.6908,
      "step": 7318
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7046211313326218,
      "learning_rate": 1.0409414830797493e-07,
      "loss": 0.6881,
      "step": 7319
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6333172768549962,
      "learning_rate": 1.0367375191414064e-07,
      "loss": 0.6945,
      "step": 7320
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.413978219554581,
      "learning_rate": 1.032541972555512e-07,
      "loss": 0.8413,
      "step": 7321
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.409451558589579,
      "learning_rate": 1.0283548440433332e-07,
      "loss": 0.8053,
      "step": 7322
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4413537418018751,
      "learning_rate": 1.0241761343246781e-07,
      "loss": 0.734,
      "step": 7323
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4365678437746379,
      "learning_rate": 1.0200058441179272e-07,
      "loss": 0.7852,
      "step": 7324
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5834066920579455,
      "learning_rate": 1.0158439741399961e-07,
      "loss": 0.7567,
      "step": 7325
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.917304444494399,
      "learning_rate": 1.0116905251063625e-07,
      "loss": 0.762,
      "step": 7326
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3461481260017183,
      "learning_rate": 1.0075454977310384e-07,
      "loss": 0.7342,
      "step": 7327
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2146375859050271,
      "learning_rate": 1.0034088927266206e-07,
      "loss": 0.7262,
      "step": 7328
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8165255456087417,
      "learning_rate": 9.99280710804229e-08,
      "loss": 0.6888,
      "step": 7329
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3207108953004871,
      "learning_rate": 9.951609526735517e-08,
      "loss": 0.7167,
      "step": 7330
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.433629032913613,
      "learning_rate": 9.910496190428164e-08,
      "loss": 0.7874,
      "step": 7331
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.46370030162552,
      "learning_rate": 9.869467106188135e-08,
      "loss": 0.8284,
      "step": 7332
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.276402797372759,
      "learning_rate": 9.828522281068787e-08,
      "loss": 0.6907,
      "step": 7333
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4425209049223926,
      "learning_rate": 9.787661722108988e-08,
      "loss": 0.8481,
      "step": 7334
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.328100177533564,
      "learning_rate": 9.74688543633323e-08,
      "loss": 0.7264,
      "step": 7335
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.258565766635221,
      "learning_rate": 9.706193430751298e-08,
      "loss": 0.7492,
      "step": 7336
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4334865018198337,
      "learning_rate": 9.665585712358704e-08,
      "loss": 0.7668,
      "step": 7337
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2546439841697716,
      "learning_rate": 9.625062288136367e-08,
      "loss": 0.8292,
      "step": 7338
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2889416999313175,
      "learning_rate": 9.584623165050766e-08,
      "loss": 0.785,
      "step": 7339
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7688059084265937,
      "learning_rate": 9.544268350053843e-08,
      "loss": 0.5939,
      "step": 7340
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4681933195575594,
      "learning_rate": 9.50399785008299e-08,
      "loss": 0.7425,
      "step": 7341
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2762680377443143,
      "learning_rate": 9.463811672061284e-08,
      "loss": 0.7718,
      "step": 7342
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7738010498999675,
      "learning_rate": 9.423709822897087e-08,
      "loss": 0.6003,
      "step": 7343
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8689839048281927,
      "learning_rate": 9.383692309484382e-08,
      "loss": 0.702,
      "step": 7344
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2507387235473661,
      "learning_rate": 9.343759138702724e-08,
      "loss": 0.77,
      "step": 7345
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7756819951685952,
      "learning_rate": 9.303910317417064e-08,
      "loss": 0.6178,
      "step": 7346
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8263193958049116,
      "learning_rate": 9.264145852477868e-08,
      "loss": 0.6354,
      "step": 7347
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4334197510460214,
      "learning_rate": 9.224465750721057e-08,
      "loss": 0.7416,
      "step": 7348
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3587573904835428,
      "learning_rate": 9.184870018968173e-08,
      "loss": 0.7988,
      "step": 7349
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2654991096822652,
      "learning_rate": 9.14535866402616e-08,
      "loss": 0.7775,
      "step": 7350
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.25589729984983,
      "learning_rate": 9.10593169268742e-08,
      "loss": 0.7577,
      "step": 7351
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2765158222827053,
      "learning_rate": 9.066589111729973e-08,
      "loss": 0.7674,
      "step": 7352
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6892473449268077,
      "learning_rate": 9.027330927917244e-08,
      "loss": 0.8195,
      "step": 7353
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3316288009655834,
      "learning_rate": 8.988157147998222e-08,
      "loss": 0.7544,
      "step": 7354
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2882115636698501,
      "learning_rate": 8.949067778707188e-08,
      "loss": 0.7224,
      "step": 7355
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3544559899850082,
      "learning_rate": 8.91006282676421e-08,
      "loss": 0.7735,
      "step": 7356
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7849323697035364,
      "learning_rate": 8.871142298874647e-08,
      "loss": 0.789,
      "step": 7357
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5101775223068572,
      "learning_rate": 8.832306201729368e-08,
      "loss": 0.6288,
      "step": 7358
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.387668348869189,
      "learning_rate": 8.793554542004756e-08,
      "loss": 0.7595,
      "step": 7359
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3961293121715088,
      "learning_rate": 8.754887326362649e-08,
      "loss": 0.7141,
      "step": 7360
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1406529367499707,
      "learning_rate": 8.71630456145045e-08,
      "loss": 0.8307,
      "step": 7361
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.51578032082396,
      "learning_rate": 8.677806253900967e-08,
      "loss": 0.672,
      "step": 7362
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3944593545684634,
      "learning_rate": 8.639392410332403e-08,
      "loss": 0.7557,
      "step": 7363
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.354279257540418,
      "learning_rate": 8.601063037348644e-08,
      "loss": 0.7539,
      "step": 7364
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4155595368403817,
      "learning_rate": 8.562818141538976e-08,
      "loss": 0.7054,
      "step": 7365
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1556956696903216,
      "learning_rate": 8.52465772947808e-08,
      "loss": 0.7401,
      "step": 7366
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5877035907287254,
      "learning_rate": 8.486581807726157e-08,
      "loss": 0.7638,
      "step": 7367
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.394586020886639,
      "learning_rate": 8.448590382829025e-08,
      "loss": 0.8183,
      "step": 7368
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4087507982795078,
      "learning_rate": 8.410683461317682e-08,
      "loss": 0.7897,
      "step": 7369
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4507573670164116,
      "learning_rate": 8.372861049708859e-08,
      "loss": 0.8067,
      "step": 7370
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5540950591182923,
      "learning_rate": 8.335123154504688e-08,
      "loss": 0.8164,
      "step": 7371
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2373628470903753,
      "learning_rate": 8.297469782192702e-08,
      "loss": 0.8213,
      "step": 7372
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4116269560371355,
      "learning_rate": 8.25990093924589e-08,
      "loss": 0.7989,
      "step": 7373
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3037779627210957,
      "learning_rate": 8.222416632122864e-08,
      "loss": 0.7332,
      "step": 7374
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3088086064013233,
      "learning_rate": 8.185016867267693e-08,
      "loss": 0.726,
      "step": 7375
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5430360505073,
      "learning_rate": 8.147701651109674e-08,
      "loss": 0.7642,
      "step": 7376
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2238397412309845,
      "learning_rate": 8.11047099006379e-08,
      "loss": 0.7423,
      "step": 7377
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.372356315940302,
      "learning_rate": 8.073324890530421e-08,
      "loss": 0.693,
      "step": 7378
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.328600494158757,
      "learning_rate": 8.036263358895402e-08,
      "loss": 0.8004,
      "step": 7379
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3307914857337018,
      "learning_rate": 7.999286401529971e-08,
      "loss": 0.6981,
      "step": 7380
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8111996894819756,
      "learning_rate": 7.962394024791043e-08,
      "loss": 0.5932,
      "step": 7381
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3626170272260347,
      "learning_rate": 7.925586235020766e-08,
      "loss": 0.7596,
      "step": 7382
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3595544577162804,
      "learning_rate": 7.888863038546801e-08,
      "loss": 0.6814,
      "step": 7383
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4358208888091217,
      "learning_rate": 7.852224441682377e-08,
      "loss": 0.6513,
      "step": 7384
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3533209157551065,
      "learning_rate": 7.81567045072601e-08,
      "loss": 0.7947,
      "step": 7385
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.944898189028827,
      "learning_rate": 7.779201071961784e-08,
      "loss": 0.6066,
      "step": 7386
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3438450929147854,
      "learning_rate": 7.742816311659185e-08,
      "loss": 0.7442,
      "step": 7387
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2621644139221255,
      "learning_rate": 7.706516176073209e-08,
      "loss": 0.6441,
      "step": 7388
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8772956083337563,
      "learning_rate": 7.670300671444309e-08,
      "loss": 0.6354,
      "step": 7389
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3353005388718244,
      "learning_rate": 7.634169803998226e-08,
      "loss": 0.7029,
      "step": 7390
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.456959976247008,
      "learning_rate": 7.598123579946382e-08,
      "loss": 0.7863,
      "step": 7391
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3896999966837604,
      "learning_rate": 7.562162005485484e-08,
      "loss": 0.7481,
      "step": 7392
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.914361682768958,
      "learning_rate": 7.526285086797813e-08,
      "loss": 0.7222,
      "step": 7393
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2609433931420602,
      "learning_rate": 7.490492830050933e-08,
      "loss": 0.7861,
      "step": 7394
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8921579999513181,
      "learning_rate": 7.454785241398033e-08,
      "loss": 0.6005,
      "step": 7395
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4970424518008745,
      "learning_rate": 7.419162326977592e-08,
      "loss": 0.7938,
      "step": 7396
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3224578766563384,
      "learning_rate": 7.383624092913655e-08,
      "loss": 0.7675,
      "step": 7397
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4780408729101848,
      "learning_rate": 7.348170545315614e-08,
      "loss": 0.8232,
      "step": 7398
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6688687292666422,
      "learning_rate": 7.312801690278426e-08,
      "loss": 0.6758,
      "step": 7399
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5810853636423952,
      "learning_rate": 7.277517533882283e-08,
      "loss": 0.793,
      "step": 7400
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4398423565463327,
      "learning_rate": 7.242318082193e-08,
      "loss": 0.8114,
      "step": 7401
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3988968486939068,
      "learning_rate": 7.207203341261792e-08,
      "loss": 0.6878,
      "step": 7402
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.517188053384217,
      "learning_rate": 7.172173317125275e-08,
      "loss": 0.8099,
      "step": 7403
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.688968098107839,
      "learning_rate": 7.137228015805519e-08,
      "loss": 0.7897,
      "step": 7404
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.304259237907046,
      "learning_rate": 7.102367443310054e-08,
      "loss": 0.8178,
      "step": 7405
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3478349445055333,
      "learning_rate": 7.067591605631752e-08,
      "loss": 0.683,
      "step": 7406
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6505291960286412,
      "learning_rate": 7.032900508749052e-08,
      "loss": 0.7207,
      "step": 7407
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.342948429635336,
      "learning_rate": 6.998294158625684e-08,
      "loss": 0.7011,
      "step": 7408
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.351513594412302,
      "learning_rate": 6.963772561210891e-08,
      "loss": 0.6457,
      "step": 7409
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.919667654226108,
      "learning_rate": 6.929335722439367e-08,
      "loss": 0.6971,
      "step": 7410
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8043902864961263,
      "learning_rate": 6.894983648231213e-08,
      "loss": 0.8038,
      "step": 7411
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4520826008954897,
      "learning_rate": 6.86071634449198e-08,
      "loss": 0.8018,
      "step": 7412
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4153149592212635,
      "learning_rate": 6.826533817112513e-08,
      "loss": 0.7581,
      "step": 7413
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4225773913739967,
      "learning_rate": 6.792436071969277e-08,
      "loss": 0.8078,
      "step": 7414
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1666817296550727,
      "learning_rate": 6.758423114924029e-08,
      "loss": 0.8187,
      "step": 7415
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3638535099165272,
      "learning_rate": 6.724494951823979e-08,
      "loss": 0.698,
      "step": 7416
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3324494542249783,
      "learning_rate": 6.690651588501795e-08,
      "loss": 0.7255,
      "step": 7417
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2913601404257862,
      "learning_rate": 6.656893030775546e-08,
      "loss": 0.709,
      "step": 7418
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.454962511684222,
      "learning_rate": 6.623219284448645e-08,
      "loss": 0.6949,
      "step": 7419
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8257308831209738,
      "learning_rate": 6.589630355310128e-08,
      "loss": 0.6804,
      "step": 7420
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5230703783985675,
      "learning_rate": 6.556126249134209e-08,
      "loss": 0.7812,
      "step": 7421
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2771434274445435,
      "learning_rate": 6.522706971680726e-08,
      "loss": 0.7351,
      "step": 7422
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3337520921161072,
      "learning_rate": 6.489372528694748e-08,
      "loss": 0.7846,
      "step": 7423
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4275087085105258,
      "learning_rate": 6.456122925906971e-08,
      "loss": 0.7957,
      "step": 7424
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8183274460376578,
      "learning_rate": 6.422958169033266e-08,
      "loss": 0.664,
      "step": 7425
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3232326200097324,
      "learning_rate": 6.389878263775129e-08,
      "loss": 0.7494,
      "step": 7426
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4583490333519606,
      "learning_rate": 6.356883215819287e-08,
      "loss": 0.7656,
      "step": 7427
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4071119966891392,
      "learning_rate": 6.323973030838037e-08,
      "loss": 0.6775,
      "step": 7428
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2526512468503295,
      "learning_rate": 6.291147714488965e-08,
      "loss": 0.6598,
      "step": 7429
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7979525856895985,
      "learning_rate": 6.258407272415223e-08,
      "loss": 0.6471,
      "step": 7430
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.797974300430539,
      "learning_rate": 6.225751710245198e-08,
      "loss": 0.7236,
      "step": 7431
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3116492240969957,
      "learning_rate": 6.193181033592788e-08,
      "loss": 0.747,
      "step": 7432
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2783429163573576,
      "learning_rate": 6.160695248057236e-08,
      "loss": 0.7631,
      "step": 7433
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5540726066441475,
      "learning_rate": 6.128294359223297e-08,
      "loss": 0.6588,
      "step": 7434
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4340747215207417,
      "learning_rate": 6.09597837266096e-08,
      "loss": 0.8057,
      "step": 7435
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4242576906616813,
      "learning_rate": 6.063747293925781e-08,
      "loss": 0.7568,
      "step": 7436
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4210714900817367,
      "learning_rate": 6.031601128558606e-08,
      "loss": 0.811,
      "step": 7437
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4924314872915996,
      "learning_rate": 5.999539882085793e-08,
      "loss": 0.7469,
      "step": 7438
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4576986508412058,
      "learning_rate": 5.96756356001904e-08,
      "loss": 0.7552,
      "step": 7439
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5259988125645623,
      "learning_rate": 5.9356721678554554e-08,
      "loss": 0.6529,
      "step": 7440
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7134919737594987,
      "learning_rate": 5.903865711077483e-08,
      "loss": 0.7802,
      "step": 7441
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3064610592787642,
      "learning_rate": 5.872144195153029e-08,
      "loss": 0.7502,
      "step": 7442
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9605913934495365,
      "learning_rate": 5.840507625535397e-08,
      "loss": 0.7296,
      "step": 7443
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.107156375011861,
      "learning_rate": 5.808956007663291e-08,
      "loss": 0.6988,
      "step": 7444
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.321973017536826,
      "learning_rate": 5.777489346960874e-08,
      "loss": 0.7151,
      "step": 7445
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3299809977947266,
      "learning_rate": 5.7461076488374844e-08,
      "loss": 0.7134,
      "step": 7446
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.50662081559836,
      "learning_rate": 5.7148109186880854e-08,
      "loss": 0.7094,
      "step": 7447
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7843896765080505,
      "learning_rate": 5.683599161892928e-08,
      "loss": 0.6327,
      "step": 7448
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2829722632444294,
      "learning_rate": 5.65247238381772e-08,
      "loss": 0.732,
      "step": 7449
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7091905964496947,
      "learning_rate": 5.621430589813459e-08,
      "loss": 0.6399,
      "step": 7450
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.161678767850644,
      "learning_rate": 5.5904737852166545e-08,
      "loss": 0.6661,
      "step": 7451
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.368274299655663,
      "learning_rate": 5.559601975348994e-08,
      "loss": 0.6677,
      "step": 7452
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4503992555626073,
      "learning_rate": 5.5288151655178427e-08,
      "loss": 0.796,
      "step": 7453
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2743743076696465,
      "learning_rate": 5.4981133610158e-08,
      "loss": 0.737,
      "step": 7454
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3191795656538006,
      "learning_rate": 5.4674965671208115e-08,
      "loss": 0.7063,
      "step": 7455
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3429335574103518,
      "learning_rate": 5.436964789096222e-08,
      "loss": 0.7208,
      "step": 7456
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3277979880533874,
      "learning_rate": 5.406518032190944e-08,
      "loss": 0.7544,
      "step": 7457
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3194457911246307,
      "learning_rate": 5.3761563016389576e-08,
      "loss": 0.6979,
      "step": 7458
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4223289548399578,
      "learning_rate": 5.34587960265992e-08,
      "loss": 0.7024,
      "step": 7459
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5385258835653777,
      "learning_rate": 5.315687940458669e-08,
      "loss": 0.8614,
      "step": 7460
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9021981575590166,
      "learning_rate": 5.285581320225552e-08,
      "loss": 0.6919,
      "step": 7461
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2017033609051853,
      "learning_rate": 5.255559747136263e-08,
      "loss": 0.6768,
      "step": 7462
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3739341163226924,
      "learning_rate": 5.2256232263517835e-08,
      "loss": 0.7932,
      "step": 7463
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3202435517831022,
      "learning_rate": 5.195771763018609e-08,
      "loss": 0.6729,
      "step": 7464
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5944746760519855,
      "learning_rate": 5.166005362268467e-08,
      "loss": 0.8103,
      "step": 7465
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5045189620164592,
      "learning_rate": 5.1363240292186535e-08,
      "loss": 0.7699,
      "step": 7466
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.27496317931825,
      "learning_rate": 5.1067277689716974e-08,
      "loss": 0.7108,
      "step": 7467
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2953718695718035,
      "learning_rate": 5.07721658661553e-08,
      "loss": 0.6478,
      "step": 7468
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.787214222501255,
      "learning_rate": 5.0477904872234804e-08,
      "loss": 0.6892,
      "step": 7469
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2870479712822196,
      "learning_rate": 5.018449475854226e-08,
      "loss": 0.7797,
      "step": 7470
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2109655011109015,
      "learning_rate": 4.9891935575517856e-08,
      "loss": 0.627,
      "step": 7471
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5508587222762344,
      "learning_rate": 4.9600227373456936e-08,
      "loss": 0.7782,
      "step": 7472
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8667907513903883,
      "learning_rate": 4.930937020250604e-08,
      "loss": 0.6282,
      "step": 7473
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2174606357230138,
      "learning_rate": 4.90193641126685e-08,
      "loss": 0.7869,
      "step": 7474
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2680940682058683,
      "learning_rate": 4.873020915379834e-08,
      "loss": 0.6887,
      "step": 7475
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2759978785649857,
      "learning_rate": 4.844190537560578e-08,
      "loss": 0.6885,
      "step": 7476
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4022914898345218,
      "learning_rate": 4.8154452827652854e-08,
      "loss": 0.7889,
      "step": 7477
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2133171134585041,
      "learning_rate": 4.7867851559356694e-08,
      "loss": 0.7402,
      "step": 7478
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2337808254166303,
      "learning_rate": 4.758210161998622e-08,
      "loss": 0.773,
      "step": 7479
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.26107586590658,
      "learning_rate": 4.729720305866603e-08,
      "loss": 0.714,
      "step": 7480
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3822373637533356,
      "learning_rate": 4.701315592437361e-08,
      "loss": 0.6902,
      "step": 7481
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3320485077006115,
      "learning_rate": 4.6729960265939344e-08,
      "loss": 0.6479,
      "step": 7482
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2746438364674917,
      "learning_rate": 4.644761613204818e-08,
      "loss": 0.681,
      "step": 7483
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2199992879377355,
      "learning_rate": 4.6166123571237955e-08,
      "loss": 0.7298,
      "step": 7484
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.238717029740634,
      "learning_rate": 4.588548263190107e-08,
      "loss": 0.7763,
      "step": 7485
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8885680485397323,
      "learning_rate": 4.560569336228338e-08,
      "loss": 0.6086,
      "step": 7486
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2500910295149874,
      "learning_rate": 4.5326755810482514e-08,
      "loss": 0.7639,
      "step": 7487
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7914109786240064,
      "learning_rate": 4.504867002445179e-08,
      "loss": 0.6236,
      "step": 7488
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3847497527522443,
      "learning_rate": 4.477143605199796e-08,
      "loss": 0.7957,
      "step": 7489
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.824711205350274,
      "learning_rate": 4.449505394078013e-08,
      "loss": 0.6324,
      "step": 7490
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2746012424085513,
      "learning_rate": 4.4219523738311396e-08,
      "loss": 0.744,
      "step": 7491
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7866175212027194,
      "learning_rate": 4.3944845491958874e-08,
      "loss": 0.6617,
      "step": 7492
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7045779255016342,
      "learning_rate": 4.3671019248943126e-08,
      "loss": 0.8,
      "step": 7493
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2628122173619105,
      "learning_rate": 4.3398045056337604e-08,
      "loss": 0.8128,
      "step": 7494
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7788085160576935,
      "learning_rate": 4.312592296106977e-08,
      "loss": 0.6468,
      "step": 7495
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3088021189486305,
      "learning_rate": 4.285465300992164e-08,
      "loss": 0.7386,
      "step": 7496
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4179686986624516,
      "learning_rate": 4.258423524952648e-08,
      "loss": 0.7447,
      "step": 7497
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.288171231016844,
      "learning_rate": 4.231466972637211e-08,
      "loss": 0.758,
      "step": 7498
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3234330079230219,
      "learning_rate": 4.2045956486800877e-08,
      "loss": 0.7859,
      "step": 7499
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.469314593741883,
      "learning_rate": 4.177809557700752e-08,
      "loss": 0.7488,
      "step": 7500
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4191779654156726,
      "learning_rate": 4.1511087043039635e-08,
      "loss": 0.7674,
      "step": 7501
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2662198048277906,
      "learning_rate": 4.12449309307994e-08,
      "loss": 0.7847,
      "step": 7502
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2710780266559178,
      "learning_rate": 4.097962728604299e-08,
      "loss": 0.8251,
      "step": 7503
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.822100427165716,
      "learning_rate": 4.0715176154378366e-08,
      "loss": 0.791,
      "step": 7504
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4613075881365027,
      "learning_rate": 4.04515775812675e-08,
      "loss": 0.8033,
      "step": 7505
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2953639497669291,
      "learning_rate": 4.018883161202691e-08,
      "loss": 0.6976,
      "step": 7506
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.533975685156444,
      "learning_rate": 3.992693829182548e-08,
      "loss": 0.678,
      "step": 7507
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3431332700289746,
      "learning_rate": 3.966589766568496e-08,
      "loss": 0.7596,
      "step": 7508
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2317456823000925,
      "learning_rate": 3.940570977848168e-08,
      "loss": 0.7631,
      "step": 7509
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2143432248670545,
      "learning_rate": 3.914637467494542e-08,
      "loss": 0.7781,
      "step": 7510
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4974340100883266,
      "learning_rate": 3.888789239965885e-08,
      "loss": 0.8002,
      "step": 7511
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8927284096409547,
      "learning_rate": 3.863026299705697e-08,
      "loss": 0.7233,
      "step": 7512
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.873533730740455,
      "learning_rate": 3.8373486511429916e-08,
      "loss": 0.7092,
      "step": 7513
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.481386195858292,
      "learning_rate": 3.811756298692126e-08,
      "loss": 0.7713,
      "step": 7514
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4517614359581683,
      "learning_rate": 3.7862492467526376e-08,
      "loss": 0.6973,
      "step": 7515
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.28591302980506,
      "learning_rate": 3.7608274997095187e-08,
      "loss": 0.7021,
      "step": 7516
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7225205527685047,
      "learning_rate": 3.735491061932994e-08,
      "loss": 0.5429,
      "step": 7517
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3264907162294035,
      "learning_rate": 3.710239937778803e-08,
      "loss": 0.6947,
      "step": 7518
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8228950209677941,
      "learning_rate": 3.685074131587863e-08,
      "loss": 0.6783,
      "step": 7519
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7041044826673852,
      "learning_rate": 3.6599936476864325e-08,
      "loss": 0.846,
      "step": 7520
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3864783934402942,
      "learning_rate": 3.6349984903861214e-08,
      "loss": 0.7541,
      "step": 7521
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3007975778325611,
      "learning_rate": 3.6100886639839904e-08,
      "loss": 0.7001,
      "step": 7522
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9944654412672018,
      "learning_rate": 3.5852641727622264e-08,
      "loss": 0.698,
      "step": 7523
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.418702172396019,
      "learning_rate": 3.5605250209885256e-08,
      "loss": 0.8195,
      "step": 7524
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.354669153109151,
      "learning_rate": 3.535871212915765e-08,
      "loss": 0.7376,
      "step": 7525
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.343586343455446,
      "learning_rate": 3.51130275278222e-08,
      "loss": 0.7025,
      "step": 7526
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.164850809020122,
      "learning_rate": 3.4868196448115675e-08,
      "loss": 0.7026,
      "step": 7527
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8301145327694871,
      "learning_rate": 3.46242189321272e-08,
      "loss": 0.7018,
      "step": 7528
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.582518328131357,
      "learning_rate": 3.4381095021798203e-08,
      "loss": 0.7771,
      "step": 7529
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4280599748790233,
      "learning_rate": 3.4138824758925826e-08,
      "loss": 0.7889,
      "step": 7530
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8924323974621753,
      "learning_rate": 3.389740818515841e-08,
      "loss": 0.5896,
      "step": 7531
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5237074719693167,
      "learning_rate": 3.3656845341998865e-08,
      "loss": 0.7254,
      "step": 7532
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.898071160351112,
      "learning_rate": 3.341713627080245e-08,
      "loss": 0.6355,
      "step": 7533
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2717320662027223,
      "learning_rate": 3.317828101277787e-08,
      "loss": 0.6987,
      "step": 7534
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3990028662882827,
      "learning_rate": 3.2940279608986714e-08,
      "loss": 0.7299,
      "step": 7535
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.7973726780727506,
      "learning_rate": 3.270313210034515e-08,
      "loss": 0.6352,
      "step": 7536
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8043113416742929,
      "learning_rate": 3.246683852762056e-08,
      "loss": 0.6119,
      "step": 7537
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6177249537329945,
      "learning_rate": 3.22313989314349e-08,
      "loss": 0.8032,
      "step": 7538
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3489920943163973,
      "learning_rate": 3.199681335226357e-08,
      "loss": 0.876,
      "step": 7539
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5271881245439707,
      "learning_rate": 3.176308183043375e-08,
      "loss": 0.7279,
      "step": 7540
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4851086379549157,
      "learning_rate": 3.1530204406127196e-08,
      "loss": 0.7259,
      "step": 7541
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3898941476133149,
      "learning_rate": 3.129818111937744e-08,
      "loss": 0.7917,
      "step": 7542
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2989527310445559,
      "learning_rate": 3.1067012010073114e-08,
      "loss": 0.6875,
      "step": 7543
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.452886530750009,
      "learning_rate": 3.0836697117954115e-08,
      "loss": 0.7227,
      "step": 7544
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.465485162959149,
      "learning_rate": 3.0607236482613764e-08,
      "loss": 0.6946,
      "step": 7545
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.331295618910518,
      "learning_rate": 3.037863014349995e-08,
      "loss": 0.7074,
      "step": 7546
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.392366739603335,
      "learning_rate": 3.0150878139912906e-08,
      "loss": 0.7125,
      "step": 7547
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.151258340761874,
      "learning_rate": 2.9923980511004645e-08,
      "loss": 0.7107,
      "step": 7548
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8834154423528698,
      "learning_rate": 2.96979372957823e-08,
      "loss": 0.6776,
      "step": 7549
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3086973263921962,
      "learning_rate": 2.947274853310589e-08,
      "loss": 0.7375,
      "step": 7550
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3143585457116758,
      "learning_rate": 2.9248414261686674e-08,
      "loss": 0.7656,
      "step": 7551
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3189090315406764,
      "learning_rate": 2.902493452009103e-08,
      "loss": 0.7204,
      "step": 7552
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2045069659756262,
      "learning_rate": 2.8802309346737666e-08,
      "loss": 0.7023,
      "step": 7553
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4494112974110733,
      "learning_rate": 2.8580538779898192e-08,
      "loss": 0.7395,
      "step": 7554
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8165043907834242,
      "learning_rate": 2.8359622857698223e-08,
      "loss": 0.677,
      "step": 7555
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4619715822187809,
      "learning_rate": 2.813956161811515e-08,
      "loss": 0.5728,
      "step": 7556
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7388855545183192,
      "learning_rate": 2.7920355098979835e-08,
      "loss": 0.6353,
      "step": 7557
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2750893081876518,
      "learning_rate": 2.7702003337977124e-08,
      "loss": 0.8224,
      "step": 7558
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6464249196401768,
      "learning_rate": 2.748450637264366e-08,
      "loss": 0.8415,
      "step": 7559
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4091231327094156,
      "learning_rate": 2.7267864240369533e-08,
      "loss": 0.7698,
      "step": 7560
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4328117637426252,
      "learning_rate": 2.7052076978398844e-08,
      "loss": 0.7134,
      "step": 7561
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4569189612110274,
      "learning_rate": 2.683714462382747e-08,
      "loss": 0.8375,
      "step": 7562
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4496152334577748,
      "learning_rate": 2.6623067213605302e-08,
      "loss": 0.7893,
      "step": 7563
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2903631293095204,
      "learning_rate": 2.6409844784533456e-08,
      "loss": 0.756,
      "step": 7564
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2965986035732218,
      "learning_rate": 2.6197477373268722e-08,
      "loss": 0.7594,
      "step": 7565
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1590889707534493,
      "learning_rate": 2.5985965016318004e-08,
      "loss": 0.6628,
      "step": 7566
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.962422571905379,
      "learning_rate": 2.577530775004389e-08,
      "loss": 0.8019,
      "step": 7567
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5300360638456652,
      "learning_rate": 2.5565505610660734e-08,
      "loss": 0.6952,
      "step": 7568
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5881292019267965,
      "learning_rate": 2.5356558634235806e-08,
      "loss": 0.7413,
      "step": 7569
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3452504489081738,
      "learning_rate": 2.5148466856689812e-08,
      "loss": 0.7759,
      "step": 7570
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1426998485895363,
      "learning_rate": 2.4941230313795252e-08,
      "loss": 0.7991,
      "step": 7571
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2661988311081518,
      "learning_rate": 2.4734849041179176e-08,
      "loss": 0.6734,
      "step": 7572
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4642586474968318,
      "learning_rate": 2.4529323074320988e-08,
      "loss": 0.8002,
      "step": 7573
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4342879267908817,
      "learning_rate": 2.432465244855242e-08,
      "loss": 0.6981,
      "step": 7574
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.344442791596343,
      "learning_rate": 2.412083719905922e-08,
      "loss": 0.7489,
      "step": 7575
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4305209818534328,
      "learning_rate": 2.3917877360879472e-08,
      "loss": 0.7471,
      "step": 7576
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8548080891456372,
      "learning_rate": 2.371577296890415e-08,
      "loss": 0.6087,
      "step": 7577
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4311604582115036,
      "learning_rate": 2.3514524057877685e-08,
      "loss": 0.7201,
      "step": 7578
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3820254425208067,
      "learning_rate": 2.331413066239685e-08,
      "loss": 0.6564,
      "step": 7579
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.415259470075818,
      "learning_rate": 2.311459281691186e-08,
      "loss": 0.8959,
      "step": 7580
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8085260551675884,
      "learning_rate": 2.2915910555725286e-08,
      "loss": 0.5797,
      "step": 7581
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.308324964355183,
      "learning_rate": 2.2718083912992573e-08,
      "loss": 0.7596,
      "step": 7582
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3793650297954052,
      "learning_rate": 2.2521112922723186e-08,
      "loss": 0.7764,
      "step": 7583
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4455517393616932,
      "learning_rate": 2.2324997618778375e-08,
      "loss": 0.8238,
      "step": 7584
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3460357527541382,
      "learning_rate": 2.212973803487284e-08,
      "loss": 0.6579,
      "step": 7585
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3445941220160305,
      "learning_rate": 2.1935334204573056e-08,
      "loss": 0.7664,
      "step": 7586
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3454071265768444,
      "learning_rate": 2.1741786161300628e-08,
      "loss": 0.7926,
      "step": 7587
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.135977477025408,
      "learning_rate": 2.1549093938327826e-08,
      "loss": 0.6466,
      "step": 7588
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9036665231901222,
      "learning_rate": 2.135725756878093e-08,
      "loss": 0.6739,
      "step": 7589
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3279223230636477,
      "learning_rate": 2.116627708563912e-08,
      "loss": 0.7899,
      "step": 7590
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5429875044589079,
      "learning_rate": 2.0976152521733905e-08,
      "loss": 0.71,
      "step": 7591
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5663540419595992,
      "learning_rate": 2.07868839097497e-08,
      "loss": 0.8158,
      "step": 7592
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3834187183084847,
      "learning_rate": 2.059847128222381e-08,
      "loss": 0.7861,
      "step": 7593
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9040453356352108,
      "learning_rate": 2.041091467154699e-08,
      "loss": 0.6629,
      "step": 7594
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3522338979250934,
      "learning_rate": 2.022421410996234e-08,
      "loss": 0.7511,
      "step": 7595
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.376774874709334,
      "learning_rate": 2.0038369629565846e-08,
      "loss": 0.6862,
      "step": 7596
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2640455188700033,
      "learning_rate": 1.9853381262306405e-08,
      "loss": 0.7115,
      "step": 7597
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.68659977085065,
      "learning_rate": 1.9669249039985794e-08,
      "loss": 0.7054,
      "step": 7598
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.568702927162674,
      "learning_rate": 1.9485972994257584e-08,
      "loss": 0.7758,
      "step": 7599
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6457014219984296,
      "learning_rate": 1.9303553156630462e-08,
      "loss": 0.7835,
      "step": 7600
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2959604643581124,
      "learning_rate": 1.912198955846323e-08,
      "loss": 0.7105,
      "step": 7601
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.363850113795898,
      "learning_rate": 1.894128223096925e-08,
      "loss": 0.7854,
      "step": 7602
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9442733435095483,
      "learning_rate": 1.8761431205214232e-08,
      "loss": 0.6789,
      "step": 7603
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4029746355299826,
      "learning_rate": 1.8582436512116776e-08,
      "loss": 0.79,
      "step": 7604
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7146093629576822,
      "learning_rate": 1.8404298182447823e-08,
      "loss": 0.7172,
      "step": 7605
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4384940117491334,
      "learning_rate": 1.8227016246831764e-08,
      "loss": 0.7877,
      "step": 7606
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5699811220513684,
      "learning_rate": 1.8050590735745334e-08,
      "loss": 0.7681,
      "step": 7607
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4385070852948028,
      "learning_rate": 1.7875021679518156e-08,
      "loss": 0.7254,
      "step": 7608
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.545684684119617,
      "learning_rate": 1.7700309108332204e-08,
      "loss": 0.7873,
      "step": 7609
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3385368043189252,
      "learning_rate": 1.7526453052223446e-08,
      "loss": 0.7425,
      "step": 7610
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.538781522032567,
      "learning_rate": 1.7353453541078534e-08,
      "loss": 0.7328,
      "step": 7611
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3180087117323547,
      "learning_rate": 1.7181310604639236e-08,
      "loss": 0.7446,
      "step": 7612
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.418070244733162,
      "learning_rate": 1.701002427249854e-08,
      "loss": 0.8018,
      "step": 7613
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4001935640659693,
      "learning_rate": 1.683959457410178e-08,
      "loss": 0.755,
      "step": 7614
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5357851971609828,
      "learning_rate": 1.667002153874886e-08,
      "loss": 0.7666,
      "step": 7615
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.320307240865144,
      "learning_rate": 1.65013051955909e-08,
      "loss": 0.7344,
      "step": 7616
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5398945899632632,
      "learning_rate": 1.6333445573632478e-08,
      "loss": 0.7102,
      "step": 7617
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3248958432793176,
      "learning_rate": 1.6166442701730513e-08,
      "loss": 0.7282,
      "step": 7618
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.755916332737291,
      "learning_rate": 1.6000296608594257e-08,
      "loss": 0.6489,
      "step": 7619
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8846609632208484,
      "learning_rate": 1.5835007322786424e-08,
      "loss": 0.6583,
      "step": 7620
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2751815238053927,
      "learning_rate": 1.567057487272261e-08,
      "loss": 0.6955,
      "step": 7621
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8350230366198381,
      "learning_rate": 1.5506999286669656e-08,
      "loss": 0.7821,
      "step": 7622
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4640051399831786,
      "learning_rate": 1.534428059274895e-08,
      "loss": 0.7562,
      "step": 7623
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2967327295547335,
      "learning_rate": 1.5182418818933676e-08,
      "loss": 0.7979,
      "step": 7624
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4370156417221454,
      "learning_rate": 1.502141399304935e-08,
      "loss": 0.7769,
      "step": 7625
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3985980259019857,
      "learning_rate": 1.4861266142775498e-08,
      "loss": 0.7979,
      "step": 7626
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4091446392450424,
      "learning_rate": 1.4701975295641768e-08,
      "loss": 0.7546,
      "step": 7627
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.459773821562005,
      "learning_rate": 1.4543541479033473e-08,
      "loss": 0.7934,
      "step": 7628
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4600175340584005,
      "learning_rate": 1.4385964720187162e-08,
      "loss": 0.719,
      "step": 7629
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3962978421501326,
      "learning_rate": 1.4229245046190609e-08,
      "loss": 0.7781,
      "step": 7630
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3469981367083113,
      "learning_rate": 1.4073382483987819e-08,
      "loss": 0.692,
      "step": 7631
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2977397620987252,
      "learning_rate": 1.3918377060371802e-08,
      "loss": 0.8138,
      "step": 7632
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4783525704720408,
      "learning_rate": 1.3764228801990686e-08,
      "loss": 0.7742,
      "step": 7633
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3247624304850292,
      "learning_rate": 1.3610937735344387e-08,
      "loss": 0.7737,
      "step": 7634
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8405771491488359,
      "learning_rate": 1.3458503886784603e-08,
      "loss": 0.6891,
      "step": 7635
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3272657409062585,
      "learning_rate": 1.3306927282517034e-08,
      "loss": 0.8097,
      "step": 7636
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2959077859902424,
      "learning_rate": 1.3156207948599176e-08,
      "loss": 0.7202,
      "step": 7637
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2631176348243955,
      "learning_rate": 1.3006345910941964e-08,
      "loss": 0.775,
      "step": 7638
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2773490160297931,
      "learning_rate": 1.2857341195308126e-08,
      "loss": 0.7754,
      "step": 7639
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3764059300697629,
      "learning_rate": 1.2709193827312727e-08,
      "loss": 0.7074,
      "step": 7640
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3974338698135969,
      "learning_rate": 1.2561903832424837e-08,
      "loss": 0.7634,
      "step": 7641
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.8623995692715214,
      "learning_rate": 1.2415471235965315e-08,
      "loss": 0.7827,
      "step": 7642
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3112668202135036,
      "learning_rate": 1.2269896063107356e-08,
      "loss": 0.7541,
      "step": 7643
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.350493382673375,
      "learning_rate": 1.2125178338876498e-08,
      "loss": 0.7824,
      "step": 7644
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4823137168658216,
      "learning_rate": 1.1981318088152283e-08,
      "loss": 0.7587,
      "step": 7645
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3405545312868175,
      "learning_rate": 1.1838315335664929e-08,
      "loss": 0.7031,
      "step": 7646
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3364599001953268,
      "learning_rate": 1.1696170105999772e-08,
      "loss": 0.8252,
      "step": 7647
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.256704882860583,
      "learning_rate": 1.1554882423591706e-08,
      "loss": 0.6438,
      "step": 7648
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4804485746037788,
      "learning_rate": 1.141445231273075e-08,
      "loss": 0.667,
      "step": 7649
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3916613251474468,
      "learning_rate": 1.1274879797558148e-08,
      "loss": 0.8086,
      "step": 7650
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3552451805827563,
      "learning_rate": 1.1136164902067481e-08,
      "loss": 0.7432,
      "step": 7651
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.359701495450146,
      "learning_rate": 1.0998307650106344e-08,
      "loss": 0.9056,
      "step": 7652
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2555636587822991,
      "learning_rate": 1.0861308065373556e-08,
      "loss": 0.737,
      "step": 7653
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9070668005887211,
      "learning_rate": 1.072516617142083e-08,
      "loss": 0.6857,
      "step": 7654
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3332675556862015,
      "learning_rate": 1.0589881991652784e-08,
      "loss": 0.7384,
      "step": 7655
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7715192105680294,
      "learning_rate": 1.0455455549326366e-08,
      "loss": 0.6939,
      "step": 7656
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9034944141086338,
      "learning_rate": 1.0321886867550868e-08,
      "loss": 0.683,
      "step": 7657
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2026075354649284,
      "learning_rate": 1.0189175969287923e-08,
      "loss": 0.7088,
      "step": 7658
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.473845560609885,
      "learning_rate": 1.0057322877352616e-08,
      "loss": 0.8015,
      "step": 7659
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4240873070973712,
      "learning_rate": 9.926327614411813e-09,
      "loss": 0.7863,
      "step": 7660
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.331326985693266,
      "learning_rate": 9.796190202985834e-09,
      "loss": 0.7842,
      "step": 7661
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3682909975613344,
      "learning_rate": 9.666910665445673e-09,
      "loss": 0.7285,
      "step": 7662
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7585951624905458,
      "learning_rate": 9.538489024016328e-09,
      "loss": 0.6156,
      "step": 7663
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3637313169640353,
      "learning_rate": 9.410925300775697e-09,
      "loss": 0.6745,
      "step": 7664
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8654843931789401,
      "learning_rate": 9.284219517652348e-09,
      "loss": 0.6724,
      "step": 7665
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.347369638053061,
      "learning_rate": 9.158371696428859e-09,
      "loss": 0.7605,
      "step": 7666
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9501983423548839,
      "learning_rate": 9.033381858740697e-09,
      "loss": 0.6549,
      "step": 7667
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4669153618051043,
      "learning_rate": 8.909250026074013e-09,
      "loss": 0.8279,
      "step": 7668
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.505939315937109,
      "learning_rate": 8.785976219768954e-09,
      "loss": 0.6754,
      "step": 7669
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1360759695291787,
      "learning_rate": 8.663560461018016e-09,
      "loss": 0.6874,
      "step": 7670
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3617987065987702,
      "learning_rate": 8.542002770865477e-09,
      "loss": 0.7656,
      "step": 7671
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.40643075438814,
      "learning_rate": 8.421303170208505e-09,
      "loss": 0.7022,
      "step": 7672
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.361461307017089,
      "learning_rate": 8.301461679796619e-09,
      "loss": 0.8163,
      "step": 7673
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2536871075104408,
      "learning_rate": 8.182478320232223e-09,
      "loss": 0.684,
      "step": 7674
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2803528841949332,
      "learning_rate": 8.06435311197007e-09,
      "loss": 0.6957,
      "step": 7675
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2878190130500988,
      "learning_rate": 7.947086075317246e-09,
      "loss": 0.6744,
      "step": 7676
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.36964446535335,
      "learning_rate": 7.830677230433181e-09,
      "loss": 0.6938,
      "step": 7677
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3716364703401125,
      "learning_rate": 7.715126597329648e-09,
      "loss": 0.6357,
      "step": 7678
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5251567360681881,
      "learning_rate": 7.600434195871864e-09,
      "loss": 0.6935,
      "step": 7679
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8541938186450578,
      "learning_rate": 7.486600045775728e-09,
      "loss": 0.6055,
      "step": 7680
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5676948945842477,
      "learning_rate": 7.373624166611137e-09,
      "loss": 0.7047,
      "step": 7681
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7994587749557917,
      "learning_rate": 7.261506577800892e-09,
      "loss": 0.6404,
      "step": 7682
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.157237025471481,
      "learning_rate": 7.1502472986179075e-09,
      "loss": 0.7471,
      "step": 7683
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8013252070241638,
      "learning_rate": 7.039846348189105e-09,
      "loss": 0.6265,
      "step": 7684
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2816672113092875,
      "learning_rate": 6.93030374549486e-09,
      "loss": 0.7042,
      "step": 7685
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4680780039989705,
      "learning_rate": 6.8216195093656624e-09,
      "loss": 0.8747,
      "step": 7686
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6757656038313098,
      "learning_rate": 6.713793658486012e-09,
      "loss": 0.7738,
      "step": 7687
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.327959351117109,
      "learning_rate": 6.606826211392192e-09,
      "loss": 0.6796,
      "step": 7688
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.48639642124718,
      "learning_rate": 6.5007171864733824e-09,
      "loss": 0.8404,
      "step": 7689
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.395674316196733,
      "learning_rate": 6.395466601971101e-09,
      "loss": 0.7168,
      "step": 7690
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4708703422797977,
      "learning_rate": 6.291074475978653e-09,
      "loss": 0.682,
      "step": 7691
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2732615289543028,
      "learning_rate": 6.187540826442795e-09,
      "loss": 0.7465,
      "step": 7692
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0756043633646508,
      "learning_rate": 6.084865671162066e-09,
      "loss": 0.6995,
      "step": 7693
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3882213795539755,
      "learning_rate": 5.983049027786791e-09,
      "loss": 0.7525,
      "step": 7694
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5755957619417156,
      "learning_rate": 5.882090913821303e-09,
      "loss": 0.8583,
      "step": 7695
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.269587054299167,
      "learning_rate": 5.781991346621163e-09,
      "loss": 0.8303,
      "step": 7696
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.384707865108297,
      "learning_rate": 5.682750343394827e-09,
      "loss": 0.7263,
      "step": 7697
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2600861935579462,
      "learning_rate": 5.584367921202538e-09,
      "loss": 0.8184,
      "step": 7698
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.181054523205192,
      "learning_rate": 5.486844096957988e-09,
      "loss": 0.7509,
      "step": 7699
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2683511913538381,
      "learning_rate": 5.390178887426101e-09,
      "loss": 0.7279,
      "step": 7700
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2931906503189325,
      "learning_rate": 5.294372309224693e-09,
      "loss": 0.6104,
      "step": 7701
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4326163428327912,
      "learning_rate": 5.19942437882448e-09,
      "loss": 0.7202,
      "step": 7702
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5634615829926222,
      "learning_rate": 5.10533511254796e-09,
      "loss": 0.7633,
      "step": 7703
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.243544485227116,
      "learning_rate": 5.012104526569972e-09,
      "loss": 0.7854,
      "step": 7704
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4396901886013742,
      "learning_rate": 4.919732636918251e-09,
      "loss": 0.7549,
      "step": 7705
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0058891062562827,
      "learning_rate": 4.8282194594723164e-09,
      "loss": 0.6136,
      "step": 7706
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4135402626916176,
      "learning_rate": 4.737565009964584e-09,
      "loss": 0.7735,
      "step": 7707
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8281529794925363,
      "learning_rate": 4.647769303979255e-09,
      "loss": 0.6912,
      "step": 7708
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3954447129880383,
      "learning_rate": 4.558832356953979e-09,
      "loss": 0.691,
      "step": 7709
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4782162919576456,
      "learning_rate": 4.470754184177084e-09,
      "loss": 0.7327,
      "step": 7710
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3801662703218947,
      "learning_rate": 4.383534800790901e-09,
      "loss": 0.7537,
      "step": 7711
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3970526170700142,
      "learning_rate": 4.297174221788991e-09,
      "loss": 0.7011,
      "step": 7712
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.358619578527554,
      "learning_rate": 4.211672462018368e-09,
      "loss": 0.8861,
      "step": 7713
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3830885857173556,
      "learning_rate": 4.127029536177274e-09,
      "loss": 0.7641,
      "step": 7714
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.428712737076774,
      "learning_rate": 4.043245458816847e-09,
      "loss": 0.7687,
      "step": 7715
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4562784540832616,
      "learning_rate": 3.960320244340565e-09,
      "loss": 0.7402,
      "step": 7716
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4925106129511834,
      "learning_rate": 3.878253907004248e-09,
      "loss": 0.7131,
      "step": 7717
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6101961352827179,
      "learning_rate": 3.797046460916054e-09,
      "loss": 0.7917,
      "step": 7718
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2149535604079755,
      "learning_rate": 3.716697920036483e-09,
      "loss": 0.6886,
      "step": 7719
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3463287881883514,
      "learning_rate": 3.637208298178929e-09,
      "loss": 0.7555,
      "step": 7720
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5605622044713083,
      "learning_rate": 3.558577609007463e-09,
      "loss": 0.7453,
      "step": 7721
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3199472551403368,
      "learning_rate": 3.480805866040715e-09,
      "loss": 0.7069,
      "step": 7722
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4939495519949324,
      "learning_rate": 3.403893082647991e-09,
      "loss": 0.7219,
      "step": 7723
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5876446642380069,
      "learning_rate": 3.3278392720514917e-09,
      "loss": 0.8193,
      "step": 7724
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4543954907820984,
      "learning_rate": 3.252644447325759e-09,
      "loss": 0.6103,
      "step": 7725
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3910154945354511,
      "learning_rate": 3.1783086213976743e-09,
      "loss": 0.7455,
      "step": 7726
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6376229096029435,
      "learning_rate": 3.10483180704646e-09,
      "loss": 0.7484,
      "step": 7727
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2806999212850143,
      "learning_rate": 3.0322140169036784e-09,
      "loss": 0.6908,
      "step": 7728
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3070963981876829,
      "learning_rate": 2.9604552634532325e-09,
      "loss": 0.7245,
      "step": 7729
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5047112893580552,
      "learning_rate": 2.88955555903081e-09,
      "loss": 0.7615,
      "step": 7730
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4103957188542449,
      "learning_rate": 2.81951491582555e-09,
      "loss": 0.6882,
      "step": 7731
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4068882112503498,
      "learning_rate": 2.7503333458778205e-09,
      "loss": 0.6579,
      "step": 7732
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3714770893705852,
      "learning_rate": 2.682010861080886e-09,
      "loss": 0.707,
      "step": 7733
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8412574400454388,
      "learning_rate": 2.614547473180351e-09,
      "loss": 0.6878,
      "step": 7734
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2452687267105818,
      "learning_rate": 2.5479431937736055e-09,
      "loss": 0.7447,
      "step": 7735
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.343172130350466,
      "learning_rate": 2.482198034310934e-09,
      "loss": 0.787,
      "step": 7736
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4639411567056047,
      "learning_rate": 2.417312006094963e-09,
      "loss": 0.7596,
      "step": 7737
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3508097221581588,
      "learning_rate": 2.353285120279547e-09,
      "loss": 0.74,
      "step": 7738
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3589023533450704,
      "learning_rate": 2.290117387872548e-09,
      "loss": 0.7339,
      "step": 7739
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0380030663921829,
      "learning_rate": 2.227808819732502e-09,
      "loss": 0.7069,
      "step": 7740
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3963221054757096,
      "learning_rate": 2.166359426570841e-09,
      "loss": 0.7418,
      "step": 7741
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8036045229411436,
      "learning_rate": 2.105769218952447e-09,
      "loss": 0.6293,
      "step": 7742
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9054426710159882,
      "learning_rate": 2.0460382072928775e-09,
      "loss": 0.8415,
      "step": 7743
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2750872723922284,
      "learning_rate": 1.98716640186003e-09,
      "loss": 0.7866,
      "step": 7744
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4033239387156184,
      "learning_rate": 1.929153812775808e-09,
      "loss": 0.8124,
      "step": 7745
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8446563115860644,
      "learning_rate": 1.8720004500122347e-09,
      "loss": 0.801,
      "step": 7746
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8310273250599696,
      "learning_rate": 1.8157063233953387e-09,
      "loss": 0.6754,
      "step": 7747
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2514256551661578,
      "learning_rate": 1.7602714426023792e-09,
      "loss": 0.699,
      "step": 7748
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.320150976621937,
      "learning_rate": 1.7056958171635108e-09,
      "loss": 0.8117,
      "step": 7749
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4858453168482433,
      "learning_rate": 1.6519794564606728e-09,
      "loss": 0.7368,
      "step": 7750
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.391173193827252,
      "learning_rate": 1.5991223697281454e-09,
      "loss": 0.8217,
      "step": 7751
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8051943411225673,
      "learning_rate": 1.5471245660531042e-09,
      "loss": 0.6687,
      "step": 7752
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2149630411616439,
      "learning_rate": 1.4959860543739546e-09,
      "loss": 0.7168,
      "step": 7753
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4670768430536376,
      "learning_rate": 1.4457068434831078e-09,
      "loss": 0.8732,
      "step": 7754
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.880225913915151,
      "learning_rate": 1.3962869420230951e-09,
      "loss": 0.7877,
      "step": 7755
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3072451507070024,
      "learning_rate": 1.3477263584904533e-09,
      "loss": 0.6982,
      "step": 7756
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3022566413526093,
      "learning_rate": 1.3000251012323939e-09,
      "loss": 0.6997,
      "step": 7757
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2767634529799445,
      "learning_rate": 1.2531831784506898e-09,
      "loss": 0.7005,
      "step": 7758
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3900982878939436,
      "learning_rate": 1.2072005981966783e-09,
      "loss": 0.7223,
      "step": 7759
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2599382133573813,
      "learning_rate": 1.1620773683757025e-09,
      "loss": 0.7592,
      "step": 7760
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6797209409567468,
      "learning_rate": 1.1178134967454456e-09,
      "loss": 0.7948,
      "step": 7761
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.295029745279327,
      "learning_rate": 1.0744089909153765e-09,
      "loss": 0.7768,
      "step": 7762
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1856256426247513,
      "learning_rate": 1.0318638583467489e-09,
      "loss": 0.802,
      "step": 7763
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2574267284581249,
      "learning_rate": 9.901781063531568e-10,
      "loss": 0.7802,
      "step": 7764
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5895290570838774,
      "learning_rate": 9.493517421022003e-10,
      "loss": 0.6276,
      "step": 7765
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4689775249343333,
      "learning_rate": 9.093847726110439e-10,
      "loss": 0.7806,
      "step": 7766
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3011011616664159,
      "learning_rate": 8.70277204751413e-10,
      "loss": 0.7692,
      "step": 7767
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.724868883173391,
      "learning_rate": 8.320290452462632e-10,
      "loss": 0.7412,
      "step": 7768
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3720054122987073,
      "learning_rate": 7.946403006703351e-10,
      "loss": 0.7311,
      "step": 7769
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.184944909159418,
      "learning_rate": 7.581109774512651e-10,
      "loss": 0.7271,
      "step": 7770
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2670423387962082,
      "learning_rate": 7.224410818695848e-10,
      "loss": 0.7221,
      "step": 7771
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7558279678186125,
      "learning_rate": 6.876306200565008e-10,
      "loss": 0.6097,
      "step": 7772
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2655833492107134,
      "learning_rate": 6.536795979966703e-10,
      "loss": 0.808,
      "step": 7773
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8963044793718236,
      "learning_rate": 6.20588021527091e-10,
      "loss": 0.8266,
      "step": 7774
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3539355723174376,
      "learning_rate": 5.883558963359903e-10,
      "loss": 0.6933,
      "step": 7775
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4106914396927863,
      "learning_rate": 5.569832279644915e-10,
      "loss": 0.8228,
      "step": 7776
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.432117556244378,
      "learning_rate": 5.264700218066133e-10,
      "loss": 0.7159,
      "step": 7777
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3766840445716109,
      "learning_rate": 4.968162831070489e-10,
      "loss": 0.7176,
      "step": 7778
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7450592905978124,
      "learning_rate": 4.680220169639427e-10,
      "loss": 0.811,
      "step": 7779
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3261825897165045,
      "learning_rate": 4.4008722832722397e-10,
      "loss": 0.7192,
      "step": 7780
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6134028010034598,
      "learning_rate": 4.1301192199971753e-10,
      "loss": 0.7712,
      "step": 7781
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8105331100332717,
      "learning_rate": 3.867961026354783e-10,
      "loss": 0.5149,
      "step": 7782
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2533714522739619,
      "learning_rate": 3.6143977474201175e-10,
      "loss": 0.7711,
      "step": 7783
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4676352330061737,
      "learning_rate": 3.369429426769433e-10,
      "loss": 0.7308,
      "step": 7784
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5285771990644819,
      "learning_rate": 3.1330561065301413e-10,
      "loss": 0.7956,
      "step": 7785
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3510830878820335,
      "learning_rate": 2.9052778273308546e-10,
      "loss": 0.7578,
      "step": 7786
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2476886100620421,
      "learning_rate": 2.686094628329139e-10,
      "loss": 0.7058,
      "step": 7787
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.29283959339595,
      "learning_rate": 2.475506547211515e-10,
      "loss": 0.7338,
      "step": 7788
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3587073041831,
      "learning_rate": 2.2735136201712527e-10,
      "loss": 0.7603,
      "step": 7789
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3068736413097366,
      "learning_rate": 2.080115881936129e-10,
      "loss": 0.7423,
      "step": 7790
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.660214278312693,
      "learning_rate": 1.895313365757323e-10,
      "loss": 0.8215,
      "step": 7791
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3759137481664327,
      "learning_rate": 1.7191061033983157e-10,
      "loss": 0.8733,
      "step": 7792
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6940686972214,
      "learning_rate": 1.5514941251570937e-10,
      "loss": 0.7408,
      "step": 7793
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.647308363518032,
      "learning_rate": 1.392477459843944e-10,
      "loss": 0.7254,
      "step": 7794
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.484609657663964,
      "learning_rate": 1.2420561348036598e-10,
      "loss": 0.6854,
      "step": 7795
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3695726469142613,
      "learning_rate": 1.1002301758822331e-10,
      "loss": 0.6948,
      "step": 7796
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3025300610142045,
      "learning_rate": 9.66999607471264e-11,
      "loss": 0.7393,
      "step": 7797
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7008947503383058,
      "learning_rate": 8.423644524691021e-11,
      "loss": 0.7975,
      "step": 7798
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4456639819544446,
      "learning_rate": 7.263247323086032e-11,
      "loss": 0.6818,
      "step": 7799
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4790224880831024,
      "learning_rate": 6.188804669349236e-11,
      "loss": 0.7542,
      "step": 7800
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4419791275263005,
      "learning_rate": 5.200316748110723e-11,
      "loss": 0.7188,
      "step": 7801
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4463394131706129,
      "learning_rate": 4.297783729456662e-11,
      "loss": 0.7614,
      "step": 7802
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.754318500550177,
      "learning_rate": 3.481205768429696e-11,
      "loss": 0.7685,
      "step": 7803
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4878703360602414,
      "learning_rate": 2.7505830054175287e-11,
      "loss": 0.7451,
      "step": 7804
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3368032582331446,
      "learning_rate": 2.105915566041894e-11,
      "loss": 0.7232,
      "step": 7805
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2685297644226834,
      "learning_rate": 1.5472035611585613e-11,
      "loss": 0.7088,
      "step": 7806
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2271154893916334,
      "learning_rate": 1.0744470868018219e-11,
      "loss": 0.6778,
      "step": 7807
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3772763481976247,
      "learning_rate": 6.876462241844906e-12,
      "loss": 0.7079,
      "step": 7808
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3285557368060592,
      "learning_rate": 3.868010399199484e-12,
      "loss": 0.6746,
      "step": 7809
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8344595523718081,
      "learning_rate": 1.7191158563356626e-12,
      "loss": 0.6622,
      "step": 7810
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.43184062841139,
      "learning_rate": 4.2977898240259554e-13,
      "loss": 0.7816,
      "step": 7811
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.237565857979131,
      "learning_rate": 0.0,
      "loss": 0.6837,
      "step": 7812
    },
    {
      "epoch": 1.0,
      "step": 7812,
      "total_flos": 3187350060744704.0,
      "train_loss": 0.7830338859048429,
      "train_runtime": 37745.7039,
      "train_samples_per_second": 26.492,
      "train_steps_per_second": 0.207
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7812,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 5000,
  "total_flos": 3187350060744704.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}